A.

Dedication
Dedicated To Teaching Fraternity

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
B. Preface
Metal forming is one of the oldest and widely used manufacturing processes. Its advantages include no or very little wastage
of material during processing, enhancement of the mechanical and metallurgical properties of the material and its suitability
for mass production. Metal forming has progressed from being a craftsman based technology to a technology requiring highly
scientific inputs. Last two-three decades have seen the extensive application of computational techniques in this area.
However, in academic institutions, there is still a dearth of suitable courses on the scientific aspects of metal forming. It is felt
that a suitable course should be offered to final or pre-final year undergraduate student and first year graduate students. At
the same time, practicing engineers should also be provided an opportunity to keep them abreast with the latest
developments in this area. With this viewpoint in mind and considering the shortage of advanced level books on metal
forming, we have decided to bring out an edited book on the subject.

The present book discusses both fundamentals and advanced topics in both bulk and sheet metal forming. In this form, the
book can be used as a textbook as well as a reference book. Chapters have been written by 18 authors from three countries
associated with academia and industry. The chapters have been arranged in a sequence of "from fundamental to advanced
topics".

Chapter 1 provides an introduction to metal forming technology and is descriptive in nature. Chapter 2 discusses the theory of
plasticity, which serves as a foundation of metal forming research as well as processes involving plastic deformation. Chapter
3 describes modeling techniques for metal forming. It includes finite element method and soft computing tools. Chapter 4
highlights the role of friction and ways to manage it. First four chapters are of fundamental nature. Chapter 5 discusses both
the fundamental and advanced aspects of metal extrusion. Chapter 6 discusses the advanced topics in die designing
particularly for extrusion. Chapter 7 emphasizes the experimental and theoretical research in metal forming with the help of
illustrative examples. Chapter 8 introduces sheet metal forming in detail. Chapter 9 discusses flange wrinkling in deep
drawing, a commonly used sheet metal forming. Chapter 10 discusses forming of tailor welded blanks and chapter 11 focuses
on the forming aspects pertaining to tailor welded blanks fabricated by friction stir welding process. Chapter 12 describes
probabilistic and mesoscale approaches to model interfacial phenomena during hot flat rolling. Chapter 13 is the concluding
chapter of the book, which discusses some of the nascent metal forming technologies and suggests direction for future
research.

We are very grateful to authors for finding time for writing the chapters in spite of their busy schedule. We are thankful to
Prof. M.S. Shunmugam of IIT Madras, Dr. P.M. Pandey of IIT Delhi, Prof. J. Paulo Davim of University of Aveiro, Portugal, Prof.
Masayoshi Akiyama of Kyoto Institute of Technology, Japan, for their comments and suggestions about the chapters. We also
thank Peter Jones, Iain McGregor, Yuguo An, Clive Sutton, Yi Gao, Ray Long, Nico Langerak, Henk Vegter, Eisso Atzema,
Matthijs Toose, Jon King of Tata Steel for reviewing all the chapters and suggesting improvements.

We have taken utmost care to avoid any errors and not to omit any relevant reference and apologise in advance for any
inadvertent errors and omissions. We request our readers to provide valuable suggestions for improving this book further. Any
feedback can be sent directly to us at our e-mail addresses mailto://[email protected] and [email protected]. We expect that
students and practicing engineers will find this book useful.

UDAY S. DIXIT

([email protected])

R. GANESH NARAYANAN

([email protected])

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
1. Metal Forming Process
R. Ganesh Narayanan, Uday S. Dixit

Department of Mechanical Engineering Indian Institute of Technology Guwahati 781 039, India

1.1. INTRODUCTION
Metal forming is a manufacturing process, in which a metallic workpiece is plastically deformed by the application of forces.
When the forces are applied on the workpiece, first the workpiece deforms elastically and then the plastic deformation takes
place. If the load is removed during elastic deformation, the workpiece returns to its original state without causing any
permanent set (deformation). In metal forming operations, usually the elastic deformation is much smaller compared with
plastic deformation and is often neglected for finding out the total load required for deforming the material. However, it must
be considered if the residual stress or spring back needs to be calculated.

The stress–strain behavior of a typical metal is shown schematically inFig. 1.1(a). A sharp yield point is shown at which the
state of metal changes from elastic to plastic. Most of the real metals will not display a sharp yield point. Figure 1.1(b) shows
the stress–strain behavior of a hypothetical rigid-plastic material neglecting the elastic deformation. In practice, no material is
rigid-plastic, but to simplify the analysis of metal forming process, it may be assumed that in a uniaxial tensile or compressive
test, the material remains rigid till the stress reaches the yield strength, after which the plastic deformation starts. The
stress–strain behavior of the material greatly influences the metal forming, but it is not the only important factor. The
controllable process parameters, lubrication, contact surface condition, and the design of the metal forming setup also affect
the metal forming process to a greater extent.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.1 Schematic diagram of stress–strain behavior (a) elastic-plastic material (b) rigid-plastic
material

A brief description of stress–strain behavior and plastic deformation properties of a typical metal are given here. A typical
engineering stress–strain behavior of metals is shown in Fig. 1.2. In annealed low carbon steel and some other materials, the
stress–strain behavior is characterized by a sharp upper yield point. The subsequent drop in yield stress is followed by a yield
plateau, in which barring small oscillations the yield stress remains constant. This phenomenon is known as the yield point
phenomenon. Here one can clearly distinguish between elastic and plastic deformation. In many metallic materials, the
transition between elastic and plastic parts is smooth as shown in Fig. 1.2. The deformation starts with linear elastic
deformation till point B. Hooke's law of stress being linearly proportional to strain is obeyed in the region OB. In the region BA,
the material is non-linearly elastic. Usually, the region BA is negligibly small and it is very difficult to locate the point A
accurately. It is customary to take the yield point as a point C at which the permanent deformation is equivalent to an offset
strain of 0.002. In Fig. 1.2, OD is equal to the offset strain. Note that the line DC is parallel to the line OB. After the onset of
plastic deformation, stress increases non-linearly with further deformation due to strain hardening. At point E, corresponding
to ultimate strength, necking starts. Finally, the necking is localized and material fails at point F. The elongation attained by
the material till necking is called uniform elongation and the elongation till failure is called total elongation of the material.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.2 Stress–strain behavior of a typical metal

It is possible to obtain some of the important forming properties from the stress–strain behavior of a material. The strain
hardening behavior during plastic deformation can be described by strain hardening relations like Hollomon relation σ = Kεn,
where n is the strain hardening exponent, and K is the strength coefficient. It is to be noted that σ is the true stress and ε is the
true plastic strain (total strain, if elastic strain is negligible) in Hollomon relation. The engineering stress–strain behavior can
be transformed into true stress–strain behavior by volume constancy principle during plastic deformation.

There are many metal forming processes for manufacturing engineering products. For example, rolling is used to convert a
thick slab into thin sheets having thickness of even in the sub-millimeter range. These sheets are then used to make
automotive, aerospace, electronic components, and various consumer products by stamping, deep drawing, stretch forming,
tube forming, bending, shearing or spinning, to name just a few sheet metal forming processes. Extrusion is used to make
rods, seamless tubes, and products with various cross-sectional profiles that are used for engineering and construction
industries. It should be noted that many a times the metal forming and other manufacturing processes are combinedly used to
manufacture engineering products. For instance, (i) bolts and screws are made by cold heading the rod and thread rolling, (ii)
cooking gas cylinders are made by deep drawing of upper and lower halves and welding, (iii) metal wash basins are made by
deep drawing and hole making process. The main motivation of the metal forming studies is to understand the change in
structural properties, design the process efficiently, and model it to have accurate performance prediction.

In general, the metal forming processes are categorized into (i) bulk metal forming, and (ii) sheet metal forming processes.
The processes like forging, extrusion, rolling, wire drawing, heading, ironing, etc., can be included in bulk forming processes;
and deep drawing, stretch forming, bending, spinning, hydro forming, etc., can be included in sheet forming processes. Bulk
forming processes exhibit high volume to surface area ratio. The workpiece sheet used in sheet forming processes has a low
volume to surface area ratio. Both the bulk forming and sheet forming processes can be performed at room temperature, high
temperature, and medium temperature, depending on the workability of the materials. If a metal forming process is carried out
at a temperature more than the recrystallization temperature, it is termed as hot working. The metal forming process
performed without heating the work-piece is termed as cold working process. In between hot working and cold working is the
warm working, in which the workpiece is heated but there is no recrystallization. In the following sections, some important
bulk forming and sheet forming processes are described.

1.2. BULK METAL FORMING PROCESS

In bulk forming, generally, the input raw material is in the form of rod, slab or billet. The raw material undergoes plastic
deformation and a shape and size change is witnessed in the output product. Other than the material properties, the working

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
temperature, strain rate, and friction play a vital role in deciding the workability of the material. The important bulk forming
processes like rolling, forging, extrusion, and wire drawing are described here.

1.2.1. Rolling
In rolling process, the semi-finished parts like slab, billet or bloom are rolled into plates, sheets, strips, rods, tubes by rollers
rotating in opposite directions. The semi-finished products are also made mainly by hot rolling of cast ingots. Hot rolling is
performed above recrystallization temperature, where large reduction in thickness, with less forming pressures is possible. In
the case of sheet, strip or foil rolling, the deformation in width direction is negligible when compared to thickness and length
direction. This type of deformation is known as plane strain deformation. This is generally seen in sheets whose deformation
zone width is more than 10–20 times larger than the length of the zone. In the case of rolling of thick sheets and slabs, there
exists considerable deformation along the width (called spread), length (called elongation), and thickness (called
compression).

The rolls are classified based on the number of rolls, number of stand, type of products produced, size of the rolling mill, etc.
Some important classifications are described here. Figures 1.3–1.5 show the schematic description of two-high, three-high,
and four-high rolling mills. The two-high mill [Figs 1.3(a) and (b)] is the simplest one, and the slab is rolled by two rolls
rotating in opposite direction— one in clockwise (CW) and another in counter-clockwise (CCW) directions. Once the rolling is
completed, the rolled sheet is taken to the same entry location, and rolling is done again for further thickness reduction. The
two-high mill can be a reversing mill as shown in Fig. 1.3(b), in which the rollers employed can rotate both in CW and CCW
directions after each stage of thickness reduction. Hence, after first stage of rolling, the sheet can be inserted from the exit
location itself for next rolling stage. The reversing mill is found to be more productive and efficient when compared with
ordinary two-high rolling mill.

Figure 1.3 Schematic diagram of two high rolling mill (a) without reversing facility, (b) with reversing
facility

The three-high rolling (Fig. 1.4) mill contains three rollers— the top and bottom rollers rotate in same direction, while the
middle one rotates in opposite direction. First, rolling is done between bottom and middle rolls in a direction and the same
sheet is rolled in the opposite direction between top and middle rolls in the second operation for more thickness reduction.
There is no reversing of roll direction in this case. In four-high roll mill (Fig. 1.5), the small work roll is involved in sheet rolling
whereas large back-up rolls support the working roll throughout its length, so that large reductions in sheet thickness is
possible without excessive deflection of working roll. Huge rolling operation involving large thickness reductions are possible
in special roll mills like cluster mills and planetary mills. In a typical 20-high Sendzimir mill (Fig. 1.6), the small working rolls
are supported by two intermediate back-up rolls which in turn are supported by three back-up rolls that in turn are supported
by four back-up rolls throughout their length. All these rolls are arranged within a rigid housing. A large thickness reduction is
possible with this rolling mill, without roller deflection. A planetary mill (Fig. 1.7) contains back-up rolls surrounded by small
work rolls that are used for large thickness reductions.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.4 Schematic diagram of three high rolling mill

Figure 1.5 Schematic diagram of four high rolling mill

Figure 1.6 Schematic diagram of Sendzimir rolling mill

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.7 Schematic diagram of planetary roll mill

For rolling operations involving different stages, rolls are arranged in "stands" as shown in Fig. 1.8. In each stand, uniform and
preset thickness reduction occurs and finally the sheets are coiled up in a "windup reel". Since different thickness reduction is
taken at each stand, the strip will move with different velocities between successive stands. The speed of each set of rolls is
controlled.

Figure 1.8 Rolling mill with four roll stands

The rolls are important components in the rolling machine as they control and plastically deform the sheet for significant
amount of thickness reduction. Selection of roll material is crucial for the performance of the process. Cast iron (as-cast and
stress relieved), heat treated high alloy iron, cast steel, hardened forged steel (typical composition: 0.85% C, 0.30% Mn, 0.30%
Si, 1.75% Cr, 0.10% V) are generally used to make rolls [1]. Recently high speed steels and semi-high speed steels are also
used widely for hot strip rolling . These rolls are more resistant to wear than conventional high chromium rolls, which make
them suitable for long rolling schedules and productivity. More details of roll materials and manufacturing can be found in the
handbook edited by Semiatin [1].

The Schematic diagram of forces acting during strip rolling is shown inFig. 1.9. A sheet material with the initial thickness of h1
enters the rolling mill from entry side with a velocity of v1. The sheet is rolled between two rolls. It has reduced thickness ofh2
at the exit side and leaves with the velocity of v2 . The sheet has intermediate thickness of h and velocity of v at a typical
location in deformation. In strip rolling, the strip width w does not vary considerably. The volume constancy principle provides

(1.1)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.9 Schematic diagram of forces acting during strip rolling

As per this equation, the exit velocity v2 is greater than entry velocity v1. In fact, the velocity of the rolled sheet continuously
increases from entry to exit. There is a point or zone on the contact surface between the roll and the sheet where the roll
surface velocity vr is equal to the sheet velocity. This point is called neutral point, shown as point N in the Fig. 1.9. Many times,
instead of a clear cut neutral point, there may be a neutral zone. At a typical point F, the radial force, Fr, and tangential friction
force, Ff, are shown. The interface friction force acts towards the neutral point from the entry and the exit sides due to the
velocity difference between the sheet and the roll surface before and after the neutral point. Before neutral point, the sheet
moves slower than the roll surface and after the neutral point, the sheet moves faster than the roll surface. The vertical
component of radial force is called the rolling load, Frv. This is also termed as roll separating force, as the pressing force
acting on the sheet by the roll and the force given by the sheet to the rolls for separating them apart are equal. The roll
pressure pr is the rolling load divided by the projected contact area between the rolls and the deforming material, i.e.,

(1.2)

where wl is the product of strip width and the projected length of the contact area. As described elaborately in [2], the
projected length, l, is given by

(1.3)

where 'Δh' is called draft. The experimental work done by many researchers, for example, Al-Salehiet al. [3] indicates that the
roll pressure rises to the maximum at the neutral point and then decreases as shown schematically in Fig. 1.10. The angle
between the entry sections to centerline of the rollers is called "angle of contact" or "angle of bite (α)". From Fig. 1.9, for the
sheet metal to enter into the rolls, the horizontal component of friction force (Ffh = Ff cos α) that acts towards the roll gap,
should be equal to or greater than the horizontal component of Fr, i.e., Fr sin α. In the limiting case,

or,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.10 Variation of roll pressure with contact length during rolling

Thus,

providing

(1.4)

From Eq. (1.4), if μ = 0, rolling cannot occur. If tanα is greater than μ, then the sheet cannot be drawn into the rolls.

From Fig. 1.9, one can write

From Eq. (1.4),

(1.5)

The maximum possible draft during cold rolling operation can be obtained from Eq. (1.5). Figures 1.11(a) and (b) depict the
effect of friction coefficient and roll radius on maximum draft based on Eq. (1.5). With increase in friction coefficient from
0.08 to 0.1, there is a substantial increase in maximum draft for same roll radius [Fig. 1.11(a)]. Similarly by increasing the roll
radius from 300 to 400 mm, for the same friction coefficient, considerable increase in maximum draft is observed [Fig.
1.11(b)].

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.11 Effect of friction coefficient and roll radius on maximum height reduction

An approximate model for evaluating the roll separating force can be developed by approximating the deformation zone (Fig.
1.9) with the homogeneous plane-strain upsetting process. The model to find the load per unit width of the strip [1, 2, 4] is
given by

(1.6)

where is the flow stress of the material, m is the friction factor (ratio of frictional traction to shear yield stress), H is the
average strip thickness {H = 0.5 (h1 + h2)}, and la is the approximate contact length (la = Rα).

The roll pressure during strip rolling with and without back and front tension is shown inFig. 1.12 [7]. The maximum stress
occurs at the neutral point. With the front and back tension, the neutral point shifts from its initial position. With the
application of back tension, the neutral point shifts towards the roll exit. A very high back tension will eventually shift the
neutral point to the roll exit. In this situation, the rolls slide over the sheet and they move faster than the sheet. The neutral
point will shift towards the roll entry with the application of front tension (Fig. 1.12).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.12 Stress distribution (or roll pressure) during strip rolling with and without back and front
tension (tb: back tension; tf: front tension; (σ y)0 : yield stress in tension; R: undeformed roll radius; h1 :
inlet thickness of strip; f: friction factor; r: reduction ratio; ϕ: angular position of a point on the
interface; α: angle of contact). With permission from [7] copyright [1996] Elsevier

There are various process and roll parameters that affect the strip rolling performance. Some of the important parameters are
friction coefficient, front and back tension, strip width, roll bending force, roll roughness, lubrication viscosity, percentage
reduction, and roll radius. The important rolling performances monitored are rolling pressure, rolling torque, forward slip, strip
roughness, etc. The variation of friction and effect of friction on the rolling parameters were studied by many researchers
including Tieu and Liu [5], Jeswiet [6], Dixit and Dixit [7], Gudur et al. [8], Jiang et al. [9], Gao et al. [10], and Pawelski [11]. For
example, Tieu and Liu [5] showed that the friction coefficient is not a constant and it varies with time and roll bite length, from
entry to exit. Similar behavior is witnessed for different rolling reductions (37.01% and 40.38%) and rolling speed (5 and 30
rpm). Likewise, it was shown by Jeswiet [6] that friction force decreases along the rolling direction, tested for a reduction of
24.2%. Dixit and Dixit [7] used Wanheim and Bays friction model in their work to analyze the flat rolling process using finite
elements. They have also applied fuzzy set theory. In their analysis, the initial yield stress, hardening coefficients, and friction
coefficients are considered as the fuzzy process parameters. As a result of fuzzy input parameters, the output parameters
also acquire fuzzy characteristics, thereby giving a more realistic simulation of experimental results which always have a
scatter. Gudur et al. [8] proposed a method for the estimation of friction coefficient based on the asymmetric rolling operation.
Using slab method, the curvature of the rolled sheet was estimated and the friction coefficient was evaluated by solving an
inverse problem. It was found that as the coefficient of friction increases the curvature also goes on increasing (Fig. 1.13).
Similarly curvature was found to increase with increase in speed ratio (Fig.1.14) [8]. An inverse problem of finding the friction
coefficient by estimating the curvature was proposed and validated by authors. Jiang et al. [9] modeled cold strip rolling taking
into account the friction along the strip rolling and transverse directions. A 3D rigid plastic finite element method was used.
Simulation results show that the friction in both directions has a significant effect on the rolling mechanics, strip shape and
profile.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.13 Influence of friction coefficient on the radius of curvature. With permission from [8]
copyright [2008] Elsevier

Figure 1.14 Influence of speed ratio on the radius of curvature. With permission from [8] copyright
[2008] Elsevier

An interesting work was done by Gao et al. [10] to understand the effect of varying friction coefficient on the rolling pressure
and force. As described by authors, there are three different asymmetrical rolling processes, one with varying roll radii and
constant angular velocity, second one with varying angular velocity and same roll radii and third one with changing friction
coefficient by keeping the other two parameters constant. The effect of different friction coefficients on the rolling
performance was studied by monitoring the roll pressure, force and torque, using slab method of analysis. The effect of back
and front tension during asymmetrical rolling is also observed. The variable friction coefficient is quantified by friction
coefficient ratio (μ1/μ2). It is observed that with increase in friction coefficient ratio, rolling pressure increases considerably.
With the application of back and front tension, the rolling pressure is found to reduce. It was also shown that with more
reduction per pass, the rolling pressure is found to increase. An interesting result is that the summation of friction coefficients
has got more effect on the rolling pressure than the friction coefficient ratio. The friction coefficient ratio affects the shear
deformation region considerably. The total rolling torque is found to decrease with increase in friction coefficient ratio.

The effect of roll roughness on the roll separating force and strip roughness can be understood from work done by Dick and
Lenard [12]. With increase in roll speed, under dry condition, the roll force is either constant or decreases and in the mean
time, with the increase of roll surface roughness, roll force is found to increase substantially. The authors also studied rolling

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
in presence of three different lubricants. The strip roughness is found to increase with roll roughness. The lubricant viscosity
had insignificant effect on the roll separating force.

As discussed earlier, Gudur et al. [8] proposed an inverse method for estimating friction coefficient by measuring the rolled
sheet curvature during asymmetric rolling operation. The strip rolling is expected to show profound effect on the mechanical
properties after rolling. Shen et al. [13] had proved that by tailoring the microstructure using rolling, one can obtain metals with
high strength and high ductility. They demonstrated the effect of cold rolling on the tensile properties of interstitial free steel.
In their work, it is observed that with increase in reduction from 13 to 75%, the ultimate strength increased significantly, with a
substantial increase in ductility [13]. Das et al. [14] investigated the change in tensile properties of SS316L steel after cold
rolling by ball indentation technique.

The elastic flattening of rolls will eventually result in more deformation of the rolls than the sheets. Thus there will be a
minimum sheet thickness, below which the sheets cannot be rolled. The thinner gage sheets can be rolled by smaller radius
rolls efficiently.

There are some rolling defects that affect the quality of sheets produced and hence their performance is affected. Theroll gap
must be uniform.If the roll deflection is excessive, edges of the sheet will be decreased more in thickness than the center
region, and since volume and width remains constant, the sheet edges will elongate more than the center region, resulting in
bow sheets. Flatness of the sheets should be maintained. Sometimes waviness develops in the sheet. If the rolls deflect in a
convex fashion, the edges of the sheet will be elongated more in the longitudinal direction as compared to the center region.
Since continuity has to be maintained between the sheet regions, the sheet edges will be compressed, while sheet center will
be stretched in tension. This results in wavy edge or edge buckling. Rolls having a ground camber or crown are usually used for
preventing this. Edge cracking occurs due to the tensile stretching of rolled sheet at the edges and compression at the center
region. This occurs mainly because of the lateral expansion of the sheet. Under severe conditions, edge cracking can lead to
center split of the sheet. Alligatoring (splitting of sheet across the thickness) type of fracture is also seen during rolling. Other
defects like scratches during cold rolling and internal fissures are also witnessed during sheet rolling.

1.2.2. Forging
Forging is a separate class of bulk forming process wherein useful products can be obtained by shape change using
hammering or pressing force. Generally the forging process is carried out in hot forming conditions (hot forging) because of
less load requirements, though sometimes metals undergo forging at room temperatures (cold forging). When hammers are
used, large impact blow is delivered on the material, while slow incremental compression is used to deform metal in the case
of forging using a press. The hammers are used to deform the material by dissipating the kinetic energy of the ram. The
mechanical presses are stroke-restricted machines since the length of the stroke and load availability at each level of the
stroke determines the capability. The hydraulic presses are load-restricted machines and the maximum tonnage of the
machines chiefly decides the ability and products produced by the machine.

The forging processes are divided into two categories namely "open die forging" and "closed die forging". In open die forging,
the metal can flow freely without any restriction from die impression. Generally, the open die forging is used when (a) the
quantity required are small to justify the cost of making dies, (b) the properties obtained after open die forging are
satisfactory, and (c) the forging is too large to be produced by closed die forging. A simple example of open die forging is the
"upsetting" operation. In this operation, a cylindrical billet is compressed between two flat dies. The billet bulges as shown in
Fig. 1.15, and this depends on the interface friction between the flat dies and the billet. More friction will not permit the upper
and lower surfaces to expand, with unrestricted expansion at the center. This forms "bulged" component. By increasing the
lubrication, the upper and lower surfaces can be made to move freely equivalent to that of billet center, which will result in
homogeneous upsetting with "un-bulged" sample. The examples discussed in [1], like making cylindrical work piece from
square stock, forging of a gear blank and hub in open dies, four diameter spindle in open dies, etc., indicate the use of open die
forging in making complex shapes with auxiliary equipment. In the case of closed die forging, the final product takes the
shape of the die impression. The impression can be in one half of the die or made in two halves of the die. An example of

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
closed die forging is shown in Fig. 1.16. The extra material is stored as "flash". The flash also controls the outward flow of
metal so that the corners can be filled properly. The schematic diagram of forging load–stroke behavior during closed-die
forging is shown in Fig. 1.17. The forging pressure increases with decreasing flash thickness and increasing flash land width.
The ideal design is to have minimum flash required to make the product. In closed die forging, heavy reductions are possible
with close tolerances as compared to open die forging. Minimal post-forging machining is required in the case of closed die
forging. Better mechanical properties are achievable. It is suitable for both low volume and high volume production. Aluminum
alloys are easily forgeable at temperatures in the range 400–500°C and superalloys and tungsten alloys are difficult to forge
even at high temperatures (1200–1300°C). Stainless steels show moderate forgeability at high temperatures (1100–1250°C).

Figure 1.15 Schematic diagram of open die forging

Figure 1.16 Schematic diagram of closed die forging

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.17 Forging load–stroke behavior during closed-die forging

A simple analysis of plane strain forging is shown in Fig. 1.18. By assuming the forging to be in equilibrium and Coulomb's law
of sliding friction (τ = μp where 'μ' is the friction coefficient), the forging pressure is given by

(1.7)

Figure 1.18 (a) Simple analysis of plane strain forging, (b) Pressure variation at different friction
conditions

Here 'σ0' is the yield strength of the material. When forging pressure is plotted with the stroke, the pressure varies

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
exponentially as per Eq (1.7). At the center of the plate the pressure is maximum which is illustrated as "friction hill" in Fig.
1.18. The center line can be seen as "neutral line" in this simple upsetting. The metal is stationary in the neutral line whereas it
moves outward away from the neutral line. The same forging pressure can be obtained by assuming sticking friction condition
at the interface, i.e., interfacial shear stress is equal to yield shear stress giving

(1.8)

The forging pressure varies linearly with the distance 'x' which is also shown in Fig. 1.18. Practically the friction conditions will
change from sliding to sticking from the edge to the neutral line. The friction will be of sticking nature close to the neutral line.
The combined friction hill is shown in Fig. 1.18 where sticking zone is visible near the neutral line showing a linear variation in
pressure.

There are varieties of lubricants that are used in cold and hot forging of metals. The selection of lubricants depends on the
forging materials, forging operation, working temperatures, interface pressures and sliding speed. However, there are several
functions that are common to lubricants for any processes and conditions [1]. Some of important functions are as follows:

the lubricating function that optimizes the metal flow and controls the die filling. This is very crucial in case of closed-die
forging that involves severe forming conditions requiring "hard surface coating" for successful operation

the tribological function that reduces the wear and tear of the die spots that has high temperatures, sliding speed and
forging pressures

the release function that facilitates the easy removal of forged part from dies without adhesion during cold and hot forging

the heat protection function that removes the heat generated during deformation without transferring to the dies. This if
not done properly will raise the local temperatures affecting the forgeability of many materials

the safety function that is responsible for minimizing the pollution caused due to lubricants, odor ill effects, fire risk, wrong
handling of materials and machines, reactiveness with man and machines, tool corrosion and wear, residue build up in the
die after continuous usage, etc.

Some of the important lubricants used for cold forging are, mineral oil, synthetic oil, and soap solutions. In the hot forging
operations, the lubricants used are molten glass, graphite contained in water and molybdenum disulfide. Often conversion
coatings are used as a lubricant carrier. Conversion coatings are the coatings obtained by the conversion of surface of the
workpiece by chemical or electrochemical means. For forging of steels, phosphate coating with soap solution or molybdenum
disulfide can be used. For forging of aluminum zinc phosphate coating, calcium aluminate coating or aluminum fluoride
coating can be used. Ganesh Narayanan et al. [15] studied the influence of different friction conditions on the hardness
distribution during a cold forging process. They also predicted the hardness variation during a cold forging-extrusion operation
by an empirical relation developed by simple upsetting for varied lubricating conditions. The empirical equation prediction is
validated with experimental results and is found to be consistent. Similar work was carried out by Kim et al. [16] also. Some of
the defects that are generally seen in forging are as follows:

i. Surface layer deformation – this defect is mainly because of the deformation being restricted to surface layer only without
much happening in the interior of the forging. Instead of using light weight, rapid hammering operation, a heavy
hammering operation can be used to minimize it.

ii. Surface cracking/defects caused mainly by the over working of the surface at too low temperatures or as a result of hot
shortness. A high sulfur content in the furnace atmosphere can cause hot shortness in steel and nickel.

iii. Flash cracking during closed-die forging is another defect that is of concern as it penetrates into the forged component
even after the flash is trimmed. This can be avoided by having a thicker flash or relocating the flash to a non-critical region

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 of the forging.

iv. Cold shut or fold— common in closed-die forging, is a discontinuity that is produced due to the folding or overlapping of
one region of the metal into another. This occurs when the metal flows past the region that is already filled partially
because of sharp die radius, excessive friction, etc. Generally sharp radius should be avoided in dies.

v. Underfilling— occurs when the material is not flown fully into the corners, edges, etc. This can be avoided by proper
cleaning of forging setup so that the residue accumulation and scale built-up can be minimized.

vi. Internal cracks— generally seen in open-die forging like upsetting of cylindrical billets because of circumferential tensile
stresses. This can be minimized by proper, continuous lubrication and using concave dies. This is not seen in closed-die
forging as the lateral compressive stresses are generated by the reaction of deforming billet with the die wall in this case.

1.2.3. Extrusion
Extrusion is the process of plastically deforming a metal block through dies to reduce its area of cross-section. The metal
billet is compressed by the ram against the die opening through which the metal gets extruded taking the shape of the
opening. The cylindrical rods and tubes are generally made through extrusion process. During extrusion, the metal undergoes
severe compressive deformation and the initial ingot microstructure is broken to obtain a part which is free of internal defects.
Many a times, extrusion is used to process difficult to deform materials like stainless steel, high strength steels, nickel based
alloys and materials deformable at high temperatures. Like any other forming processes, the extrusion can be cold, warm or
hot.

The extrusion process can be classified into forward extrusion, backward extrusion, and lateral extrusion, depending on the
metal flow with respect to ram displacement. Many of the extrusion processes are combination of these. In forward extrusion,
the metal flows in the same direction as that of punch displacement. The billet is pushed through the die opening for the area
reduction. In this case, there is a relative movement between billet and the die wall. Friction exists between the billet and the
die wall and the billet–punch interface. In backward extrusion, the metal flows in the opposite direction as that of ram
displacement, through a hollow ram. During backward extrusion, there is no relative motion between the billet and the die wall
and hence there exists less friction during forming. This eventually results in less load requirement in backward extrusion
when compared to forward extrusion (Fig. 1.19). Sometimes both backward and forward extrusions are combined to produce
formed parts. The double cup extrusion test (DCET) is an example of this type. In lateral extrusion, the metal flow is
perpendicular to the punch displacement. The billet is placed in the die container and pushed sideward through die opening. In
Hooker extrusion, a variation of the forward extrusion process, a tube billet is pressed through a forward extrusion die that acts
a pusher and mandrel, such that the outer diameter is reduced significantly and the tube is elongated. Similarly in the ironing
operation, the wall thickness of the tubes is reduced by forcing it under tension/compression force into the forward extrusion
die. The Schematic diagram of above described processes are shown in Fig. 1.20. Impact extrusion, is predominantly used to
make short hollow products, like collapsible toothpaste tubes. As the name implies, the process is done on a high speed
mechanical press, deforming the material by impact pressure. The process is restricted to softer materials like lead,
aluminum, copper. Some of the advantages of cold extrusion over hot extrusion are dimensional control, superior surface
finish, high production rates, clean and hygienic work environment, and net shaped formed components. The disadvantages
are the manufacture of less complex parts, high load requirement, lubrication usage and cost, and limited deformation.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.19 Variation of extrusion pressure with stroke for backward and forward extrusion

Figure 1.20 Schematic description of different extrusion processes

The extrusion process is generally carried out in mechanical or hydraulic presses. The machines are expected to have high
rigidity, accurate alignment, and long working strokes. The mechanical presses are suitable for high production rates,
production volume, and large batch sizes. The extrusion process can be performed in single or multiple press stations
depending on the machine capacity and reductions. The capacity of such machines varies from 1000 to 2000 tons. There are
horizontal mechanical presses with coil feeds and billet shearing facility. The drive mechanism of the mechanical press also

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
varies like crank, knuckle, link, and eccentric [1]. It is important to analyze the force–displacement curve of the press and of
the extrusion process to ensure that sufficient deformation energy is available during the cycle. In multi-press station transfer
presses, careful study of time–displacement curves of the press, transfer, part, and ejector is essential to ensure proper
transfer. The hydraulic presses are generally suitable for low volume production, long working strokes and large extrusions.
The tool design depends on the part dimension, alignment, friction, lubrication, tool setup and assembly, load distribution, and
tool elastic deflection. The ram, die and shrink rings are important parts of an extrusion setup. The tools are made of AISI tool
steels M2 and M4 heat treated to hardness of 60–70 HRC. Tungsten carbide tools are also used for extruding hard materials
requiring large extrusion ratios. There is an optimum design of punch for favorable working of extrusion. A tapered punch with
170° included angle is found to be optimal to avoid lubricant escape during forming and prevent failures.

The die extrusion angle (half angle) that is dependent on extrusion ratio varies between 5 and 30°. Lower angles are preferred
for higher reductions. If higher angles are used, care should be taken that chevrons (arrow shaped internal defects) are not
formed during multiple extrusion. An angle of 30–60° is used for hollow parts. A soap lubricant formulated to react with zinc
phosphate coating, resulting in a layer of water insoluble metal soap (zinc stearate) on the surfaces of the billets are used that
gives excellent lubricity even for high volume production and heavy reductions. Soaps with and without additive, polymer
lubricants, oil and water based lubricants are also used in industrial practice.

The extrusion pressure is an important parameter which is monitored during extrusion and it depends on the amount of
deformation, material properties, die and punch design, initial billet geometry and tool–billet interface friction. There are
analytical and empirical equations that were developed to predict the extrusion pressure. For instance, Onuh et al. [17]
developed empirical equations for predicting the normalized extrusion pressure as a function of die reduction (r), loading rate
(V), and die angle for Al and Pb alloys. The equations are of the form:

(1.9)

(1.10)

The values for A, B, C and D are different for different materials and conditions. Here 'Pmax' is the maximum extrusion pressure,
'σ0' is the mean yield stress.

Ajiboye and Adeyemi [18] analyzed the impact of die land on the extrusion pressure by including the ironing effect through
extended upper bound method. The dimensionless extrusion pressure [Eq. (1.11)] is given by,

(1.11)

where V0 is the punch velocity, A0 is the original billet area, σ0 is the mean yield strength of the material and P* is the extrusion
pressure. The extrusion power (or total powers of deformation, P*) is given by,

(1.12)

In the above equation, Wi is the internal power of deformation, WS is the shear loss at the boundaries of velocity discontinuity,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
and Wf is the total friction power which is the summation of power dissipated in ironing, friction dissipated in punch–metal
interface and billet–die interface. It is observed from their work that the normalized extrusion pressure increases almost
linearly with die land dimension. The theoretical calculations agree well with the experimental results. Similar work was done
by the authors for varied die opening geometries like square, rectangular, I-shaped and T-shaped sections [19]. Analytical
equations were developed for three different power losses that constitute the extrusion pressure by considering ironing effect
for all die opening sections. The normalized extrusion pressure is found to vary linearly with the die land length.

Kudo [20] has developed the following expressions based on upper bound analyses for normalized extrusion pressure for
smooth and rough square dies:

(1.13)

(1.14)

These equations predict error less than ±5% for R = 0.3 – 0.97, where R is the extrusion ratio (ratio of initial cross-section area
of the billet to the final area of cross-section after extrusion).

Most of the time, the extrusion is combined with forging operations to make parts. Choi et al. [21] studied the influence of
different parameters like gap height, die corner radius, and friction factor on the forming load and volume ratio. It is observed
that as compared with friction factor, gap height and die corner radius showed profound effect on the forming load and
volume ratio. Tube extrusion from a rod or another tube is of interest to research community and is analyzed through
analytical methods, finite element methods and experiments. It is known that during the manufacture of tubular products
employing extrusion, the extrusion pressure, the state of the stresses and the flow pattern are influenced by parameters such
as, reduction ratio, the shape of the die and the mandrel, the frictional conditions on the tool–metal interfaces and the
strength and flow properties of the material.

A generalized slab analysis was presented by Chitkara and Aleem [22] for extrusion–piercing process through a shaped die
using a general shaped mandrel starting from a hollow or solid circular billet. As per their analysis, the total extrusion force
(Fe) is given by,

(1.15)

where σz, is the axial stress component. The mean extrusion pressure on the punch/ram is estimated as,

(1.16)

Here Ri and Ro are inner radius and outer radius of the hollow billet, respectively. The analyses were performed for different
mandrel types. Mandrel diameters were varied as 12.8, 16.5, and 19.0 mm, for the conical mandrels and had the included
cone angles of 70°, 90° and 120°. An ogival-shaped mandrel and another curved-shaped mandrel of 16.5 mm shank diameter

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
were also fabricated. On each of the mandrel shanks, relief of 0.1–0.12 mm and land length of 3 mm was also provided.
Besides these, three parallel shank mandrels of diameters 12.8, 16.5 and 19 mm were also made. Tellurium lead specimens
(size: 12.5 mm diameter and 15 mm height) are used for experiments. The empirical equations (developed for extrusion
pressure are as follows:

For extrusion of tubes with straight mandrel

(1.17)

where R is the extrusion ratio,

and for extrusion–piercing of solid billet through shaped dies/mandrels

(1.18)

An upper bound analysis on tube extrusion process can be seen in [23] also. In this work, a kinematically admissible velocity
field is developed and the effect of process variables on the relative extrusion pressure is investigated. Based on this model,
the equation for optimum die angle, dead zone cone angle, and relative extrusion pressure are also derived. The extrusion
behavior analysis and prediction for bimetallic tubes [24–26] and composites [27–29] were presented by many researchers.
The visio-plasticity method of analysis has been used to study the grid distortion, deformation zone analysis, velocity profiles,
etc. A simple analysis of bimetallic rod after extrusion process can be obtained from work done by Berski et al. [24]. The work
done by Avitzur et al. [25, 26] describes the prediction of fracture in the drawing of bimetallic rods and wires. The visio-plastic
analysis of layered composites for different component ratio (volume of sleeve and core) using various dies is conducted by
Irena and Andrzej [27]. The composite material consisting of the soft lead and hard lead are used for testing. The grid patterns
were used to obtain the optimized die angle for efficient extrusion of composite material. The influence of die angle on the
extrusion load is also studied. The simulation analysis of composite extrusion is studied by Schikorra and Kleiner [28], while
analytical model using stream functions that can satisfy the velocity boundary conditions is proposed by Hwang and Hwang
[29] for evaluating the deformation of composite rods consisting of copper sleeve layer and aluminum core layer. Experiments
are conducted to validate the analytical method.

In order to increase the production rate of small components, the multi-hole extrusion process was developed. Multi-hole
extrusion process is the extrusion process carried out through a die having more than one hole. The requirement of ram force
is lesser in this process compared to single-hole extrusion process. The research on multi-hole extrusion process is several
decades old. Dodeja and Johnson [30] carried out experiments to extrude pure lead, tellurium lead, pure tin and super pure
aluminum at room temperature through square dies containing up to four holes arranged in varied fashion. They provided
empirical expressions for the calculation of the ram force. Keife [31] has carried out upper bound analysis of extrusion
through two die openings. Their theoretical work was validated by extruding model material like plasticine. Ulysee and
Johnson [32] presented analytical and semi-analytical upper bound solutions for plane-strain extrusion through a die
containing eccentric hole and unsymmetrical multi-holes. They have validated the results from upper bound analyses with the
literature results and with the finite element method results. Recently Sinha et al. [33, 34] proposed simplified models to
evaluate the single-hole and multi-hole extrusion process behavior. The ram and die pressures obtained from this are used to
design a multi-hole extrusion process. Later the model proposed was validated by many experiments using lead and candle
wax through various die configurations.

There are a few defects that occur during extrusion because of the way the metal flows and this depends on the friction and

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
lubrication at the billet–tool interface. For a well lubricated container, the metal flows smoothly along the die wall and
equivalent amount of deformation is seen at the billet center and the billet–die interface. The case with high friction (or low
lubrication) between the billet and the die wall is characterized by a dead zone of stagnant material with little strain. The
elements near to die wall undergo shear deformation resulting in redundant work which is not related to final extruded product
dimensions. The high friction at the die wall concentrates the metal flow towards the center and an internal shear plane
develops. Due to sticking friction condition, the metal shears internally along the shear plane and a thin layer of metal will
remain in the container. The dead metal zone extends along the extruded outer surface. After about 75% of the extrusion is
completed, the surface layer moves towards the billet center and deforms near the axis of the rod. Since the outer surface
contains oxide layer, this metal flow forms internal oxide stringers and it looks like an annular ring of oxide. This defect is
generally seen as internal pipe. This defect formation increases with increase in sticking friction condition. The usage of
follower block (has slightly less diameter than the internal diameter of die) pushes the billet forward, leaving the thin oxide
layer in the container itself. Center burst or chevron cracking is an arrow shaped defect that is formed during forward extrusion
internally because of low extrusion ratio or high die angle. Surface cracking (Fig. 1.21) is formed on the extrude billet surface
in the transverse direction by the longitudinal tensile stresses generated when the billet passes through the die. In cold
extrusion, the defect is understood to occur because of the intermittent sticking of the billet to the die land resulting in sudden
shoot up of load, and then breakaway.

Figure 1.21 Surface cracking during extrusion

1.2.4. Wire, Rod and Tube Drawing

Drawing operations involve pulling metal through a die hole by means of tensile force applied at the exit side of the die. This is
just opposite to extrusion where metal billet is pushed through the die hole from the entry side. The area of cross-
section/shape of the billet is changed during drawing operations. The metal is plastically deformed by tensile and
compressive forces generated by the reaction of deforming billet with the die wall. The drawing operations are carried out in
multi-stages causing heavy reductions of the input material. Generally the drawn material takes the shape and size of the die
hole. Mostly the operation is done at room temperature, while some materials like tungsten (to make incandescent lamp
filaments) are drawn at warm temperatures. In rod drawing, a rod is pulled through shaped die hole. The initial shape of the
rod is round and is converted into another round rod or to other shapes like square, triangle, etc. In the case of tube drawing,
the tube billet is drawn over a mandrel or without using a mandrel. In tube drawing without mandrel (known as tube sinking),
the outer diameter of the tube is reduced, with increase in length and tube thickness. The increase in tube wall thickness and
length depends on the flow stress of the material, friction conditions and die geometry. In tube drawing with fixed mandrel,
generally used for drawing large to medium diameter straight tubes, the mandrel when pushed into the deforming zone, is
pulled in the forward direction due to the friction generated by the translation of tube during forming. Sometimes while making
long, small diameter tubes, the mandrel will also elongate and even undergo failure. In order to avoid this, floating mandrels
are used for tube drawing. The Schematic diagram of rod and tube drawing (with, without mandrel) is shown in Figs 1.22
(a)–(d).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.22 Schematic diagram of rod and tube drawing (with, without mandrel)

The rod and wire are not strictly distinguished. Generally the term "wire" refers to a small diameter product less than 5 mm.
The rod and wire drawing operations are generally designed such that uncoiling, drawing and recoiling occur continuously at
different stages. Rod coils, when ready for drawing, are usually butt welded for continuous drawing operation. The rod or wire
drawing involves intermittent annealing in order to reduce the effect of strain hardening and flow stress increase during
drawing.

The selection of tool materials for cold drawing of raw billet into rod, bar, and tube is dependent on the size, composition,
shape, stock tolerance, and quantity of metal drawn. The dies used for drawing undergo severe abrasion. Dies used for
drawing are made of hardened steel, tungsten carbide or diamond. The tube mandrels are usually fitted with carbide nibs.
Lubrication is required for wire, bar, and tube drawing operations for reducing friction and wear. It should also dissipate heat
during drawing, especially in high speed forming. It is important in tube drawing operations with fixed mandrels also
(lubrication is required in floating mandrel any way). In dry drawing, the dry soap powder is used as lubricant that is placed in a
die box through which the material will be passed before it is drawn in the die . This is used for drawing a steel wire of
diameter 0.5–1 mm. The conversion coating consisting of sulfate or oxalate is also used. For drawing titanium, polymers are
used as lubricant. For wet drawing, the lubricants are of oil-base or aqueous nature. This wet drawing practice is typical of all
non-ferrous metals and steel wires less than 1 mm in diameter. The commonly used lubricant is sodium stearate. The average
friction coefficient ranges from 0.01 to 0.1 in a practical wire drawing process.

A highly approximate analysis of cold drawing a wire (rigid, perfectly plastic, incompressible material) throughfrictionless
conical die with the reduction in cross-sectional area from A0 to A1 delivered the wire drawing force 'F' and stress 'σw' as,

and

(1.19)

where σ0 is the constant flow stress. Here σw should be less than σ0 so that wire does not break. The drawing stress in a die
with semi-cone angle (θ) and friction (μ) in the presence of back tension (σb) is given by

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(1.20)

In the absence of σb, the drawing stress, and force to give a reduction of r' ' where

(1.21)

and

(1.22)

where

Similar equation like Eq. (1.21) was used by Rubio et al. [35] to investigate and compare the results with finite element results
and experiments. Finite element modeling and analyses of wire drawing was performed by Dixit and Dixit [36] to investigate
the influence of parameters like reduction ratio, semi-die angles, friction, back tension, and strain hardening. With increase in
semi-die angle, the normalized drawing stress decreases to an optimum value and then increases. It is found that the
optimum-die angle increase with increasing reduction for a given friction coefficient. The drawing stress increases with
increase in percentage reduction. It is also seen that the optimum die angle increases with increasing coefficient of friction
from zero to 0.1. Further, the variation of drawing stress with coefficient of friction is more significant at smaller die angles in
comparison to that at larger die angles (Fig. 1.23). The drawing stress is observed to increase with increasing back tension. It
is observed that the pattern of die-pressure distribution does not change greatly with strain-hardening. However, its value is
greater with strain-hardening than without strain-hardening, along the whole taper length of the die. The influence of above
said parameters on die pressure and separating force were also studied for three different materials viz., steel, aluminum, and
copper.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.23 Variation of drawing stress with die angle at different coefficients of friction during wire
drawing [36]. With permission from [36] copyright [1995] Elsevier

Similar analyses were performed by Vega et al. [37] by comparing the experimental results with finite element simulations.
They have quoted Siebel's equation for normalized drawing stress of wire drawing as,

(1.23)

where σf is the drawing stress, σ0 is the yield stress, Ri is the initial wire radius, Rf is the wire radius after deformation, θ is the
die semi-angle, and μ is the friction coefficient. An interesting behavior is the effect of semi die angle on the equivalent plastic
strain distribution across the radius of the wire from center to surface, both at the entry and exit side of the die. It is observed
at the entry side that with decrease in semi-die angle, the plastic strain is uniformly distributed. The strain distribution is
uniform along the wire radius for semi-die angle of 7° and 8°, whereas it is not uniform at higher angles 14° and 18°. Similar
behavior is observed at the exit side of the die also.

Avitzur [38] has proposed the normalized drawing stress as,

(1.24)

where σb is the back tension and P is the bearing length.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
In the case of tube drawing, the plug or mandrel decides the plastic deformation of tube wall. Under plane strain conditions,
for tube drawing with a plug, the draw stress [2] is given by,

(1.25)

Here μ1 is the friction coefficient between tube and the die wall, μ2 is the friction coefficient between tube and plug, α is the
semi-cone angle of die, β is the semi-cone angle of plug (= 0° for cylindrical plug).

There are a few defects that are generated during wire drawing process. A comprehensive analysis of the surface defects can
be found in Shinohara and Yoshida [39]. The important defects are scab, transverse crack, burnt, scratch, overlap, scale,
rolled-in material, over filled, slip mark, rust. The experimental and finite element simulations were compared to study the
evolution of various flaws during stainless steel wire drawing. Similarly the prediction of center burst, an arrow shaped
internal defects, is studied by Ko and Kim [40] for extrusion and wire drawing. In this they have used Cockcroft–Latham
criterion to predict the center burst defects and compared it with experiments and criterion given by Avitzur. Avitzur's criterion
is based on an upper bound analysis assuming that when the central burst occurs, energy required for the flow with defect will
be lesser than that for a defect-less flow By finite element analysis, Dixit and Dixit [36] observed that during the occurrence of
central burst, the plastic zone narrows down at the center and strain-rates near to the die surface are much higher compared
to the center of the wire.

1.3. SHEET FORMING PROCESS

The sheet metal forming processes involve plastic deformation of metallic sheets with the help of tools like punch, die, blank
holder, draw bead to make useful sheet products. The material properties (like n, m, r); process parameters like friction
between the tools and sheet, blank holding force, working temperature, strain rate; die and punch geometry influence the
formability of sheet parts in a compounding fashion. Formability is nothing but the ability of material to be formed under the
action of forces till occurrence of neck/failure. The important sheet forming operations like deep drawing, stretching (or
stretch forming), bending, spinning are described here.

1.3.1. Deep Drawing

Deep drawing of sheet materials is used to make sheet products by pushing the sheet at the center portion into the die hole
while holding it with the help of blank holder. The sheet is drawn in, to make a cup like component without wrinkling and
failure. The blank holder is used to avoid or minimize wrinkling during this operation. The deep drawing operation is
schematically shown in Fig. 1.24 for making simple cylindrical cup. The deformation is complex with tension in wall and
bottom, compression and friction at the flange region (un-deformed region of sheet lying on the die surface), straightening in

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
the die wall and bending at the punch and die radius. The process is utilized to make beverage cans, washbasins, cooking
utensils, cylinders, automotive body parts, pressure vessels, etc. The clearance between the punch and die wall is decided as
per the sheet thickness such that ironing is prevented. Some times ironing is also included as part of forming process which is
performed subsequently. Optimum radii in punch bottom to side wall, and approach to die opening has to be provided to have
safe drawing operation without tearing. The radii should be smooth and designed such that strain localization does not occur
in those regions.

Figure 1.24 Deep drawing of cylindrical cup

In most of the deep drawing operations, the cup is partially drawn, the flange region of the sheet is trimmed if not required,
and the straight cup region is used. A desirable, but suitable, blank holding force (BHF) is applied on the sheet such that
neither tearing (near the holding region) not wrinkling occurs during the deep drawing operation. Wrinkling occurs because of
the compressive hoop stresses developed in the flange region. Two broad approaches are used to analyze the flange
wrinkling: (i) energy method, and (ii) bifurcation method. These methods predict the critical values of certain process
parameters at the initiation of wrinkling. In the energy method, a certain wave-form of the flange is assumed and the critical
values are determined from the condition that this wave-form must be in equilibrium. In the bifurcation method, the critical
values are obtained from the condition at which the solution of the (non-linear) governing equations becomes non-unique.
These approaches will be described in detail later in the book.

In some cases, the cup shape is fully drawn into the die opening, and a straight wall cup shape is ejected through the die
opening. Both mechanical and hydraulic presses are used for deep drawing operations. Hydraulic presses are generally
preferred because of the controllability of punch movement. This operation can be performed in single or in multiple stages,
with or without annealing in between stages.

At the start of the drawing process, the bottom of the punch touches the sheet surface. The sheet wraps around the punch
bottom corner and on the die opening corner. The punch pushes the sheet into the die cavity that forms a straight side wall
with some sheet part in the flange region. If the metal has a high resistance to thinning and thickening, the bottom radius and
the upper sidewall areas remain close to their original thickness, and the radial and circumferential strains obtained in the
drawn-in flange are increased to accommodate the deep-drawing process. Once the bottom is formed, the clearance between
the punch and die is such that the metal in the cup side is free to move without excessive rubbing on the die walls. It is
observed that slight roughening of the punch radius and minimizing the lubrication of this area contribute to better drawability,
while the die opening should be smooth and well lubricated.

The sheet properties that are important for proper deep drawing operations are, the composition of the material; the
mechanical properties like elongation— obtained from tensile tests, plastic strain ratio (r), strain hardening exponent (n), yield
strength of the material; and properties like die geometry, and elastic modulus. The drawability of sheet material depends on

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
two important factors: (i) the ability of sheet in the flange region to move freely in the radial direction in the plane of the sheet
under shear deformation, (ii) the ability of sidewall material to resist deformation in the thickness direction. In order to have
good deep drawability, maximum material flow is desirable in the plane of the sheet and maximum resistance to material flow
in a direction perpendicular to the plane of the sheet. Since it is difficult to evaluate the flow strength in the thickness
direction, plastic strain ratio (r) is used to compare strengths in the plane and thickness directions by determining true strains
in these directions in a tension test. The plastic strain ratio is defined as the ratio of true width strain (εw) to true thickness
strain (εt) for a material strained in longitudinal (perpendicular to width and thickness) direction, i.e.,

(1.26)

The sheet materials are anisotropic in nature, i.e., the properties of the sheet are different in different directions. Therefore, it
is necessary to use the average of the strain ratios measured along 0°, 90°, and 45° to the rolling direction of the sheet to
obtain an average strain ratio, , which is expressed as

(1.27)

where r0, r45, r90 are plastic strain ratios in 0, 45, and 90° to the rolling direction of the sheet. Here is the measure of normal
anisotropy of the sheet. For isotropic material, = 1. For higher resistance to thinning, > 1. The higher the normal
anisotropy value, higher the resistance to thinning and hence deeper draws can be achieved. For many types of steel, for >
1, the flow stress in thickness direction will be more than the in-plane flow stress. The reverse is the case for < 1. However,
for some commercial aluminum alloy and steel sheets an anomalous behavior is observed in which the flow stress in
thickness direction is greater than in-plane flow stress, even when < 1.

There is another measure known as planar anisotropy (Δr) expressed as

(1.28)

which is a measure of how different the 45° directions are from the symmetry axes. Drawability can also be expressed in
terms of limiting draw ratio (LDR) or percentage of reduction based on cup tests. The limiting draw ratio (LDR) is the ratio of
the diameter D of the largest blank that can be successfully drawn to the diameter of the punch d, i.e.,

(1.29)

The percentage reduction is then defined as

(1.30)

The drawing operation is carried out either in hydraulic or mechanical presses. A double action hydraulic press is favored so

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
that a uniform blank holding force can be applied over the entire stroke of the deep drawing operation. Double action hydraulic
presses with die cushion arrangement are also preferred because of their constant drawing speed, stroke adjustment, and
uniform application of clamping pressure.

The process parameters that affect the deep drawing behavior include punch and die radii, punch to die clearance, press
speed, lubrication, and type of restraint of metal flow used. Material properties such as sheet thickness and anisotropy also
affect deep drawing. The deep drawing force (maximum value and load–stroke behavior) and failure analyses are the
important drawing behaviors that are predicted for a successful deep drawing operation. A typical punch force (or load)–
stroke diagram during deep drawing operation is shown in Fig. 1.25. The following equation was developed through
equilibrium analyses of deep drawn cup to evaluate the total punch force [2]:

(1.31)

where PT is the total punch load, σ0 is the average flow stress, Dp and D0 are punch and blank diameters, Pb is the blank
holding force, h is the wall thickness, μ is the friction coefficient, and B is the force required to bend and re-straighten blank.
The load in deep drawing is applied to the cup bottom and then it is transmitted to the die wall. Generally failure occurs in the
annular region just above the punch radius. This region undergoes tensile stretching without any bending and radial drawing.
Failure occurs through necking, later converted into tearing, at a stress approximately equal to tensile strength, increased by a
plane strain factor [2]. This is given by

(1.32)

where Fu is the tensile strength.

Figure 1.25 Punch force–stroke diagram during deep drawing

Also the theoretical upper limit of LDR is given by

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(1.33)

where η is an efficiency term accounted for friction losses. For η = 1, LDR = 2.7, while LDR ≈ 2 forη = 0.7. This agrees with the
observation of how difficult it is to draw a cup with height more than its diameter even for a ductile material. More
comprehensive analyses of deep drawing for evaluating stress, strains can be seen in the work done by Budiansky and Wang
[41] and Woo [42].

The selection of punch to die clearance is important in deciding the smooth metal flow without burnishing of side wall and
punching out of cup bottom. This can be guaranteed with a clearance per side of 7–15% of sheet thickness [1]. The drawing
force will be minimal when the clearance per side is 15–20% of sheet thickness and the cupped portions of the part are not in
contact with the walls of the punch and die. The drawing force increases as the clearance decreases. It should be noted that a
secondary peak occurs on the load–stroke curve where the sheet metal thickness is slightly greater than the clearance
provided and thereafter ironing starts. The clearance between the punch and die for a rectangular cup (at the sidewalls and
ends) is same as that for a circular cup. Clearance at the corners may be as much as 50% of sheet thickness to avoid ironing
in these regions of the cup and to increase drawability. Redrawing operations require greater punch to die clearance as
compared to first draw in order to compensate for thickening during cupping.

The restraint given to the metal flow during deep drawing is provided by the sheet thickness and the die radius. Additionally for
successful operation, the blank holding force is also applied that prevents wrinkling and controls the metal flow. The ratio (l/t)
between the supported length (the length from the edge of the blank to the die cavity) and the sheet thickness is a critical
parameter to decide the usage of blank holding during drawing operation. The usage of blank holder prevents wrinkling, which
if present, would deteriorate the deep drawability of the sheet. Much greater drawability is possible when a blank holder is
used. The blank holders can be used in single action or double action presses. In double action presses, the blank holding is
done first and then the punch is moved down to deform the sheet. The holding process is continued till the process is
completed. By using die cushion arrangement and inverted die, similar action can be obtained in a single action press. The flat
holders are used in many cases that restrict the movement just because of the friction between the sheet and the holder.
Some times the blank holder can be used with drawbead facility that gives more restriction to the sheet movement required
for making a product. The usage of drawbead will provide more stretching component rather than pure drawing. The
drawbeads are used to lock the sheets in which case only stretching is present without any drawing mode of forming. There
are flexible blank holders that are used to control the metal flow locally in specific locations. For example, Kinsey et al. [43]
developed a novel method to control the weld line movement locally and this delivered a Tailor Welded Blank (TWB) product
with increased formability.

The press speed for deep drawing operation is dependent on the sheet material formability, lubrication, toolings, and blank
holder pressure. Speed is of greater significance in drawing stainless steels and heat resistant alloys than in drawing softer,
more formable materials. It is observed that excessive press speeds have caused cracking and more wall-thinning in drawing
a stronger, less ductile metals. By decreasing the speed, the stretching part is reduced and gives the metal enough time to
undergo plastic deformation.

As discussed in other metal forming operations, lubrication significantly affects the performance of deep drawing operation. In
deep drawing operation, lubrication serves two purposes. Firstly it reduces friction and prevent/minimize galling (pressure
welding) of tools and sheet material, wrinkling, and tearing; secondly, it should not enhance corrosion. Water based and oil
based (like fatty oil, mineral oil, mill oil, etc.) lubricants are commonly used for drawing applications.

Material properties like anisotropy and sheet thickness also affect the deep drawing behavior. Planar anisotropy (defines the
variation of properties in sheet plane), if very high, causes earing defect, which is nothing but the result of differential
drawability of regions in the sheet plane. The thicker metal in the valleys between the ears forces the die open against the
blankholder pressure, allowing the metal in the relatively thin areas near the ears to wrinkle. Sheet thickness affects the

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
drawability through pressure requirement and wrinkling. Thinner blanks are more susceptible to wrinkling than thicker ones.
Hence the blank holding pressure requirement for thicker sheets is minimum or even less than that required for thinner
sheets. The finite element simulation by Padmanabhan et al. [44] suggests that by using variable blank holding force (BHF)
and friction during drawing of LPG bottles, the metal flow and minimum thickness achieved can be enhanced. It is indicated by
them that a low constant BHF in the starting stage prevents necking failure between punch and die radius and it plays a
prominent role in the thickness distribution of the drawn part. When the punch force remains constant, an increasing blank
holder force restrains the wrinkling tendency and enables a smooth flow of material into the die cavity [44].

Other than wrinkling, tearing, earing, the defects like orange peeling and stretcher strains also destroy the structural integrity
of the formed parts. Orange peeling is the generation of higher surface roughness in the region of sheet that has undergone
appreciable deformation. This is seen generally in large grain size material. This is due to the fact that the individual grains
tend to deform independently of each other, and hence the grains stand out in relief on the deforming surface. This is better
minimized by using finer grain sized material so that the grains deform as a whole. The stretcher strains are depressions on
the sheet surface, mostly seen in low carbon steel sheets. This is directly related to presence of yield point in the stress–
strain curve and the non-uniform deformation arising from the yield point elongation. This is eliminated by small amount of
cold reduction (2%) in thickness. The temper rolling or skin rolling treatment cold works the material to eliminate the stretcher
strains.

1.3.2. Stretch Forming

Stretch forming of sheets (or stretching) is a sheet forming process in which the sheet is formed over the die by keeping it
under tension. The sheet is held at certain locations with the help of jaws or drawbeads, and then is wrapped around the form
block or die to give a permanent shape change to the part. It is primarily used for making automotive door panels, automotive
roof, aircraft skin panels, window frames, etc. Some of the advantages of stretch forming over other conventional forming
operations are reduced springback, less likelihood of wrinkling, less force requirement, increase in strength, and hardness of
material, lower residual stresses, no requirement of female die part, and usage of less expensive die materials for room
temperature operations. Some of the limitations are as follows: the operation is not suitable for high volume production; it is
best suited for shallow or flat components; not suited to progressive or transfer operations etc. The stretch forming is
categorized into four different categories namely stretch draw forming, stretch wrapping, compression forming, and radial
forming. The difference between these categories lies in the manner by which the sheets are stretched.

In stretch draw forming, the sheet is held in tension by grippers at the ends. A die or form block is moved against the sheet and
is stretched taking the shape of the form block surface. Alternatively the form block is kept stationary; the grippers are moved
such that the sheet is stretched over the block. The stretch draw forming can be used to make small contour shapes with the
help of mating dies. In stretch wrapping, sufficient tension is applied to one end of a workpiece, while the form block revolves
into the workpiece with the help of a turning table. The other end of the workpiece is clamped in a table gripper to the end of
the form block.

The sheet stretching over the die is influenced by material parameters like strain hardening exponent (n), strain rate sensitivity
(m), sheet anisotropy (r); friction conditions, and tool geometries. The stretching is used in many standard formability tests
like limit dome height test (LDH) in which the sheet is gripped using drawbead and stretched using a hemispherical punch till
necking occurs. The height (or strain called limit strain) at which failure/necking occurs is the measure of formability of
sheets. In a simple manner, the elongation at failure during tensile testing is the measure of ductility of sheet metals. The
necking phenomenon in sheet stretching operation occurs in two steps — diffused necking and localized necking. In diffused
necking, the neck region is extended over larger sheet region. For diffused necking, Considere has proposed a criterion during
deformation of thin sheet, as early as 1885, in which the condition of instability resulting in localized deformation is defined by
dP = 0. By assuming volume constancy relationship, the tensile instability condition is given by,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(1.34)

In terms of engineering strain, the necking criterion can be expressed as,

(1.35)

A simple form of the criterion is derived as below. From the definition of n' ',

(1.36)

By further substitution, for instability or necking to start,

(1.37)

where 'εu' is the true uniform strain.

(1.38)

The localized necking is the case when the neck is restricted over a small, narrow band with a width equal to sheet thickness,
inclined at θ = ±55° (approximately) to the major straining direction in the case of isotropic material. In the case of anisotropic
material,

(1.39)

where , ε2 is the strain along width direction, and ε3 is the strain along

thickness direction [45].

Similarly for a balanced biaxial tension condition of sheet metal, the instability occurs when, , by assuming power law
( ) as strain hardening law. In the case of thin walled sphere under internal pressure, instability occurs when,

[2].

The impact of strain hardening exponent (n) and strain rate sensitivity (m) is significant during sheet stretching operation. The
strain hardening exponent basically decides the strain gradient in the sheet material during forming. The strain gradient arises
whenever the deformation is not uniform. The deformation becomes more uniform (reduced strain gradient) with increase in
strain hardening exponent (larger n). This is expressed as Eq. (1.40) [2] by assuming Hollomon strain hardening law, i.e.,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(1.40)

Once the material region is strain hardened, the load is passed on to the neighboring regions. This will make the material to
strain more and in doing so, the strain gradient is reduced. The uniform strain gradient will result in increased formability of
sheet material, i.e., sheet materials with larger n value exhibit better formability when compared with sheets with lower n
value. The influence of strain rate sensitivity (m) on the forming extent is similar to the effect of n value. The m value governs
the post-necking behavior. The conversion of diffused necking to localized necking is characterized by this value. The higher
the m value, the more the diffused necking, and greater the deformation beyond the maximum load before fracture occurs.
Thus, higher the m value, larger will be the forming limit.

During stretching operation, the formability of sheet metal is best described by forming limit diagram (FLD). In this diagram,
forming limit curve (FLC), plotted between major strain (in Y-axis) and minor strain (in X-axis), is the index that says the amount
of safe strains that can be incorporated into the sheet metal. A typical FLC is shown in Fig. 1.26. The FLC is the locus of all the
limit strains in different strain paths of the sheet material. The plane-strain condition possesses the least forming limit, when
compared with deep drawing and stretching strain paths. The influence of n, m, and r values on the predicted forming limit by
Marciniak and Kuczynski (M-K) model is described in [46, 47]. It is seen that the positive 'm' value (= 0.012) improves the
forming limit as compared to negative 'm' value (= – 0.012) by delaying the conversion of diffused necking to localized
necking and then to failure [47]. The influence of 'r' value on the forming limit curve is dependent on the yield criterion used for
modeling. By using Hosford non-quadratic yield criterion, it is shown that 'r' value has got negligible effect on the forming limit
strains of sheet metal. On the other hand, by assuming planar isotropy, it is predicted that on decreasing the 'r' value the level
of the FLC is increased on the right hand side of FLD [46, 47]. This is explained by Sowerby and Duncan [47] that on increasing
the 'r' value from 0.5 to 3.0, the forming limit strain is found to decrease considerably. Like the effect of m'
' value, on increasing
'n' value from 0.1 to 0.4, the forming limit is found to increase considerably [47].

Figure 1.26 Forming limit diagram (FLD) of sheets

1.3.3. Sheet Bending

Bending of sheets is a common sheet forming operation in which the deformation is concentrated in a small curved region,
resulting in a non-uniform deformation across the thickness and curvature of the sheet. The different bending operations like
V-bending, L-bending and U-bending are shown in Fig. 1.27. These operations are distinguished based on the dies used, with
U-bending resulting in a channel type component. The sheet outer surface undergoes tension and inner surface undergoes
compression during bending as shown in Fig. 1.28. When the metal is deformed beyond the elastic limit, permanent set

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
occurs. However, some elastic deformation usually remains, so that the metal tends to spring back as the loading is released.
This springback is compensated by overbending. If failure occurs during bending, it will occur on the outer surface, and
wrinkling is witnessed on the inner surface. The friction forces act in the opposite direction to the sliding direction of sheet
during bending and hence proper lubrication is required for the efficient running of the process. The identification of neutral
axis in a bending operation is crucial in bending analyses. The neutral axis is the line of zero stress and strain in a bend with
tension on one side and compression on the other side of the line. Before bending, the outer surface, inner surface lengths
and neutral axis lengths are same as that of sheet length. After bending, the outer surface length is increased, while inner
surface length is decreased. But the neutral axis length remains the same and hence it is considered as a reference for blank
development calculations.

Figure 1.27 Bending operations.

Figure 1.28 Sheet bending showing tension and compression on the surfaces

When the sheet is bent for the first time, the neutral axis is at the center of the sheet thickness. As bending progresses, the
neutral axis shifts toward the inside, or compression side, of the bend.

In bending a thin sheet, it may be assumed that a plane normal section in the sheet will remain plane and normal and converge
on the center of curvature as shown in Fig. 1.29. The line A0B0 at the middle surface may change its length to AB, if the sheet
is under stretching during bending. The original length lo becomes, ls = ρθ. A line C0D0 at a distance y from the middle surface
will deform to a length,

(1.41)

where ρ is the radius of curvature.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.29 Deformation of longitudinal fibres during bending and stretching

The axial strain of the fiber CD is

(1.42)

where 'εa' and 'εb' are the strains at the middle surface and bending strain, respectively. In the case of bending with radius of
curvature larger compared to the thickness, the bending strain is approximated as

(1.43)

The strain distribution is linear [48] across the thickness. For the strain distribution given by Eq. (1.43), the stress distribution
can be evaluated by following a stress–strain law. This stress–strain law is usually approximated to have several material
models like elastic bending, rigid-perfectly plastic bending, elastic-perfectly plastic bending, strain hardening sheet, etc. [48].
The strain also depends on the bend ratio, which is defined as the ratio of the radius of curvature to sheet thickness, ρ/t.
Bending without tension is done by pure moment without any tension being applied. The neutral axis is at the mid-thickness in
this case.

Spring back is considered as an important phenomenon during bending operations. The tensile stress, A, above the neutral line
cause plastic deformation and it is less than ultimate tensile strength [Fig. 1.30(a)]. Upon load removal after first bending, the
elastic band near the neutral axis [Fig. 1.30(b)] tries to return to the original flat condition but cannot, due to the restriction
given by the plastic deformed regions. Some return occurs which is named as springback. Increase in elastic limit/yield
strength, increases the springback and hence stronger sheets have greater degrees of springback. Springback is lower, when
elastic modulus is reduced and plastic strain in increased. The springback prediction will be helpful in proper and accurate die
design, minimizing springback, standardizing methods to reduce springback, accurate formability prediction, etc. There are
many numerical and analytical methods by which springback can be predicted. More information on springback prediction
can be obtained from the work done by Mullan [49], Gau and Kinzel [50] including Bauschinger effect, Li et al. [51] using finite
element simulations, Wagoner et al. [52] simulating the effect of through thickness integration points, Bahloul et al. [53] using
response surface methodology and analyses done by many others. The die design prediction by including springback errors
can be understood from the work of Karafillis and Boyce [54] and Gan and Wagoner [55].

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.30 (a) Changing stress patterns in a bend

Figure 1.30 (b) Elastic and plastic deformation zones during bending

1.3.4. Spinning
In spinning, the blank of certain thickness is held by a mandrel of certain shape, while the tool is made to press the blank
against the mandrel. At the end of spinning process, blank takes the shape of the mandrel that is axi-symmetric. There are
three types of metal spinning operations— manual spinning, power spinning, and tube spinning. In manual spinning, the tool
used for spinning operation is manually or hydraulically operated as shown in Fig. 1.31. Generally the manual spinning is
carried out at room temperatures. However, at elevated temperature metal spinning is performed for components with thick
sections or for alloys with low ductility. It is meant for low volume production (say less than 1000 pieces). Larger volume parts
are usually produced at lower cost by power spinning. Metal spinning is used to make components made of aluminum alloys,
stainless steels, super alloys, brass, copper, and platinum. Power spinning can be carried out at high temperatures also. Since
high shear stresses are involved in this, the heat generated can be removed only with proper lubrication. The surface
roughness produced in this is of the order of 0.5–1.5 μm. Buckling, wrinkling, and formation of circumferential cracks are
some of the defects seen during spinning.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 1.31 Manual spinning operation

There are process variables like feed ratio, roller path, roller design, spinning ratio affecting the performance and ability to
make successful spun components [56]. Feed ratio is defined as the ratio of the roller feed rate to the spindle speed. By
keeping the feed ratio as constant, the roller feed and the spindle speed can be varied without any considerable influence on
the quality of the product. High feed ratios generate higher forces, resulting in cracking. On the other hand, too low a feed
ratio reduces the workability and severe thinning occurs. There are different roller paths such as linear, concave, convex,
involute and quadratic, and they show significant influence on the deformation of the blank. With correct roller path, the
formation of wrinkles and cracks can be avoided. The roller design is found to show profound influence mainly on the
thickness uniformity of the blank. Smaller the nose roller radius, poor thickness uniformity is witnessed. Spinning ratio is
defined as the ratio of blank diameter to mandrel diameter. The larger the spinning ratio, more difficult is the spinning
process. If the spinning ratio is too large, circumferential splitting along the transition from the flange to the wall occurs [56].
A review on the mechanics of spinning process can be seen in Music et al. [57].

1.4. SOME OTHER FORMING PROCESS

In this section, some of the recently developed and researched metal forming processes like Equal Channel Angular Extrusion
(ECAE), microforming, ring rolling, hydroforming, incremental forming, rubber pad forming, and age (or creep) forming are
described briefly.

In equal channel angular extrusion, the deformable material undergoes severe plastic deformation creating large plastic strains
by the application of simple shear. The tool used for ECAE contains two intersecting channels of identical cross-sections. A
lubricated billet is pressed into the first channel and extruded through the second channel. The deformed billet can be
removed from the end of second channel. Since the billet dimensions are same, the process can be repeated few times,
changing the orientation of the billet between each pass. ECAE processed materials show improved mechanical properties,
extremely fine grain sizes, and varying textures [58]. This process is very suitable for the production of bulk ultra fine grained
materials. Though the application of this process is restricted, it is being researched widely by many researchers. The
suitability of this technique for processing of tubular specimen geometries has also been investigated by Nagasekhar et al.
[58].

The microforming is the process of manufacturing a part or feature by plastic deformation, whose at least one orthogonal
view can be enclosed in a square of 1 mm size. These parts are used mostly in making electronic parts and in micro-electro
mechanical systems (MEMS). Typical examples are microscrews, fasteners, etc. [59]. Microforming can be done with both
bulk and sheet materials. In the bulk metal forming, a typical micro forming process is wire bending for the production of
filaments and springs as used in medical applications and electronic industry. In this process, the influence of size effect on

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
the flow behavior is significant. With decreasing specimen size, in general, a decrease in flow stress is observed. According to
the surface layer model, the grains located at the surface of a tensile or upsetting specimen are less restricted than grains
inside the material. Dislocations moving through the grains during deformation pile up at grain boundaries but not at the free
surface. This results in less hardening and lower resistance against deformation of the surface grains. With decreasing
specimen size and a size invariant microstructure, the share of surface grains increases, which leads to lower flow stress.
Details about micro deep drawing of thin sheets can be obtained from the work of Saotome et al. [60].

In ring rolling, seamless ring shaped components are produced. The input material for the ring rolling process is of do-nut
shape (shaped like a thick ring) made by hydraulic or mechanical presses or hammers. The donut shaped blank is placed over
the undriven mandrel and the mandrel is forced with pressure against the rotating main roll. Once contact occurs between the
rotating roll and the blank and the mandrel, since the main roll is rotating, the blank and the mandrel will also rotate in the
opposite direction. The mandrel is then moved towards the main roll, shaping the ring, and reducing the wall thickness.
Simultaneously the diameter is increased by circumferential extrusion that occurs during this process. The ring height is
controlled by applying pressure through top and bottom rolls on the ring surface. This technology can be used in the
manufacture of bearings, ring gears, final drive gears, transmission components, clutch components, and wheel blanks. Parts
used in aerospace sector include rotating and non-rotating rings for fans, engine casings, and engine disks.

In hydroforming, the parts are made by plastically deforming the sheet or tube by the application of fluid pressure through
movable punch. The sheet is deformed and it takes the shape of the die cavity. In tube hydroforming, the axial force is applied
to the tube as well as internal pressure, which creates compressive stress in one direction so that the tube is deformed
without thinning and necking is delayed. Some of the advantages of this process are requirement of simpler die construction
and being economical for small number of parts. Disadvantages include presence of very high pressures and greater cycle
time than for stamping in mechanical presses.

In incremental sheet forming, the sheet is clamped at the edge and a small tool is moved on the sheet surface. If an
intermediate support is provided during forming, it is called two point incremental forming, otherwise it is called single point
incremental forming. Both symmetric and asymmetric shapes are manufactured by this technique. A review by Emmens et al.
[61] details the variants of incremental sheet forming process.

Rubber pad forming, also called as flexible die forming, involves forming of sheet metal with one of the tools made of flexible
diaphragm or rubber pad, with another tool like that used in conventional forming operations. The flexible tool could be a
punch or die cavity. The rubber pad acts like hydraulic fluid in exerting nearly equal pressure on all workpiece surfaces as it is
pressed around the form block [1]. When compared with conventional forming or deep drawing, thinning of the sheet metal is
reduced considerably, and excellent surface finish is obtained by using this method.

Age forming (or creep forming) is a shaping process mainly for heat treatable aluminum alloys that has gained popularity in
recent times. In this process, parts are given aging treatment while simultaneously being deformed by the application of
mechanical loads. It is used mainly for making large parts that have to be imparted a curvature. The part shape is obtained due
to the creep that occurs at the aging temperatures. Since the part shape is achieved by creep, age forming is sometimes
referred to as creep forming. The most common application of age forming is the shaping of upper wing skins in the
aerospace industry. During age forming, the forming loads are usually less than the yield strength of the material, and the part
shape is obtained due to the low temperature creep that occurs during the aging process. Hence, there is less non-uniformity
in the microstructure as compared to cold formed parts. The age formed components have lower residual stresses and
consequently better stress corrosion resistance.

1.5. EVALUATION OF FRICTION COEFFICIENT IN METAL

FORMING
Friction conditions play a vital role in deciding the forging behavior of billets. There are various methods by which the friction
factor m can be evaluated. Some of the methods like ring compression test (RCT), double cup extrusion test (DCE), forward

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
extrusion, bucket cup extrusion test, spike test, upsetting sliding test, injection-upsetting method, and T-shape compression
test are explained briefly.

In ring compression test, a ring shaped billet of size outer diameter : inner diameter : height–6:3:2, is compressed between
two platens (upper and lower) using a hydraulic press. The variation in internal diameter with height reductions is plotted with
specimen height reduction for different friction conditions. Depending on the friction between the ring sample and the
platens, the inner diameter will decrease or increase with plastic reduction in height. Under low friction conditions, the inner
diameter is found to increase (or bulge outwards), while under high friction conditions, the inner diameter is observed to
decrease (or bulge inwards) (Fig. 1.32). The experimental "friction calibration curves" is plotted between percentage reduction
in internal diameter and percentage reduction in height. Lots of analysis was performed by Male and Cockcroft [62], Yamada
et al. [63], Carter and Lee [64], Robinson et al. [65] and many others. The RCT was conducted on porous metals to evaluate the
interfacial friction factor. Tabata and Masaki [66] investigated the friction curves based on RCT on copper powder metals.
The overall review of the ring compression test is presented by Rao and Sivaram [67]. The influence of sample geometry,
barreling, changing friction conditions, strain rate, and temperature can be understood from the available literature. Recently,
some researchers have observed that material properties also influence RCT. In RCT, the die geometry and metal flow are
simple, the forging pressure is relatively low, and the new surface generated during deformation is small. Therefore, in a ring
test the frictional characteristics of the forging lubricant cannot be adequately evaluated. Also, the friction factor evaluated
from RCT can be used mainly for open die forging operations, where degree of deformation is less. In order to overcome the
problem involved in RCT, a double cup extrusion (DCE) test was developed.

Figure 1.32 Sample bulging during RCT

In double cup extrusion test, a cylindrical billet is plastically deformed between two punches, in which the bottom punch is kept
stationary. The top punch which is attached to the ram of the hydraulic press is moved downwards. A combined forward–
backward extrusion is performed on the billet and finally "H-shaped" sample is produced. The ratio of backward (h1) to
forward (h2) extrusion cup heights is controlled by the friction conditions at the container–billet interface. The friction
calibration curves can be obtained by plotting the height ratio (h1/h2) with stroke or height reduction for varied friction
conditions (or friction factors, lubricants). The DCET is found to have following advantages over RCT.

i. The test simulates severe deformation and friction conditions as seen in actual cold forging operations

ii. The test is more suitable for closed-die forging operations, and

iii. The test setup can be fabricated easily and reliable ranking of lubricants can be done

Like RCT, lots of work has been done on DCET for friction evaluation. The work done by Ghobrialet al. [68] shows that the
different reductions in area have significant effect on the height ratio (h1/h2). In turn, the influence on friction factor 'm' is also

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
significant. The FE simulations performed also agreed well with the experiments for different area reductions. Barcellona et al.
[69] from their work suggested that the 'm' value obtained from RCT is more suitable for simple upsetting process simulation,
whereas the 'm' value from DCET is applicable for closed-die forging and extrusion processes. A critical evaluation of DCET
was performed by Schrader et al. [70] in which the effect of different tool geometry and process parameters on the cup height
ratio and contact pressure were analyzed. The impact of strain hardening exponent, strength coefficient, prior work hardening
of the billet, billet height, extrusion ratio (d2 punch /d2 billet), punch die land, punch friction on the cup height ratio, contact
pressure are studied. It is found that almost all the parameters show profound effect on the friction characteristics during
DCET. Similar investigation was carried out by Buschhausen et al. [71] and they recommended the optimum usage of
reduction ratio and billet height during DCE test.

Another important test to evaluate the friction is spike forging test, in which the cylindrical billet is compressed by a top die
with a central bore, while the bottom die is kept stationary. This process involves both backward extrusion and radial forging.
The height of "spike" produced describes the friction between the billet and the die. The Schematic diagram of spike forging
test is shown in Fig. 1.33. Xu and Rao [72] studied the effect of ram velocity, specimen temperature, friction, billet and die
geometry on the spike height of the deformed billet. The ram velocity and specimen temperature showed insignificant effect
on the spike height of the formed billet. The interface friction and die geometry (like fillet radius, vertical draft) showed
significant effect on the spike height. The optimum billet height for successful and efficient evaluation of friction conditions at
the tool–billet interface has been suggested. The metal flow is more in the axial direction than in the radial direction, when the
friction is large. Similar spike forging tests were conducted by Im et al. [73] to investigate the effect of heat loss on the metal
flow through simulations in a Ti alloy. The comparison between simulation and experimental results is done for spike forging
tests by Rao et al. [74] to investigate the impact of friction on the strain and strain rate distribution.

Figure 1.33 Spike forging test

In bucket cup extrusion test, a cylindrical billet is compressed by a punch in a closed die. The billet undergoes backward
extrusion and the final product is in "bucket form". The forging load during deformation and the bucket thickness quantifies
friction between the tool and the billet. Shen et al. [75] compared the RCT and bucket test for their validity to evaluate the
friction factors for four different lubricants. They found that the bucket test is advantageous over RCT in the following ways:
the bucket test is complex representing large surface generation that can be seen in industrial forging operations, and for the
same sample thickness, the forging pressure is more in the bucket test that in RCT that is generally seen in practical forging
operations. They identified a new factor "load multiplied by bucket thickness" which is sensitive to changing friction and hence
used to quantify friction in bucket cup test. They also compared the experimental results with simulations done in DEFORM, a
finite element code.

Recently a T-shape compression test is developed by Zhang et al. [76] in which a cylindrical billet is compressed in a die with
'V' shaped groove. The curved surface of the billet touches the groove before forming. The billet undergoes extrusion in the V-
groove and compression radially resulting in a T-shaped formed product. The height of deformation inside the V-groove
determines the friction characteristics of the lubricant (schematic shown in Fig. 1.34). The authors have identified the slope of
the load curve as a sensitive parameter to quantify friction factor and coefficient. The effect of die geometries like V-groove
angle and corner radius on the height of the extruded part is analyzed. Through FE simulations, the friction factor and friction

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
coefficient of different lubricants (oil, solid, mixed) were evaluated. It can be seen from their work that the friction factor 'm' is
nearly 2.3 times the friction coefficient 'μ' for the same slope of load curve. Therefore, the friction coefficient can be easily
obtained by knowing the friction factor of the lubricant. The amount of bulging in open die forging can also be used as a
measure of coefficient of friction at the die-workpiece interface [77].

Figure 1.34 T-shape compression test

1.6. CONCLUSION
Metal forming processes are used to make automotive, aerospace, household components, etc. through plastic deformation
of billets and sheets. In this chapter, an overview of different bulk and sheet metal forming processes has been presented.
Subsequent chapters will discuss various aspects of metal forming in detail. The information presented in this chapter has
been compiled from a number of research papers and industry catalogues. For more information, about technological aspects
of metal forming, the references cited in this chapter can be referred. It is also advisable to consult the latest data from
industrial sources.

1.7. REFERENCES
1. Semiatin, S. L. (ed.) (2005). ASM Handbook, Metal working: Bulk forming (#06957G), Vol. 14A, ASM International.

2. Dieter, G. E., Mechanical Metallurgy, SI Metric Edition, McGraw-Hill, ISBN 0071004068.

3. Al-Salehi, F. A. R., Firbank and T. C., Lancaster, P. R. (1973). An experimental determinations of the roll pressure
distributions in cold rolling, International Journal of Mechanical Sciences, 15; 693.

4. Altan, T., Oh, S. I. and Gegel, H. L. (1983), Metal forming: Fundamentals and applications, in Gegel, H. L. (ed.),American
Society for Metals, Metals park, OH, USA, ISBN 0871701677.

5. Tieu, A. K. and Liu, Y. J. (2004). Friction variation in the coldrolling process, Tribology International, 37; 177.

6. Jeswiet, J. (1998). A comparison of friction coefficients in cold rolling, Journal of Materials Processing Technology, 80–81;
239.

7. Dixit, U. S. and Dixit, P. M. (1996). A finite element analysis of flat rolling and application of fuzzy set theory,International
Journal of Machine Tools and Manufacture, 36 (8); 947.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 8. Gudur, P. P., Salunkhe, M. A. and Dixit, U. S. (2008). A theoretical study on the application of asymmetric rolling for the
estimation of friction, International Journal of Mechanical Sciences, 50; 315.

9. Jiang, Z. Y., Xiong, S. W., Tieu, A. K. and Wang, Q. J. (2008). Modelling of the effect of friction on cold strip rolling,Journal
of Materials Processing Technology, 201; 85.

10. Gao, H., Ramalingam, S. C., Barber, G. C. and Chen, G. (2002). Analysis of asymmetrical cold rolling with varying
coefficients of friction, Journal of Materials Processing Technology, 124; 178.

11. Pawelski, H., Friction inhomogeneities in cold rolling (2002). Journal of Materials Processing Technology, 125–126; 392.

12. Dick, K. and Lenard, J. G. (2005). The effect of roll roughness and lubricant viscosity on the loads on the mill during cold
rolling of steel strips, Journal of Materials Processing Technology, 168; 16.

13. Shen, Y. F., Xue, W. Y., Wang, Y. D., Liu, Y. D., and Zuo, L. (2008). Tensile behaviors of IF steel with different cold-rolling
reductions, Materials Science and Engineering A, 496; 383.

14. Das, G., Sabita Ghosh, T. and Sahay, S. K. (2005). Use of ball indentation technique to determine the change of tensile
properties of SS316L steel due to cold rolling, Materials Letters, 59; 2246.

15. Ganesh Narayanan, R., Gopal, M. and Rajadurai, A. (2008). Influence of friction in simple upsetting and prediction of
hardness distribution in a cold forged product, Journal of Testing and Evaluation, 36 (4); 1.

16. Kim, H., Lee, S. M. and Altan, T. (1996). Prediction of Hardness Distribution in Cold Backward Extruded Cups, Journal of
Materials Processing Technology, 59; 113.

17. Onuh, S. O., Ekoja, M. and Adeyemi, M. B. (2003). Effects of die geometry and extrusion speed on the cold extrusion of
aluminium and lead alloys, Journal of Materials Processing Technology, 132; 274.

18. Ajiboye, J. S. and Adeyemi, M. B. (2006). Effects of die land on the cold extrusion of lead alloy,Journal of Materials
Processing Technology, 171; 428.

19. Ajiboye, J. S. and Adeyemi, M. B. (2007). Upper bound analysis for extrusion at various die land lengths and shaped
profiles, International Journal of Mechanical Sciences, 49; 335.

20. Kudo, H. (1960). Some analytical and experimental studies of axi-symmetric cold forging and extrusion–I, International
Journal of Mechanical Sciences, 2; 102.

21. Choi, H. J., Choi, J. H. and Hwang, B. B. (2001). The forming characteristics of radial-backward extrusion,Journal of
Materials Processing Technology, 113; 141.

22. Chitkara, N. R. and Aleem, A. (2001). Extrusion of axi-symmetric tubes from hollow and solid circular billets: a generalised
slab method of analysis and some experiments, International Journal of Mechanical Sciences, 43; 1661.

23. Ebrahimi, R., Reihanian, M., Kanaani, M., and Moshksar, M. M. (2008). An upper-bound analysis of the tube extrusion
process, Journal of Materials Processing Technology, 199; 214.

24. Berski, S., Dyja, H., Maranda, A., Nowaczewski, and J., Banaszek, G. (2006). Analysis of quality of bimetallic rod after
extrusion process, Journal of Materials Processing Technology, 177; 582.

25. Avitzur, B., Wu, R., Talbert, S., and Chou, Y. T. (1986). Analysis of core fracture in drawing of bimetal rods and wires,
Journal of Engineering for Industry, 108 (2); 133.

26. Avitzur, B., Wu, R., Talbert, S. and Chou, Y. T. (1985). An analytical approach to the problem of core fracture during
extrusion of bimetal rods, Journal of Engineering for Industry, 107 (3); 247.

27. Irena, N. and Andrzej, S. (2009). Influence of die geometric parameters on plastic flow of layer composites during extrusion
process, Journal of Materials Processing Technology, 209; 1943.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
28. Schikorra, M. and Kleiner, M. (2007). Simulation based analysis of composite extrusion processes,Annals of the CIRP,
56(1); 317.

29. Hwang, Y. M. and Hwang, T. F. (2002). An investigation into the plastic deformation behavior within a conical die during
composite rod extrusion, Journal of Materials Processing Technology, 121; 226.

30. Dodeja L. C. and Johnson, W. (1957). The cold extrusion of circular rods through square multiple hole dies,Journal of the
Mechanics and Physics of Solids, 5; 281.

31. Keife, H. (1993). Extrusion through two die openings: a 2D upper–bound analysis checked by plasticine experiments,
Journal of Material Processing Technology, 37; 189.

32. Ulysse, P. and Johnson, R. E. (1998). A study of the effect of the process variables in unsymmetrical single-hole and multi-
hole extrusion processes, Journal of Material Processing Technology, 73; 213.

33. Sinha, M. K., Deb, S. and Dixit, U. S. (2009). Design of a multi-hole extrusion process,Materials and Design, 30; 330.

34. Sinha, M. K., Deb, S., Das, R., and Dixit, U. S. (2009). Theoretical and experimental investigations on multi-hole extrusion
process, Materials and Design, 30; 2386.

35. Rubio, E. M., Camacho, A. M., Sevilla, L., and Sebastian, M. A. (2005). Calculation of the forward tension in drawing
processes, Journal of Materials Processing Technology, 162–163; 551.

36. Dixit, U. S. and Dixit, P. M. (1995). An analysis of the steady-state wire drawing of strain-hardening materials,Journal of
Materials Processing Technology, 47; 201.

37. Vega, G., Haddi, A. and Imad, A. (2009). Investigation of process parameters effect on the copper-wire drawing,Materials
and Design, 30; 3308.

38. Avitzur, B. (ed.) (1968). Metal Forming: Processes and Analysis, McGraw-Hill, New York; also, (1979). Krieger, Huntington,
NY.

39. Shinohara, T. and Yoshida, K. (2005). Deformation analysis of surface flaws in stainless steel wire drawing, Journal of
Materials Processing Technology, 162–163; 579.

40. Ko, D. C. and Kim, B. M. (2000). The prediction of central burst defects in extrusion and wire drawing,Journal of Materials
Processing Technology, 102; 19.

41. Budiansky, H. and Wang, N. M. (1966). On the swift cup test, Journal of Mechanics and Physics of Solids, 14; 357.

42. Woo, D. M. (1968). On the complete solution of the deep-drawing problem,International Journal of Mechanical Sciences, 10;
83.

43. Kinsey, B., Liu, Z. and Cao, J. (2000). A novel forming technology for tailor welded blanks, Journal of Materials Processing
Technology, 99; 145.

44. Padmanabhan, R., Oliveira, M. C., Alves, J. L., and Menezes, L. F. (2008). Numerical simulation and analysis on the deep
drawing of LPG bottles, Journal of Materials Processing Technology, 200; 416.

45. Hosford, W. F. and Caddell, R. M. (eds.) (1993). Metal forming: Mechanics and Metallurgy, Prentice-Hall, New York, ISBN 0-
13-588526-4.

46. Zhao, L., Sowerby, R. and Sklad, M. P. (1996). A theoretical and experimental investigation of limit strains in sheet metal
forming, International Journal of Mechanical Sciences, 38 (12); 1307.

47. Sowerby, R. and Duncan, J. L. (1971). Failure in sheet metal in biaxial tension, International Journal of Mechanical Sciences,
13; 217.

48. Marciniak, Z., Duncan, J. L. and Hu, S. J. (eds.) (2002). Mechanics of Sheet Metal Forming, Elsevier, New Delhi.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
49. Mullan, H. B (2004). Improved prediction of springback on final formed components,Journal of Materials Processing
Technology, 153–154; 464.

50. Gau, J. T. and Kinzel, G. L. (2001). A new model for springback prediction in which the Bauschinger effect is considered,
International Journal of Mechanical Sciences, 43; 1813.

51. Li, K. P., Carden, W. P. and Wagoner, R. H. (2002). Simulation of springback,International Journal of Mechanical Sciences,
44; 103.

52. Wagoner, R. H. and Li, M. (2007). Simulation of springback: Through-thickness integration, International Journal of
Plasticity, 23; 345.

53. Bahloul, R., Ben-Elechi, S. and Potiron, A. (2006). Optimisation of springback predicted by experimental and numerical
approach by using response surface methodology, Journal of Materials Processing Technology, 173; 101.

54. Karafillis, A. P. and Boyce, M. C. (1992). Tooling design accommodating springback errors, Journal of Materials Processing
Technology, 32; 499.

55. Gan, W. and Wagoner, R. H. (2004). Die design method for sheet springback, International Journal of Mechanical Sciences,
46; 1097.

56. Wong, C. C., Dean, T. A. and Lin, J. (2003). A review of spinning, shear forming and flow forming processes,International
Journal of Machine Tools and Manufacture, 43; 1419.

57. Music, O., Allwood, J. M. and Kawai, K. (2010). A review of the mechanics of metal spinning,Journal of Materials
Processing Technology, 210; 3.

58. Nagasekhar, A. V., Uday Chakkingal and Venugopal, P. (2006). Equal Channel Angular Extrusion of tubular aluminum alloy
specimens analysis of extrusion pressures and mechanical properties, Journal of Manufacturing Processes, 8 (2); 112.

59. Engel, U. and Eckstein, R. (2002). Microforming— from basic research to its realization,Journal of Materials Processing
Technology, 125–126; 35.

60. Saotome, Y., Yasuda, K. and Kaga, H. (2001). Microdeep drawability of very thin sheets,Journal of Materials Processing
Technology, 113; 641.

61. Emmens, W. C., Sebastiani, G. and van den Boogaard, A. H. (2010). The technology of Incremental Sheet Forming— A brief
review of the history, Journal of Materials Processing Technology, 210; 981.

62. Male, A. T. and Cockcroft, M. G. (1993). A method for the determination of the coefficient of friction of metals under
condition of bulk plastic deformation, Journal of Institute of Metals, 64–65; 38.

63. Yamada, R., Suzuki, M. and Harayama, Y. (1977). Application of finite element method to ring compression test, Nuclear
Engineering and Design, 44; 75.

64. Carter, W. T. Jr. and Lee, D. (1985). A finite element analysis of cylinder and ring compression and its experimental
verification, Computers & Structures, 21 (1–2); l.

65. Robinson, T., Ou, H. and Armstrong, C. G. (2004). Study on ring compression test using physical modeling and FE
simulation, Journal of Materials Processing Technology, 153–154; 54.

66. Tabata, T. and Masaki, S. (1978). Determination of the coefficient of friction in forging of porous metals from ring
compression, International Journal of Mechanical Sciences, 20; 505.

67. Rao, K. P. and Sivaram, K. (1993). A review of ring compression testing and applicability of the calibration curves, Journal
of Materials Processing Technology, 37; 295.

68. Ghobrial, M. I., Lee, J. Y. and Altan, T. (1993). Factors Affecting the Double Cup Extrusion Test for Evaluation of Friction in
Cold and Warm Forging, Annals of the ClRP, 42 (1); 347.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
69. Barcellona, A. and Cannizzaro L. (1996). Validation of frictional studies by double cup extrusion tests in cold forming,
Annals of the ClRP, 45 (1); 211

70. Schrader, T., Shirgaokar, M. and Altan, T. (2007). A critical evaluation of the double cup extrusion test for selection of cold
forging lubricants, Journal of Materials Processing Technology, 189; 36.

71. Buschhausen, A., Weinmann, K., Lee, J. Y., and Altan, T. (1992). Evaluation of lubrication and friction in cold forging using a
double backward extrusion process, Journal of Materials Processing Technology, 33; 95.

72. Xu, X. L. and Rao. K. P. (1997). Analysis of the deformation characteristics of spike forging process through FE simulations
and experiments, Journal of Materials Processing Technology, 70; 122.

73. Im, Y. T., Vardan, O., Shen, G., and Altan, T. (1988). Investigation of metal flow in non-isothermal forging using ring and
spike tests, Annals of the ClRP, 37 (1); 225.

74. Rao, K. P., Doraivelu, S. M. and Sivaram, K. (1991). Physical modeling studies using spike forging to verify analytical
predictions, Journal of Materials Processing Technology, 28; 295.

75. Shen, G., Vedhanayagam, A., Kropp, E., and Altan, T. (1992). A method for evaluating friction using a backward extrusion
type forging, Journal of Materials Processing Technology, 33; 109.

76. Zhang, Q., Felder and E., Bruschi, S. (2009). Evaluation of friction condition in cold forging by using T-shape compression
test, Journal of Materials Processing Technology, 209; 5720.

77. Ebrahimi, R., and Najafizadeh, A. (2004), A new method for evaluation of friction in bulk metal forming, Journal of Materials
Processing Technology, 152; 136.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
2. Fundamentals of Plasticity
Arbind Kumar Singh

Department of Civil Engineering, IIT Guwahati, India

2.1. INTRODUCTION
The subject of study of motion and deformation of a body under the action of forces acting on it is called mechanics. Solid
mechanics primarily deals with motion and deformation of deformable solid under action of forces. There are two approaches
to understand these physical behaviors (motion and deformation) of solid body:

a. Physical approach

b. Phenomenological approach

a. Physical approach: In this approach, body is assumed to be made of atoms, molecules and crystals. The behavior of the
body is studied based on interaction of these atoms and molecules. This is the subject matter of solid state physics. It is
very difficult (almost impossible) to apply this approach to engineering structures such as aircrafts, machine tools or
buildings. But the approach gives a very good physical insight into the deformation of the body.

b. Phenomenological approach: Assumptions are made based on macroscopically observed behavior of the body and the
basic principles of physics are applied. The obtained response is checked with the experimental results and hence theory
is developed. Solid mechanics dealt in this chapter is based on phenomenological approach.

2.2. UNIAXIAL TENSION TEST

This is the simplest test conducted on metals. The results of the test provide basic information which has been widely used in
the design of structures. In this test a cylindrical circular rod is tested by applying continuously increasing load and measuring
the elongation of the specimen (gauge length, specified portion) simultaneously. Also, diameter of the specimen is measured.
The specimen is tested in a machine (simplest case) called universal testing machine. The rate of loading is kept constant
during the experiment and adjusted in such a way to avoid creep or dynamic effect. (The rate of loading should not be high
such that inertia effect comes and also it should not be low such that the time dependent deformation comes into existence.
To keep these points a standard has been developed for the test.) The tension test is conducted at constant temperature.

Although the uniaxial tension test is simple and easy to conduct, but the interpretation of its data is not so easy as it looks at
the first glance. The complexity of the test can be judged by the following statement of E. Orowan [1].

"The extension of a piece of metal (is) in a sense more complicated than the working of a pocket watch and to hope to derive
information about its mechanism from two or three data derived from measurement during the tensile test is perhaps as
optimistic as would be an attempt to learn about the working of a pocket watch by determining its compressive strength."

In spite of this disheartening statement, tensile test has been a basic test to understand the primary mechanical behavior of
the material. In most of the design procedures, data from the tensile test is widely employed. A typical specimen generally
used in the tension test is shown in Fig. 2.1. We shall define the following terms to understand the results of the tension test.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 2.1 A typical specimen used in the tension test

The initial distance between two points is called initial gauge length or gauge length.L0 is the initial gauge length and ΔL is
the increase in the gauge length. P is the applied force and A0 is the initial cross section area of the gauge length portion and
A is the area at any instant of time when length is L. We shall define the following stress and strain.

a. Nominal stress or engineering stress: This is ratio of force applied on the specimen divided by initial area of the specimen
[2, 3, 4].

(2.1)

b. Nominal strain or engineering strain: This measures the change in length of the specimen per unit of initial length of the
specimen.

(2.2)

One can note that this is not the one dimensional equivalent of Green Lagrange strain tensor although it is a measure of
strain with respect to initial configuration. In the case of small deformation this strain is approximated with Green
Lagrange strain tensor.

c. True stress or Cauchy stress: This is the ratio of force per unit of current area of the specimen.

(2.3)

This is one dimensional equivalent to Cauchy stress tensor.

d. Logarithmic strain or true strain: Let us consider the specimen at current time. Let dl be the infinitesimal change in length
when length was l, then infinitesimal strain introduced is

(2.4)

Total strain accumulated during stretching of the specimen from length L to l is called logarithmic strain defined as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.5)

Engineering strain and engineering stress measure the deformation of a body with respect to original configuration whereas
Cauchy stress and logarithmic strain deal with current deformed state.

The graphical relation between one dimensional stress and strain is called stress–strain diagram. A typical stress–strain
diagram for the mild steel under uniaxial case is shown in Fig. 2.2. The stress and strain used are engineering stress and
engineering strain. Following observation can be made from the curve.

Figure 2.2 Typical stress strain curve for mild steel

The relationship between σe and εe is linear up to point A. The point "A" is called proportionality limit beyond which stress-
strain relations are not linear. The ratio of the stress to strain is called Young's modulus. During this stage of loading, ratio of

lateral strain to the longitudinal strain remains constant and this ratio is called Poisson's ratio, i.e., . From point

A to B, loading–unloading curve is same but stress–strain curve is non linear.

If we load beyond B, there is a drop in stress at almost constant strain and stress–strain curve drops to B′. Then stress
remains constant with increase in strain from point B′ up to point C. If we unload the specimen after loading beyond B′, the
unloading path is different than the loading curve. The point B is called upper yield point and B′ is called lower yield point. The
behavior during B′–C is called perfectly plastic behavior. Generally, lower yield point is considered as the elastic limit. If we
unload the specimen after loading beyond lower yield point, the unloading curve is parallel to the initial elastic curve. A
permanent deformation remains on the specimen even if we completely remove the load. This permanent deformation is
called plastic deformation. If we reload the specimen it follows the unloading curve as shown in Fig. 2.2. Hence we can
observe that the unloading is elastic and elastic properties are not affected by plastic deformation. If we load the specimen
beyond C, stress increases with strain and reaches a maximum value at D and stress starts decreasing after D and continues
till it breaks at E. The stress level at D is called ultimate stress and E is called breaking point.

The phenomenon of increase in stress level during C–D is called strain hardening. The decrease in yield stress is called strain
softening which is shown in Fig. 2.2 by curve segment D–E.

Sress-strain diagram is assumed to be independent of the size of the specimen and its gauge length. Stress–strain diagram

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
for different materials differ widely. Even for the same material the stress–strain curves depend upon rate of loading, ambient
temperature and many other variables during loading. For the material like aluminum, yield stress discussed above is not
clearly obtained. For such materials, we define a stress called proof stress which limits the elastic domain. Proof stress is the
offset stress corresponding to the 0.2 percent of plastic strain, i.e., if the material is loaded up to proof stress and we unload
along the line parallel to the initial elastic line, 0.2 percent plastic strain is obtained as shown in Fig. 2.3.

Figure 2.3 Proof stress for a material like aluminum

It has been observed that during plastic deformation beyond ultimate load, cross section starts decreasing drastically at a
particular section of the specimen as compared to other section and specimen breaks with the development of necking as
shown in Fig. 2.4. Reduction of cross section area occurs in the localized portion of the specimen and it does not result in the
entire portion of the specimen. The development of necking causes instability to the testing. At the breaking point specimen
fractures. Some materials such as steel and aluminum exhibit a typical "cup and cone" fracture.

Figure 2.4 A typical cup-cone failure

Remarks:

1. As the specimen elongates in length and there is a very slight change in the volume, specimen contracts in diameter. Thus
the actual area on which force is acting is less than the initial area of the specimen. The engineering stress calculated
based on the initial area is always lower than the true stress. But for small deformation, the difference between these two
stresses are negligibly small and the stress–strain diagram will be the same for both the stresses during initial stage of
loading and the true stress curve diverges from the engineering stress curve. Also, it is very difficult to measure the lateral

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 contraction. Hence for all practical purposes, engineering stress is used for engineering design. We can say that there is
very insignificant difference between σe and σtrue up to yield point (see Fig. 2.5).

Figure 2.5 Stress-strain curve for engineering stress and true stress (Not to the scale)

2. The initial straight line in stress–strain curve is not followed by all the materials. For material like rubber, the stress–strain
curve is nonlinear even at the initial stage but it has been found that the loading–unloading curves are the same as shown
in Fig. 2.6.

Figure 2.6 A loading unloading curve for rubber

3. Hardening: We can see from Fig. 2.2 that the loading beyond C increases the yield stress of the material. If we unload the
specimen and load it, it follows almost the same unloading curve up to the hardening curve and then it follows the original
path. Suppose we load the material up to point F and unload it, we shall get permanent strain. If we load the material again,
it behaves elastic up to point F and then follows exactly the same behavior as virgin curve. This process of increasing the

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 yield stress by giving permanent deformation is called strain hardening. Although stress increases during hardening,
ductility of the material decreases. The increased yield stress is called subsequent yield stress.

Energy during deformation  A stress–strain curve is shown in Fig. 2.7 for linear elastic material. During loading, energy is
supplied to the material which is stored as the energy of deformation. The stored energy is called strain energy. The energy
can be equated with the shaded area in Fig. 2.7.

Figure 2.7 Energy stored during elastic deformation

During elastic deformation, strain energy density is defined for linear behavior as

(2.6)

where Φ is the strain energy in the material per unit volume called strain energy density. Total energy is given by

(2.7)

For general nonlinear case, energy density of deformation is

(2.8)

which is the area under stress–strain curve. Total energy of deformation is

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.9)

Elastic deformation  During complete cycle of loading–unloading in elastic deformation, there is no dissipation of energy.
Even if the material returns to its original shape but there is a dissipation of energy during deformation, then material behavior
is not elastic.

Anelastic deformation  There are materials which return to original configuration but through different path as shown in
Fig. 2.8 and hence we get dissipation of energy. This type of material be-havior is called Anelastic. We can see that there is a
loop formation in the stress–strain curve which shows the dissipated energy during the loading–unloading process.

Figure 2.8 Stress–strain curve for anelastic material

Time independent behavior  If deformation takes place as soon as we apply the load then the material behavior is called
time independent behavior.

Elastic degradation  It has been observed that elastic modulus during unloading decreases during deformation for many
materials. Decrease in elastic property of the material due to deformation is called elastic degradation.

2.3. YIELD CRITERIA

Basic aim in the design of a structure is to provide adequate stiffness and strength under loading and environment. One of the
objectives is to analyze how strong a structure is and how much load can be applied on it before it fails or collapses. Mode of
failure of a material/structure can be classified into two categories

i. Ductile failure

ii. Brittle failure

A failure is said to be ductile if it undergoes perceptible deformation before it fails whereas brittle failure is characterized by
sudden failure without perceptible deformation. Ductile material shows elastic behavior up to certain point and loading beyond
this point, material no more remains in elastic state. We can define yield point as the stress limit beyond which material does
not remain elastic on loading. In other word, yield point for uniaxial loading is the stress limit beyond which path independent

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
characteristics is no more valid. Under complex loading above criteria is termed yield criteria which provide the limit of elastic
behavior.

Yield criterion is characteristics of ductile materials. After yielding, ductile material does not loose its strength. But brittle
material fails suddenly. For granular or brittle materials, we refer failure criteria as failure strength because these materials
cannot take load beyond the above limit whereas ductile material can take load even beyond the yield stress.

The failure analysis of a structure up to collapse is necessary to assess all safety aspect of the structure and find its
deformation characteristics. This type of analysis is particularly desirable for certain structures such as reactor vessel, nuclear
containment structures, parts of offshore structures, aircrafts structures, etc. Experimental studies on these structural
systems are very expensive and empirical studies are not sufficient for the adequate evaluation of safety with respect to limit
state. At least number of experiments can be reduced if we successfully predict the failure to certain level of accuracy. Failure
criterion is one of the basic requirement for analysis of structure such as non linear finite element analysis.

In the present chapter, overview of failure criteria/yield criteria has been presented. In a general framework, yield criteria for
ideal plastic isotropic material have been discussed and its geometric representation has been explained. Some commonly
used yield criteria for isotropic, homogeneous material have been described first and their limitations have been presented.
Brief review of failure criteria for anisotropic materials, high performance composite materials and granular materials like
concrete have been presented.

2.4. PLASTICITY AND YIELD CRITERIA

2.4.1. Ideal Plastic Material
A material is said to be ideal plastic [5] if it has following properties.

a. The theory developed is valid only at temperature for which recovery, creep and thermal phenomena are neglected. Ideal
plastic material should obey the above properties.

b. Non-uniformity on microscopic level is neglected, i.e., the Bauschinger effect and hysteresis loop in loading and reloading
are neglected. Material is assumed to be isotropic.

c. Size effects are neglected.

d. Yield strength in compression and tension are assumed to be the same.

e. Yielding is independent of hydrostatic stress.

f. Time plays no role during loading and reloading.

We shall first provide the mathematical and geometric aspect of yield criteria of isotropic materials.

2.4.2. Yield Criteria of Ideal Plastic Material

i. Existence of the yield criteria  The yield criterion limits the elastic region. It is a mathematical expression to define the
combination of components of stress such that when it reaches material no more behaves elastically. Yield criterion gives
the onset of plastic deformation. In other words, if a state of stress satisfies yield criterion, we can say that the
plastification may start. It is assumed that initial yielding is dependent only on the state of stress and not on how the stress
is reached. We can assume that there exists a function f (σij) called yield function such that

Material is elastic

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (2.10)

or

(2.11)

where f (σij) = 0 defines the yield surface in stress space and indicates unloading. The latter combination
tells us that onset of plastification has taken place, but unloading is going to take place elastically. As the yield criterion
does not depend on the path of loading, it does not tell anything about deformation. If state of stress is already satisfied f
(σij) = 0, it tells us only that plastification has just started or taken place. But it does not say whether plastic deformation
has taken place or not. But the yield function gives the information about loading.

Material behavior is plastic

(2.12)

Mathematically yield function is the closure of open set formed by elastic region. Also, one can observe that yield function
is non-positive function. In general, function f maps nine dimensional stress space to non-positive real number f : ℜ9 → ℜ–
where ℜ– is the set of non-positive real number. For non polar media, stress is symmetric. Yield function is function of six
components of stress, i.e., f : ℜ9 → ℜ– which can be written in expanded form as

(2.13)

State of stress can be determined by specifying principal stress and orientation of principal axes. The three principal stress
and three angle constitute the six dimensional space. We can write yield function in terms of principal space. Here
functional form of yield function will change, but we keep the same function f. The yield function can be written as

(2.14)

where σ1, σ2, σ3, are principal stresses and α1, α2, α3, are orientations of principal axes.

ii. Material is isotropic  For isotropic material, yield function must be independent of orientation of principal axes, i.e., yield
function should be independent of α1, α2, α3. Yield function f is function of σ1, σ2, σ3 only which can be written as

(2.15)

Also, for isotropic material, if we interchange the direction of σ1, σ2 and σ3, yield function does not change. Function f
should be independent of sequence of principal stress, i.e., yield function is symmetric with respect to principal stress.
Thus, we can write

(2.16)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 We can represent the yield function by taking any three independent symmetric function of principal stress. We know that
stress invariants are symmetric function of principal stresses. The above function can be written also in terms of stress
tensor invariants as

(2.17)

where I 1, I 2, I 3 are symmetric function of principal stress. The yield function can also be written as in terms ofρ, ξ, θ or σoct,
τ oct, θ as

(2.18)

These combinations are used in developing yield function for geomaterials such as soil, rock concrete.

iii. Yield function is independent of the hydrostatic stress  We know that the state of stress at a point can be split into
hydrostatic stress/volumetric stress and deviatoric stress. It has been observed (Bridgman) that yield function does not
depend upon hydrostatic stress for metals and alloys. In these cases, we can write yield function as

(2.19)

where s ij is the components of deviatoric stress tensor. The function f depends only on deviatoric stresses, we can write
for isotropic materials

(2.20)

where s 1, s 2, s 3 are the principal stresses of deviatoric stress tensor, defined as

(2.21)

where

We can write Eq. (2.20) in terms of stress invariants

(2.22)

where J2, J3 are stress invariants of deviatoric stress tensor. In terms of ρ, ξ, θ and σoct, τ oct, θ, we can write yield function
as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
iv. Identical tensile and compressive yield strength  It has been observed that yield stress in tension and compression is
same for many of the metallic materials. Thus, the function form for tension–compression independent yield function is
even function. Physically, the assumption requires that yield criteria should be independent of direction of loading. It
should satisfy following condition

(2.23)

For isotropic material, we have expressed yield function in terms of stress invariant. If we take material with hydrostatic mean
independence and tension–compression independent, yield function will be even function of J3. J2 is always positive number.

(a) Geometric representation

There exists three principal stresses and their orientation for given state of stress at a point. Hence one can represent state of
stress at a point in three dimensional vector space whose bases are principal stresses for isotropic material where orientation
of axes is not important. For isotropic materials, f (σ1, σ2, σ3) = 0 represent a surface in the principal stress space. A yield
function represents a surface in three dimensional vector space.

If material is acted upon by only deviatoric stress, then yield surface can be represented as

(2.24)

subjected to

(2.25)

Here, σ1 + σ2 + σ3 = 0 is a plane passing through origin and equally inclined to the principal axes. This plane is calledπ plane.
The above equation represent a curve C on σ1 + σ2 + σ3 = 0 plane (see Fig. 2.9). If the yield function is independent of
hydrostatic stress then yield function is right cylinder with generator perpendicular to π plane and whose cross section from
hydrostatic line is same as curve C.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 2.9 Intersection of yield surface with π plane

In order to discuss possible form of curve, let us consider state of stress where hydrostatic components is zero. The locus on
π plane may be convex or concave to origin but obviously not such that radius cut it twice. Due to isotropy, we havef (σ1, σ2,
σ3) = f (σ1, σ3, σ2). Hence, we can say that curve is symmetric about AA′ axis. Similarly it is symmetric about BB′ and CC′ axes.
Hence the curve can be split of into six mutual symmetric parts.

If yield stress in tension and compression is same then locus is not only symmetric about the principal axes diameter but also
about the three diameters orthogonal to them. Hence we can say that yield curve is made of 12 symmetric 30° segment on π
plane.

There are other ways of representing yield function or stress in terms of (ρ, ξ, θ) and (σoct, τ oct, θ).

2.4.3. Some Commonly used Yield Criteria

The yield criteria of materials limit the elastic domain during loading whereas failure criteria give the maximum stress that can
be applied. Traditionally, we use the term yield criteria for metals, alloys and failure criteria for geomaterials such as soil
concrete. We shall discuss some of the yield criteria or failure criteria which have been widely used in design of structures and
forming process. Suitable selection of these criteria depends upon the experimental behavior of the materials. There are a
number of yield criteria proposed for different materials. Most of them are only of historical importance. Those which are
widely used in the design are also important from the point of view of development of new criteria. Some of the commonly
used yield criteria/failure criteria have been described which are as follows

a. von Mises yield criteria,

b. Tresca yield criteria,

c. Mohr Coulomb yield criteria,

d. Drucker–Prager yield criteria.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
2.4.4. (a) von Mises yield criteria
von Mises (1913) suggested that yielding will occur when second invariants of deviatoric stress tensor,J2 reaches a critical
value. He does not take J3 into account in the yield criteria. We can write at onset of yielding

(2.26)

where s 1, s 2, s 3 are principal deviatoric stress. We can also write von Mises criteria in terms of principal stresses as

(2.27)

In terms of components of stress tensor, von Mises yield criteria can be written as

(2.28)

where k is a parameter which is constant for virgin material and depends on the amount of plastic strain. Geometrically, von
Mises yield criterion represents right circular cylinder whose generator is equally inclined to the principal stress axes as
shown in Fig. 2.10. Physical explanation of von Mises yield criteria was given by Hencky (1924) and Nadai (1937). Hencky
interpretation of von Mises's law is that yielding begins when the elastic energy of distortion reaches a critical value. Nadai
proposed that yielding begins when the shear stress acting on the octahedral plane reaches a critical value. This criterion
sometimes also refers to as Hencky–Mises or the Huber–Mises criteria.

Figure 2.10 von Mises yield criteria

Let us define effective stress σeff corresponding to stress tensor σ as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
where s ij is the components of deviatoric stress tensor s. von Mises criteria can be written as

where σy is the yield stress of material in uniaxial tension/compression.

2.4.5. (b) Tresca yield criterion

According to the Tresca yield criterion (1864), yielding of the material will begin to occur when maximum shearing stress at a
point reaches a critical value. If σ1, σ2, σ3 are principal stresses arranged in descending order σ1 ≥ σ2 ≥ σ3, we can write Tresca
criterion as

(2.29)

where k T is material dependent parameter determined experimentally. If σy be the yield stress, the maximum shear is .

Tresca condition can be written as

(2.30)

or in terms of ρ and θ

(2.31)

The maximum shear stress at a point does not change when the state of stress at the point is changed hydrostatically. Tresca
yield criterion represents a hexagonal cylinder in principal stress space.

This cylinder is inside the von Mises right circular cylinder. One can see that Tresca yield surfaces have not unique normal at
the corner (see Fig. 2.11) which poses problem in finite element formulation. But the expression is very simple. Due to simple
expression of Tresca, it is preferred to von Mises for analytical solution although von mises criterion has been found
experimentally closed for metal. But von Mises criteria has been used comprehensively for finite element implementation.
This criteria is sometimes called Coulomb–Tresca or the Saint Venant–Tresca criteria.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 2.11 Locus of Tresca and von Mises yield criteria on deviatoric plane

Remarks : Both von Mises and Tresca criteria does not depend upon the hydrostatic mean stress. They are called pressure
insensitive yield criteria. They model consequences of Bridgman experiments correctly.

2.5. BASICS FOR CONSTITUTIVE RELATIONS

Analysis of structure beyond yielding plays an important role in understanding the behavior of failure of the structure. It helps
us to decide the factor of safety in design. Understanding the behavior beyond yield point of material is also needed in
calculation of load to obtain desired shape in forming process. The relationship between stress and strain is not unique
beyond yield point. Given a state of stress, there exists a unique state of strain for elastic material. Unique stress–strain
relation does not exist if we load a material beyond yield point. If material has loading beyond yield point A and
unloading/reloading, it is not possible to get unique stress–strain relations. Suppose if the stress level is σB, any strain along
line BC can exists (see Fig. 2.12). This non-unique nature of stress–strain behavior poses problem in constitutive modeling.
The strain at a point not only depends on the level of stress, but also on the history of loading. Thus, the load history plays an
important role in describing the behavior of material beyond yield point. But we can talk about incremental relations for these
materials. It is convenient to formulate governing equations of plastic deformation in rate form.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 2.12 A typical uniaxial stress strain relation

In subsequent sections, we shall discuss the basic equations of plasticity. We shall describe in general yield criteria, flow rule
and hardening rule. We shall also talk about how to integrate these equations for given strain history. The treatment is general
and brief. Our presentation is more toward development of basic equations and its computational aspects which can be
applied in non-linear finite element analysis.

2.6. CONSTITUTIVE RELATIONS FOR ELASTO-PLASTIC

MATERIALS
As we have discussed, the plastic behavior of material can be described in the rate form. For time independent deformation,
these equations are homogeneous in derivatives with respect to time. Time is a pseudo parameter. Time plays a role in
viscous material models such as viscoplasticity. During loading beyond yielding, the strain has two parts (a) recoverable parts
(b) permanent deformation. These concepts will be utilized in the formulation of the theory. To describe the elasto-plastic
behavior of material, we shall discuss the following points:

a. Decomposition of strain into its elastic and plastic part,

b. Elastic stress–strain relation,

c. Initial yield criteria,

d. Evolution of plastic strain (flow rule),

e. Change in yield function or loading function during loading–unloading (hardening rule),

f. Loading/unloading criteria.

These six topics form the backbone of plasticity theory. Our discussions are primarily restricted the infinitesimal deformation.
Out of these six topics, following three are dependent on the nature of particular materials.

Yield criterion

Flow rule and

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Hardening Rule.

Yield criterion, flow rule, and hardening rules are material dependent. Various yield criteria, flow rule and hardening rules have
been proposed for different material based on their observed experimental behavior. Our discussions here will be to develop
the basic ingredients of the theory of plasticity.

2.6.1. Elastic and Plastic Strain

If we load a material beyond yield point and unload it, material does not regain its shape. But it has permanent deformation
(see Fig. 2.13). We can see that at any level of loading, a portion of strain is recoverable whereas other part is permanent in
nature. This separable nature of strain into recoverable and permanent part in tension test has lead to the assumption of
decomposition of infinitesimal strain tensor into its elastic and plastic parts. If the strain is small, we can assume that the total
strain is sum of elastic strain and plastic strain

(2.32)

where εije is elastic part of strain and εijp is plastic part of strain. This intuitive separation of elastic and plastic strain based on
one dimensional theory is valid only for small strain. If deformation is large, Eq. (2.32) does not hold good.

Figure 2.13 Decomposition of strain into its elastic and plastic parts

2.6.2. Elastic Stress–Strain Relations

We can observe from Fig. 2.13 that state of stress can be uniquely determined if the elastic strain is known at the load level.
This observation is also assumed to hold in the case of general state of stress. This is based on the assumption that the
elastic properties of material does not change during plastic deformation, i.e., slope of loading/unloading curve is parallel to
the initial loading curve. The elastic parameters do not change. We can write the stress–strain relations as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.33)

where Cijkl is component of elastic stiffness tensor C of virgin materials. This assumption holds fairly well at moderate strain.
Elastic degradation may takes place in the case of large deformation in metal and even at small deformation for brittle
materials. An interested reader may consult book on damage mechanics [6, 7, 8]. We shall assume that Cijkl does not depend
upon the straining. For metal and alloy, it has been observed that the plastic deformation induces the change in only deviatoric
part strain. Considering this, it is convenient to write stress–strain relation in its deviatoric and volumetric components as

where and s is deviatoric part of stress tensor and e is deviatoric part of strain tensor. The elastic constantsK and G

are called bulk modulus and shear modulus and they are independent of strain in this case. We can write the constitutive
relations in terms of K and G as

2.6.3. Initial Yield Criteria and Subsequent Yielding

The yield function defines the domain of elastic deformation. The yield function is non-positive function. A material is said to
have onset of plastic deformation if

(2.34)

where, f is a non-positive function called yield function and αi, β ij are hardening parameters. The hardening parameters are
related to inelastic deformation of the material. For a given material, satisfaction of Eq. (2.34) represents the initial yield
surface. With inelastic deformation, hardening parameter changes. Thus the yield surface changes. These yield surfaces are
called subsequent yield surfaces.

In generalized isotropic hardening, yield surface changes its size whereas yield surface translates for kinematic hardening.
Both isotropic as well as kinematic hardening can take place together called mixed hardening. Hardening rules are discussed
in detail in Section 2.6.5. We shall illustrate initial and subsequent yield surfaces with the help of von Mises yield criterion in
the following paragraphs.

von-Mises yield criterion can be written as

(2.35)

where effective stress is defined as

(2.36)

Initial yield function is given

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.37-a)

Subsequent yield function considering isotropic hardening can be written as

(2.37-b)

where, is the effective plastic strain which is equal to uniaxial plastic strain in the case of

incompressible deformation. Effective plastic strain has been taken as isotropic hardening parameter.

2.6.4. Flow Rule

Deformation can be split up into its elastic and plastic parts during infinitesimal elastoplastic deformation. We have already
discussed that total strain is sum of elastic strain plus plastic strain. This statement is valid only when strain is infinitesimally
small. For finite strain, splitting of elastic and plastic deformation has to be treated based on either deformation gradient
tensor or deformation rate tensor. We shall restrict ourself to infinitesimal strain only. Elastic part of strain is governed by Eq.
(2.33). Flow rule defines the evolution of plastic strain. It relates the evolution of plastic strain rate with current state of
loading or deformation. Plastic strain rate is proportional to the gradient of a function of stress given as

(2.38)

where, Q (σij) is function of stress called as plastic potential. The name plastic potential has been taken similar to elastic
potential which defines the stress for hyper-elastic materials

If yield function is taken as plastic potential Q ≡ f, the flow rule is known as associated flow rule. Otherwise, it is called as non-
associated flow rule. Associated flow rule has been employed for metals and alloys whereas non-associated flow rule is used
for geomaterial. Geomaterial such as concrete and soil shows dilatancy which is taken care by non-associated flow rule. For
associated flow rule

(2.39)

The associated flow rule is also called normality rule or normality criteria because gradient of a function is normal to the
tangent plane of the surface at a point on the surface. The plastic straining takes place in the direction of normal to the yield
surface in the case of associated flow rule. Please note that plastification takes place only when stress is on the yield surface,
i.e.,

Equation (2.38) (Flow rule) should satisfy together with consistency condition f (σij – β ij, αi) = 0, but reverse is not true, i.e., if f
(σij – β ij, αi) = 0, may not exist.

We shall discuss the particular case of the flow rule for metals and alloys considering von Mises yield criterion.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Using von Mises yield criterion

we can write flow rule

(2.40)

Using f (σij, εp) = 0, we can substitute in the above equation for flow rule as

(2.41)

We can see that plastic strain rate is proportional to the deviatoric stress. This flow rule was used by Prandtl (1927) and Reuss
to derive the stress–strain relation for elastoplastic materials.

(2.42)

which can be written as

(2.43)

Using consistency of yield criterion, above equation reduces to

(2.44)

Substituting from above equation in Eq. (2.41), we can write flow rule as

(2.45)

substituting R = σy, we get

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.46)

We can note that in this case. Plastic deformation is isochoric and volumetric strain is purely elastic. This fact
facilitates to write stress–elastic strain relation in volumetric and deviatoric part separately.

The above equation can be written explicitly in the incremental form as

(2.47)

(2.48)

(2.49)

(2.50)

(2.51)

(2.52)

In the analysis of metal, elastic strain can be neglected as compared to plastic strain in many of the engineering analysis.
Thus and we can write flow rule as

(2.53)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
This assumption is employed by Levy and Mise in their analysis. The corresponding plasticity constitutive equation thus
derived is called Levy Mises equations.

2.6.5. Hardening Rule

Hardening rule tells us how the evolution of yield surfaces (which is also called loading surfaces) take place. The yield
surfaces may expand or translate even distort during elasto-plastic deformation. In other words, hardening rules say how
elastic domain changes during the deformation. For elastic-perfectly plastic material model, the domain of elastic deformation
remains the same which gives rise to the fixity of yield surface. Depending upon the material under consideration, hardening
rules are developed based on the observed behaviors of the material. Thus the experimental observation forms the basic
ingredients to develop the hardening rules. There have been many theories to mathematically model the evolution of loading
surfaces. The popular among them are

a. Isotropic hardening

b. Kinematic hardening

These are the simplest hardening rules to describe the most commonly observed behavior related to change in elastic
domain.

a. Isotropic hardening  The expansion of loading surface is modeled by appropriately choosing hardening parameter (see
Fig. 2.14). We shall consider only one isotropic hardening parameter although there has been multiple hardening parameter
model for materials such as concrete. To model uniform expansion, yield function can be written by separating function of
stress from the function of hardening parameters as

(2.54)

where F (σij) is only function of components of stress and R (α) is function of hardening parameter alpha. We can observe
that with the increase of the value of R (α), the yield surface expands. This can be understood by taking the example of the
equation x2 + y2 + z2 = a2. With the increase with a2, the sphere expands. With the above discussion, it is clear that
modeling of uniform expansion of yield surfaces reduces to two sub problems (a) selection of appropriate hardening
parameter α and (b) selection of hardening function R (α). Isotropic hardening parameter αi can be either effective plastic
strain or plastic work in the yield surface equation f (σij, αi). If we choose effective plastic strain as hardening parameter, we
can write

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 2.14 Isotropic hardening

Substituting in above equation, we can write

If we choose plastic work as hardening parameter, we have

Substituting from flow rule in above equation, we can write

Many investigators have suggested different functions for hardening function R (α). The most popular among these
hardening functions are power hardening law and exponential hardening law. In the power hardening law, R (α) is
expressed by algebraic function as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (2.55)

where K and n are material constants. For exponential hardening, R (α) is defined as exponential function of α as

(2.56)

where R∞ and K are material constants.

Please note that in power hardening, yield surface keeps on expanding with the increase with α whereas in exponential
hardening it increases exponentially and saturates at R∞. Many of the materials show saturation during hardening. The
material parameters K, n, R∞, in the hardening expressions are obtained from uniaxial tests, primarily by curve fitting.

It can also be noted that hardening parameter , but it is obtained by integrating where α

is defined as .

b. Kinematic hardening  Kinematic hardening models the movement of the yield surface in the stress space. It is assumed
that the yield surface translates as a rigid body without changing its shape and size. In the kinematic hardening, the center
of the yield surface is chosen as hardening parameter. It is assumed that the center of the yield surface changes with
plastic deformation in the stress space (see Fig. 2.15). The center of the yield surface is called back stress. Kinematic
hardening models the Bauschinger effect. Thus considering only kinematic hardening, we can write yield surface

(2.57)

where β ij are components of the back stress tensor. The back stress tensor β is second order tensor. The hardening
parameter β ij is related to the plastic deformation. Thus our objective in the kinematic hardening modeling is to obtain the
appropriate expression for the evolution of components of back stress β ij. There have been various models for the
evolution of back stress tensor. Some of them are

Figure 2.15 Kinematic hardening

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
a. Prager's model: Prager proposed that the rate of back stress tensor is proportional to rate of plastic strain tensor.

where c is the material parameter. Thus according to the Prager's kinematic hardening model, yield surface translates in
the direction of plastic strain rate or along the normal to the yield surface in the case of associated flow rule. Substituting
plastic strain rate in the above equation, we can write

Ziegler pointed out the deficiency in the hardening law proposed by Prager. According to Ziegler, Prager's kinematic
hardening does not give consistent result in three dimensional, two dimensional and one dimensional cases. It can be
proved that it introduces hardening in one direction and softening in other two directions in uniaxial case which has not
been observed experimentally [9, 10].

b. Ziegler's model:  Ziegler proposed another kinematic hardening rule which states that the movement of yield surface
takes place in the direction determined by the vector α – β. Mathematically it can be stated that

where is the proportionality constant determined by particular yield criteria.

Material behaves very complex manner under cyclic loading. There has been developed many approaches to model
behavior of material under cyclic loading. An interested reader can consult book on plasticity [11, 12, 13].

2.6.6. Loading–Unloading Criteria

Loading-unloading criteria can be written as

a. Elastic loading

b. Elasto-plastic loading

c. Neutral loading

d. Elastic unloading

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Loading-unloading criteria can be combined as Kuhn Tucker optimality criteria as

and consistency condition

2.7. CONSTITUTIVE EQUATIONS FOR ELASTO-PLASTIC

MATERIALS
Constitutive relations for elastoplastic deformation can be given as

(2.58-a)

(2.58-b)

(2.58-c)

(2.58-d)

(2.58-e)

(2.58-f)

Equations (2.58-a), (2.58-b) and (2.58-d) are algebraic equation and Eqs (2.58-c), (2.58-e) and (2.58-f) are differential
equation which is homogeneous in time. For a given load history, these equations are integrated to obtain the variables
. We can combine all the equations. One can write the relations between stress and strain rate as

follows.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Differentiating Eq. (2.58-d), we obtain

(2.59)

Substituting and from Eq. (2.58-c) and (2.58-f) in the above equation, we get

(2.60)

Differentiating Eq. (2.58-b) and obtaining with Eq. (2.58-a), we get

(2.61)

Substituting from Eq. (2.58-c) in the above equation, we can write

(2.62)

Subsituting from above equation in Eq. (2.60), we get

(2.63)

Solving for , we can write

(2.64)

Subsituting , from above equation in Eq. (2.62), we can write

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.65)

which can be simplified as

(2.66)

which can be written as

(2.67)

where is the component of elasto-plastic continuum tangent modulus which is given as

(2.68)

which can be written in the vector notation as

(2.69)

Remarks: The above derived modulus is called continuum tangent modulus because it has been derived using continuum
equation (in time) of elasto-plasticity. There is another modulus which is called consistent tangent modulus which has been
derived from discretized equations (in time) of plasticity [14,15]. Equation (2.67) is equivalent to Eqs (2.58-a)–(2.58-f) for
continuous time. Integrated equations represent different consistency discussed in the next section.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
2.8. INTEGRATION OF CONSTITUTIVE RELATIONS
The boundary value problems in plasticity are highly nonlinear. They are generally solved by numerical methods. In the
numerical methods, one of the objectives is to integrate the constitutive equations discussed earlier. In this section, we shall
discuss the integration of the constitutive relations between time tn and tn + t. In the rate independent plasticity, time t is a
pseudo-parameter which can be taken as loading parameter. Strain driven algorithm has been popularly used in the literature.
We shall discuss the strain driven integration algorithms which has been widely employed in nonlinear finite element analysis.

2.8.1. Definition of the Problem

We shall assume that all the quantities (strain field εn, plastic strain field εnp, elastic strain field εne, back stress β n, isotropic
hardening parameter αn and stress field σn) is known at time tn, i.e., {εn, εne, εnp, β n, αn, σn} are known at tn and strain field εn + 1
= εn + Δεn is known at time tn + 1. Our objective is to obtain the field at time tn + 1 taking into account constitutive relations
given by Eqs (2.58-a)–(2.58-f). Except for few cases, these equations are so complex that they are integrated numerically. The
problem of integration of constitutive relations can be thought at an initial value problem. But some equations are algebraic
whereas others are differential in time. This poses the problem in the solution. Many investigators have carried out work to
obtain the accurate and stable solution [15, 16, 17].

We shall briefly discuss some of the issues here

2.8.2. Integration Algorithm

Numerical integration of constitutive relations can be grouped as

a. Explicit integration scheme

b. Implicit integration scheme

Explicit integration scheme does not require the solution of nonlinear algebraic equations, but obtained stress field may drift
from the yield surface. We need to apply correction for drift from yield surface or sub-incrementation to increase the accuracy
and stability of solution. The method is conditionally stable. Implicit algorithms are unconditionally stable but requires
solution of nonlinear equations. These resulting nonlinear equations are solved by Newton–Raphson method.

A large number of algorithms can be described by so called generalized mid point method or generalized trapezoidal method.

Equations (2.58-c), (2.58-e), and (2.58-f) are integrated to give

(2.70-a)

(2.70-b)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(2.70-c)

where can be calculated from (a) Generalized Trapezoidal rule as

(2.71)

or from (b) Generalized Mid-point rule as

(2.72)

We need to satisfy the algebraic part of constitutive Eqs (2.58-a)–(2.58-f). These equations are satisfied at time tn + 1 as

(2.73-a)

(2.73-b)

(2.73-c)

Thus we have obtained algebraic equations in field variables at time tn + 1. The strain field at time tn + 1, (εij) n + 1 is known. The
unknowns in Eqs (2.70-a) –(2.70-c) and Eqs (2.73-a)–(2.73-c) are (εp) ij, (εe) ij, (σij) n + 1, (β ij) n + 1, λn + 1 and αn + 1. We also have
four tensor equations and two scalar equations. The number of equations are equal to number of unknown. These equations
are solved by Newton–Raphson method.

2.8.3. Remarks on Integration

1. Equation (2.73-c) is called consistency condition. In the equation, we have imposed the consistency at time tn + 1. It has
been reported that we need to impose consistency at the time tn + p to obtain the B-stability, i.e., f ((σij) n + p – (β ij) n + p, αn +p)
[18].

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 2. The algorithms presented above is explicit for p = 0 otherwise it is implicit. The algorithm has second order accuracy at p =
0.5.

3. It has been pointed that the tangent modulus should be obtained by linearization of integrated Eqs (2.70-a)–(2.73-c) to
preserve the quadratic convergence of Newton–Raphson method used to solve discretized boundary value problem in
FEM. Tangent modulus thus obtained is called consistent tangent modulus. The efficient calculation of tangent modulus
and accurate as well as stable integration algorithm has been active area of research for past two decades and many
issues are still to be solved.

4. The integration of Eq. (2.67) is not the same as the integration of Eqs (2.58-a) and (2.58-f) numerically to obtain stress
field σn + 1. The integrated Eq. (2.67) satisfy the consistency approximately whereas we impose the exact consistency in
the later case in terms of approximately calculated field quantities.

2.9. CONCLUSION
Plasticity plays pivot role in failure analysis of structures and obtaining required load in forming process. We have discussed
some of the points of plastic modeling of material. Flow rule, Hardening rule and summary of constitutive relations have been
presented. Boundary value problems in plasticity are highly nonlinear. Integration of constitutive relations plays an important
role in the solution of boundary value problem. We have also presented numerical methods to integrate constitutive relations
which are at central to the implementation of plasticity in numerical methods such as finite element method.

2.10. REFERENCES
1. Orowan, E., (1944), Proceedings of Institutions of Mechanical Engineers, 151, pp.133.

2. Jog, C.S., (2001), "Foundations and applications of mechanics", Volume-I, Continuum Mechanics, Narosa Publishing
House, New Delhi.

3. Gurtin, M.E., (2005), An introduction to continuum mechanics, Academic Press, Indian print.

4. Chatterjee, R., (1999), Mathematical theory of continuum mechanics, Narosa Publishing House, New Delhi.

5. Hill R. (1998), The mathematical theory of plasticity, Oxford University Press.

6. Lemaitre, J., (1996), A course on damage mechanics, Springer.

7. Krajcinovic, D., (1996), "Damage mechanics", North Holland series in Applied math and mechanics.

8. Lemaitre, J. and Desmorat, R., (2005), Engineering damage mechanics: ductile, creep, fatigue and brittle failure, Springer.

9. Ziegler, H. (1959), A modification to Prager's hardening rule, Quarterly of applied mathematics, 17, pp. 55–59.

10. Shield, R.T. and Ziegler, H., (1958). On Prager's hardening rule, Zeitschrift für Angewandte Mathematik und Physik (ZAMP),
99, pp. 260–276.

11. Khan, A.S. and Huang, S., (1995), Continuum theory of plasticity, John Wiley and Sons.

12. Lubarda, V.A. (2002), Elastoplasticity theory, CRC Press.

13. Jirasek, M. and Bazant, Z.P. (2002), Inelastic analysis of structures, John Wiley and Sons.

14. Simo, J.C. and Taylor R.L. (1985), "Consistent tangent operators for rate independent elasto-plasticity", Com. Meth. Appl.
Mech. Engineering, 48, pp. 101–118.

15. Simo, J. and Hughes, T.J.R., (1998), Computational inelasticity, Springer.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
16. Dunne, F. and Petrinic, N. (2005). Introduction to computational plasticity, Oxford University Press.

17. D.R.J. Owen and Hinton, E. (1980). Finite element in plasticity: theory and practice, Pineridge Press, Swansea.

18. Simo, J. C. and Govindjee, S. (1991). Nonlinear B-stability and symmetry preserving return mapping algorithms for
plasticity and viscoplasticity, International Journal Numerical Methods of Engineering, 31, pp. 151–176.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
3. Modelling of Metal Forming Process
Uday S. Dixit, R. Ganesh Narayanan

Department of Mechanical Engineering Indian Institute of Technology Guwahati, Assam 781 039, India

3.1. INTRODUCTION
A comprehensive description about various metal forming processes is given in Chapter 1. There are various material and
process parameters that affect the forming behavior in a synergistic fashion. It is advantageous to perform experiments in
order to understand the metal forming processes. In order to reduce the cost associated with conducting experiments and
dependence on resources, theoretical models were developed for predicting the metal forming behavior and process
performance. Modeling methods like slab method, upper and lower bound methods, slip line field theory, finite element
method, and recently soft computing methods like neural networks, fuzzy sets, genetic algorithm are used in an integrated
manner to predict the forming behavior under varied working conditions. The main objective of the modeling exercise is to
predict the metal forming behavior at the design stage itself so that only limited trial experiments can be planned. Later on to
improve the accuracy of prediction, the modeling methods are fine tuned for day-to-day life usage.

The application of various modeling methods can be understood from the published literature. For example, Slab method has
been used for analyzing the sheet and foil rolling process by Salimi and Kadkhodaei [1], Salimi and Sassani [2], Kumar and
Dixit [3], etc. This method has been used by Chitkara and Aleem [4] for tube extrusion process analysis. Similarly upper bound
method is used to-analyze the sheet hydroforming process [5], evaluate friction in bulk metal forming [6], analyze axi-
symmetric extrusion process assisted by cyclic torsion [7], predict the influence of die angle on the relative extrusion pressure
during tube extrusion process [8] and many other works. Many times the modeling methods are used in an integrated fashion
for forming process analyses. For instance, slab method and finite element simulation together are used to investigate the
ring forming operation by Tzou et al. [9]. The effects of frictional coefficient, rotating angular speed, reduction and aspect ratio
on compression force, effective stress, and the effective strain are predicted and discussed in their work. The compression
behavior of double layer clad sheet compression forming have been investigated using the upper bound and slab methods by
Hsu and Tzou [10]. They compared the effect of friction factor ratio on the compression force using both the methods.

The soft computing methods like neural network, fuzzy sets, genetic algorithm are also used to predict the metal forming
behavior. Lorenzo et al. [11] and Sivakrishna and Ganesh Narayanan [12] predicted the fracture in cold forming operations and
forming limit of welded blanks, respectively, using neural networks trained with data from finite element simulations. Hans Raj
et al. [13] predicted the forging load during hot upsetting and hot extrusion operations using back propagation neural
networks. The neural network predictions are compared with simulations results. Chan et al. [14] predicted the forging load
and maximum effective stress during a typical metal forming operation using ANN integrated with finite element methods.
Kazan et al. [15] developed the prediction model of springback in wipe-bending process using ANN approach. Fuzzy sets are
being used for metal forming equipment design and process analysis as described by Dixit et al. [16], Abburi and Dixit [17],
Shashi Kumar [18], Hsiang and Lin [19] and many others in the field of rolling, machining, material deformation studies, etc.
ANN-GA hybrid system is also used for modeling and prediction of various material forming operations like minimizing
springback during sheet forming [20] and reducing flatness values of hot rolled strips [21].

In this chapter, a brief description about the various modeling methods are given with their applications in the field of metal
(or material, in general) forming.

3.2. SLAB ANALYSES

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
This method, also called as force equilibrium approach, is based on equating forces acting at an elemental region in a
deforming billet in one direction. This produces a differential equation that is solved along with the boundary conditions.
Assumptions usually followed while using this method for modeling metal forming operations are as follows:

1. The deformation is considered homogeneous, i.e., a plane section remains plane throughout forming.

2. The principal axes are the same as that of loading axes and friction does not change the principal axes direction for the
purpose of applying yield criterion.

The slab method is applied to analyze various metal forming operations like plane strain compression, axially symmetric
compression, rolling, sheet drawing, etc., and some of them are briefly explained here.

3.2.1. Plane Strain Compression of Solid Billet

Consider the compression of a solid billet under plane strain condition (no deformation in z-direction) shown in Fig. 3.1.
Assuming sliding friction at the interface with a constant friction coefficient of μ, force balance across the elemental region in
x direction gives

(3.1)

or,

(3.2)

Figure 3.1 Slab analysis of plane-strain compression of solid billet

For plane-strain condition, by considering σx and σy (or – p) as principal stresses, and following von Mises yield criterion, we
can obtain

or,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(3.3)

where σ0 is the flow stress of the material in tension or compression. Differentiating Eq. (3.3), we get

(3.4)

Now Eq. (3.2) becomes,

or,

(3.5)

Integrating Eq. (3.5) by taking σx = 0, i.e., p = 2k at x = L provides

or

(3.6)

The maximum value for p can be obtained by putting x = 0 at the centerline. From Eq. (3.6),

(3.7)

The graph between and distance x is shown in Fig. 3.2 and is generally referred to as "friction hill".

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.2 Friction hill (die pressure variation) during plane-strain compression under sliding friction

The average pressure during plane strain compression (pa) is given by

(3.8)

In sticking friction condition, the friction stress on the interface is equal to yield strength of the material in shear, k. In this
case, Eq. (3.2) becomes,

and by following the same procedure as described above for sliding friction condition, Eq. (3.6) becomes,

(3.9)

This shows the linear variation of with distance x from the edge to

the centerline (Fig. 3.3). The maximum value which occurs at the centerline (x = 0) is given by

(3.10)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.3 Friction hill (die pressure variation) during plane-strain compression under sticking
friction

3.2.2. Axially Symmetric Compression of Solid Billet

Similar to the analysis followed for plane strain-compression, assuming sliding friction exists at the interface with a constant
friction coefficient of μ and force balance across the elemental region (Fig. 3.4) in the radial direction gives,

or

(3.11)

(by neglecting higher order terms).

Figure 3.4 Slab analysis of axi-symmetric compression of solid billet

Assuming that as a result of deformation R and r get increased by the same factor, εr = εθ and consequently σr = σθ . At the
onset of yielding, σr + p = σ0 or dσr = – dp.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Substituting this in Eq. (3.11) gives

or,

(3.12)

The solution of Eq. (3.12) can be obtained by taking σr = 0 at the edge of the disk (r = R). Hence with boundary condition p = σ0
at r = R provides the solution

(3.13)

By assuming sticking friction condition at the interface and following the above procedure, we get

(3.14)

Similarly slab analysis is applied for many other forming processes like deep drawing of sheet, strip drawing, extrusion, wire
drawing, etc. [22, 23].

Salimi and Kadkhodaei [1] analyzed the asymmetrical sheet rolling using slab method for predicting the work roll pressure and
mean pressure distributions, mean shear stress distribution, and variation of rolling force per unit width with initial thickness.
They have also predicted the effect of friction coefficient and percentage reduction on the rolling force and torque per unit
width. Some of the predicted results are compared with experiments. Salimi and Sassani [2] used a modified slab analysis
method to predict the strip curvature during asymmetric sheet rolling operation. This modified method describes an
enhancement to the rolling theory where friction becomes an integral part of deformation mechanics in the roll gap. The slab
analyses predictions are compared with experimental results and a few other literature results. The slab method is also used
for analyzing the extrusion of tubes from hollow and solid circular solid billets by Chitkara and Aleem [4]. The extrusion
pressures predicted are agreeing well with the experiments. Kumar and Dixit [3] have used strain hardening law and a realistic
friction model, Wanheim and Bay's friction model, during slab analysis of cold foil rolling process. A comparison between the
roll torque and roll force predicted by Wanheim and Bay's friction model and Coulomb's friction model is shown in Fig. 3.5.
Also the effect of strain hardening on the roll force and roll torque for different percentage reductions are discussed. It is
shown that the model with strain hardening effect not only changes the magnitude of roll pressures, but also the distribution
pattern.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.5 Comparison between (a) roll force, and (b) roll torque predicted by Wanheim–Bay's and
Coulomb's friction model [3]. With permission from Elsevier, Copyright [2005]

3.3. UPPER BOUND ANALYSES

Machine and tool designers are often interested to find the load required for plastic deformation in a metal forming process.
The upper bound analyses predict the load requirement that is either equal to or greater than the actual load required for
causing the plastic deformation. The upper bound theorem is stated as : "The upper bound of the load required for plastic
deformation can be obtained by equating the internal rate of energy dissipation plus the energy spend in overcoming friction
to the rate at which external forces produce work in some particular pattern of deformation." The load calculated in this
manner will be greater than or equal to the actual forming load.

In upper bound analyses,

1. First, an internal flow field satisfying the incompressibility condition is assumed that should account for the required shape
change. The flow field can also be assumed to consist of some rigid blocks that may have the tangential velocity
discontinuities with the neighboring blocks. There is no velocity discontinuity in the normal direction on the surface of the
block.

2. The power consumed internally in the deformation regions can be calculated. If the strain-rate is continuous, the power per
unit volume is given by the scalar product of stress and strain-rate tensor. It is to be noted that actual stress tensor is not
known. The stress tensor is calculated from the assumed strain-rate tensor using the constitutive relation for the material.
At a surface of the velocity discontinuity, the power per unit area is equal to shear yield stress of the material multiplied by
the magnitude of the velocity discontinuity. The total internal power is taken as the sum of the powers in continuous
strain-rate zone and velocity discontinuity zone.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
3. The internal power consumed plus the frictional power is equated to the external power for evaluating the external forces.

The energy dissipated on a plane of discrete shear or velocity discontinuity can be calculated as described elaborately in the
book by Hosford and Caddell [24]. The same is briefly described here. Consider a rigid metal regionABCD moving at an initial
velocity V1 at an angle θ1 and having unit width perpendicular to the paper [Fig. 3.6(a)]. The metal region after crossing the
plane (or line) OO′ becomes A′B′C′D′ with different velocity V2 at an angle θ2. Thus, there is a change in velocity, shape, and
direction of the region ABCD after crossing OO″ plane. The velocity vector diagram shown in Fig. 3.6b depicts that the velocity
discontinuity V12 occurs along the plane OO′. There will not be any velocity discontinuity normal to OO′. Now the rate of energy
dissipation at OO′ will be equal to the work done per volume times the volume per time crossing OO′ This after few
substitutions gives

(3.15)

where k is the shear strength, H is the length of the line crossing AA′ (for unit width).

Figure 3.6 (a) Plane of velocity discontinuity (b) Velocity vector diagram

By considering many planes of discrete shear, we can write Eq. (3.15) as,

(3.16)

where Hi and Vi belongs to that particular plane. In Section 3.3.1, the application of this approach is presented.

3.3.1. Plane Strain Extrusion

Figure 3.7(a) describes the extrusion of billet through the die without considering friction between them. One half of the
deformation (initial half thickness = H0) is considered and the velocity discontinuities occur along PQ and QR. The velocity
vector is constructed as shown in Fig. 3.7b in which the entry and exit velocities, V0 and Ve, are drawn horizontally. As the
material crosses line PQ the velocity is changed to, say V1, and is parallel to PR. The velocity discontinuity along PQ is VPQ
which is drawn parallel to PQ starting at V0. Now the velocity V1 at plastic deformation zone will be the summation of V0 and
VPQ. The material in this region then crosses line QR, wherein it undergoes second velocity discontinuity, VQR, which is
represented by a vector VQR parallel to QR, starting at VPQ. Its magnitude is calculated by representing the exit velocity, Ve,
parallel to V0.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.7 (a) Velocity fields for plane strain, frictionless extrusion, (b) Velocity vector diagram

The work done per unit time (for unit width normal to the paper) is calculated as

(3.17)

where Pe is the extrusion pressure.

This initial energy is completely dissipated along the discontinuitiesPQ, QR and from Eq. (3.16), energy balancing gives,

(3.18)

This equation can be solved graphically and numerically By considering 50% reduction (therefore,H0 = 1; exit half thickness,

He = 0.5), α = 30°, θ = 90° (then β = 30°), we have .

By varying θ and hence β, will vary yielding a minimum value of 0.78

when θ = 72° [24]. Similar values can be obtained from other methods like slip line field theory, slab analyses, lower bound
analyses, etc.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
By taking sticking friction condition at the interface, i.e., along PR, an additional term will be included in Eq. (3.18).

(3.19)

For different values of θ, the minimum value of occurs at θ = 83° [24]. It is also possible to propose more

complex deformation field to obtain lower values of upper bound.

3.3.2. Other Forming Operations

An elaborate discussion about the application of upper bound analyses to other forming operations can be read from any
standard textbook, viz. [24]. The following are the general upper bound solutions of different forming operations:

(3.20)

Here h is the height and L is the width of the billet (Fig. 3.1).

(3.21)

Here m is the friction factor, α is the semi-die angle, and ε = ln(to /tf), where t0 is the initial thickness and tf is the final
thickness.

(3.22)

The upper bound analysis is used to understand and predict the forces involved in various metal forming operations. For
instance, Wang and Rao [25] predicted the effect of friction during bulk forming using the upper bound analysis by considering
velocity fields in single surface asperity. They derived expressions for normal pressures (p/σf) and studied the influence of
some parameters on the friction coefficient, normal pressures, asperity deformation, etc. Wilson [26] established a general
upper bound analysis of axi-symmetric metal forming problems and showed some examples for velocity discontinuities
during axi-symmetric extrusion through square dies, conical dies, parabolic dies, ball indention, backward extrusion, and few
other axi-symmetric problems. Choi et al. [27] analyzed the rotary forging of a cylindrical billet with the help of upper bound
method. The total power dissipated during the operation is divided into three parts namely—power dissipated by internal
deformation, power dissipated by velocity discontinuity, and power dissipated because of friction at the interfaces. In order to
determine the forming force, the total power is minimized for representing the actual velocity distributions. The analysis is

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
validated with experiments on carbon steel (AISI1020 and AISI1045) and Al 6061. The forming force predicted is compared
with results presented in literature and good agreement is witnessed. The influence of spiral feed on the forming force is also
studied. The velocity fields involved in various metal flows through converging dies was described by Avitzur and co-workers
[28]. The metal flow involved in conically converging radial flow, cylindrically converging radial flow, parallel flow, and toroidal
flow are described. The spherical, trapezoidal, triangular velocity fields are considered for their derivations.

3.4. SLIP LINE FIELD THEORY

Slip line field is a method of modeling plane-strain rigid-plastic metal forming problems, although there have been attempts to
apply the method to strain hardening materials and for axi-symmetric problems as well. The method is based on the fact that
through each point in the plane of plastic flow, one can consider a pair of orthogonal curves along which the shear stress has
its maximum value. These curves are called slip lines or shear lines and material particles are assumed to flow along these
lines.

In essence, the following equations are solved in the slip line method:

(3.23)

(3.24)

(3.25)

(3.26)

(3.27)

where σx, σy and τ xy are the stress components, vx and vy are the velocity components and k is the shear yield stress. Figure
3.8 shows a set of orthogonal slip lines. These are designated as α and β slip lines. By convention, the algebraically greatest
principal stress direction bisects the angle between α and β direction taken as a right handed pair of curvilinear axes.
Following equations and theorems are useful in analyzing and constructing the slip lines:

i. Hencky equations:

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (3.28)

(3.29)

where ϕ is the counter clockwise angle of α line with the x-axis and p is the pressure (negative of hydrostatic stress).

ii. Geiringer equation:

(3.30)

(3.31)

iii. Hencky's first theorem: The angle between the tangents to a pair of slip lines of one family at the points of interaction with
a slip line of the other family is constant along their lengths. In other words, if we pass from one slip line to another of
same family, the angle turned through and the change in hydrostatic pressure are the same along each intersecting slip
lines. In Fig. 3.9:

(3.32)

Figure 3.8 A set of α and β slip lines

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.9 Intersection of slip lines

iv. Hencky's second theorem: Let the radii of curvature of the α and β lines be denoted by R and S, respectively, i.e.,

(3.33)

It can be shown that

(3.34)

Alternatively,

(3.35)

(3.36)

Usually, slip line field satisfying the stress boundary conditions is assumed. Then the solution is tested to ensure that the
stress boundary conditions are also satisfied. In general, one may have to use trial and error, although the procedure may be
computerized. Applicability to only plane strain case and the difficulty in the construction of slip line are two main causes that
inhibit the use of slip line field in metal forming.

Just to give some idea, a very simple problem of flat punch indentation using slip line method is illustrated.Figure 3.10 shows
the proposed slip line field for indentation of a semi-infinite medium by a flat punch. Punch is considered perfectly smooth;
hence the shear stress on the punch face is zero. Therefore, the maximum principle stress is along the punch travel. It is clear
that slip lines that are the lines of maximum shear stress must meet the punch face at 45°. Similarly, the slip lines are
meeting at free surface (which is a principal plane) at 45°. In the triangular regions ABD, ACE and EFG the stresses are
constant. These regions are joined by centered fan regions.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.10 Slip line field for the indentation by a flat punch

In the region ABD, one principal stress (along y-direction) is zero. Hence, the other principal stress has to be (– 2k). The Mohr
circle is shown in Fig. 3.11. The hydrostatic which happens to be the mean of the two principal stresses in the plane strain
case is (– k). Now, by Hencky's equation, the mean stress in region ACE is (– k – kπ). A Mohr circle with this value as center is
constructed. It is seen that the greatest (magnitude wise) principal stress is (– 2k – kπ). Hence, the uniform pressure on the
punch is given by

(3.37)

Figure 3.11 Mohr circles for the slip line field of indentation by a flat punch

A hodograph showing the possible velocity field can also be constructed.

3.5. FINITE ELEMENT METHOD

Finite element method (FEM) is a numerical method of solving differential equations. As the behavior of metal forming

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
processes can be expressed in the form of differential equations, FEM can be used to model metal forming processes. There
are two prominent FEM approaches to model metal forming processes—updated Lagrangian approach and Eulerian approach.
In updated Lagrangian approach the domain consists of a set of particles, which may keep changing their position in space.
Lagrangian approach is a control mass approach that considers the initial configuration as the reference configuration. In
updated Lagrangian approach the current configuration is taken as the reference configuration. Thus, the reference
configuration keeps updating with the time. Here, the measure of deformation is an incremental strain tensor and the
constitutive equation is expressed in terms of incremental stress and incremental strain tensor.

Three set of equations are required for modeling a metal forming process by updated Lagrangian FEM: incremental strain–
displacement relations, incremental elastic–plastic stress–strain relations and incremental equations of motion [29]. These
are briefly described below.

i. Incremental strain–displacement relations: These are the set of six scalar equations. In index notations, they are
represented by

(3.38)

where i and j vary from 1 to 3 for a three-dimensional case.

ii. Incremental elastic–plastic stress–strain relations: These are six scalar equations. There are two types of equations: (a)
equations applicable during loading with yielding and (b) equations applicable during loading without yielding and after
unloading. During loading with yielding:

(3.39)

where

(3.40)

where G is the shear modulus, ν is the Poisson's ratio, S ij is the component deviatoroic stress tensor, σy is the flow stress
and H′ is the rate of change of flow stress with respect to equivalent strain.

Following equation is applicable for loading without yielding and after yielding:

(3.41)

where

(3.42)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Here, the superscript 'o' on the stress increment indicates that it is the product of the Jaumann stress rate and the time
incremen. Its relation with Cauchy stress (σij) is as follows:

(3.43)

where

(3.44)

iii. Incremental equations of motion: These are three scalar equations given by

(3.45)

where dai is the component of acceleration vector, dbi is the component of incremental body force (per unit mass), andj in
subscript denotes differentiation with respect to current coordinates xj. Note that Einstein summation convention is
employed here. As j is a dummy index:

(3.46)

The Eulerian formulation is convenient for the analysis of metal forming processes like rolling, drawing, extrusion, etc. In this
formulation, a region fixed in space (called the control volume) is chosen as the domain for the analysis. The governing
equations used in Eulerian formulation are as follows:

i. Strain rate–velocity relations: These are six scalar equations given by

(3.47)

ii. Elastic–plastic stress–strain rate relations: In plastic zone, the following equations are used:

(3.48)

where K is the bulk modulus of elasticity and denotes the deviatoric part of strain-rate tensor. In elastic zone, the
following equations are used

(3.49)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Here, the superscript 'o' denotes that it is the Jaumann stress rate. The Jaumann stress rate is related to the Cauchy stress
rate through spin tensor. Thus

(3.50)

and

(3.51)

where

(3.52)

Note that the time derivative of the Cauchy stress has to be the material time derivative.

iii. Equations of motion: Three scalar equations are

(3.53)

These equations can be modified for rigid-plastic case considering K to be infinite. Thus, we have incompressibility
constraint:

(3.54)

The equations can be solved by Galerkin FEM, in which the weighted residual of differential equation is minimized. In
Galerkin formulation, the weight function is taken of the same form as that of approximating function.

3.6. SOFT COMPUTING METHOD

Soft computing methods are tolerant of imprecision, vagueness and incomplete information. That means that some inference
can be drawn about the process in absence of proper input data by using soft computing tools. The most popular modeling
constituents of soft computing are neural networks and fuzzy sets. The optimization can be carried out using genetic
algorithm, an evolutionary algorithm considered to be part of soft computing. In the modeling by neural networks, a lot of
training data in the form of input–output pairs is required. The trained network is tested by a number of testing data. A number
of network architectures are tried and one which gives the minimum of training and testing errors is chosen. Two most
common neural networks for function approximation are described in this section. Fuzzy set theory can be used for different
purposes including input–output mapping. If the rule base is already available and the rules can be generated by experts, the
fuzzy set based modeling does not require data. Automatic rule generation based on the data, will need a huge amount of
data.

3.6.1. Multi Layer Preceptron (MLP) Neural Networks

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
3.6.1. Multi Layer Preceptron (MLP) Neural Networks
This neural network is a feed forward neural network with one or more hidden layers. This consists of an 'input layer' to which
input data are given, an 'output layer' corresponding to the number of outputs to be predicted and one or more 'hidden layers'
that relates input to output layer. The presence of hidden layers is required as they provide complexity to the network
architecture that is suitable for non-linear functional relationship. The input, output and hidden layers are arranged
sequentially and they contain neurons that connect one layer to another layer. The output of neurons of the previous layer
becomes input for neurons in the next layer.

An important step in neural network modeling is to train the selected neural architecture by providing data in the form of
several input–output pairs. The main aim in the training process is to minimize the error between the predicted and desired
outputs by adjusting the weights. A most common method to adjust the weights is the back propagation algorithm. In the
forward pass, the input signals propagate from input layer to output layer through hidden layers. The output result is
compared to the desired result, and in the backward pass, the error is propagated back towards the input layer where they are
used to adjust the weights. Any optimization method can be used to minimize the error at the end of first iteration. During the
backward pass, the weights are adjusted for minimizing the error, layer by layer. The weights in the output layer are adjusted
first as the target value of output neurons are available for the weight adjustment. The weights of next hidden layer are
adjusted after proper transformation of errors from the previous layer and are propagated through the network, layer after
layer. This forward and backward pass is continued till the desired output is achieved or error is close to zero. Hence the
algorithm is known as 'Back propagation algorithm'. After training the network, it is tested for efficiency by supplying the
testing data. The schematic description of the steps involved in Feed forward back propagation (FFBP) neural network is
shown in Fig. 3.12. This neural network consists of two inputs, two hidden layers containing four and three neurons,
respectively, one output neuron as shown in the figure.The following are the steps involved in training a FFBP neural network.

Figure 3.12 Description of a neural network with two hidden layers

Step 1: The learning rate η > 0 and desired maximum error are chosen. The initial weights (W1 and W2) are chosen at small
random values.

Step 2: Training of network starts here with the propagation of outputs from layer to layer till it reaches the output layer. For
example, Y1, Y2, Y3, …, Y5 are evaluated and propagated till the output layer and the output Y is evaluated as shown in Fig.
3.13.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.13 Forward pass of FFBP algorithm: (a) signals sensed by first hidden layer, (b) signals
sensed by second hidden layer, (c) signals sensed by output layer (d) error calculation

Step 3: The output Y is compared with the desired value D and the error δ is calculated. The main aim hereafter is to minimize
the error.

Step 4: The error δ is propagated backwards as shown in Fig. 3.14 and the weights are modified.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.14 ((a)–(e)) Propagation of error backwards for modification of l weights between (a)
output layer and second hidden layer (b) first and second hidden layer (c) first hidden layer and input
layer

Step 5: The output of each neuron is calculated with the modified weights and is propagated to the output layer. The output is
compared to the desired output and error is calculated. The error is propagated backwards once again and initial weights are
modified. This training process is continued till acceptable error is attained. The final weights are stored later.

The learning rate η decides the efficiency and convergence of the algorithm. However, the η values depend on the problem
modeled and there is no single value for learning rate for different training cases. The error minimization can be carried out by
steepest descent method. Other optimization methods are also used for the purpose.

There are standard activation functions g(e) like linear function, threshold function (or step function), sigmoid function,
hyperbolic tangent function that are used during the training process. The summed signal e is processed as below depending
on the activation functions chosen:

(3.55)

(3.56)

otherwise, where eh is the constant threshold value,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(3.57)

(3.58)

3.6.2. RBF Neural Network

The architecture of a typical radial basis function (RBF) network is shown in Fig. 3.15. It has one input layer, one output layer
and one hidden layer. Associated with each neuron of the hidden layer is a center, which is a vector of same size as input
vector. The weights are associated between hidden layer and output layer but not between the input and hidden layer. If there
are n centers in the hidden layer, then the output of RBF for input vector x is calculated as

(3.59)

where ϕj is a radial basis function from set of all positive real numbers to set of real numbers, ‖.‖2 denotes the Euclidean norm,
wj is the weight associated with the jth neuron of the hidden layer and the output neuron. The most common radial basis
function is Gaussian function given by

(3.60)

where σj2 is called the variance, which controls the spread of the distribution about the center. Some other functions are as
follows:

(3.61)

(3.62)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(3.63)

Figure 3.15 A typical RBF neural network

It is clear that for a fixed center, the output o is a linear combination of weights. The weights can be determined by the least
square method.

The radial basis function neural networks can be trained at a faster speed compared to MLP neural networks. However,
generally, they require relatively more training data. This is because the influence of a center reduces as the input data moves
far away from it. For better input–output mapping, there should be lots of centers covering the entire input domain and
enough training data around each center.

3.6.3. Fuzzy Set

A fuzzy set consists of elements whose membership grade can be any positive real number that is less than or equal to 1. A
membership grade denotes the strength of the association of the element within the set with 1 indicating the perfect
association. As an example, consider three types of steel: s1, s2, and s3. Suppose s1 is the best material for the work rolls of
a rolling mill followed by s2 and s3. The suitability of the material to act as roll material may be graded in the scale of 0–1. If
an expert grades them as 0.9, 0.7, and 0.5, respectively in the set S of suitable roll material, then the fuzzy set S may be
represented as

(3.64)

There is subjectivity in the allocation of membership grade and because of it the fuzzy set is often criticized. However, on the
positive side, given the subjectivity present in day to day life, fuzzy set theory provides a mathematical frame work of tackling
it. Many a times, only qualitative information in the form of language is available, which can be used in computation using the
fuzzy set theory. Thus, fuzzy set theory provides a way to carry out computation with language.

There are various ways to assign the membership grades. The most popular among them seem to be using the expert's
estimates. Various psychological methods may also be employed. One can also construct the membership grade from data
using neural networks or similar tools.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Like for conventional crisp sets, various operations have been defined for fuzzy sets. For example, union of the fuzzy sets
denoted by "∪" associates the maximum membership grade with an element. In the previous example, assume that s1 is the
costliest steel followed by s2 and s3. Assume that with the help of an expert, a fuzzy set C indicating the suitability from the
point of view of cost has been formed and is given by

(3.65)

The union of fuzzy set S and C will be a set which will have the maximum of its membership grades in C and S. Thus,

(3.66)

In language form, 'C ∪ S' is interpreted as "C or S". Here, we have members that are good in either of the attributes. (Of course
in fuzzy set we will have elements that are good in neither of the attributes, but their memberships will be low.) In the similar
way, the intersection of fuzzy sets C and S can be constructed; in which each of its element will have the minimum of its
membership grades in C and S. Thus,

(3.67)

In language form, C ∩ S means "C and S" and is collection of members that are good in both the attributes.

Two prominent branches of fuzzy set theory are fuzzy arithmetic and fuzzy logic. Fuzzy arithmetic deals with mathematical
operations on fuzzy numbers. Fuzzy numbers are generalization of interval numbers, where corresponding to each
membership grade, there is an interval in which the fuzzy number lies. A fuzzy number is a convex set indicating that as the
width of the interval reduces, membership grades keep on increasing. Figure 3.16 shows some examples of fuzzy numbers.

Figure 3.16 Examples of fuzzy numbers: (a) symmetric linear triangular fuzzy number (b) non-
symmetric linear triangular fuzzy number (c) trapezoidal fuzzy number (d) bell-shaped fuzzy number

An α-cut of a fuzzy number is the set containing members having membership grades greater than or equal toα. Figure 3.17
shows an α-cut. Note that all the values contained between A and B constitute an α-cut. At an α-cut, a typical fuzzy number
may be represented by

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(3.68)

where Aα is the membership grade of α with a1α and a2α being the lower and upper bound.

Figure 3.17 An example of an α-cut

The four basic fuzzy operations are as follows:

a. Fuzzy addition: Addition of two fuzzy numbers at anα-cut in R is defined by

(3.69)

b. Fuzzy subtraction: Subtraction of two fuzzy numbers at anα-cut in R is defined by

(3.70)

c. Fuzzy multiplication: Multiplication of two fuzzy numbers at anα-cut in R+ is defined by

(3.71)

d. Fuzzy division: Division of two fuzzy numbers at anα-cut in R+ is defined by

(3.72)

For the domain R, the definition of multiplication and division operations are modified as

a. Fuzzy multiplication:

(3.73)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
b. Fuzzy division:

(3.74)

Fuzzy logic is an approach of solving the problem by accepting the partial truth. A perfectly true statement has a truth value
equal to 1 and a perfectly false statement has a truth value 0. In general, a statement can have the truth value between 0 and
1. In solving a problem, by fuzzy logic, inference is made with a rule base. The rule base consists of IF-THEN rules, in which
either the antecedent (IF part) or the consequent (THEN part) or both contain fuzzy sets. In the following subsections the
application of fuzzy set theory in metal forming is reviewed.

3.6.3.1. Application of fuzzy set operations

Dixit et al. [30] have discussed the design of a rolling mill using fuzzy set theory. In their approach, the design specifications
are described by fuzzy membership functions, with membership grade indicating the customer satisfaction. Figure 3.18
shows the linear membership functions of desired strip velocity, reduction and power. A customer wants high value of
possible reduction and strip speed but low value of power. But increasing the reduction and strip will lead to an increase in the
power. Thus, customer's requirements can be satisfied simultaneously and a compromise solution is to be obtained. The
overall membership of customer satisfaction can be expressed as a function of membership grade in different attributes.
Thus,

(3.75)

where μ1, μ2, μ3 are the membership grades for velocity, reduction and power, respectively, and μo is the overall membership
grade. The function may mean either a fuzzy set theoretic operation similar to Eq. (3.67), where the overall membership grade
is the minimum of the three membership grades, in which the overall membership grade is decided by the lowest membership
grade or a function in which poor performance of one attribute can be compensated by the better performance of other
attributes. Out of various available solutions, the solution maximizing the overall membership is chosen.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.18 Membership functions of desired (a) velocity, (b) reduction, and (c) power. From 3
[ 0],
copyright 2002, Elsevier

Figure 3.19 shows another application of choosing the proper roll diameter. Two criteria have been considered—minimization
of power and defect prevention. Membership functions for these criteria have been plotted. The optimum diameter that will
maximize the overall membership grade will be at the intersection of two membership functions.

Figure 3.19 Membership functions of roll diameter from the viewpoints of low power and defect-free
rolling process. From [30], copyright 2002, Elsevier

3.6.3.2. Application of fuzzy arithmetic

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
In a typical metal forming process, material and process parameters are fuzzy. If the computation is carried out based on
fuzzy input data, the prediction will also be in the fuzzy form, which is more realistic. A better inference can be made from a
fuzzy prediction than from a crisp prediction without any knowledge about the uncertainly and accuracy of the solution.

Figure 3.20 shows the flow stress, hardening parameters, and friction as fuzzy numbers [31]. Figure 3.21 shows the torque
required at 24% reduction based on this data. The predicted torque is a fuzzy number.

Figure 3.20 Membership functions of (a) yield stress, (σy)o (b) hardening parameter b, (c) hardening
index n, and (d) coefficient of friction. From [31] copyright [1996] Elsevier

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.21 Typical fuzzy output torque at 24% reduction. From [31] copyright [1996] Elsevier

3.6.3.3. Application of fuzzy logic

Fuzzy logic can be used to make the inference system for prediction. A typical system will have a rule base. Depending on the
input parameters, the rules are fired. Aggregate of different fired rules is obtained to obtain the fuzzy output. Finally, the
output is defuzzified for making prediction.

Gudur and Dixit [32] used the first order Takagi–Sugeno (TS) fuzzy model for predicting the roll force and roll torque in a cold
flat rolling process. Figure 3.22 shows the fuzzification of input variables for modeling the cold flat rolling process. The input
variables are ratio of roll radius to inlet thickness (R/h1), coefficient of friction (f), percentage reduction (r), hardening
parameter (b) and hardening index (n). A huge amount of data was used for making a rule base. Generation of data by FEM
takes a lot of computational time. Therefore, first the authors fitted a neural network (NN) model for which the training data
was generated by an FEM code. Later NN was used for producing the required data for TS fuzzy model. Figure 3.23 compares
the fuzzy model predicted roll force with NN predicted roll force. The validation data set consisted of the data that was not
used in the development of fuzzy models. Figure 3.24 compares the fuzzy model predicted roll torque with NN predicted roll
torque. Both roll force and roll torque could be predicted within ±5% accuracy. The fuzzy models were also developed to
predict the most likely, lower and upper estimates of the roll force and roll torque. Figures 3.25–3.26 show the validation
results for roll force and roll torque, respectively. An algorithm to detect and suppress the outliers in the data was also
proposed by the authors.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.22 Fuzzification of input variables for a typical cold rolling process. From [32] copyright
[2009] Springer

Figure 3.23 TS fuzzy model predicted versus NN predicted roll force. From 32]
[ copyright [2009]
Springer

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.24 TS fuzzy model predicted versus NN predicted roll torque. From 32]
[ copyright [2009]
Springer

Figure 3.25 Lower, most likely and upper estimates of roll force. From [32] copyright [2009] Springer

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.26 Lower, most likely, and upper estimates of roll torque. From [32] copyright [2009]
Springer

3.6.4. Hybrid Methods

Hybrid methods are the methods that use a combination of two are more methods. There are various ways in which two
methods can be combined. The broad classification can be as follows: (i) Assisting a method by another method, (ii) Using
two are more methods together and (iii) Developing a method that is inspired by more than one method. A brief discussion of
these is provided below:

1. Assisting a method by another method: In this type of application, a well-known method is used for solving the problem,
but the method is assisted by one or more methods. For example, the weights of a neural network can be optimized using
genetic algorithm. For solving the problem, neural network plays the primary role and genetic algorithm plays an assistive
role. The situation can be other way round also. The genetic algorithm may be used for optimizing a problem, where the
functions are evaluated by neural networks. In that application, genetic algorithm plays the primary role and is assisted by
neural networks.

FEM and soft computing techniques can also be combined. Gudur and Dixit [33] have used neural network for assisting the
FEM code. A general overview of their scheme is shown in Fig. 3.27. Here, a radial basis neural network is trained by the
velocity field generated by an FEM model. Trained FEM model is used to provide a guess value of velocity field and
location of neutral point. From this guess value, the FEM model is used to generate an accurate solution. A tenfold
decrease in computational time is observed by this method. Figure 3.28 shows the comparison of roll force and roll torque
by NN assisted FEM and simple FEM prediction. Figures 3.29–3.30 show the contours of strain rate.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.27 A general overview of neural network assisted FEM model. (A label in the figure has
been slightly modified). From [33] copyright [2005] Elsevier

Figure 3.28 NN assisted FEM predictions versus FEM predictions for (a) roll force, (b) roll torque.
From [33] copyright [2005] Elsevier

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 3.29 Equivalent strain contours for high strain-hardening material (b = 0.02, n = 0.6) plotted by
(a) FEM code and (b) NN-assisted FEM code. From [33] copyright [2005] Elsevier

Figure 3.30 Equivalent strain contours for high strain-hardening material (b = 0.02, n = 0.6) plotted by
(a) FEM code and (b) NN-assisted FEM code. From [33] copyright [2005] Elsevier

2. Using two are more methods together: Two or more methods can be used together for modeling. For example, Abburi and
Dixit [34] trained a neural network with experimental data for modeling the surface roughness in turning. The trained
neural network was used to generate the rule base for fuzzy rule based system. The fuzzy rule based system has
transparency, ability to do inverse prediction and some extrapolation capability.

Garg et al. [35] used a combination of MLP and RBF to model the natural circulation boiling water reactor. Initially, an MLP
network was used to model the mass flow rate and void fraction with limited data. The trained network was used for
providing the training and testing data for an RBF as it requires more data, but provides better prediction. Geerdes et al.
[36] have used a combination of physics-based and artificial neural-network based model for the prediction of temperature

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 in a hot strip mill. It was observed that combination of these two models provides better results than their individual
performances.

3. Developing a method that is inspired by more than one method: In the neural network, the prediction is carried out by
adjusting the weights. However, it acts as a black-box and requires a large number of data for training. The fuzzy set based
system is transparent, but assignment of membership grades is arbitrary. An adaptive-network-based fuzzy inference
system (ANFIS) [37] has been developed that combines the positive features of both the techniques.

A typical architecture of ANFIS is shown in Fig. 3.31. It consists of five layers. Layer 1 carries out fuzzification. A1, A2, B1, B2
are fuzzy subsets. The output of this layer is membership grades in different fuzzy subsets. In layer 2, the membership grades
corresponding to IF part of a rule are multiplied. The output of this layer is the firing strength of various rules. As layer 1
fuzzifies the input variables x and y into two fuzzy subsets each, there are fourpossible rules. Layer 2 provides the firing
strength of each rule. Layer 3 carries out the normalization. If the firing strength of the ith rule is denoted by wi, then the
normalized strength is given by

(3.76)

Figure 3.31 ANFIS architecture

The output of layer 4 is the product normalized weight and a linear function of the input variables. Thus, a typical neuron in
this layer emits the output oi given by

(3.77)

Layer 5 contains a summing node that adds all the inputs. The parameters corresponding to membership grades of layer 1
and parameters ai, bi, and c i in layer 4 are adjustable. These are adjusted to minimize the error based on the training data. The
procedure for training is similar to the supervised training of neural network.

3.7. CONCLUSION
In this chapter, the various methods for modeling of metal forming processes are overviewed. The slab method, the upper
bound method and the slip line field method are the popular methods for estimating the required load for deforming the
material. The detailed information about the process can be obtained by FEM. However, it requires a large amount of

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 computational time and needs the accurate data for material and friction. Soft computing techniques consisting of mainly the
neural network and fuzzy set help to carry out modeling at faster computational speed and also with imprecise data. But these
techniques are dependent on experimental data and/or expert's knowledge. A combination of FEM, soft computing and other
suitable techniques can be a better strategy.

3.8. REFERENCES
1. Salimi, M. and Kadkhodaei, M. (2004). Slab analysis of asymmetrical sheet rolling, Journal of Materials Processing
Technology, 150; 215.

2. Salimi, M. and Sassani, F. (2002). Modified slab analysis of asymmetrical plate rolling, International Journal of Mechanical
Sciences, 44; 1999.

3. Kumar, D. and Dixit, U. S. (2006). A slab method study of strain hardening and friction effects in cold foil rolling process,
Journal of Materials Processing Technology, 171; 331.

4. Chitkara, N. R. and Aleem, A. (2001). Extrusion of axi-symmetric tubes from hollow and solid circular billets: A generalised
slab method of analysis and some experiments, International Journal of Mechanical Sciences, 43; 1661.

5. Assempour, A. and Emami, M. R. (2009). Pressure estimation in the hydroforming process of sheet metal pairs with the
method of upper bound analysis, Journal of Materials Processing Technology, 209; 2270.

6. Ebrahimi, R. and Najafizadeh, A. (2004). A new method for evaluation of friction in bulk metal forming, Journal of Materials
Processing Technology, 152; 136.

7. Maciejewski, J. and Mroz, Z. (2008). An upper-bound analysis of axi-symmetric extrusion assisted by cyclic torsion,
Journal of Materials Processing Technology, 206; 333.

8. Ebrahimi, R., Reihanian, M., Kanaani, M., and Moshksar, M. M. (2008). An upper-bound analysis of the tube extrusion
process, Journal of Materials Processing Technology, 199; 214.

9. Tzou, G. Y., Hsu, H. H. and Hsiao, Y. H. (2006). Investigation of a slab method analysis and FEM simulation on rotating
compression forming of ring, Journal of Materials Processing Technology, 177; 150.

10. Hsu, H. H., and Tzou G. Y. (2003). Two analytical models of double-layer clad sheet compression forming based on the
upper bound and the slab methods, Journal of Materials Processing Technology, 140; 604.

11. Lorenzo, R. D., Ingarao, G. and Micari, F. (2006). On the use of artificial intelligence tools for fracture forecast in cold
forming operations, Journal of Materials Processing Technology, 177; 315.

12. Siva Krishna, K., Ganesh Narayanan, R. and Saravanakumar, G. (2010). Tailor welded blank forming limit computation by
neural network based expert system, International Journal: Computer and Experimental Simulations in Engineering and
Science (CESES), Issue No. 8,

13. Hans Raj, K., Sharma, R. S., Srivastava, S., and Patvardhan, C. (2000). Modeling of manufacturing processes with ANNs for
intelligent manufacturing, International Journal of Machine Tools & Manufacture, 40; 851.

14. Chan, W. L., Fu, M. W. and Lu, J. (2008). An integrated FEM and ANN methodology for metal-formed product design,
Engineering Applications of Artificial Intelligence, 21; 1170.

15. Kazan, R., Firat, M. and Tiryaki, A. E. (2009). Prediction of springback in wipe-bending process of sheet metal using neural
network, Materials and Design, 30; 418.

16. Dixit, U. S., Robi, P. S. and Sarma, D. K. (2002). A systematic procedure for the design of the cold rolling mill,Journal of
Materials Processing Technology, 121; 69.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
17. Abburi, N. R. and Dixit, U. S. (2006). A knowledge-based system for the prediction of surface roughness in turning process,
Robotics and Computer-Integrated Manufacturing, 22; 363.

18. Shashi Kumar, Sanjeev Kumar, Prakash, Ravi Shankar, and Tiwari, M. K. (2007). Shashi Bhushan Kumar, Prediction of flow
stress for carbon steels using recurrent self-organizing neuro fuzzy networks, Expert Systems with Applications, 32; 777.

19. Hsiang, S. H. and Lin, Y. W. (2008). Application of fuzzy theory to predict deformation behaviors of magnesium alloy
sheets under hot extrusion, Journal of Materials Processing Technology, 201; 138.

20. Liu, W., Liu, Q., Ruan, F., Liang, Z., and Qiu, H. (2007). Springback prediction for sheet metal forming based on GA-ANN
technology, Journal of Materials Processing Technology, 187–188; 227.

21. Shylu, J., Sikdara, S., Kumar Swamy, P., Das, S., and Maity, B. (2008). Hybrid neural–GA model to predict and minimize
flatness value of hot rolled strips, Journal of Materials Processing Technology, 195; 314.

22. Rubio, E. M., Camacho, A. M., Sevilla, and L., Sebastian, M. A. (2005). Calculation of the forward tension in drawing
processes, Journal of Materials Processing Technology, 162–163; 551.

23. Rojas, H. G. A, Calveta and J. V., Bubnovich, V. I. (2008). A new analytical solution for prediction of forward tension in the
drawing process, Journal of Materials Processing Technology, 198; 93.

24. Hosford, W. F. and Caddell, R. M. (2007), Metal Forming Mechanics and Metallurgy, 3rd edition, Cambridge University Press,
USA, p. 1.

25. Wang, S. L. and Rao, K. P. (1997). An upper bound analysis of friction in metal forming,International Journal of Mechanical
Sciences, 39 (2); 201.

26. Wilson, W. R. D. (1977). A simple upper bound method for axi-symmetric metal forming problems, International Journal of
Mechanical Sciences, 19; 103.

27. Choi, S., Na, K. H. and Kim, J. H. (1997). Upper-bound analysis of the rotary forging of a cylindrical billet,Journal of
Materials Processing Technology, 67; 78.

28. Avitzur, B., Narayan, C. and Chou, Y. T. (1982). Upper-bound solutions for flow through conical converging dies,
International Journal of Machine Tool and Manufacture, 22 (3); 197.

29. Dixit, P.M. and Dixit, U.S. (eds.) (2008). Modeling of Metal Forming and Machining Processes by Finite Element and Soft
Computing Methods, Springer–Verlag, London.

30. Dixit, U.S., Robi, P.S. and Sarma, D.K. (2002). A systematic procedure for the design of a cold rolling mill,Journal of
Materials Processing Technology, 121(1–14), 69.

31. Dixit, U.S. and Dixit, P.M. (1996). A finite element analysis of flat rolling and application of fuzzy set theory,International
Journal of Machine Tools and Manufacture, 36 (8); 947–969.

32. Gudur, P.P. and Dixit, U.S. (2009). An application of fuzzy inference for studying the dependency of roll force and roll
torque on process variables in cold flat rolling, The International Journal of Advanced Manufacturing Technology, 42; 41.

33. Gudur, P.P. and Dixit, U.S. (2008). A neural network-assisted finite element analysis of cold flat rolling,Engineering
Applications of Artificial Intelligence, 21(1); 43.

34. Abburi, N.R. and Dixit, U.S. (2006). A knowledge-based system for the prediction of surface roughness in turning process,
Robotics and Computer-Integrated Manufacturing, 22(4); 363.

35. Garg, A., Sastry, P. S., Pandey, M., Dixit, U.S., and Gupta, S.K. (2007). Numerical simulation and artificial neural network
modeling of natural circulation boiling water reactor, Nuclear Engineering and Design, 237(3); 230.

36. Geerdes WM, Alvardo MAT, Cabrera-Rĺos, M., and Cavazos, A. (2008). An application of physics-based and artificial neural
network-based hybrid temperature prediction scheme in a hot strip mill. J. Manuf. Sci. Eng., 130, 014501.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
37. Jang, J.-S.R. (1993). ANFIS: adaptive-network-based fuzzy inference system, Systems, Man and Cybernetics, IEEE
Transactions, 23(3); 665–685.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
4. Friction in Metal Forming
R.K. Uyyuru, Shimjith M., Satish V. Kailas

Department of Mechanical Engineering, Indian Institute of Science, Bangalore 560 012, India

4.1. INTRODUCTION
From the understanding of basic physics, friction has been defined as the resistance encountered when a body is moved in
contact with another [1]. In technical terms, friction force is the resisting force tangential to the common boundary between
two bodies when, under the action of an external force, one body moves or tends to move relative to the surface of the other
[2]. Friction is not a material property; it is property of a tribological system [3]that consists of at least two bodies in contact
along with surrounding atmosphere and interface. It can be said that friction of a material is dependent on the mating
material, surface preparation and operating conditions. The most obvious consequence of friction is that work has to be
expended which would not otherwise be necessary. It is understandable that there is always a need for extra energy in
addition to the energy needed to perform a physical task. And so, in general perception, friction is always detrimental!
However, it is also true that in absence of friction this World would be in chaos. Thus there are many advantages of friction in
daily life too. The usefulness of existence of friction along with its characterization and role in metal forming processes is
explained in the following sections.

Since man discovered metal, he has had a need to shape it to make tools, machinery, weapons and jewelry. There are three
basic methods to shape the metal: Casting—to give a shape by pouring in liquid state into a mold that holds the required
shape, and solidifying the metal with or without external pressure; Forming—to give shape in solid state by applying pressure;
and Machining—in which material is removed in order to give it the required shape. Thus metal forming can simply mean the
act or process of shaping things out of solid metal using usually very high external forces.

Metal forming—the act of shaping metal invariably involves relative movement between tool(s) and metal to be shaped
(workpiece) under very high pressures that entail contact between two bodies—the tool and the workpiece. Thus friction, or in
fact tribology, has an important role to play in metal forming. Friction and wear of metal forming tools (as a consequence of
friction) have great influence on the economics of most metal forming processes. However, one of the last frontiers in the
study of metal forming processes is a quantitative understanding of phenomenon called friction, i.e., the friction phenomenon
in the metal forming, thus, is one of the most important aspects of the field with which the industry has not been able to fully
understand or cope up yet!

The additional energy requirement due to friction arises mainly from three sources, namely: (1) friction loss along workpiece–
tool interfacing surfaces of velocity discontinuity, (2) tangential shear along surfaces of velocity discontinuity within the
workpiece, and (3) normal fracture losses across generated fracture surfaces within the workpiece. The extra energy spent
during an operation is usually dispersed into the tribological system either in form of heat, sound, vibrations or due to plastic
deformation across the interface. Heat generated at the interface might result in softening of the materials in contact, or may
lead to phase changes; while an example for sound is squeal in brakes. Plastic deformation across the interface causes the
softer material to tear away from its base either as wear debris or as a transfer film on to the harder surface.

Apart from its contribution to the additional energy requirement in a metal forming process, friction causes inhomogeneity in
workpiece during the process. This inhomogeneity appears in two ways. The frictional stress causes a rotation of the
directions of principal stresses, which in turn determines which of the possible crystallographic planes are most favorably
oriented for atomic slip to occur. There is also a macroscopic inhomogeneity, observable using a grid scribed on a cross-
section of the workpiece before working. The second inhomogeneity effect of friction is nothing but uneven material flow
during a metal forming process, represented by unevenly distributed strain over a cross-section of the part formed.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Uneven material flow, i.e., inhomogeneous strain (e.g., bulging) results in a tri-axial state of stress. This, in turn, causes
inhomogeneous distribution of mechanical properties, apart from controlling the incidence of failures due to cracking, scoring,
wrinkling, tearing, and geometric variability. Thus it is important to have the knowledge of the friction effects on material flow
to produce defect-free products. Since forces can be relatively easily measured, friction effects on load economics could
readily be available. But the friction effects on material flow are not easily accessible for obvious reasons as no simple means
of looking into friction interface is available till today. Hence a satisfactory solution for this problem must be found with the
help of different modeling techniques and perhaps a closer look at the parameters that control the friction.

4.2. HISTORY
Metal forming processes are almost as old as mankind. Like many other branches of technology, scientific analysis and
development have accompanied their evolution. The complexities of the theories involved, together with the amount of
parameters, have, however, precluded what was achievable in any other cases, i.e., elegant, easy-to-use closed-form
solutions, or reasonably accurate equations describing the phenomena and which can be immediately used in applications.

As it is for metal forming, the problem of friction between solid bodies is as old as the human race, yet the understanding of
the basic mechanism of friction is still in its infancy. One of the earliest inventions of the human mind is the art of producing
fire by friction between solid bodies. The cave man produced a spark by rubbing a piece of flint with a stone of pyrite. He
aspired to make a fire at will. Later generations learned to make a fire by generating sufficient frictional heat with a spinning
wooden rod to ignite wood shavings strewn around the bearing in which the rod was revolving. These two observations
contain the essential features of all friction processes between solid bodies: Work must be done to move relatively to each
other two solid bodies which are in frictional contact; Heat is always generated in the course of friction processes, i.e.,
mechanical energy is converted into heat; Sparks can be produced by friction between solid bodies.

Our modern civilization depends to a large extent on the control of frictional forces. It is generally appreciated that without
friction there would be no walking, no driving, no riding and no railway traffic; it would be impossible to write with a piece of
chalk on a blackboard, or to use a pencil or to paint with a brush or to turn the pages of this book; music could not be made by
stroking of strings with a bow. On the other hand, with too much friction a prohibitively large motive power would be required,
and the wear of moving parts would be excessive [4].

In the literature, a study of friction by Thermistius (350 BC) has been regarded as the oldest ever attempt to understand the
friction. He found that the friction for sliding is greater than that for rolling. This finding should have led to the conclusion that
the static friction coefficient is greater than the kinetic coefficient of friction. First noted by Leonardo da Vinci in the 1500's,
later on re-discovered by Amontons in 1699, verified by Euler in 1750 and Coulomb in 1781 are the other empirical laws of
friction stated as: friction is proportional to load and independent of the area of sliding surfaces. Thus the coefficient of
friction is independent of load, and in the case of dry (un-lubricated) sliding, independent of velocity [5].

4.3. GENESIS OF FRICTION

Amontons, Coulomb and other early investigators proposed that metallic friction arises due to contact between two bodies in
focus. Under applied loads, this contact may lead to mechanical interaction of asperities. As proposed by Coulomb in his
theory, the action of the wedge shaped asperities causes the two surfaces to move apart as they slide from one position to
another and then come close again. Potential energy stored, thus, will be recovered and only a small fraction of energy is
dissipated. However, a realistic friction theory should include mechanisms of energy dissipation as friction is a dissipative
process.

The widely accepted friction theory by Bowden and Tabor [1] states that during the contact, under external load, energy is
required to move two bodies relatively. This is due to the work to be done against mechanical interlocking of asperities or to
break micro-welds, where the junction may get fractured during relative motion or ploughing of softer material by asperities of

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
harder material leading to physical changes (micro- and macro-scale deformation) of contacting surfaces. Thus it can be said
that when two surfaces are brought into contact, the contact occurs at the tips of asperities and the load is supported by the
deformation of contacting asperities.

Suh and Sin [6, 7] elaborated the genesis of friction in light of a new theory. Contrary to the postulates of the adhesion theory
of friction, this theory postulates that the frictional force (and thus the friction coefficient μ) is affected by the sliding distance
and the environment because of the changing contributions of three components of friction, i.e., that due to the deformation
of surface asperities (denoted μd), that due to plowing by wear particles and hard asperities (denoted μp) and that due to the
adhesion of the flat portions of the sliding surface (denoted μa). Of these three basic contributing factors, the plowing of the
surfaces by hard and wear particles is found to be the most important in most sliding situations [8]. However, relative
contribution from these factors depends up on surface preparation, i.e., surface roughness and texture.

Material loss because of sliding actions is termed as wear, which is certainly detrimental except in few cases where it is
intended for, e.g., metallographic polishing, sharpening knife edges by grinding, run-in of newly built motors. Thus friction and
wear, in fact, are functions of one another, can be related with each other [9].The frictional behavior of materials is important
in tribology not only because the frictional force between sliding surfaces is of interest but also because it generally affects
the wear behavior [10–11]. To minimize, if not to avoid, the wear of surfaces, surfaces need to be protected. Surfaces can be
protected mainly in two ways: separation of surfaces, or sacrifice by material coated on softer base surface. Lubrication,
action of applying a lubricant, serves the purposes of separating the two surfaces. Lubricants are nothing but soft solid or fluid
materials introduced at interfaces to minimize the friction. This article is intended to explain the phenomena of friction in
metal forming.

4.4. BASIC UNDERSTANDING

As discussed earlier, in metalworking processes friction arises from sliding of the workpiece against the tool. Deformation of
the workpiece usually proceeds with relative motion at the interface between workpiece and tool. This movement may
originate in imposed velocity differences (as in wire drawing), or it may develop as a consequence of the deformation process
itself (as in the upsetting of a cylinder), or it may result from a combination of both (as in the rolling of a strip).

In a metal forming process, the pressure and shear stresses transmitted through the tools affects plastic deformation of the
workpiece. The frictional conditions prevailing at the tool–material interface greatly influence distribution of these pressures
(or stresses), and thereby the material flow. The heavy loading of tools in metal forming processes makes it important to be
able to estimate the local tool stresses accurately in order to design tools appropriately. As friction plays a central role in
these calculations, due to its significant influence on the local pressures, it is important to determine the friction as accurately
as possible. Although a tremendous amount of work and discussion has gone into this subject over the decades, especially
during last few decades, friction in metal forming is still not very well understood.

In summarizing the effects of friction in material processing, magnitude of friction needs to be known for three important
reasons:

– Pressures, forces, or energy requirements can be calculated only if interface conditions can be described by one of the
variables like τ i, μ, or m.

– Reduced friction is often the main criterion in choosing a lubricant that will be applied at the tool–workpiece interface.
Lubricants are commonly used to minimize or prevent tool wear, thus increasing life of tools by reducing friction. However,
to choose lubricant comparative values of friction coefficient are often sufficient. For example in case of steel friction
coefficient value ranges from 0.3 to 0.5 and in case of aluminum the friction coefficient value ranges from 0.4 to 0.8. The
ultimate choice of the lubricant may have to be based on full-scale operating experience, which usually is disruptive and
prohibitively expensive. Therefore, a preliminary evaluation is common in simulating tests or small-scale laboratory
versions of actual metalworking processes.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 – Interface friction has an important role to play in material flow during the forming operation.

– Stable and low friction conditions are attractive to produce products with desired qualities like good surface finish,
internal microstructure, etc.

Friction between a workpiece and tools dominates the strain patterns and performance of many forming operations, and yet is
often the least quantified of all phenomena involved in forming. This status is a result of two conditions: (1) friction laws
depend sensitively on a host of variables that are not themselves usually well characterized, and are quite diverse for various
forming operations, and (2) until recently, forming analysis was qualitative and there was little motivation to derive
quantitative laws for friction under large-strain forming conditions [12].

4.5. FUNDAMENTAL CHARACTERIZATION

One may question as to why we know so little about friction processes? Stated simply by Tabor, the answer is "we do not have
a way of seeing what is actually taking place at the interface while sliding is taking place". Thus, one of the experimental
difficulties in investigating friction is that friction occurs at a buried interface. The second difficulty is that friction events can
take place at very rapid rate: less than nanoseconds for atomic events and microseconds for micrometer-sized asperity
events. New approaches are therefore needed to investigate, in real time, the prompt loss processes buried in the interface of
a sliding junction. The new ways of probing friction are Optical spectroscopy, Electromagnetic wave probes and Phonon
spectroscopy. Atomic Force Microscopy and related tools are also being used to understand the fundamentals of friction.

Although three areas of tribological phenomena (thermal, chemical and mechanical) are represented (as in Fig. 4.1), the
mechanical branch will be discussed in order to illustrate the hierarchy of levels of understanding. Moving away from the
center, one first encounters a continuum elastic/plastic description with a single yield criterion (Y = constant). At the next
level, there are polycrystalline grains which can produce inhomogeneous and anisotropic behavior in a material. Then, within
each grain, deformation is controlled by dislocations and other defects. These defects are ultimately controlled by the inter-
atomic interactions at the outermost perimeter of the branches. From this example it can be incurred that the study of friction
phenomena is very complex.

Figure 4.1 Hierarchy of inputs required to calculate friction coefficient at an engineering level 1[ 3]

Metal forming is suitable for micro scale products because of high production rate, minimized material loss and high quality
manufactured products. The main drawback in using the forming tool in this process is the die–work piece interface friction
become very significant. It has been shown that the use of traditional friction coefficient can lead to erroneous results [13].
Furthermore, the drawback of most numerical simulation tools is that it is only possible to give global friction to the interface
of die and work piece. In most of the metal forming simulations friction which has been traditionally used is Coulomb friction
[14– 17] and the constant friction law [18]. Variations in these have been developed and used [19– 21].

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
In the case of sheet metal forming effective mathematical friction model is very important [22], especially in case where the
surface to thickness ratio is large [23]. It does not yield the best results despite the well-developed material behavior models.
This is due to the usage of simple friction model in which only a constant friction coefficient is considered. The most
frequently used model is Coulomb friction model in which the ratio of friction force and normal load is called coefficient of
friction, which is considered to be a constant [24] as used by Cheng et al. [25] and Huang [26]. However, friction depends on
large number of parameters like micro geometry, sliding velocity temperature and normal load. If any of these parameter
changes the coefficient of friction will also change [27].

Effective lubrication systems result in low friction levels which reduce the loads imposed on tooling and workpiece. Lower
force levels also reduce tooling deflection and can improve the dimensional accuracy of the product. In most of the metal
forming processes, friction between the workpiece and tooling has an important effect on the mode of deformation and the
required forces. In analyzing such processes it is important to have some method of characterizing the frictional stress at the
interface. However, "dry metal forming", i.e., metal forming without lubrication has become possible in recent days, although it
causes high friction at the tool–work piece interface. To achieve this it is necessary to understand the mechanism involved in
dry friction. Dry metal forming is advantageous since the effect on the environment could effectively be reduced [28].

During the conceptual understanding of friction over the centuries, as explained in previous section, friction has been
characterized by empirical laws. The laws state that (1) the force of static friction is proportional to the normal load between
the surfaces and (2) Friction is independent of the areas in contact, and that (3) kinetic (or sliding) friction is less than the
limiting static friction. The first two laws can be explained on the basis of the minute roughness present on all real surfaces
resulting in plastic deformation and possibly local welding, and is therefore, dependent upon the load applied and the surface
finish and not upon the superficial areas involved. The third law can be explained by considering that the heat generated
during sliding reduces the yield strength of the contacting high spots.

The above empirical laws were summarized, and put forward by Coulomb in a mathematical form as:

(4.1)

where F is the frictional force at interface, P is the normal force, μ is Coulomb coefficient of friction, τ is average frictional
shear stress, and p is the normal pressure.

In the Eq. (4.1), τ and p are derived from F and P by dividing them with apparent area of contact, Aa. For a constant μ, the
interface shear stress, τ, must increase at the same rate as the interface normal pressure, p. It is known to be valid for many
physical contacts where the normal pressures are considerably less than the flow stress of either of the materials in contact.

Once the interface shear stress is increased, in proportion with increase of normal pressure, beyond the limit of shear flow
stress, k, of either of materials in contact, material has a further option than sliding along the interface. As it tries to move
relatively with the other, it will take less energy for the material to shear inside the body, while the surface remains immobile
at the interface. This condition is termed as "sticking friction". Then Eq. (4.1) is modified in the following way to represent the
condition of sticking friction:

(4.2)

From theory of plasticity, shear flow stress of a material is constant and is related with flow stress, σ, as follows:

(4.3)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
according to von Mises yield criterion, and

(4.4)

according to Tresca yield formulation. Thus μ assumes a maximum value of 0.577.

Let us assume a metal forming process where it is usual that the interface pressure reaches a multiple of flow stress. Ask is
constant for metals, with increasing pressure the calculated μ actually will have to drop. This is in contradiction to the friction
law and can lead to misinterpretation of reported values of μ, and thus it can be concluded that definition of Coulomb
coefficient of friction becomes meaningless when μp > k.

To overcome the difficulty with μ, in later developments interface friction is made a function of shear flow stress of the softer
material. This is described by the following equation:

(4.5)

where m is the frictional shear factor which has a value m = 0 for a frictionless interface, and m = 1 for sticking friction. This
formulation of friction has great mathematical convenience as τ is defined with the aid of k, the value of which is known . In
contrast, the use of μ can lead to complications because the value of p (which in itself is dependent on μ) has to be found.

As with the coefficient of friction, the friction factor is usually assumed to be a property of the workpiece, tooling materials,
and lubricant, and independent of other parameters. However, the friction factor has an important advantage over the
coefficient of friction as far as metal forming analyses are concerned. With the former characterization, the friction stress at a
point is independent of the normal stress. This factor, coupled with the relative ease with which m may be measured, has led
to increased use of the interface friction factor for describing friction in metal forming processes. It appears to be particularly
suited for hot-working processes involving large deformations, such as forging and extrusion. In processes such as cold-
rolling and wire-drawing the use of coefficient of friction appears to be well established. Some other simple characterizations
of friction have been used in the literature. However, their advantages, if any, are usually limited to a narrow range of operation
of a particular process and the coefficient of friction and friction factor approaches account for the vast majority of existing
analyses.

The heavy loading of tools in bulk metal-forming processes makes it important to be able to estimate the local tool stresses
accurately in order to design tools appropriately. Friction plays a central role in these calculations due to its large influence on
the local normal stresses. The development and application of the finite-element method has allowed, when compared to the
classical analytical methods, a more accurate calculation of process parameters including the tool stresses. However,
increased accuracy in the prediction of the material flow at the tool–workpiece interface, and consequently in the calculation
of the tool stresses, is unrealistic as long as inadequate friction models are applied.

When analyzing pressure distributions in bulk metal-forming processes, it is conventional to adopt either Amonton's friction
law, τ = μp or the law of constant friction stress, τ = mk. Neither of these two laws, however, is applicable universally as
explained in earlier section. But commercially available finite element programs for the simulation of metal-forming processes
usually model the frictional constraint acting at the interface between the workpiece and tools according to either of these
friction laws. Unfortunately such description is often inadequate and does not represent the actual interface relation. In the
simulation of metal-forming processes, the use of Amonton's law gives occasion for an overestimation of the friction stresses
at the tool–workpiece interface, because the normal pressure often is considerably greater than the yield stress of the
material. Consequently, the friction stress becomes greater than the yield stress of the material in pure shear.

The difficulties with Amonton's law prompted several researchers to include the law of constant friction in finite-element
programs for bulk metal-forming. However, as the friction does not depend on the current state of stress at the tool–
workpiece interface, but simply on a material property, the friction stress is overestimated at low normal pressures. Wanheim

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
and Bay [29] developed a general model for expressing friction at the tool–workpiece interface. In practical terms the model
assumes friction to be proportional to the normal stress at low normal pressure (p/σo < 1.5), but going towards a constant
value at high normal pressure (p/σo > 3), (Fig. 4.2), the two ranges being combined via an intermediate transition region. It has
been proved that the proposed general model by Wanheim and Bay is a major improvement in the ability to simulate
processes where low tool–workpiece interface stresses may prevail. This is done through experimental and numerical
investigations into the upsetting of a semi-tapered specimen between parallel dies [30].

Figure 4.2 Schematic representation of general friction model [15]

4.6. BENEFICIAL EFFECTS OF FRICTION

As mentioned in earlier sections, it is not always desirable to reduce the friction to a minimum, but even in these instances
pickup should be avoided. Some beneficial effects of friction are listed in the following [31–32]:

a. As a motivation force to supply required power: In several metal forming processes including strip rolling, ironing, hydro-
static extrusion with viscous drag. In rolling the energy required for the operation is transmitted to the strip through the
friction or surface traction between the rolls and the strip. The improvement in roll bite by increasing friction permits
heavier drafts in cold-rolling of sheet and strip, i.e., it allows the metal to be drawn into the gap between the rolls,
according to the relation (4.6) the roll bite is directly proportional to coefficient of friction and diameter of the roll.

(4.6)

where Δhmax is maximum roll bite, μ is interface coefficient of friction, and R is radius of rolls.

b. As a stabilizing factor: In the process of strip rolling the position of the neutral point is automatically determined by the
power required. Whenever the neutral point moves away from its stable position, the power balance is disturbed. For
example, should the strip slow down temporarily, the neutral point will move toward the exit. The driving force increases,
and the resisting friction decreases, causing the total power supplied to rise and restore the higher exit speed.

c. As a deterrent to defect formation: In extrusion, higher friction values serve to deter formation of central burst. In drawing,
however, higher friction values promote the same defect. Moreover, in extrusion, the back end of the billet tends to flow
over the face of the pressure pad during the final stage when the extrusion pressure becomes too high. The rear section of
the billet thus becomes entrained along the axis of an extruded bar, and even a hollow pipe may form. This feature is so
common that it is usually referred to as the extrusion defect, without further qualification, the defective material has to be
sawn off, or the extrusion must be terminated at an earlier stage, reducing the yield. The lower the friction at the face of

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 the pressure pad, the easier it is for metal to flow inwards and the sooner the extrusion defect starts. This is particularly
noticeable when a plain graphite dummy-block is used to obtain a complete, discard less extrusion. Considerable lengths
of rear-end pipe may then be produced, so the friction is deliberately increased by placing asbestos pads between the
surfaces.

d. In punch-bench manufacture of tube on a mandrel, it is advantageous to maintain a reasonably high friction on the
mandrel, to carry part of the drawing force. In this way, the tensile stress on the leading end of the finished tube is reduced,
and heavier reductions of area can be made without fracture. A somewhat analogous situation is found in deep drawing or
pressing. To prevent the onset of necking in the walls while deep-drawing a cup there should be high friction on the punch
and low friction against the die.

e. In open-die forging it prevents the metal escaping from between the tools and in closed-die forging it provides the back
pressure in the flash to ensure filling of the die cavity.

f. High friction is desirable in varieties of tensile grips used in metalworking.

The primary role of friction in metal forming processes is evident. Friction not only affects the forces required but also the
outcome of an operation, whether it will end up successfully or fail, may depend on the value of friction. Many operations such
as rolling depend entirely on friction for their implementation. It is, therefore, obvious that sophisticated and more accurate
and dependable means of measuring friction will aid the metal forming industry.

4.7. MEASURING FRICTION IN METAL FORMING

Friction in metal forming processes comprises a very complex situation with many parameters interacting with it. Some of the
important variables which affect the friction between die-material interface are relative velocity, temperature, lubrication
condition, roughness, and mechanical properties of material or die [33]. Ideally, one would prefer to include all the variables
affecting friction in a simple friction test. However, unfortunately, it is impossible because of the complex interaction between
many of these variables [34]. This is also due to the great variety of process conditions. To put it in simple words, as
mentioned elsewhere [35], a universal test to measure friction in metal forming processes has not been and will never be found.
This lack of universal test to measure friction can also be attributed to the facts that: (1) the friction laws depend sensitively
on a host of variables that are not themselves usually well characterized, and are quite diverse for various forming operations,
and (2) until recently, forming analysis was qualitative, and there was little motivation to derive quantitative laws for friction
under large-strain forming conditions.

There is at present no generally accepted method of measuring the value of the coefficient of friction for given surfaces and
lubricant. Various factors can influence the result, and it is essential that test conditions should match the actual conditions of
operation as closely as possible. This implies that the only strictly reliable friction data to use are those obtained from
measurements during the operation considered.

Because of the difficulty in measuring the friction and its distribution, friction at the tool–material interface is usually
measured globally, i.e., an average value for friction coefficient (or friction factor) will be used in calculations. Hydrodynamic
lubrication, Coulomb friction, and constant shear factor are few assumptions to describe (global) friction phenomena [36]. A
popular tribological guide to the material forming scientists is due to Schey [37].

An excellent overview of friction measuring methods (or evaluating the lubricant), imitating the particular working conditions,
is due to Lange [35]. It also gives a brief explanation of different friction measuring methods and their pros and cons. On the
basis of this overview, friction tests can be broadly classified according to means of measuring friction– by measuring loads
or forces using sophisticated instrumentation, or just by measuring geometrical changes without resorting to instrumentation
(e.g., ring compression). Again the first category of tests was classified, though not in very strict sense, into groups of tests
depending on the—geometry of contact (e.g., twist compression), possibility of generating new surfaces with bulk
deformation (e.g., wire drawing), etc. The basic principles guiding the selection of a friction test are simplicity (to separate

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
frictional effects from others) and similitude (to assure that the many variables are similar to actual operation of interest)
apart from tooling cost, specimen size, and so on [12].

Metal forming processes are basically categorized into two groups, bulk metal forming and sheet metal forming, for simplicity
of analysis on the basis of physical changes the workpiece will undergo during a process. Thus it would be appropriate to
discuss various friction measuring methods applicable to these two kinds of metal forming processes separately as explained
in the following sections.

The earliest experimental work of measuring friction (published in the 1930s) is due to Ekelund [38] who investigated the
influence of temperature on the coefficient of friction (inverse relation between them) through contact angle calculations in
his hot rolling experiments with steel over the temperature range 700°–1100°C. An empirical relation between the coefficient
of friction and rolling temperature resulting from his investigations is given by Eq. (4.7). Later, this relation has been modified
for rolling-speed-effect by Geleji [40], Eq. (4.8).

(4.7)

(4.8)

where T is the temperature in Kelvin, v is the rolling speed in ms–1.

After this early work on friction measurement in rolling, there was no further work for a couple of decades. This may be
probably due to the dominance of rolling process in metal forming industry where the above equations might have served the
purpose. Then in late 1950's and later, more work on friction measurement was documented as a result of search for
alternative possibilities that can lead to a simplified test to carry out. The results of this work apply to a wider range of metal
forming operations.

In early investigations, the technique of inserting feeler pins through the tool to the tool–workpiece interface was a popular
method to measure interface loads and calculate friction values. Some short comings of this technique include the following:
the method is expensive and delicate; measurements may not be reliable; method cannot be applied easily outside the
laboratory, i.e., under production conditions. Such laboratory studies can indeed provide some much needed insight into the
phenomenon and its effects on processing. It is well recognized today, however, the values existing in the plant would be
better measured under production conditions. Furthermore, it is recognized that different friction values will be measured for
identical lubricants under differing conditions of pressure, temperature, speed, and many other undefined variables. Friction
values for extrusion may differ from those for drawing, forging, rolling, or other metal-forming processes.

4.8. FRICTION TESTS FOR BULK FORMING

Bulk metal forming processes basically involve high pressures ranging multiples of yield stress of workpiece material, larger
strains than those usually found in sheet metal forming processes. Over the last few decades, numerous simulative bulk
friction tests were developed such as twist compression test (TCT), upsetting, ring compression test (RCT), bucket test, open
die backward extrusion test (ODBET), plane strain compression test (PSCT), rolling test, combined forward and backward
extrusion test, double-backward extrusion test, and more. The choice of individual tests depends partly on the mechanisms
judged important, and may be partly on individual experience and prejudices of the researcher.

4.8.1. Twist Compression Test

There are many variations of the test, and in these tests normal pressure is combined with continued sliding over the same

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
surface. In one version of the test [Fig. 4.3(a)], two cylinders are pressed together on their end faces, and then one is rotated
against the other. This has been used extensively in studies of adhesion and of lubricant properties at high normal pressure. In
the twist compression, the pre-applied lubricant film is exposed to prolonged rotation under pressure, and its resistance to
breakdown and potential for die pickup are evaluated. In modified version [Fig. 4.3(b)]—one cylinder will be hollow—when
pressures are high, it shows remarkable correlation with lubricant performance in severe industrial operations such as cold
extrusion. This is probably because the limited lubricant supply is exposed to continued sliding, and its durability and
resistance to breakdown are evaluated. This test can be readily adapted to high temperatures. Pressures are, however, limited
to the flow stress of the hollow specimen, and this limitation has been overcome by rotating a punch against a confined
workpiece. In another version of the test [Fig. 4.3(c)], the principle of simultaneous twisting and compression is used where
the interface pressure is increased by confining the workpiece in a container.

Figure 4.3 Schematic presentation of geometries for different version of twist compression test

4.8.2. Upsetting
In ordinary upsetting of a billet to evaluate strength of a metallic material, the billet will be compressed between two flat dies.
However, due to existence of friction at the interface between billet and dies, uneven deformation occurs across the
longitudinal cross-section of the billet. This barreling phenomenon could be used to evaluate friction for open die forging
operations. The method employs conical platens with a certain semi-cone angle to compress the billet as shown in Fig. 4.4.
The cone angle which produces no barreling is equal to the friction angle, α = tan– 1μ, from which friction parameters can be
calculated. However, this method is tedious and may not be very accurate.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.4 Schematic view of upsetting using conical platens

4.8.3. Ring Compression Test

This test is now the most commonly used experimental method for quantitative evaluation of friction in bulk metal forming
processes for a given tool, material and lubricant combination. This is probably because only simple interpretation of
experimental results is enough to find friction values from calibration charts, and usefulness of these friction values in many
metal forming processes, especially in forging operations, over a wide range of strain rates and temperatures. The popularity
of the test can be attributed to its practical convenience including the fact that friction can be judged from deformation, i.e.,
geometrical changes alone.

RCT was first used by Kunogi [40] as a qualitative method of evaluating different lubricants in cold extrusion experiments with
steels. The concept involved in these friction experiments is that an increase in the coefficient of friction causes an increase
in the radius of the no-slip circle (Rn) as shown in Fig. 4.5 where Ri is inner radius of the ring, Ro is outer radius of the ring, and
H is height of the ring. Later, Male and Cockcroft [41] (and others) developed the basic ring compression testing method of
evaluating lubricants into today's more user-friendly technique.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.5 Schematic representation of neutral line position for different friction conditions

The principle involved in RCT is as follows. When a short, hollow cylinder is compressed axially between flat, rigid platens, the
diameter of the hole may instantaneously increase, decrease, or remain constant according to the amount of frictional
constraint imposed by the platen–cylinder interface.

After compressing a ring between two parallel platens, the changes in geometry are recorded in terms of percentage change
in inner diameter as a function of deformation, measured as percentage reduction in height of the ring. Then the friction
parameter can be obtained from calibration curves (Fig. 4.6). The calibration curves are developed theoretically, and relate the
concerned variables—percentage change in inner diameter, percentage reduction in height, and the friction parameter.
Mathematical analyses leading to the development of calibration curves are due to many researchers [e.g., 41–43].

Figure 4.6 Analytically derived calibration curves for ring compression test [37]

The theoretical analysis of RCT was based on the following assumptions: (1) the deformation of the ring is uniform
throughout the thickness, implying no bulging takes place on the background of uniformly distributed frictional constraint.
Though this assumption is not valid in strict practical sense, barreling effects can be minimized by implementing appropriate

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
geometrical conditions, (2) interface friction value is constant, which is not necessarily a valid assumption, (3) the ring
material is rigid and perfectly plastic.

Based on theoretical studies [43–44], in the absence of data on best specimen geometry, the maximum accuracy can be
obtained by using a ring of smaller height and large internal diameter as compared with the external diameter. Too large an
internal diameter, however, makes deformation asymmetrical. Too low height of the cylinder makes the measurement of
deformation inaccurate because of the presence of relatively thick lubricant related coatings. Thus it is important to choose a
particular geometry ratio between outer diameter (OD), inner diameter (ID), and height of cylinder (H), for example—OD:ID:H =
6:3:2 or 6:3:1 or 6:3:0.5. For deformations under low friction conditions, i.e., in experiments with good lubrication, a somewhat
greater thickness (e.g., OD:ID:H=6:3:2) may be used [45].

The major advantage of using the RCT for the evaluation of interface friction is that it involves only the physical measurement
of shape changes. Thus, friction may conveniently be measured at the high temperatures and strain rates typical of hot-
working processes. However, if deformation pressure is measured during compression of the ring, it is possible to calculate
the flow stress in addition to the friction. Thus, the RCT can be used for two purposes: (1) to evaluate the friction parameter
and (2) to evaluate the flow stress of a given material [46–47].

On the other hand, there exists certain drawbacks of RCT such as (1) the theoretical analysis from which the calibration curves
are derived may not be valid in all cases (2) μ or m values will change depending on the choice of theory (3) the necessary
interruptions may cause problems in boundary conditions.

There exists practical difficulty in conducting experiments where the unavoidable barreling or bulging of the cylindrical
surface of rings due to friction causes inaccuracies in calculations of test-after geometry. The other difficulties involve the
non-circularity of holes after compression due to anisotropy of the work material and 'tapering' [48]. However despite
difficulties due to experimental nature and inaccuracies from material properties, RCT is still one of the most commonly used
methods of measuring friction for metal forming processes. One of the recent critical reviews of ring compression test and
applicability of calibration curves was due to Rao and Sivaram [49]. Other kinds of development of RCT involve use of rings
with different cylindrical shapes [50–51], in an attempt to extend the practical range of the test to allow the evaluation of
friction under low tool–workpiece interface pressures.

Due to the geometrical design of the conventional ring-test, the interface stresses will, however, always be greater than the
flow stress of the material, thereby impeding quantification of friction, and evaluation of the behavior of lubricants, for
processes where interface stresses below the flow stress of the material occur. A new complementary ring-test geometry,
which allows the characterization of friction under low pressure conditions, has been proposed by Petersen et al. [50]. Given
the convenience of the ring compression test, modification to the conventional geometry is proposed to include a double-
conical outer profile (as shown in Fig. 4.7) has been proposed in an attempt to extend the practicable range for the test to
allow the evaluation of friction under low tool–workpiece interface pressures (p ≤ σ). Finite-element analysis in conjunction
with metal experiments, applying both the conventional and the modified geometry for different lubricants, provides the
validation for the general feasibility of the proposed test geometry.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.7 Schematic presentation of complementary geometry for ring compression test [50]

Friction is difficult to measure even though there exist a large number of tests which have been developed, very few are
suitable for hot metal forming applications. Friction is one of the most significant physical phenomena influencing the forming
of metal, yet in comparison with metallurgy, heat transfer and mechanics it remains the least understood despite the large
number of tests which have been developed over the decades. The importance of friction in hot metal deformation processes
can hardly be over-emphasized as it contributes to the success or failure of the process to fulfill its objectives. For instance in
hot rolling, friction is necessary at the interfaces of the roll bite in order to transmit deformation energy from the work rolls to
the workpiece, though excessive friction tends to restrain the level of deformation which results in unacceptably high roll
forces and spindle torques. Moreover, if the level of friction is too low, roll slippage or failure of the workpiece to enter the roll
bite will occur. With the increasing use of finite elements to solve hot metal forming problems, a need for improved knowledge
of how friction behaves under changing interface conditions, i.e., variations in strain rate and temperature, is required in order
to improve the degree of accuracy of process models.

Few studies have been undertaken to understand friction in hot metal forming, especially when addressing the critical issue of
whether changing magnitudes of strain rate and temperature at the interface influence the level of friction. The article by
Rudkins et al. [52] describes an experimental investigation into friction under hot forming conditions using the RCT. In their
work [52], the authors investigated the possibility of using the ring compression test to assess how friction is affected by
variations in temperature. Additional observations were made with regard to the suitability of using the RCT for hot metal
forming, especially their ability to reproduce process conditions in hot rolling. It was found that RCT is a good simulative
friction test for hot metal forming and rolling; increasing the temperature which causes the level of friction to increase.

RCT is also used for different purposes such as to investigate the influence of scale on friction at the work–tool interface in
hot metal working. Although it has been recognized that scale reduces interfacial friction, it has not previously been possible
to separate the effects of scale thickness and of scale composition. In the work by Luong et al. [53], techniques were
developed to obtain specimens in which the scale consisted exclusively of oxide. This enabled the authors to investigate the
influence of both scale thickness and scale composition on interfacial friction. In the work by Luong et al. [53], specimens
were heated in different atmospheres in order to obtain scales with different thicknesses and compositions. In a series of
tests designed to study the influence of scale thickness on friction, specimens were heated in a CO2 atmosphere to obtain
scales which consisted essentially of wüstite. Heating times were varied to obtain different wüstite thicknesses. In another
series of experiments, both the atmosphere and the heating time were varied to obtain specimens with scales of similar
thicknesses but with different compositions. It was found that the friction at the work–tool interface during hot metal working
is influenced by both the thickness of the scale and the bulk properties of the scale. The scale thickness appears to be more

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
important in this respect than the scale composition. Scale reduces friction and, up to a certain limit, thicker scales give lower
friction values.

In a ring test, both the forging load and the inner diameter are influenced by friction. However, the friction-calibration curve of
RCT considers only the latter as a measure for interface friction. The reason is that the former, i.e., the forging load, is not
sensitive to friction in ring tests because of the influence from the following two factors: (1) the forging pressure, i.e., the
average value of the normal stress at the interface; (2) the contact area between the ring and the upper die. These two factors
have a tendency to cancel each other at relatively low value of reduction in height. There is no "universal" ring-test friction-
calibration curve. The friction-calibration curves must be defined for the specific ring material under specified ring- and die-
temperatures and ram-speed conditions.

4.8.4. Bucket Test

In a ring test, metal flow and die geometry are simple, and the new surface generated, i.e., surface extension during
deformation is small. Therefore, in a ring test the friction characteristics of a forging lubricant cannot be adequately
evaluated. A new test, named as bucket test [54] (Figs. 4.8–4.9) has been formulated where surface extension is expected to
be relatively high in addition to large amount of material flow.

Figure 4.8 Bucket set-up used for evaluation of forging lubricants [54]

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.9 Proposed method for estimating friction coefficient in bucket tests 5[ 4]

The quantitative determination of the friction factor, or coefficient, requires: (1) a friction sensitive parameter that can be used
as a friction-calibration factor (for example, the internal diameter in the ring test), and (2) theoretically determined standard
friction-calibration values. For the "bucket" test, the forging load and the bottom thickness together can be chosen as the
friction-sensitive parameter because a good lubricant results in both lower load and thinner bottom thickness in this forging.

Very recent development of friction evaluation using backward cup extrusion is due to Uyyuru et al. [55]. The method is based
on analysis of deformed physical grid pattern, embedded into the slug before cup extrusion. An ingenious physical pattern has
been developed and used during the technique as it allows calculation of the extension of the surface all over the inner wall of
the cup with ease. Degree of surface extension over the head was observed to vary along the cup length, which was
compared with expansion curves obtained from numerical simulations. It was found that BCE with new physical pattern
technique can be used for qualitative friction prediction, as well as quantitative evaluation of friction.

4.8.5. Open Die Backward Extrusion Test

The disagreement or inconsistency found while using RCT can be attributed to assumptions such as no barreling of the
deforming specimen and a constant friction coefficient across the interface. Although the ring compression test is an
effective method for determining the friction parameter for many deformation processes, the use of generalized friction
calibration charts regardless of the material type and test conditions must be avoided.

The new technique, named open-die backward extrusion test (ODBET), was developed as an alternative method to evaluate
the coefficient of friction in metal forming processes [56]. In this technique, a cylindrical billet was compressed between two
rigid, upper and lower, platens (dies). Instead of making a hole in the billet as in the ring compression test, the hole is made in
the top platen for materials to flow upward through this hole (Fig. 4.10).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.10 The boundary condition used in ODBET [56]

As the top platen moves down to plastically deform the cylindrical billet, part of the material flows into the hole and generates
the extruded height which is the main concern of this study while the other flows outward resulting in an increase in the
diameter. Finite Element (FE) simulations of ODBET were performed for different aspect ratios (H/D) and different die
geometries (d/D) by utilizing an elastic/plastic finite element code in order to obtain the friction calibration curves (FCCs). The
results indicated that the extruded height is a function of the friction conditions at die–workpiece interface. The FCCs are,
thus, generated by plotting the percentage increase in the extruded height versus the percentage reduction in the specimen
height for numerical results.

Once the reduction in the height of billet and the increase in extruded height are known, the coefficient of friction, m, can be
found from the appropriate plot. It was also shown that the extruded height was affected by the billet and the die geometry
such as H/D and d/D ratios. When d/D is highest, the extruded height becomes largest and decreases as the ratio decreases
for the same coefficient of friction. On the other hand, the extruded height increases as H/D decreases. It can therefore be
concluded that the ODBET technique can be used for obtaining reliable data regarding the coefficient of friction encountered
in large deformation processes as an alternative method to the ring compression test. However, the results should only be
used in conjunction with calibration curves specifically derived for the material under investigation and the specific test
conditions.

4.8.6. Plane Strain Compression Test

It is very interesting to note that the development of PSCT was related to solving the stress problems in rolling by taking
advantage of plane-strain conditions. It was then used to evaluate yield properties of materials at high strain levels from
1950's. Though it was developed to solve the problems with stress and mechanical properties of materials, from early days of
its development itself, PSCT has been also used in studies of tribological problems. A schematic picture of PSCT set-up is
shown in Fig. 4.11.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.11 Schematic 3D-presentation of PSCT [34]

Probably the earliest work in the direction of use of PSCT to quantify friction was from mid 1950's [57]. The first qualitative
use of PSCT in order to distinguish between different lubricants, to find better lubricant for cold rolling processes, was
reported at the end of the 1950's [58]. Alexander [57] suggested two methods to measure friction using plane strain
conditions. First, the coefficient of friction was derived from geometric slip-line-field method for two integral ratios of (w/t) =
3–7. In the other experimental method, the coefficient was derived for all (w/t) ratios by comparing the mean pressures with
the basic yield stress of the material using an approximate theory based on the simplifying assumption that plane sections
remain plane.

Later on, Takahashi and Alexander [59] also showed experimentally that the PSCT could be used to derive considerable
amount of information about the behavior of lubricants in metalworking processes. They were also successful in
implementing the experimental method of comparing the mean die pressures with the basic yield stress of the material, and
in obtaining quantitative values for the coefficient of friction. However, in that work, the basic yield curve of the material was
derived at room temperature using graphite grease, which gives a very low coefficient of friction.

At elevated temperatures, problems may arise because of lack of suitable lubricants that can give very low friction. This
problem was addressed by Bailey and Singer [60] in their work on PSCT. Here they adopted a different method called cylinder
compression test to obtain basic yield curve developed by Cook and Larke [61]. In that method stress values were
extrapolated to zero ordinate as they were decreased as the ratio (d0/h0) decreased to find the basic yield stress (d0, initial
diameter, h0, initial height of the specimen). These methods of indirect measurement were tested to obtain good results
though the process might be tedious.

One of the reported attempts to measure friction directly was due to Mizuno et al. [62]. In their attempt, they modified the
common PSCT. Friction in this modified version of the test was measured directly by simultaneous sliding of one of the dies
in the strip-width direction. Important experimental constraint was to make sure that V is much larger than either U or W, when
V being the velocity of die in strip-width direction, U being the outflow velocity of workpiece, and W being the press ram
velocity. Thus, the measured friction force component can be considered as true friction force. The procedure may not work at
large thickness, and for higher reductions.

In early 1980's, the PSCT has been further developed for the direct evaluation of lubricants for metalworking purposes. In their
research work, Lonn and Schey [63] calculated the magnitude of friction from the deformation of the edge in the compression
zone instead of measuring compression forces. By doing that, they tried to put PSCT at par with ring compression test, which
is used to measure friction directly from physical measurements instead of force measurements. They found that bulging of
the specimen edge under the anvil zone is a reliable quantitative measure of friction. As in other attempts of using PSCT to
measure friction, they had some constraints while standardizing the test procedure. One of them is to keep initial (w/t) > 1 to

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
avoid non-homogeneous indentation in which the effect of in-homogeneity may overshadow friction effects. The other
constraint is (b/w) ≥ 2 to assure that the friction hill develops fully. Here they also found that larger (b/w) ratios do not offer
any advantage. Thus, they were able to quantify friction coefficient by standardizing the test procedure up to a 75% reduction
in case of hot and 87.5% reduction in cold working conditions. The principle involved in their calculations is that though the
total spread is almost independent of friction, the shape of the edge zone under the anvil varies from almost straight for low
friction to a very peaked profile for high friction. That shape has been quantified as bulging (defined as the differences in
widths over the contact zone) as shown in Fig. 4.12.

Figure 4.12 Schematic representation of bulging

Another kind of PSCT called 'constrained PSCT' has also been used in some applications. In this test, it is always questionable
if it is possible to achieve the real plane strain conditions during the test. As shown in Fig. 4.13, in this test the side surfaces
of the work material are in contact with the grooved die walls during the whole test, thus frictional effects cause
inhomogeneous material flow. However, the problem of sideways bulging is, of course, avoided.

Figure 4.13 Schematic presentation of constrained PSCT

A study concerned with use of PSCT for identification of flow stress is by Becker [64]. He tried changing the boundary
conditions at edges of the tools according to plasto-mechanics theory, in order to obtain the flow properties of the workpiece
material. In addition, he proposed a method that is based on an iterative procedure to calculate the interface friction. However,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
this method must be considered as inaccurate. His friction measuring method includes experiments with different dies with
varying widths thus obtaining different force–stroke curves.

In another study concerned with tribology in metalworking, Wang and Ramaekers [65] developed a tribo-meter that produces
important characteristics of bulk forming processes. The constrained PSCT has been used in their work with certain
reservations, for example, neglecting the friction force on the side surfaces of the die, etc. Though good lubrication has been
applied, and the side surface area was much smaller than the contact area between platens and work material, this
assumption certainly leads to some inaccuracies. However, their study highlighted the use of PSCT in measurement of
friction and flow stress of the material, especially at larger strains thus bringing in the PSCT as an alternative to RCT and/or
the Rastegaev test whose specific specimen design render perfect lubrication at higher strains between the platens and
workpiece.

Gurrero [66] applied FE-codes to analyze the hot plane strain compression testing, where she focused on friction and lateral
spread. In this study, the author found that deformation patterns are sensitive to both initial geometry and friction, while they
are insensitive (or dependency is inconsiderable) to the other major test variables like material type, strain rate, thermal
effects, etc. It was suggested that a more complex relationship exists between friction and lateral spread in contrast to the
view expressed by others [63, 67]. At the same time, an attempt has been made [68] to interpret the hot PSCT results to obtain
true stress–true strain curve for Al-samples. This interpretation includes methods to correct for friction and lateral spread.

Another interesting dimension of PSCT study is concerned with homogeneity of material deformation. As interpreted by many
authors like Lonn and Schey [63], Dieter [69], it was found that tool and workpiece geometries must agree to a specific relation
in order to achieve homogeneous deformation. Recently Valberg [70] confirmed this relation and went a step ahead to find the
best possible region of workpiece where the material would deform at a constant strain rate, i.e., subjected to homogeneous
deformation. He found that location is on vertical mid-axis, and lies at halfway between the center and the surface of the
workpiece. Similar investigation has been reported earlier by Gurrero [66] where it was found that strain distribution does not
change significantly in the region, comprised center plane and its surroundings up to 20% away from it in the lateral direction.
Metallurgists will be looking for this kind of information, which is useful in studies of micro-structural changes during forming
operation, which in turn is important for understanding the whole process.

An interesting study of friction characterization during plane strain compression was due to Uyyuru et al. [71]. In that study,
PSCTs were conducted on Al-alloy work-pieces embedded with internal physical grid patterns for the purpose of studying
lubrication effects through deformed physical grid patterns. Experiments were complimented by numerical simulations. It was
found that both the interface friction and die stroke levels influence the deformation of physical patterns in Plane Strain
Compression. A new friction formulation was put forwarded for PSCT based on Inverse Parameter Identification approach,
where deformed patterns in experiments are compared with corresponding patterns from FE-simulations to predict friction. By
quantifying the deformed patterns in FE simulations, a set of friction calibration curves were developed, and were used with
reasonable success.

4.8.7. Rolling Test

In rolling the friction conditions are unusual, since the direction of the frictional force reverses at the neutral line, and the
interface is not subjected to pure slip. Frictional balance in the roll gap determines the position of the neutral point, where
strip moves with the roll at input speed v0. The strip emerges at a higher speed v1 from the roll gap. The speed difference,
called the forward slip (v1 – v0), increases with increasing friction. A method which has been used in research is to apply a
steadily increasing back tension until the neutral point coincides with the exit point, and the strip begins to skid. Under these
conditions, the friction force acts in the same direction over the whole roll surface [Fig. 4.14(a)], and the coefficient of friction
(μ) can be deduced as it will be simple function of the torque (T), the roll load (P), and roll radius (R) as given by Eq. (4.9).

(4.9)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.14 Friction measuring techniques in rolling

This method was first used by Whitton and Ford [72] for an evaluation of lubricants. If the rolling mill is not equipped with
torque meters, the strip may be stopped by attaching it to a spring dynamometer [Fig. 4.14(b)] according to the technique of
Pavlov [73]. If B is back tension, and α is the angle of bite, then,

Even though these techniques are attractive because μ is derived without the aid of theory, interface conditions are not
necessarily representative of those prevailing under normal rolling conditions.

On the other hand, the simplest practical assessment of friction is to determine the maximum draft which the rolls will accept,
Eq. (4.10), thus

(4.10)

where μ is coefficient of friction, Δh is maximum draft, and R is the roll radius.

4.8.8. Combined Forward and Backward Extrusion Test

A new test method for cold-forging lubrication method was proposed by Sanchez et al. [74]. This test is based on a combined
forward rod and backward cup extrusion process. A global friction factor is estimated quantitatively by relating it to the length
of the forward extruded bar for a specified displacement of the punch. The procedure involves comparison of the extrudate
length with theoretically computed lengths keeping all other variables constant. The procedure was analyzed using an upper
bound approach, the calibration curves for different friction factors were determined by analyzing the flow characteristics of
the extruded material. Calibration curves, based on upper-bound calculations, deliver m-values between m = 0 and m = 1. This
test was developed for the cold extrusion of aluminum. Corresponding experiments conducted by Popilek [75] showed,
however, that the test is not very sensitive to small changes in the actual lubrication conditions. Figure 4.15 depicts the
schematic view of the test; while a set of friction curves are shown in Fig. 4.16 [12]. In Fig. 4.15, α is die angle for forward
extrusion, R1 is radius of forward extrudate, R2 is radius of punch, and Ro is billet radius/outer radius of backward extruded
cup.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.15 Combined forward and backward extrusion test

Figure 4.16 Schematic of friction curves for a particular test geometry, along with lubricants closely
simulating different friction conditions [12]

The advantages of the test are: it can closely simulate extrusion operation and no load measurements are required to evaluate
friction. However, disadvantages of the test are that it is not very sensitive to small changes in lubricating conditions and
accuracy of test results depends on assumptions used in the analysis.

With the advent of FEM and better computational tools, a combination of computational methods and experiments has been
used to get better understanding of many variables, including interface friction. The simultaneous usage of computational and
experimental work makes it easier to find the process parameters more accurately as one method can be used to validate the
other.

4.8.9. Double Backward Extrusion Test

In cold-forging operations, the material flow and the quality of forged parts are functions mainly of the tool geometry, the

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
amount of deformation or strain, and the conditions at the workpiece–tool interface (e.g., friction, relative surface velocity,
surface finish and heat transfer). While many of these parameters are well known and controllable, the influence of friction is
often difficult to predict and depends on a variety of factors. A lubrication test that reflects the conditions of industrial
production, i.e., high surface pressure, severe material flow, substantial surface enlargement, etc. has been developed.
Geometry of the test is shown in Fig. 4.17. A billet is forced against a stationary lower punch by a moving upper punch in a
stationary die to produce two cups.

Figure 4.17 Double backward extrusion test geometry [76]

This test was optimized using the finite element method to design test geometry such that maximum sensitivity in test results
was obtained for small changes in lubricating conditions. The larger the billet height-to-diameter ratio, the bigger is the
maximum difference in cup heights. The ratio of the two cup heights was used as a measure of the friction existing in the
system. The difference in the cup heights arises due to different relative velocities resulting in different friction conditions and
is a measure of lubrication. The height of the upper cup h1 is bigger than the height of the lower cup h2. To find calibration
curves, an FEM analyses using program DEFORMTM was carried out. The experimental value of the cup height ratio can then
be compared to the predicted ratios of the friction coefficient. The advantages of this test are: geometry is very sensitive to
small changes in lubricating conditions, and the test is simulative and does not require measurement of loads and strains to
evaluate the friction. [76].

4.8.10. Forward Rod Extrusion Test

A simulative friction measuring test for extrusion processes was proposed by Tan et al. [77]. Friction stresses were obtained
from measurements of slopes of extrusion pressure–punch travel curves at the steady state stage. Experimental
investigations of measuring the friction stress at the container–workpiece interface were carried out. Extrusion pressure–
punch travel curves before the steady state can be divided into four stages: elastic deformation, filling container deformation,
filling die aperture deformation and exiting inhomogeneous deformation as shown in Fig. 4.18. It is found that the relationship
between normal pressure and friction stress appears linear, and therefore, Coulomb's friction model fits the experimental data
very well. It should be noted that the experimental data obtained here were under limited cold-forming conditions: at constant
room temperature and constant strain rate. It is also to be noted that the temperature at the tool–workpiece interface, which
might have a significant influence on friction, was not considered in friction stress calculations.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.18 Schematic representation of four stages in punch load-displacement curve for forward
rod extrusion [17]

In Fig. 4.18 the following nomenclature is observed: O is initial point, A is point of the maximum elastic (linear) deformation, B
is point of the maximum expanded deformation filling the container, C is point of the maximum extrusion pressure, D is start
point of steady state, and E is final point of steady state.

4.8.11. Barrel Compression Test

Analogous of RCT, where only measures of geometry of the specimen before and after compression would suffice to calculate
friction parameter, a new method for evaluation of friction in bulk forming has been reported recently [78]. As the name
suggests, in this test, a cylinder with specific aspect ratio was compressed under appropriate test conditions representing real
forming process. Equation (4.11) based on upper-bound analyses, modified to include velocity discontinues [79], gives a
minimum requirement of power for the actual velocity field:

(4.11)

where J* is upper-bound power, σ0 is flow stress of material, V is volume of deformation zone, is components of strain rate
tensor, S, discontinuity and friction surface, K, shear yield stress of material, m is constant friction factor, and T i is the external
traction.

After some simplifications, an equation for m can be written as follows:

(4.12)

where

(4.13)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
and R is average radius of cylinder after deformation,H is height of cylinder after deformation, b is the barrel parameter, ΔR is
the difference between maximum and top radius, ΔH is reduction of height of cylinder after deformation,R0 is initial radius of
cylinder, H0 is initial height of cylinder.

After measuring top radius (RT), maximum radius (RM) and height of cylinder after deformation (H), Eqs (4.12)–(4.13) can be
used to calculate the constant friction factor. However, the accurate measurement of the top radius of cylinder after
deformation is difficult in practical conditions. Therefore, with approximation of the profile of the barreled specimens with an
arc of a circle, top radius can be determined by Eq. (4.14):

(4.14)

Thus, using this method the constant friction factor, m, can be calculated by measuring only maximum radius, RM, and height
of the cylinder after deformation.

The advantages of the above method are as follows:

i. It is a very simple method.

ii. There is a high sensitivity of geometrical shape of cylinder to friction condition.

iii. It does not need force measurement.

iv. The constant friction factor can be determined for forming processes corresponding to actual processing condition.

4.8.12. Tip Test

A very specific test that is based on backward extrusion of a material is the tip test. The special experimental set-up of this
method induces the formation of a sharp radial tip on the extruded end of the workpiece [80]. It has been found that the
simple geometrical measurement. the radial distance from the radial tip to the side wall of the specimen, b in Fig. 4.19,
increases with higher levels of friction, and thus can be used to effectively determine the friction condition. Both numerical
simulations and physical experiments were employed during the study and found to correlate well with each other.

Figure 4.19 Schematic view of tip position in backward extrusion process

4.8.13. Strip Drawing Test

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
4.8.13. Strip Drawing Test
This test was proposed by Le and Sutcliffe [81] in relation with friction model in cold rolling process. The test procedure
involves pulling a strip through two stationary dies (with specific wedge angle, Φ) placed in close approximation by the
application of indentation force, N. Force, F, required to pull the strip through the dies is measured in presence of different
lubrication mediums, i.e., this test describes the measurements of friction in the bite during strip drawing under thin film
conditions. The simplicity of the strip drawing process makes it attractive for this type of investigation. The tool geometry
makes it easier to control tool roughness and the absence of a reversal in sliding direction, as compared with rolling, allows a
more confident evaluation of friction. A schematic drawing of strip drawing test is shown in Fig. 4.20.

Figure 4.20 Schematic representation of strip drawing test [81]

It was assumed that the friction coefficient, μ, is constant through the bite, the force, F, required to pull and indentation force,
N, are used to calculate the constant friction coefficient according to the Eq. (4.15).

(4.15)

All the tests discussed until now are based either on force measurement or geometry measurements. There have been few
attempts to predict friction parameter mathematically by which it might be possible to avoid errors in measurement and thus
in the reported friction parameter. A method has been proposed by Lee et al. [82] to find the friction factor at the die–
workpiece interface for the forging process without the need for measurement of the shape changes of the workpiece. It is
based on the concept of the inverse analysis and requires only the definition of the problem inversely and does not need
complicated mathematics to invert the problem. Using this method, it was shown that friction factor decreases when the die
velocity is increased, and thus confirming the validity of the proposed method.

4.9. FRICTION TESTS FOR SHEET FORMING

In most of the sheet forming processes friction between the workpiece and tool surface has an important effect on the

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
formability by affecting the strain distribution in the various regions of the sheets. In analyzing such processes, it is important
to have some method of characterizing the frictional stress at the interface. Especially, in recent years, more computational
programs based on finite element schemes have been developed and used to study plastic deformation in sheet forming
processes. The accuracy of such analysis is limited by the knowledge of boundary conditions. The lack of knowledge of
friction behavior during forming operations is a serious deficiency in the current state of the art. A good understanding of
friction is crucial for improving accuracy of simulation of sheet forming processes.

The friction behavior is known to be a complex function of interacting variables that constantly change during the forming
processes. Therefore, it is important to study friction behavior and estimate friction coefficients under the conditions that can
closely represent those in real sheet forming processes. Such conditions include contact pressures, deformation magnitudes,
tool geometry, forming speed, surface roughness, sliding velocities and lubrication conditions. Because it is experimentally
difficult to isolate and identify all of the variables independently, an alternative method of studying friction behavior of sheet
metal is needed for production condition. Although no single test can be used to simulate the wide range of practical sheet
metal forming conditions, some areas have received a great deal of attention.

There are currently several different types of friction tests for sheet metal operations which have been used to evaluate the
interfacial sliding behavior of materials. There are certain basic elements inherent in all friction tests, namely two surfaces in
sliding contact with one another. There must be a force to create a relative sliding motion between the two contacting bodies,
and a normal force to create a frictional condition at the interface. To calculate a friction coefficient these forces must be
measurable.

4.9.1. Pinch-Type Tests (Sheet Friction Tests)

Several types of geometries for pinch-type friction tests of sheet metal are shown in Fig. 4.21. These tests may be one-sided,
two-sided, and may have a flat, cylindrical, on inclined contact geometries. A back force may be applied to introduce a degree
of tensile deformation that can be controlled independently from the normal force, fN. These tests have the advantage of
simple interpretation and friction coefficient between the sheet and die is found out using Eq. (4.16). However, the sticking
case is complicated by the unknown contact area if curved dies are used. Tests are much more similar to conditions in rolling,
drawing, or extrusion operations. Disadvantages of these tests are: geometry and deformation patterns induced are not similar
to the tensile deformation and sliding in press dies, where through-thickness compression is nearly always absent.

(4.16)

Figure 4.21 Pinch-type friction tests for metal sheets

4.9.2. Simulative Tests for Deep Drawing and Stretch Forming

The effectiveness of boundary lubricants can be determined by finding the operative coefficient of friction between sheet
metal and the tool. For this purpose, two different methods for the determination of coefficient of friction were developed by

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Kaftanoglu [83]. The first method is applicable to the radial drawing region (between sheet metal and die) in deep-drawing
(Fig. 4.22). The second method is applicable by finding the operative coefficient of friction between sheet metal and the tool.
From experimentally measured quantities such as loads and strain distributions, the coefficients of friction can be calculated
using analytical equations. The results indicated that the coefficient of friction varied as a function of the plastic deformation.
The coefficients of friction in stretch forming were found to be much higher than those in radial drawing. The experimental
results indicated that plastic films could be more effective in reducing friction than grease type lubricants. The methods
developed can be used in the evaluation of the lubricants in such processes. They can also be used to provide the necessary
coefficients of friction to be used in metal forming theories.

Figure 4.22 Schematic view of deep drawing [83]

To determine the coefficient of friction, in the radial drawing zone, according to Eq. (4.17), blanks with the same diameter
were tested under the same conditions using different blank-holding loads. During the experiment, the punch load was
accurately measured. The increase in the punch load due to the increase in the blank-holder load was determined for different
bulge depths. Then the coefficient of friction could be computed for different δH's, and μ versus δH curve could be plotted and
extrapolated to zero δH to find the correct μ. The coefficients of friction for the stretch forming zone were found to be higher
than those for the radial drawing region.

(4.17)

Where μ is coefficient of friction, P is the punch load, and H is blank holding load.

The main drawback of the test is that while measuring large numbers which are close, small errors in measurement can lead
to considerably larger errors in friction values.

4.9.3. Simulative Tests for Sheet Forming Operations

In sheet metal forming frictional forces are important boundary conditions and so must be known for accurate analytical and
numerical process modeling. Accurate measurements of friction forces can provide valuable information for process model
development and validation. The importance and unique features of friction in sheet metal forming have led to the
development of friction measurement apparatus which simulate the workpiece deformation in forming operations.

The tensile strip test developed by Duncan et al. [84] is widely used. In this test, shown schematically in Fig. 4.23, a strip

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
specimen of sheet metal is pulled over the cylindrical surfaces of pins to simulate stretching and drawing processes. The
pulling force on one side of the pin is measured along with the strain in a section of the test specimen on the other side of the
pin. The strip force on this second side of the pin is calculated from the measured strain using the stress–strain
characteristics of the test material. Assuming that the coefficient of friction is constant over the pin surface, the strip tensions
are used in the capstan friction model to calculate the coefficient of friction. There are questions that arise when the strip
friction test, other process simulation tests, and experimental results are considered. Some of the results reported indicate a
dependence of coefficient of friction on the test configuration, e.g., pin radius. In terms of the test apparatus itself, a
drawback of the tensile strip test is that the load on one portion of the strip is estimated from measurement of strain.

Figure 4.23 Tensile strip test for measurement of coefficient of friction [84]

In a successful effort to overcome the above mentioned difficulty, Hao et al. [85] proposed two different simulative tests for
measuring friction in sheet forming operations. A major advantage of these test apparatus is that strain does not have to be
measured to measure coefficient of friction, but by measuring the strip tensions of each side of the pin using capstan or belt
friction model given by Eqs (4.18)–(4.20). A schematic representation of 'U' shape sheet forming friction test, is shown in Fig.
4.24. The strip specimen is pulled over two pins and around a surface which is mounted on a load cell. This apparatus is set
up in a tensile test machine and the pulling force, F1, is measured by the test machine load cell. The strip tension on the other
side of the pins, F2, is measured with the load cell in the apparatus. The radius and location of the pin can be changed to
investigate the effects of the pin radius and wrap angle on friction.

Figure 4.24 Schematic view of 'U' shape strip friction test [85]

Figure 4.25 shows the setup for the 'L' shape sheet forming friction test. A test strip is held at one end in a grip supported by a
load cell. The specimen is wrapped around a cylindrical pin and loaded in a tensile testing machine. The strip tensions and F1
and F2 are measured simultaneously during the test. An extensometer is used to monitor strain and to determine strain rate
during the test.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.25 Schematic view of 'L' shape strip friction test [85]

Force and moment equilibrium on an element of the strip result in the following descriptions of the average contact pressure,
p, the average friction stress, f, and the coefficient of friction, μ.

(4.18)

(4.19)

(4.20)

where F1 and F2 are the strip tensions, W is the width of the strip, R is the radius of the pin and Φ is the wrap angle of the strip
on the pin.

4.9.4. Sheet Drawing Test

A test device was constructed [86] which allow one to measure the normal force FN and the tangential force FT whilst drawing
a strip of sheet between two cylindrical-surface (23.4 mm diameter) dies clamping the sheet. A schematic representation of
the test is presented in Fig. 4.26.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.26 Schematic view of sheet drawing test for friction evaluation [86]

The test was carried out in such a way that a strip of the sheet was clamped with forceFN so that the clearance between the
dies was smaller than the sheet's thickness. During the recording of the drawing force FT and the normal force FN the sheet
was drawn for a distance of about 10 mm. Then the clamping force FN was increased and the action was repeated until, as a
result of too great a clamping force, the drawing of the sheet turned into the stretching of it, at which point the test was
stopped. In this way the values of forces FN and FT were determined for a range of strains. The strains of the sheet were
determined after the friction test by measuring the thickness of the drawn sheet. As the specimen starts to be forced through,
the sheet undergoes plane strain.

The coefficient of friction, μ, defined as a ratio of the resultants of the normal force and the tangent force on the contact
surface, was determined using the relationship

(4.21)

where α is the angle of the arc of the surface contact of the cylindrical dies with the sheet and tan (α/2) is represented by

(4.22)

where R is the radius of the cylindrical surface of the dies and Δh is characterized by the difference between the thickness of
the sheet before deformation and that after deformation in the friction test.

4.9.5. Sheet Metal Forming Simulator

A sheet metal forming simulator (SMFS) was developed by Wilson et al. [87], which stretches the sheet around a cylindrical pin
as shown in Fig. 4.27. SMFS was used [88] to investigate the relationships between friction and process variables including
sliding speed, angle of wrap and strain rate, in the boundary lubrication regime, with electro-galvanized steel sheet and
commercially pure aluminum sheet.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.27 Sheet metal forming simulator [88]

The average coefficient of friction, μ, can be calculated from Eq. (4.23).

(4.23)

where F1 and F2 are the inlet and outlet tensions, respectively, Fb is the Swift bending tension, and θ is the angle of wrap. If the
workpiece thickness is small compared to the pin radius, Fb may be neglected compared with the difference (F2 – F1).

4.10. FRICTION TESTS FOR HYDRO FORMING

Hydro forming is comparatively a new class of sheet forming method. This class of forming technique has found many
applications especially in automotive industry where there is a need for thin-walled hollow complex shaped parts. The main
advantages of hydro formed parts are low weight and high stiffness. Tribology plays an important role in this class of forming
methods too. The available tests to experimentally determine friction parameter in sheet forming are not satisfactory as they
do not simulate the hydro forming process.

As shown in Fig. 4.28, a typical hydro forming process consists of two areas—feed area and forming area. The feed area is
meant to feed material into forming zone, i.e., it acts like a reservoir for material which is pushed into the forming zone. Actual
gross deformation takes place in deformation area where sheet attains its complex shape because of pressure induced by the
fluid. The different classes of tests are developed to find friction parameters in these two areas.

Figure 4.28 Different zones in hydro forming

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
One class of tests, flow-through tests, is meant for measuring friction in feed area. In these tests a tube will be expanded by an
internal pressure against the tool wall. Later on the tube will be forced out of the tool by die against the friction force at the
interface between tube and the tool. This force is measured either as a difference of the punch forces or as a resulting force
on the tool, measured in a horizontal or vertical test layout, as shown in Fig. 4.29. The coefficient of friction is calculated by
dividing the friction force with the nominal area of contact.

Figure 4.29 Schematic presentation of feed area friction tests for hydro forming processes [89]

The first method is based on the direct measurement of the difference in punch force, while the second is based on the force
measurement on the tool, i.e., only one force has to be determined, which minimizes the possible errors. In the case of the
horizontal layout, the die is split parallel to the tube axis. The closing force Fp is measured and load cells between the two die
halves measure the tool contact force. The difference of the two forces is taken as contact force between the tube and the
die. The disadvantage of this method is that it is more complex than others.

Figure 4.30 shows the different tests for measuring friction parameter in hydro forming at the forming area. In first method
(tube expansion test), a round tube is expanded by inner pressure in a rectangular tool. Depending on the friction conditions
the development of the radius and the wall thickness in the edges differs. The test can be terminated at a certain pressure,
using the edge radius as a measure for the friction parameter. The radius will be proportional to the friction parameter [90].
Alternatively, the wall thickness distribution can be analyzed at the end of the test and correlated, e.g., by using FEM
calculations, with the friction parameter. The advantages of the test are the plastic deformation of the specimen during the
whole test and that the forming process is not affected by the testing method itself, like it is the case in the direct
measurement, explained in the following section. The disadvantage is that there is very little sliding of the specimen against
the tool surface and that the contact pressure is not constant but increases during the test.

Figure 4.30 Schematic presentation of forming area friction tests for hydro forming processes 8
[ 9]

In direct measurement of friction for hydro-forming process, a ring segment is placed at the position where the dome of the
part is formed, which will be shifted by the friction force against load cells. The disadvantages of the method are very typical,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
and arise due to the elasticity of the load cell and gap between the ring section and the remainder of the tool. This cause
severe disturbances or even errors in the measurement. The dome must have a minimum length, as the material has to be
formed so that it covers the tool radius. Before the radius is formed, it is not possible to measure any friction force in this test.

The third test, the tube upsetting test, was developed to overcome these difficulties. The principle of the test is similar to that
of the ring compression test or the combined forward–backward extrusion test. The friction parameter is determined from the
geometry of the test piece after forming, using a suitable diagram (Fig. 4.31). Such diagrams are again calculated from
analytical solutions, or can be found from FEM calculations.

Figure 4.31 Tube upsetting test

During the test, a tube is continuously upset in a closed die, while it is elastic/plastically expanded towards the tool wall by an
internal pressure. Due to the friction forces the wall thickness increases non-homogeneously. The biggest increase of the wall
thickness occurs near the punch with the higher velocity. The shape of the wall cross-section will be a function of the friction
coefficient, and strain hardening behavior (characterized strain hardening behavior C and strain hardening exponent n) of the
material under deformation and the compression ratio. Depending on these parameters, different shapes can develop [89].

a. μ = 0, n = 0: a homogeneous thickening of the whole tube occurs, the wall thickness is a function of the compression ratio
only

b. μ = 0, n > 0: same as in case (a).

c. μ > 0, n = 0: the wall thickness increases linearly from one end to the other, while the largest thickness occurs at the tube
end which is in contact with the punch having the higher velocity.

d. μ > 0, n > 0: the wall thickness is similar to that of the case (c), but the difference in thickness along the tube wall is smaller
as the hardening coefficient increases.

Compared to the other test principles which are mentioned above, this method has the following advantages: the processing
parameters (internal pressure, sliding velocity, deformation speed and upsetting grade) can be chosen freely during this test;
the equipment used for this test is simple compared to some of the other tests and does not interfere with the process itself;
by changing the length of the tube, the test can be adapted to different friction conditions. In the case of low friction, long
tubes would be beneficial, as the friction stresses are accumulated to forces which significantly influence the thickening of
the tube.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
4.11. PARAMETERS AFFECTING FRICTION
4.11.1. Surface Roughness
Figure 4.32 depicts the variation of friction coefficient as a function of surface roughness. With very smooth surfaces, the real
area of contact grows rapidly, and so does the friction, whereas with very rough surfaces the friction is high again because of
the need to lift one surface over the asperities on the other. In the intermediate range of roughness (normally used in
engineering practice), the friction is at minimum and almost independent of the roughness.

Figure 4.32 Schematic depiction of variation of friction against surface roughness

With sheet metal forming industrial backdrop, Bello and Walton [91] studied the combined effect of surface roughness and
lubrication on friction at the tool–metal interface during sliding contact in presence of lubricant. In their experiments, strips of
commercial pure aluminum were pulled through steel dies designed to give partial simulation of the conditions, which exists in
the flange and die radius profile regions of the deep drawing process. They found that the conventional surface roughness
parameters do not provide a satisfactory functional characterization of the surface roughness in the context of the friction
developed in sliding contact. They coined a new roughness parameter, Bearing Length Product (BLP), that is equivalent to the
product of average peak width (wp) and RMS roughness parameter (Ra) for the purpose. It was pointed out that BLP was a
better functional characteristic of the surface. However, the nature of the proportionality between BLP and friction parameter
(μ) depends on the way lubricant behaves, for example as shown in Fig. 4.33 where lubricant Tallow acts as boundary
lubricant because of adsorption on surface, and Tellus-27 was effective in hydrostatic lubrication regime due to its viscosity.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.33 Variation of friction parameter with BLP for different lubricants [91]

4.11.2. Effect of Surface Topography

Lubricants are widely used in metal forming operations for obvious reasons like for reducing friction by avoiding metal-to-
metal contact thus easing material flow. Lubricants have been conveniently characterized in many ways, popular ones being
ring compression test and plane strain compression test. However, not much attention has been given to the surface
characteristics that lubricants are applied to and the influence of the surface topography of workpiece surfaces on friction
reduction.

Early research in this direction mostly dealt with aspects like real area of contact and lubricant trapping. For example, in his
theoretical work, using the plane-strain theory of plasticity for rigid-non-hardening materials, Kudo [92] proposed a
methodology to estimate the effect of trapped lubricant in microscopic surface pockets of a work material by the tool surface
on the apparent coefficient of friction. It was also pointed out that the effect was more pronounced for lubricant having a
higher bulk modulus. Thus Kudo tried to highlight the significance of surface topography. Later-on, Nelleman et al. [93]
investigated the effect of different surface topography geometries by varying asperity angles and concluded that normal
pressure and bulk modulus have a great influence on the real area of contact, whereas the asperity slope and friction factor
are of minor importance. The friction factor is also of minor importance to the real area of contact, but has a strong influence
on the friction stress.

In one of the earliest experimental attempts to understand the effect of surface topography, Schey [94] demonstrated the
effect of different machined surfaces though he did not examine in detail the effect of surface topography as such. More
recently, Geiger et al. [95] identified two types of lubricant pockets, closed and connected, in their investigation and were able
to determine the proportion of surface area at the interface they comprised by analysis of the surface topography. They
compared the difference between two types of surface topography but not in terms of surface roughness. Lakshmipathy and
Sagar [96] in their industrially relevant study tried to understand the influence of die grinding marks directionality on friction in
open die forging under lubricated conditions. Two sets of dies, one with unidirectional grinding marks and other with criss-
cross grinding marks were used. It was found that, for the same percentage of deformation, the dies with the criss-cross
ground pattern required reduced forging loads when compared with the die having uni-directionally ground pattern (Fig. 4.34).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Lower friction value was also observed for the forging process when the die with the criss-cross surface pattern was used.
They concluded that the lubrication breakdown tendency is increased when pressing is done with uni-directionally ground die
compared to that done with a criss-cross ground die.

Figure 4.34 Forging load versus deformation at different die surfaces 9

[ 6]

Hu and Dean [97] studied the relationship between friction behavior and surface topography using ring upsetting test. The
tests were carried out using either a liquid lubricant or under clean dry conditions. Two types of workpiece surfaces, random
and directional, were prepared by either shot-blasting, or EDM or turning to obtain different levels of surface finish. It was
found that, for random surfaces, smoother ones could retain more lubricant and decrease friction resistance. Rasp and
Wichern [98] studied the effect of surface topography on frictional resistance using different kinds of surfaces. In their
experiment, the five different specimen surfaces used were (i) as received, (ii) etched, (iii) coarse ground perpendicular to test
direction, (iv) coarse ground parallel to test direction and (v) polished. They found that the arithmetic roughness value (Ra) and
lubrication regime has greater influence than the directionality of the surface lay. It was also pointed out that the effects of
surface lay on frictional behavior during upset without lubricant vary with upset velocity.

Experiments were conducted using an inclined scratch tester to study the influence of surface texture on friction [99, 100]. A
special feature of the inclined scratch tester is that in a single experiment a range of loads can be studied. Various kinds of
surface textures—namely unidirectional grinding marks, 8-ground (steel plate ground in 8 shaped motion), and random
(polished)—were prepared using simple metallographic techniques. Roughness, represented by Ra, of surfaces was varied
over a range as they were prepared using different grit emery papers and abrasive powders. Both surface texture and sliding
angle (angle between unidirectional grinding marks and sliding direction), and their effects on friction and transfer layer
formation were studied. It was observed that the plowing component varies significantly with both surface texture and sliding
angle, whereas the adhesion component of friction does not vary with either. It was found that a surface texture that promotes
plane strain conditions near the interface causes a higher plowing component and thus the higher coefficient of friction. On
the other hand surface texture that promotes plane stress conditions at the interface results in lower value for plowing
component of friction. It was found that the sliding perpendicular to unidirectional grinding marks gave maximum friction
force contributed by higher plowing component, and at the other extreme random (polished) texture resulted in lower friction
values. In addition it was observed that friction value varies with sliding angle, it was highest for 90° and lowest for 0° sliding

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
angles. It was observed that the roughness as given by Ra within the range of the present tests does not significantly affect
the friction values.

4.11.3. Effect of Crystal Structure

The effect of material crystal structure on friction depends on how easily material can undergo plastic deformation. The ease
of plastic deformation of a material depends on the number of slip systems available. It is well known that a minimum of five
independent slip systems are required for plastic deformation to take place by slip. Hexagonal metals have a limited number
of slip systems, and thus the chance for forming higher true areas of contact is reduced due to lack of plastic deformation
when compared with FCC. Hence, HCP metals experience low friction coefficient than FCC and BCC metals in which large
number of slip systems are available for plastic deformation [101]. Crystal structure can affect friction behavior directly and
indirectly through texture evolution. The factors that influence the friction behavior of metallic materials include evolution of
crystallographic texture, plastic deformation and strain hardening. Preferred orientation, texture, can arise during fabrication
or can be induced by deformation processes such as wear. Strain softening was also associated with texture development in
many instances. The concept of strain softening with texture evolution was used by Kuhlmann–Wilsdorf [102] to explain the
transition in the coefficient of friction with the sliding distance.

As early as 1940, Ernst and Merchant [103] recognized the influence of crystal structure on friction. Alison and Wilman [104]
suggested that there is a fundamental difference in wear and friction behavior of hexagonal and cubic metals. As it could be
expected from the above discussion, coefficient of friction for cubic metals is larger than that for hexagonal metals (Fig.
4.35). Buckley and Johnson [101], in their study of the influence of crystal structure on frictional behavior of HCP metals,
found that crystal indeed have strong influence on friction by comparing two metals that are very close in nature except for
crystal structure (Rhodium atomic number 45 with FCC crystal structure, and Ruthenium, atomic number 44 with HCP crystal
structure). Since crystal structure affects friction behavior, it could be expected that allotropic metal will exhibit a similar
influence on either side of its transition temperature of it. Experiments were performed using Cobalt which possesses HCP
structure at room temperature and transforms to FCC structure at 417°C. A significant change in friction value was observed
correspondingly on either side of the temperature, as shown in Fig. 4.36, which was attributed to the phase transformation.
They also found that for HCP metals, the lattice parameters influence the number of operating slip systems, and thus ease of
plastic deformation which in turn influences the friction. HCP metals with close to ideal atomic stacking ratio (c/a = 1.633),
e.g., Co, Rh, Mg, primarily slip along basal planes (i.e., less number of slip systems), and metals which deviate considerably
from ideal stacking such as titanium (c/a = 1.587) will exhibit primarily non-basal slip mechanisms (i.e., high number of slip
systems). Variation of the friction with lattice parameter is shown in Fig. 4.37.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.35 Variation of coefficient of friction for cubic and hexagonal metals [104]

Figure 4.36 Change in friction coefficient with allotropic transformation in cobalt 1

[ 05]

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.37 Variation of friction with lattice parameters in HCP metals [101]

Texture development and its effect on friction were studied extensively over the decades. Wheeler and Buckley [106] found
that during sliding for (1) FCC metals, a texture develops where (111) was tilted about 10° towards the sliding direction, (2)
BCC metals, similar texture develops along (110) with [111] direction oriented in direction of sliding, and (3) HCP metals, no
texture development was found at lower loads. However, basal texture evolution along (0001) for HCP metals was observed
by many other researchers, for example [107]. It can be stated that texture evolution and thus the crystal structure of a metal
plays an important role in its friction behavior. The lower coefficient of friction for hexagonal metals when compared with
cubic metals can be attributed to the ease of metal shearing through higher number of slip systems.

4.11.4. Effect of Strain Hardening

It is well established fact that severe plastic deformation occurs in the surface regions (e.g., asperity contacts) of a metal
sliding over a harder surface. This results in the surface metal being progressively work hardened and reaches a maximum
hardness that depends on the method by which it has been deformed. The maximum hardness attainable by abrasion can
differ appreciably from that resulting from trepanning or heavy compression [108].

There are very limited sources available in literature about the effect of material strain hardening capacity on friction. In one of
these, Gaivoronskii et al. [109] formulated a methodology to relate the contact pressure with the friction stress and studied the
hardness and strain hardening, and found that the degree of influence of these material bulk properties on friction and their
interconnection with the surface layer properties can be used to determine friction stress.

On the other hand, Lin [110] studied the friction, wear, and micro-hardness of some metals as a function of degree of work
hardening. It was found, as shown in Fig. 4.38, that there is a systematic relation between the friction coefficient and the
surface micro-hardness as a function of increasing degree of work hardening for the face-centered cubic metals Al, Cu, Ag
and Au. These loci are close to the locus for different fully work hardened cubic metals of a similar range of hardness [104],
and they may be represented approximately in the range 40 < Hd < 140 kg/mm2 by an equation of the form

(4.24)

where a and b are constants.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.38 Variation of coefficient of friction with degree of strain hardening [110]

Reid and Schey [111] performed experiments, and tried to correlate friction with bulk hardness, surface hardness and strain
hardening rate. They found no correlation between bulk hardness and friction; however, they found satisfactory correlation
between surface hardness and strain hardening. Other results indicate that a greater increase in surface hardness or a higher
strain hardening rate do not imply lower friction. It has been concluded that hardness cannot be used as a criterion for
predicting coefficient of friction in un-lubricated sliding. However, it was stated that metallurgical compatibility between
materials plays an important role over their tribological properties.

Mokhtar [112] experimentally showed that hard metals have lower frictional resistance than softer metals owing to stronger
inter-atomic bonds. He postulated that the atomic bonds in harder metals are strong and hence the resistance to adhesion is
increased providing low frictional characteristics. Surfaces hardened by heat treatment were also characterized by low friction
owing to phase transformation and increase in surface energy and the induction of internal stresses. Figure 4.39 depicts the
marked decrease in coefficient of friction with increase in hardness for pure metals. Hardness has less but recognizable
influence on friction in case of heat-treated alloys, as shown in Fig. 4.40. The influence of hardness on friction was attributed
in two ways—firstly, strong inter-atomic bonds that minimize the plastic deformability of hard metals with a subsequent
decrease in the ability of metals to adhere, and secondly higher surface energy associated with metals of higher hardness
leading to formation of weak asperity contacts. This experimental evidence obeys theoretical prediction of coefficient of
friction as an inverse function of hardness of metals. The effect of hardness on friction is attributed to the lack of plastic
deformation of hard metals with subsequent decrease in the ability of metals to adhere resulting in low friction.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.39 Variation in friction coefficient with hardness for pure metals while sliding against
themselves and against steel [112]

Figure 4.40 Variation in friction coefficient with hardness for heat-treated beryllium bronze 1
[ 12]

It was also claimed by Mohtkar [113] that the correlations between the coefficient of friction and the mechanical and physical
properties of metals [114], which depend on inter-atomic bonds, and the periodic variation of all these phenomena with the
atomic number, allow the coefficient of friction to be accepted as an intrinsic property of metals.

4.11.5. Effect of Grain Size

The effect of grain size on frictional characteristics of many materials has been studied in combination with their wear
behavior. According to available results, it can be stated that friction coefficient decreases with decrease in grain size. One of
the important consequences of this is the improved mechanical and chemical properties of nanocrystalline materials. Thus
nanocrystalline materials are of great industrial importance, and their wide spread use in modern technology is quite evident,
for example in MEMS. These materials are used as bulk, as well as coatings to engineering substrates.

Bregliozzi et al. [115] in their experimental study of effect of atmospheric humidity and grain size on friction and wear of
austenitic steels measured friction coefficient using a tribometer in a ball-on-disk (BoD) configuration, whereby a steel ball
(AISI 52100) slides over the austenitic stainless steel with a constant linear speed of 10 cm/s and under a load of 2 N. The
effect of relative humidity and the grain size on the friction coefficient of AISI 304 are shown in Fig. 4.41. The fine grained
steel is found to have more wear resistance than the large grained steel. The hardness of the steel is the predominant factor:
the fine grained steel (with a higher hardness) shows a weight loss and a friction coefficient lower than the relatively softer
large grain steel.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.41 Effect of humidity and grain size on friction coefficient of austenitic steels [115]

Wang et al. [116] created a thin nanocrystallized (nc) layer by surface mechanical attrition method on a low carbon steel plate,
and studied the tribological characteristics of it. They found that the friction coefficient decreases and the wear resistance
increases with the nc surface layer. The improvement in friction and wear properties were attributed to the harder nc surface
layer, which reduces the degree of plowing and micro-cutting under the lower load, and alleviates plastic removal and surface
fatigue fracture under the higher load.

Farhat et al. [117] produced nanocrystalline aluminium samples using radio frequency magnetron sputtering technique to
study the wear characteristics as a function of grain size. They found that during early stages of sliding, the coefficient of
friction increased to a peak value followed by a decrease to a steady state value. The peak coefficient of friction decreased
from 1.4 for aluminum with a coarse grain size in order of 10 6 nm to 0.6 for nanocrrystalline aluminium with a grain size of
16.4 nm.

Similar results for nanocrystalline nickel over copper substrate were reported by Mishraet al. [118]. Nanocrystalline nickel
was electrodeposited using Watt's bath (NiSO4·6H2O 240 gm/l, NiCl2·6H2O 30 gm/l, H3BO3 30 gm/l). The wear tests were
performed at a normal load of 1N, an oscillation frequency of 8 Hz, tangential displacement amplitude of 100 μm and number
of fretting cycles was 10,000 for all samples. All the wear tests were performed against steel ball (SAE 52100 grade, 8 mm
diameter) without lubrication, at room temperature and in ambient air of approximately 40% relative humidity. The coefficient
of friction (COF) was recorded continuously during the wear tests. The coefficient of friction was determined as a function of
number of cycles. Figure 4.42 shows the variation of COF with number of cycles for different systems. In all the cases, it was
noticed that the COF increased from a low value to a very high value within the running-in period (500 cycles) and thereafter,
COF marginally decreased and attained a steady state value within the next 1000 cycles. By comparing the steady state COF
values, it is clear that extremely low COF (down to 0.16 for 8 nm grain size Ni) is obtainable with electrodeposited Ni coatings.
The coefficient of friction apparently decreases with decreasing grain size.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 4.42 Coefficient of friction for nanocrystalline Ni of grain sizes 8, 22 and 61 μm 1[ 18]

The grain size effect on friction, thus, is quite evident for metals and their alloys. Similar grain size effect on friction is
observed in ceramic materials as in metals and their alloys. A weak relation between grain size and coefficient of friction was
observed for alumina at elevated temperatures [119]. Zum Gahr et al. [120] studied the effect of the average grain size on
friction and wear of self-mated Al2O3 and 3 mol%Y2O3-ZrO2 in dry unidirectional and reciprocating sliding wear at ambient
conditions. The stationary values of the friction coefficient were measured after a relatively short running-in period of less
than about 1 km sliding distance. The average values of the friction coefficients were found to increase slightly with the
average grain size.

4.12. CONCLUDING REMARKS

The difficulties in the determination of the friction parameter lie in the complexity of the friction phenomena itself, and in the
inability to accurately measure shear stresses. Friction in metal forming processes comprises a very complex situation with
many parameters interacting with it, for example, normal pressure, relative displacement (velocity), surface roughness,
surface strain, lubricant, temperature, material (both tool- and work-material) properties, work hardening, surface coatings,
surrounding atmosphere, etc., all play a part. It is impossible to formulate the friction in a universal way because of the
complex interaction between many of these variables. Therefore, far-reaching approximations are used to describe friction
behavior during metal forming. These approximations deal with apparent friction rather than with the fundamental
phenomenon. One of the consequences of this approach is that friction must be measured separately for each forming
process.

For example, for the last two decades it is evident from industrial and domestic applications that Al extrusions are very much
in demand. Most of these Al extrusions are made in un-lubricated conditions. As expressed by Storen [121], 3D numerical
analysis shall be able to give good prediction of material flow, and however, an important question left unanswered is: how
can one model the friction condition and the formation of the surface of the extrudate in the bearing channel in un-lubricated
extrusion?

The trend in material forming technology is characterized by increasing demand for complex shapes, close tolerances (near-
net-shapes) and new materials such as powder metallurgy or refractory metals which have low ductility and demand a high
forming load. The computer aided design, analysis, and simulation have proved to be of great utility in providing valuable
information about metal forming processes. A trend for using numerical simulations in combination with physical

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 experiments to analyze friction in metal forming processes is being used today. However, the accuracy of numerical
simulation is strongly influenced by the tool–workpiece interfacial friction condition which is a major concern in metal forming
design and analysis. On the other hand, a universal test to measure friction in metal forming processes has not been and will
never be found [51]. There is a need for the use of different modeling techniques and/or a closer look at the parameters that
control the friction. There is a growing trend in this direction, for example, analysis of surface roughness and texture effect on
sliding friction [115–116].

One of the major concerns in manufacturing is the friction and wear of machines, tools, dies, and devices, for they control the
accuracy, life, power, and size of machines as well as the quality of the final products. Yet the field of tribology has relied
heavily on empiricism in solving technological problems, because until the recent decades scientific understanding has
usually followed technological development. As assessed by Suh [122], we now have a fair scientific understanding of the
first-order tribological effects, thanks to the significant scientific advances made during the past few decades. Nevertheless,
major new scientific breakthroughs are still needed for creating nearly frictionless mechanisms for nanotechnologies, high-
speed and high-stiffness spindles without ball bearings, and low-friction and non-wearing machines and dies. Authors stand-
by Suh who believes that such scientific breakthroughs can be brought forth through research, and there are scientific
indicators that support this optimism.

For example in the process of wire drawing, the independent parameters like reduction, die angle can be measured directly.
Not so for friction. There is no instrument called Friction-meter to measure friction. Friction is not directly measurable, nor is it
really an independent parameter. But, in many metal forming processes the effect of friction is as strong as that of reduction,
die geometry, etc., and thus the importance of friction parameter and its elusive characteristics. Friction, it seems, is the last
frontier in the study of metal forming! [123].

4.13. REFERENCES
1. Bowden, F. P. and Tabor, D. (eds.) (1950). The friction and lubrication of solids, Part-I, Clarendon Press, Oxford, UK.

2. (1969) Glossary of terms and definitions in the field of Friction, Wear and Lubrication: Tribology, OECD, Paris.

3. Bharat Bhushan (ed.) (2002). Introduction to Tribology; John Wiley & Sons, Inc.

4. Schnurmann, R. (1962). Friction and Wear, Wear 5; 31–42.

5. Tichy, J. A. and Meyer, D. M. (2000). Review of solid mechanics in tribology,International Journal of Solids and Structures,
37; 391–400.

6. Suh, N.P. and Sin, H.C. (1981). The genesis of friction,Wear 69 91–114.

7. Suh, N.P. and Sin, H.C. (1980). On the genesis of friction and its effect on wear,ASME Applied Mechanics Division, 39; 167–
183.

8. Kim, D.E. and Suh, N.P. (1991). On microscopic mechanisms of friction and wear, International Conference on Wear of
Materials, 1; 475–480.

9. Suh, N.P. and Sridharan, P. (1975). Relationship between the coefficient of friction and the wear rate of metals; Wear 34;
291–299.

10. Tohkai, M. (1979). Microstructural aspects of friction; M.S. Thesis, Department of Mechanical Engineering, Massachusetts
Institute of Technology, USA.

11. Rabinowicz, E. (1977). Proc. Int. Conf. on Wear of Materials, St. Louis, MO, American Society of Mechanical Engineers, New
York, pp. 36 – 40.

12. Wagoner, R. H. and Chenot, J. L. (eds.) (1996). Fundamentals of Metal Forming; John Wiley & Sons Inc.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
13. Geiger, M., Kleiner, M., Eckstein, R., Tiesler, N., and Engel, U. (2001). Microforming, Annals CIRP 50 (2); 445–462.

14. Tran, D.V., Lewis, R.W., Gethin, D.T., and Afriffin, A.K. (1993). Numerical modelling of powder compaction processes:
displacement based finite element method. Powder Metall, 36; 257–266.

15. Wanheim, T., Bay, N. (1978). A model for friction in metal forming processes, Ann. CIRP 27; 189.

16. Joun, M.S. (1999). Iterative and non-iterative schemes for coulomb frictional law in metal forming simulation and their
numerical characteristics, Commun Numer Methods Eng; 15 (7); 515–522.

17. Fedorciuc-Onisa, C. and Farrugia, D.C.J. (2003). Simulation of frictional conditions during long product hot rolling.The 6th
ESAFORM conference on material forming, pp. 763–766.

18. Sivaprasad, P.V. and Davies, C.H.J. (2005). An assessment of the interface friction factor using the geometry of upset
specimens, Model Simul Mater Sci Eng, 13; 355–60.

19. Fedorciuc-Onisa, C. and Farrugia, D.C.J. (2006). Through process characterisation of frictional conditions under lubrication
for long product hot rolling, Steel rolling 2006—9th international and 4th European conferences, Paris. June 19–21.

20. Fereshteh-Saniee, F. (2004). Friction modelling for the physical simulation of the bulk metal forming processes, J Mater
Proc Technol; 153–154;151–156.

21. Lebon, F. (2003). Contact problems with friction: models and simulations. Simul Model Pract Theory, 11; 229–463.

22. Lee, B.H. and Keum, Y.T. (2002). Wagoner RH Modeling of the friction caused by lubrication and surface roughness in
sheet metal forming. J Mater Process Technol 130–131; 60–63.

23. Ramezani, M., Ripin, Z.M. and Ahmad, R. (2010). Modelling of kinetic friction in V-bending of ultra-high-strength steel
sheets. Int J Adv Manuf Technol 46 (1–4); 101–110.

24. Ramezani, M., Ripin, Z.M. and Ahmad, R. (2009). A static friction model for tube bulge forming using a solid bulging
medium, Int J Adv Manuf Technol 43 (3–4); 238–247.

25. Cheng, H.S., Cao, J. and Xia, Z.C. (2007). An accelerated springback compensation method,Int J Mech Sci 49; 267–279.

26. Huang, Y.M. (2007). Finite element analysis on the V-die coining bend process of steel metal. Int J Adv Manuf Technol 34;
287–294.

27. Matuszak, A. (2000). Factors influencing friction in steel sheet forming. J Mater Process Technol 106; 250–253.

28. Leu, D.K. (2009). A simple dry friction model for metal forming process. J Mater Process Technol 209; 2361–2368.

29. Bay, N. and Wanheim, T. (1976). Real area of contact and friction stress at high pressure sliding contact;Wear, 38; 201–
209.

30. Petersen, S.B., Martins, P.A.F. and Bay, N. (1997). Friction in bulk metal forming: a general friction model vs. the law of
constant friction; Journal of Materials Processing Technology, 66; 186–194.

31. Avitzur, B. (1980). Metal forming—The application of limit analysis; Marcel Dekker Inc.

32. Rowe, G.W. (1977). Principles of industrial metalworking processes; Edward Arnold Publishers Ltd.

33. Joun, M.S., Moon, H.G., Choi, I.S., Lee, M.C., and Jun, B.Y. 2009. Effects of friction laws on metal forming processes,
Tribology International, 42 (2); 311–319.

34. Uyyuru, R.K. (2003). Engineering the friction and forming mechanics in cold Aluminium forming—Use of modeling
techniques, Ph.D. dissertation, Dept. of Machine Design and Materials Technology, Norwegian University of Science and
Technology, Trondheim (Norway), ISBN: 82-471-5584-2, ISSN: 0809-103x.

35. Lange, K. (1985). Handbook of Metal Forming; McGraw-Hill, New York.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
36. Evans, W. and Avitzur, B. (1968); Measurement of friction in drawing, extrusion, and rolling; Journal of Lubrication
Technology/Transactions of ASME, 90; 72–80.

37. Schey, J. A. (1983). Tribology in Metalworking: Friction, Lubrication, and Wear; American Society for Metals, Metals Park,
Ohio.

38. Ekelund, S. (1933). Steel, 93; 38.

39. Geleji, A. (1960). Metal Forming in computation and Experiment (Bildsame Formung der Metalle in Rechnung und
Versuch), Akademie Varleg, Berlin (in German).

40. Kunogi, M. (1956). New method of cold extrusion, Journal of Scientific Research Institute (Tokyo), 50; 215–246.

41. Male, T. and Cockcroft, M. G. (1964–65), A Method of the Determination of the Coefficient of Friction of Metals under
Conditions of Bulk Plastic Deformation; Journal of the Institute of Metals, 93; 38–46.

42. Avitzur, B. (ed.) (1968), Metal Forming: Processes and Analysis, McGraw-Hill, New York.

43. Kunogi, M. (1953). Reports/Scientific Research Institute (Tokyo) (in Japanese), 30; 63–93.

44. Kudo, H. (1955). An Analysis of Plastic Compressive Deformation of Lamella Between Rough Plates by Energy Method,
Proceedings of the 5th Japan National Congress for Applied Mechanics, Science Council of Japan /Tokyo, V; 75–78.

45. Mielnik, E. M. (ed.) (1991). Metalworking Science and Engineering, McGraw-Hill.

46. Bramley, A. N. and Abdul, A. (1975). Stress Strain curves from ring test,Proceedings of the 15th International Machine Tool
Design and Research Conference, Birmingham/Macmillan, London, 431–436.

47. Saul G., Male, A. T. and Depierre, V. (1971), Metal Forming: Interrelation between Theory and Practice, in Hoffmanner, A. L.
(ed.), Plenum Press, New York.

48. Bhattacharya, D. and Brown, R. H. (1980). Sensitivity of Ring-Compression test, Metalworking Lubrication/ASME, 23–30.

49. Rao, K.P. and Sivaram, K. (1993). A review of ring-compression testing and applicability of the calibration curves, Journal
of Materials Processing Technology, 37; 295–318.

50. Petersen, S.B., Martins, P.A.F. and Bay, N. (1998). An alternative ring-test geometry for the evaluation of friction under low
normal pressure, Journal of Materials Processing Technology, 79; 14–24.

51. Tan, X., Martins, P.A.F., Bay, N., and Zhang, W. (1998). Friction studies at different normal pressures with alternative ring-
compression tests, Journal of Materials Processing technology, 80; 292–297.

52. Rudkins, N.T., Hartley, P., Pillinger, I. and Petty, D. (1996). Friction modelling and experimental observations in hot ring
compression tests, Journal of Materials Processing Technology, 60; 349–353.

53. Luong, L. H. S. and Heijkoo, P. T. (1981). The influence of scale on friction in hot metal working, Wear, 71, 93 – 102.

54. Shen G., Vedhanayagam A., Kropp E. and Altan T. (1992). A method for evaluating friction using a backward extrusion-type
forging, Journal of Materials Processing Technology, 33; 109–123.

55. Uyyuru, R.K. and Valberg, H. (2006). Physical and numerical analysis of the metal flow over the punch head in Backward
Cup Extrusion of Aluminium, Journal of Materials Processing Technology, 172(2), 312–318.

56. Sofuoglu, H. and Gedikli, H. (2002). Determination of friction coefficient encountered in large deformation processes,
Tribology International, 35; 27–34.

57. Alexander, J.M. (1955). The effect of Coulomb friction in the Plane-Strain-Compression of a plastic-rigid material,Journal
of the Mechanics and Physics of Solids, 3, 233–245.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
58. Guminski, R.D. and Willis, J. (1959–60). Development of Cold-Rolling Lubricants for Aluminium Alloys, Journal of the
Institute of Metals, 88; 481–492.

59. Takahashi, H. and Alexander, J.M. (1961–62). Friction in Plane-Strain Compression test; Journal of the Institute of Metals,
90; 72–79.

60. Bailey, J.A., and Singer, A.R.E. (1963–64). The determination of the coefficient of friction at elevated temperatures using a
Plane-Strain Compression test, Journal of the Institute of Metals, 92; 378–380.

61. Cook, M., and Larke, E.C. (1945). Resistance of Copper and Copper alloys to homogeneous deformation in compression,
Journal of Institute of Metals, 71; 371–390.

62. Mizuno T., Kamiya, S. and Okamoto, M. (1978). Lubricant Behaviours in the Compression– Friction Test of Strip Metals,
Bulletin of the JSME, 21; 1424–1430.

63. Lonn, A.H., and Schey, J.A. (1974). Development of the plane-strain compression test, in North American Metalworking
Research Conference-II (NAMRC-II), Dearborn-Michigan pp. 165–178.

64. Becker, N. (1994). Further development of methods for accusation of flow curves in regions of large deformations, PhD
Thesis, no. 122, University of Stuttgart/Springer–Verlag (Germany), (in German).

65. Wang, S.L., and Ramaekers, J.A.H. (1996). Measurement of friction and material flow-stress by a plane strain compression
tribometer, Journal of Materials Processing Technology, 57; 345–350.

66. Gurrero Mata, M.P. (1996). Finite element modelling of hot plane strain compression testing, PhD Thesis, University of
Sheffield / UK, .

67. Gelin, J.C., Ghouatio, O. and Shahani, R. (1994). Modelling the plane strain compression test to obtain constitutive
equations of Aluminium alloys, International Journal of Mechanical Sciences, 36; 773–796.

68. Slik, N.J. and van Der Winden, M.R. (1999). Interpretation of hot plane strain compression testing of Aluminium specimens,
Materials Science and Technology, 15; 295–300.

69. Dieter, G.E. (ed.) (1988). Mechanical Metallurgy, McGraw-Hill, Singapore.

70. Valberg, H. (2001). The non-homogeneous nature of plastic deformation in cylinder- and plane strain- compression
investigated by FEM-analysis with experiment; OPROTEH-2001 (TSTM7-Romanian Academy-2001), Bacau, Romania, pp.
334–352.

71. Uyyuru, R.K., and Valberg, H. (2005). Friction induced material flow to characterize lubricating conditions in Plane Strain
Compression Test, International Journal of Forming Processes, 8; 77–115.

72. Whitton, P.W. and Ford, H. (1955). Surface friction and lubrication in cold strip rolling Proceedings of Institution of
Mechanical Engineers, 169; 123–140.

73. Pavlov, I.M. (ed.) (1938). Theory of Rolling and Fundamentals of Metalworking, Gonti, Moscow, (Russia).

74. Sanchez, L.R., Jeinmann, K. and Story, J.M. (1985). Proceedings of the.13th North American Manufacturing Research
Conference, Society of Manufacturing Engineers, 110–117.

75. Popilek, M. (1991). A friction test for extrusion using combined flow with emphasis on backward extrusion, M.Sc. Thesis,
Department of Mechanical Engineering, Michigan Technological University, Houghton, MI,.

76. Buschhausen, A., Weinmann, K.L., Lee Y. and Altan T. (1992). Evaluation of lubrication and friction incold forging using a
double backward-extrusion process, Journal of Materials Processing Technology, 33; 95–108.

77. Tan, X., Bay, N. and Zhang, W. (2003). Friction measurement and modeling in forward rod extrusion tests,Journal of
Engineering Tribology (Proceedings of Institute of Mechanical Engineers), 217; 71–82.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
78. Ebrahimi, R. and Najafizadeh, A. (2004). A new method for evaluation of friction in bulk metal forming; Journal of Materials
Processing Technology, 152; 136–143.

79. Druker, D.C. and Providence R.I. (1954). Coulomb friction,Journal of Applied Mechanics, 21; 71–74.

80. Im, Y.T., Cheon, J.S. and Kang, S.H. (2002). Determination of friction condition by geometrical measurement of backward
extruded aluminium alloy specimen, Journal of Manufacturing Science and Engineering (Transactions of ASME), 124; 409–
415.

81. Le, H.R. and Sutcliffe, M.P.F. (2002). Measurement of friction in strip drawing under thin film lubrication, Tribology
International, 35; 123–128.

82. Lee, C.D, Weng, C.I. and Chang, J.G. (2001). A prediction of the friction factor for the forging process; Metallurgical and
Materials transactions, 32 B; 137–143.

83. Kaftanoglu, B. (1973). Determination of coefficient of friction under conditions of deep drawing and stretch forming;Wear,
25; 177–188.

84. Duncan, J.L., Shabel, B.S., Filho, J.G. (1978). A tensile strip test for evaluating friction in sheet metal forming;SAE
Technical Paper, No. 780391, 8.

85. Hao, S., Klamecki, B.E., and Ramalingam, S. (1999). Friction measurement apparatus for sheet metal forming, Wear, 224;
1–7.

86. Matuszak, A. (2000); Determination of frictional properties of coated steel sheets,Journal of Materials Processing
Technology, 106; 107–111.

87. Wilson, W.R.D., Malkani, H.G. and Saha, P.K. (1991). Boundary friction measurements using a new sheet metal forming
simulator, Proc. NAMRC XIX, SME; 37–42.

88. Saha, P.K., Wilson, W.R.D., Timsit, R.S. (1996). Influence of surface topography on the frictional characteristics of 3104
aluminium alloy sheet, Wear, 197; 123–129.

89. Vollertsen, F. and Plancak, M. (2002). On possibilities for the determination of the coefficient of friction in hydroforming of
tubes, Journal of Materials Processing Technology, 125–126; 412–420.

90. Nielsen, M. (ed.) (2000); in Proceedings of Symposium on New Coatings for Hydroforming Tooling, Next Generation Solutions,
Nashville, USA, 10.

91. Bello, D. O. and Walton, S. (1987). Surface Topography and Lubrication in Sheet–Metal Forming; Tribology International, 20;
59–65.

92. Kudo, H. (1965). A note on the role of microscopically trapped lubricant at the tool–work interface; International Journal of
Mechanical Sciences, 7; 383–388.

93. Nelleman, T., Bay, N., Wanheim, T. (1977). Real area of contact and friction stress—The role of trapped lubricant,Wear, 43;
45.

94. Schey, J.A., Mysliwy, R.E. The effect of die surface finish on lubrication in ring upsetting; Euro Trib'77, 2nd European
Tribology Congress, Paper 67.

95. Geiger, M., Engel, U., Niederkom, S., Pfestorf, M. (1995). Experimental investigation of contact phenomena in cold forging,
ICFG plenary meeting, Denmark.

96. Lakshmipathy, R. and Sagar, R. (1992). Effect of die surface topography on die-work interfacial friction in open die forging,
International Journal of Machine Tools and Manufacture, 32; 685–693.

97. Hu, Z. M. and Dean, T. A. (2000). A study of surface topography, friction and lubricants in metal forming, International
Journal of Machine Tools and Manufacture, 40; 1637–1649.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 98. Rasp, W. and Wichern, C.M. (2002). Effects of surface-topography directionality and lubrication condition on frictional
behaviour during plastic deformation; Journal of Materials Processing Technology, 125–126; 379–386.

99. Menezes, P. L., Kishore, and Kailas, S.V. (2006). Studies on friction and transfer layer using inclined scratch, Tribology
International, 39(2); 175–183.

100. Menezes, P.L., Kishore and Kailas, S.V. (2005). Effect of directionality of grinding marks on friction at different surface
roughness using inclined scratch test, to be presented at World Tribology Congress III (WTC2005), Washington DC.

101. Buckley, D.H. and Johnson, R.L. (1968). The influence of crystal structure and some properties on hexagonal metals on
friction and adhesion; Wear,11; 405–419.

102. Kuhlmann-Wilsdrof, D. (1980); Dislocation concepts in friction and wear, in Rigney D.A. (ed.), Fundamentals of friction and
wear of materials, ASM, Metals Park, Ohio, 119.

103. Ernst, H. and Merchant, M.E. (eds.) (1940). Chip formation, Friction and Finish, Cincinnati Milling Machine Co.

104. Alison, P. J. and Wilman, H. (1964). The different behaviour of hexagonal and cubic metals in their friction, wear and work
hardening during abrasion, British Journal of Applied Physics, 15; 281–289.

105. Karapetian, S.S. and Korostelin, Y.I. (1983). The dependence of the friction coefficient on the crystal structure of solids:
Physical principles, Wear, 85; 133–141.

106. Wheeler, D.R. and Buckley, D.H. (1975). Texturing in metals as a result of sliding,Wear, 3365–74.

107. Farhat, Z.N. (2001). Contribution of crystallographic texturing to the sliding friction behaviour of fcc and hcp metals,Wear,
250; 401–408.

108. Richardson, R.C.D. (1967). The maximum hardness of strained surfaces and the abrasive wear of metals and alloys, Wear,
10; 353–382.

109. Gaivoronskii, A.T. (Acad. of Sciences of the USSR); Prokop'ev, G.A.; Pin'kovskii, V.P. (1988). Influence of material hardness
and strain hardening on friction stress, Soviet Journal of Friction and Wear (English translation of Trenie i Iznos), 9; 109–111.

110. Lin, D.S. (1969). Effect of the degree of work hardening on the friction and wear of metals during abrasion,Wear, 13; 91–
97.

111. Reid, J.V. and Schey, J.A. (1987). The effect of surface hardness on friction, Wear, 118; 113–125.

112. Mokhtar, M.O.A. (1982). The effect of hardness on the frictional behaviour of metals, Wear, 78; 297–304.

113. Mokhtar, M.O.A. (1981). Friction: Is it intrinsic property of metals? Wear, 72; 287–293.

114. Mokhtar, M.O.A., Zaki, M. and, Shawki, G.S.A. (1979). Effect of mechanical properties on frictional behaviour of metals,
Tribology International, 12; 265–268.

115. Bregliozzi, G., Di Schino, A., Kenny, J.M., and Haefke H. (2003). The influence of atmospheric humidity and grain size on
the friction and wear of AISI 304 austenitic stainless steel, Materials letters, 57; 4505–4508.

116. Wang, Z.B., Taao, N.R., Li, S., Wang, W., Liu, G., Lu, J., and Lu, K. (2003). Effect of surface nanocrystallization on friction and
wear properties in low carbon steel, Materials Science and Engineering A352: 144–149.

117. Farhat, Z.N., Ding, Y., Northwood, D.O., and Alpas, A.T. (1996). Effect of grain size on friction and wear of nanocrystalline
aluminum, Materials Science and Engineering, A206; 302–313.

118. Mishra, R., Basu, B., and Balasubramaniam, R. (2004). Effect of grain size on the tribological behavior of nanocrystalline
nickel, Materials Science and Engineering; A373; 370–373.

119. Senda, T., Yasuda, E., Kaji, M., and Bradt, R.C. (1999). Effect of grain size on the sliding wear and friction of alumina at
elevated temperatures; Journal of the American Ceramic Society 82;1505–1511.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
120. Zum Gahr, K.-H., Bundschuh, W., and Zimmerlin, B. (1993). Effect of grain size on friction and sliding wear of oxide
ceramics, Wear, 162–164; 269–279.

121. Storen, S. (1993). The theory of extrusion—Advances and challenges, International Journal of Mechanical sciences, 35;
1007–1020.

122. Suh, N.P. (1991). Tribology in manufacturing: A retrospective and prospective assessment, ASME Production Engineering
Division (Tribological Aspects in Manufacturing), 54; 61–75.

123. Avitzur, B. (1990). A personal look on my involvement in the recent history of metal forming research, A keynote
presentation at Advanced Technology of Plasticity, 4;1990.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
5. Metal Extrusion
Santosh Kumar

Department of Mechanical Engineering, IT-BHU Varanasi 221 005, India

5.1. INTRODUCTION
Extrusion is a near-net shape chipless manufacturing process, of forcing the work material to flow through a die opening of
desired shape through a single-hole or a multiple-hole die to obtain the final shape of a component. Forward extrusion causes
the product to be in the same direction as that of the ram/billet, whereas, in backward extrusion, the product comes out in
opposite direction of the room/billet. Forward extrusion is very common for all kind of extrusions, while cold backward
extrusion is popular to produce aluminum (compressed) gas cylinders. Dies of appropriate shape and sizes are required to
impart the right shape and specified tolerance in their respective passes. A welding chamber die (Porthole, bridge, spider,
etc.) is used under hot condition for common extrusion of hollow components. Square dies are widely used for multi-hole
extrusion under hot condition. Continuous dies such as streamlined, conical, elliptical, hyperbolic, and cosine dies are
becoming important and popular for manufacture of better mechanical and metallurgical property products at lower cost.
Extrusion through multiple-hole dies, using a high capacity extrusion press is preferred for reasons of economy and high
productivity. Some of the common solid and hollow extrudable metallic sections are shown in Fig. 5.1 [1].

Figure 5.1 Common extrudable sections. With permission from [1]

Cold and hot extrusion may be defined in terms of working temperature (T) and metal melting point temperature (T m). Cold
Extrusion (T/T m < 0.3) refers to extrusion that occurs usually, but not necessarily, at room temperature. For example, extruding
lead at room temperature is a hot working process because the recrystallization temperature of lead is below room
temperature (20°C). Hot Extrusion (T/T m > 0.6) refers to extrusion carried out above the recrystallization temperature. Warm
extrusion (T/T m in range of 0.3–0.6) is carried out at intermediate temperatures. It is a compromise between cold and hot
extrusion where T is the working temperature and T m is the melting point of the metal, both on the absolute scale homologous
temperature.

Extrudability (W) indicates the ease with which a metal can be deformed and the degree of deformation it can withstand. It is a
ratio of cold ductility to flow stress in MPa (average of upper and lower field point). The two parameter of W are very difficult
to measure in extrusion, and therefore, W is determined by laboratory studies, the results of which cannot replace extrusion
trials but merely complement them. The values of W for Aluminum alloy may be classified as a practical guide: < 2 (poor), 2–4
(average), 4–15 (good) and >15 (very good) extrudability. The extrudability of many metals, including aluminum and

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
magnesium alloys, nickel alloys and high-strength steels, falls between narrow limits and extrudability in individual cases is
assessed according to various parameters like: extrusion temperature, permitted temperature range, load or specific pressure
required, exit speed at a constant extrusion load, maximum extrusion speed (to the onset of hot shortness), and maximum
extrusion ratio, etc.

Deformation using cold working results in higher stiffness and strength, but reduces malleability and ductility of the metal. In
hot working, annealing takes place while the metal is worked and therefore is worked without getting work hardened. At high
temperature, scaling and oxidation exists which generates undesirable surface finish. Most ferrous metals need to be cold
worked after hot working in order to improve the surface finish. The amount of force needed to perform hot working is less
than that for cold work. The mechanical properties of the material remain unchanged during hot working. The metal usually
experiences a decrease in yield strength when hot worked. Therefore, it is possible to hot work the metal without causing any
fracture. Quenching is the sudden immersion of a heated metal into cold water or oil. It is used to make the metal very hard.
To reverse the effects of quenching, tempering is used (reheating of the metal for a period of time). Warm working depends on
how close is the warm process to be to a cold or hot process. This means that the effects and characteristics get similar to
the ones in a cold process when the warm process gets closer to 0.3. This also applies for hot processes. The effects and
characteristics get similar to the ones in a hot process when the warm process gets closer to 0.5.

The extrusion process can produce simple as well as intricate components of different shapes and sizes with good accuracy,
surface finish, mechanical as well as metallurgical properties at a lower unit cost. It is also recognized for its capability to
produce even those components which are being fabricated by welding, casting, machining, etc. Recent technological
developments have led the extrusion process to manufacture various types of domestic, consumer, aircraft, medical,
engineering, and even food products.

Until recently both single and multiple-hole extrusion dies of complicated shapes are being designed on the basis of expertise
gained through experience. Before designing such dies the designer has to identify the layout of component shapes that could
be extruded simultaneously using the available press capacity. Non-symmetric component shapes both re-entry and non re-
entry types, along with their locations are usually identified in such a way that the output speeds of each extruded component
are the same. Even with a properly designed and manufactured die, several trial runs are necessary before the desired
components, free from defects such as central burst, fir-tree cracking, roller mark, die spot, broken die, blisters, dross, speed
cracks, water stains, die line, weld line, chatter mark, Orange peal, oil stains, graphite mark, joint inside crack, bulging and off
flatness are obtained. Controlling all these defects in extrusion products is very difficult.

Trial and correction process is adopted in industries to control it completely. The process of trial and correction is tedious and
expensive. These difficulties have motivated researchers and practitioners to look for an automated design and
manufacturing practice using existing or new modeling techniques, optimization procedures, rules and algorithms, etc. with
the help of computers. With the development of highly reliable and accurate FEM based plastic deformation techniques, it has
now become possible commercially to model almost any metal forming process including extrusion. Availability of huge
information in handbooks, online journals and material data websites, and CAD based large scale cheap computational
facilities, it is now possible to design, optimize, and finalize the extrusion die design and process study for any complex
profile with the help of CAE activity.

5.2. AN OVERVIEW OF ANALYTICAL MODELING

TECHNIQUES FOR EXTRUSION
Modeling of extrusion process is required for designing the process parameters of extrusion and related tooling. It also helps
to decide the tooling materials and process condition of extrusion process at desired output rate. Common modeling
techniques are: Slab method, Slip line method, Upper bound method and Finite element method. Slab method has been used
for axi-symmetric simple cases, whereas slip line method has been used for plane strain situations in 2D case. Extrusion of 3D

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
cases has also been solved by slip line method but it becomes very tedious process. Upper bound method has been used in
many ways for extrusion.

The slip-line field method is used to obtain the stress field in forming problem by constructing slip lines. Originally the method
was proposed for plane strain analysis of rigid and perfectly plastic materials. However, lately there have been some attempts
to extend the method to axisymmetric problems as well as to incorporate strain hardening.

Many researchers have analyzed the extrusion process by the slip-line field method for axisymmetric extrusion. Extension of
slip-line method for non-axisymmetric problems is not possible mainly because the construction of slip lines in 3D space has
not yet been established. It, therefore, appears that this method is limited only for axisymmetric extrusion.

5.2.1. Upper Bound Analysis for Solid Sections

The upper bound method is a technique for determining an estimate of the load which generally results into an over estimate,
obtained from an assumed kinematically admissible velocity field that satisfy the incompressibility conditions and velocity
boundary conditions. An extensive literature exists on upper bound analysis of metal forming processes and has been
reviewed by many researchers. The plane strain and axisymmetric extrusion processes have been analyzed by many
researchers. Many upper bound solutions are available for three-dimensional extrusion of solid sections as well as tubular
hollow components. Lubricated cold extrusion of sections through continuous dies has been widely used for the manufacture
of high strength products with good surface finish, minimum defects, and high productivity at lower unit cost.

Two kind of extrusion dies commonly used are continuous and square. Square dies are used in hot extrusion, whereas
continuous dies are used for cold extrusion. Continuous dies may be divided into two categories. First category consists of
streamlined continuous dies, e.g., third order, fourth order, fifth order polynomial dies and cosine dies, whereas second
category encompass conical, hyperbolic, elliptical, Blazynasky's CRHS die (constancy of the ratio of successive generalized
homogeneous strain increments), and modified CRHS dies etc.

CRHS dies are constructed in such a fashion that it minimizes the internal power of deformation (IPD) but there are power
losses at the entry as well as exit sections. As a result, the total power consumption of CRHS die is almost as high as that for
conical dies. In the modified CRHS dies, the discontinuities at the end sections are removed (i.e., it is made streamlined) while
retaining the advantage of minimum IPD. Even then the total power consumption of the modified CRHS is more than the third,
fourth order, etc., polynomial dies and cosine dies. Therefore, the concept of CRHS does not seem to be useful in minimizing
the total power consumption.

It has been found that the first category dies require comparatively less amount of power for extrusion; and the extruded
products, due to uniform deformation, have absence of intensive shear bands resulting in better microstructure and
mechanical properties. However, second category dies are also preferred as a conventional practice. Despite the increased
demand for 3D extrusion of generalized shaped sections through continuous dies, so far very little systematic work has been
carried out for the determination of forming loads and flow characteristic of generalized shape, taken into consideration
simultaneously, in order to design the extrusion process.

5.2.2. Constitutive Equations

The system of equations describing the behavior of a solid under various types of response is referred to as the constitutive
equations. In steady-state forming processes such as extrusion, drawing, rolling, etc., the measure of deformation is the strain
rate tensor , which is expressed as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.1)

where Vi and Vj represent the velocity components along xi and xj direction, respectively, p is the hydrostatic part, δij is
Kronecker's delta and σ′ij is the deviatoric part of the total stress.

In bulk metal forming processes such as extrusion, drawing, rolling, etc., the total strains are large as compared to the elastic
strains. Therefore, elastic deformation can be considered as negligible compared to plastic/visco-plastic deformation. The
constitutive law for rigid plastic–plastic material relating the deviatoric stress tensor σ′ij and the strain rate is expressed as

σ′ ij = 2μ . For a material yielding according to von-Mises criteria, the Levy-Mises coefficient μ is given by , where

generalized yield stress and the generalized strain rate . The generalized strain is therefore,

defined as . The integration is to be carried along the particle path.

In general depends on and temperature T. Since cold extrusion is carried out at room temperature, effect of
temperature on can be neglected, i.e., . The specific functional form of Γ for the material under
consideration is mentioned at respective places.

5.2.3. Upper Bound Formulations

Determination of extrusion power is very important for die design. Normally, the upper bound method provides an over
estimate of extrusion power. However, the upper bound model proposed here assumes a kinematically admissible velocity
field which is quite close to actual velocity field. Another distinguishing characteristic of the proposed model is that the
kinematically admissible field is expressed in terms of Cartesian coordinate system of streamlines which makes it easy to
incorporate variable yield stress. The proposed model is appropriate for generalized non re-entry shape extrusion through a
fairly general class of die profiles.

An upper bound theorem is required to satisfy only the kinematic conditions in terms of strain increments, strain rate and
velocities in a plastically deforming medium and does not necessarily satisfy the stress equilibrium equations. An important
concept involved is that of a kinematically admissible velocity field. Velocity fields that satisfy the constraint of volume
constancy and the velocity boundary conditions are called kinematically admissible velocity fields. A kinematically admissible
velocity field may have discontinuities in the tangential component along certain surfaces, but the normal component must be
the same on both sides of such surfaces in order to satisfy the constraint of incompressibility. The unknown parameters in
kinematically admissible velocity field are determined using the upper bound solution. The upper bound states that among all
possible kinematically admissible velocity fields, the one which minimizes the total power field represents true velocity field.

5.2.4. Choice of Deformation Zone and Velocity Boundary Conditions

Since the upper bound analysis does not satisfy stress equilibrium, the prediction of complex shapes of plastic boundariesS 1
and S 2 as shown in Fig. 5.2 are never accurate. The upper bound solution obtained by earlier researchers [8,10] using straight
and arbitrarily shaped plastic boundaries indicate that there is little effect of the shapes of surfaces S 1 and S 2 on the overall
solution. Hence, in the present work, the deformation zone Ω is assumed to be bounded by straight plastic boundaries at the

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
end sections of the die. Material is assumed to be rigid outside the entry and exit sections of the die. Therefore, the axial
velocity at the entry and exit sections of the die should be uniform. These conditions are given by

(5.2)

Figure 5.2 Deformation zone and typical stream surfaces in extrusion. With permission from [3]
publisher

At cross-section S 2, each point has different velocities. A point at the center has maximum velocity and a point on the
periphery has minimum velocity. Corresponding to N different points, the common velocity of extrudate Vave is defined as

There should not be any metal flow across boundary S 3 and the axis of symmetry. This condition on these boundaries can be
expressed as

(5.3)

on die surface S 3 and the axis of symmetry.

5.2.5. Proposed Kinematically Admissible Velocity and Strain Rate

Fields
Since deformation is under certain assumptions and streamlines are the basic flow path during the deformation process, it is
necessary to build the proposed kinematically admissible velocity using the concept of streamlines. The geometry of die and
stream lines in the steady flow process of extrusion is shown in Fig. 5.3. One power element OEGG′O′E′EO is chosen to
demonstrate the analytical construction of velocity and strain rate fields. In Fig. 5.2 the billet with an initial radius (R0) at entry,
is extruded through a shaped die (continuous) constructed by a number of pre-defined streamlines representing the surface
of die, to the final generalized given section at the exit. An arbitrary point E on the die surface at the entrance to the die can be
combined with the corresponding point E′ on the die surface at the exit. It is assumed that the material of the billet passing
through sector OEG at the entrance goes through the triangle O′E′G′ at exit, preserving the extrusion ratio. Then the surface
defined by the points O, E, E′, and O′ becomes a three-dimensional stream surface.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.3 Geometry of die and stream lines in an extrusion die. With permission from [3] publisher

The following assumptions are made throughout the analysis described in this work:

i. The material is incompressible, rigid, perfectly-plastic and follows a particular strain hardening curve.

ii. The deformation takes place under homogeneous and steady state conditions.

iii. The deformation zone is assumed to be bounded by straight plastic boundaries at the end sections of the die.

iv. The neutral line is the line joining CG of billet (i.e., point O) and C.G. of the extruded product cross-section (i.e., point O′).

v. The stream surface OEE′O′ consists of a number of predefined streamlines, straight or curved.

vi. The particle passing through an arbitrary point F (Fig. 5.3) on line OE goes towards the corresponding point F′ on line O′E′
maintaining the proportionality of position.

vii. The elastic strain is small and can be neglected.

viii. In the present analysis, friction factor between the die and workpiece material is assumed to be independent of slip.

Using assumption (vi), the extrusion ratio is constant for any element, i.e.,

(5.4)

Let length O′G′ = a1 and O′E′ = a2. If the coordinates of points G′ and E′ are (x1, y1) and (x2, y2), respectively, and Aa is the cross-

sectional area of the extruded product shape, then and Angle ψ on the product

side is found by cos ψ = b1, where , b = (x1x2 + y1y2), and .

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Let streamlines on the stream surface be represented by a fourth order polynomial to satisfy the smooth entry and exit of the
material flow. Any coordinate on the stream line is therefore, defined as

(5.5)

where bi and c i (for i = 1, 2, 3, 4, and 5) are constants determined by the boundary conditions. Since the streamline does not
produce any abrupt change in flow direction along the extrusion axis at entry and exit, the boundary conditions can be written
as

(5.6)

(5.7)

where Ro is the billet radius, n is the distance from the axis formed by joining the CG of the product and the billet shape (i.e., O-
O′) to an arbitrary point F at the die entry, ϕ and ψ are the angles between the plane of symmetry and the stream surface at
entry and exit of the die, respectively, and L is the length of the die. Defining Constant, C

(5.8)

Substituting these boundary conditions from Eqs. (5.6) and (5.7) into Eq. (5.5) gives,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.9)

Equation (5.9) can be rewritten in the following form

(5.10)

(5.11)

In the present analysis, function f is represented by the following fourth order curve.

(5.12)

Equation (5.12) describes not only the coordinates inside the plastically deforming region but the relationship between the
Cartesian and n, ϕ, z coordinate systems also. Although the present analysis employs a fourth order curve represented by Eq.
(5.12) for the description of die profile and the assumed streamlines of particles, it is to be noted that functionf in Eq. (5.12)
can be any general function of z provided the function satisfies the given boundary conditions as given in Eq. (5.11).

The Jacobean of Eq. (5.10) can be found as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.13)

Determinant of the Jacobean is therefore, written as det J = – n . g (ϕ, z), where

Assuming that the plastic zone is bounded by entry and exit shear surfaces, the velocity field components can be obtained.
Because of volume continuity, the velocity component along z-direction in the Cartesian coordinate system (Vz) should be Vo
at the entrance, or Vf at the exit of die. Vo and Vf are the speeds of the billet and the outgoing product, respectively. Vf can be
described in terms of Vo as

These requirements are satisfied using Eqs (5.11) and (5.12). Therefore, other velocity components for incompressible
material are determined as [1]:

(5.14)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Using the velocity fields given by the above equations, it is found that the velocity boundary conditions (Eq. 5.11) are satisfied.
Different strain rate components have been found in Appendix [5]. It is found analytically that the proposed velocity and strain
rate satisfy the incompressibility condition, i.e., . Hence the proposed velocity field model
fulfills the stringent requirement for the construction of a kinematically admissible condition.

5.2.6. Strain Hardening

The generalized yield stress at any point in the deformation zone is calculated using the value of generalized strain at
that point (section 5.2.2). The value of in the deformation zone is obtained by integrating the generalized strain rate along

the streamline. The generalized strain rate is obtained from the proposed strain rate field along any streamline, ,

the generalized strain can be written as Since the material enters the die as a rigid body, the generalized

strain is taken as zero on the entrance boundary of the die. Integration of is performed using ten point Gauss Quadrature
rule to find the cumulative generalized strain along each streamline.

5.2.7. Estimation of Extrusion Power and Average Ram Pressure

From the proposed velocity and strain rate fields and the calculated field, the extrusion power and the average ram pressure
are determined.

The total power consumed inside the die is the sum of total power consumed within different power elements. One such
element shown in Fig. 5.3 is OEGG′O′E′EO. Total power consumed within the element is the sum of power losses due to plastic
deformation (ϕi), the velocity discontinuities at entry (ϕe) and outlet (ϕo ) of the die, and the friction power loss along the
interface between the material and the die (ϕf). The predicted total power obtained through the present velocity field would be
higher (if no other redundant power losses are present) than the power actually consumed. Each power in the power element
is computed numerically using velocity, strain rate components, the generalized yield stress of the material and the given
friction condition. The volume and surface integration are carried out numerically using ten point Gauss quadrature rule after
extracting the necessary data from the geometry of die profile and the cross-section of the given component to be extruded.

Let and be generalized yield strain and generalized yield stress for the given strain hardening material without
consideration of the redundant work factor, and let σo be the yield stress of the given material without considering strain
hardening effects. Then, various powers required for the calculation of total power (ϕT) are calculated as shown below.

a. Internal power of deformation : By considering an element of volume dv in the deformation zone subjected to a stress
system σij which causes strain rates , the incremental power of deformation (dϕi) can be expressed as

(5.15)

The total power of deformation can be obtained by integrating dϕi over the entire volume of the deformation zone. Thus,

(5.16)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 The above equation in expanded form can be written as

(5.17)

b. Frictional power : If the frictional resistance of the material along the total-work interface is τ and the slip velocity (or, the
tangential velocity discontinuity) along the interface is ΔVt, then the incremental frictional power at the interface can be
expressed as

(5.18)

where ds is an elemental surface area. The total frictional power can be obtained by integrating the above equation along
the total interface length. Thus,

(5.19)

Here s is the area of the die–workpiece interface.

In the present work, the frictional resistance on the interface is assumed to be a constant times the yield shear stress of
the material i.e., , where m is the friction factor whose value is chosen on the basis of die–

workpiece interface and the lubrication conditions. The friction factor varies from zero (for frictionless condition) to unity
(for sticking friction condition). In the present analysis friction factor is assumed to be independent of slip.

Figure 5.4 shows the relationship between the die surface and the projected surface for frictional power calculations. A
small unit square element s 1 on the die surface is selected. Let α be the maximum angle of inclination of the element (s 1)
of the die surface with respect to the projected surface on the x-z plane. This angle (α) is obtained from the geometry of
the die by knowing angle γ produced by the direction of a streamline (say EE′) and angle δ created by the position of this
streamline. From Eq. (5.10), at the die surface,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (5.20)

(5.21)

(5.22)

Therefore,

(5.23)

Figure 5.4 Die surface and projected surface for frictional power calculations. With permission from
[3] publisher

The length of diagonal (ll) of the small unit square element on the die surface is found as
. Angles γ and δ can be found from Eqs (5.20) and (5.23). Therefore, angle α, which is

subtended by ll with the x-y plane is found as

(5.24)

Based on Eq. (5.19), frictional power for the considered element is obtained from the above as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (5.25)

c. Shear power : In case the die profile has non-zero slope at the entry and exit sections, there will be shear losses due to
velocity discontinuity. These planes of discontinuity can be called as the planes of sudden shear. The power dissipated
along these surfaces can be calculated by employing the yield shear stress as the material resistance. Thus, the shear
power can be expressed as

(5.26)

where ΔVt is the tangential velocity discontinuity along the ds and s is the area over which integration has to be carried out.
Using Eq. (5.26) the shear power at entry (ϕe) and exit (ϕo ) side are obtained as:

(5.27)

(5.28)

d. Power due to interface shear

Here Ai is the area of the ith interface formed by the coordinate points (0, 0, L), (xi, yi , L), (Ro sin 0°, Ro cos 0°, 0) and (0, 0, 0).
The velocity discontinuity (Δvi) developed at the ith interface due to relative velocity of ith and (i–1)th elements is evaluated
as Δvi = (vft – vf – 1) where vft and vft – 1 are average tangential velocity of ith and (i–1)th elements. For ith element, vft is

evaluated as . Therefore, for n power elements, the total interface power is (ϕint-tot) is

e. Total power and average extrusion ram pressure : The total power (ϕT) is the sum of its constituents including the
redundant powers. Therefore, the total power (ϕT) is given as

(5.29)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Once the total power (ϕT) consumed during extrusion is obtained, the upper limit to the average ram pressure (Pave) for
extrusion is found as

(5.30)

The average ram pressure (Pave) can be converted to relative forming stress (Rs) as

(5.31)

where σo is the effective stress at zero hardening condition.

5.2.8. Upper Bound Anlaysis Applications

The above analysis is generalized in nature and can be used for any shape (re-entry and non re-entry) extrusion for calculation
of load/power requirement and can be used for design of extrusion dies and processes under forward extrusion. These are
explained below.

5.2.8.1. Application to non re-entry component shapes

Application of the proposed upper bound model is demonstrated for a general shape given inFig. 5.4. The following steps are
carried out for any given shape before going for the proposed upper bound solutions of power and other process parameters.

1. The Geometry of the component to be extruded is drawn using AutoCAD, Release-12. DXF file is created and processed to
separate the data of line and arc entities [3].

2. The center of gravity (xc, yc) and cross-sectional area (Aa) of the given component is found using Advanced Modelling
Extension (AME) of AutoCAD, Release-12.

3. Each arc and line entity is further divided into smaller lines to decide the total number of minimum triangular power
elements, using the pseudo vertex concept. The total number of power elements corresponding to each line and arc
entities are decided judicially such that the average error in incompressibility condition required in the upper bound
solution does not exceed one percent for each power element and thus incompressibility condition during
extrusion is satisfied.

4. From step 3, coordinates of each triangular power elements are obtained. For the example given inFig. 5.3, the coordinate
points of power element one (1) are O, E1, E2 (not required to be known) on the billet side and on the
extruded product side. Since coordinates of are now known from steps 2 and 3, angle ψ1 for the power
element is calculated using the relationships obtained from Eq. (5.4). Since percentage reduction (red) per pass is given,

the billet radius Ro is calculated as .

5. From the assumption that the extrusion ratio in each power element is maintained constant, i.e.,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
6. The corresponding angle (ϕ1) on billet side is also found using Eq. (5.4). Remaining (n′ – 1) included angles, corresponding
to various power elements on the billet side (i.e., ϕ1, ϕ2, ϕ3, ϕn) and product side (ψ1, ψ2, ψ3, ….. ψn) can be found in the
similar manner.

7. It is found numerically from step 6 that sum of angles (ψ) of all the power elements is very close to 2π radians (i.e., ψ1 + ψ2
+ ψ3 + ……… + ψn ≅ 2π). Then, ϕ + ϕ2 + ϕ3 + … + ϕn ≅ 2π and average incompressibility,

is also found to be within the tolerance limits. Here,

, are individual incompressibility in various power elements within the deforming material.

8. The domain is divided into n′ number of triangular power elements using step 7. For each power element (1 to n′) the total
power consumed (ϕT) can be found. Let ϕ1, ϕ2, ϕ3, … ϕn be the total power consumed in 1, 2, 3 …n′th power elements, then
the overall power (ϕo ) consumed in the extruding the given shape through the given fourth order die profile is found as the
sum of individual powers in different power elements, i.e., ϕo = ϕ1 + ϕ2 + ϕ3 + …….. + ϕn.

9. The minimum power (ϕmin) to ϕo is at an optimum die length that can be found using any single variable optimization
technique. In this work Golden Section Unimodel Optimization Algorithm. Once the optimum power is found, the minimum
average ram pressure (Pave) and relative stress (Rs) on the die can be found easily.

Figure 5.5 Construction of power element for a general non re-entry shape. With permission from [3]
publisher

5.2.8.2. Application to re-entry components shapes

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
5.2.8.2. Application to re-entry components shapes
Application of the proposed upper bound model is demonstrated for a general re-entry shape shown inFig. 5.6. The following
steps are carried out for the given re-entry to find the process parameters such as optimum power, average extrusion
pressure and optimum die length.

1. The geometry of the re-entry component to be extruded is drawn using Auto CAD. The drawn component is then broken
into minimum number of sub-components such that each of them represents a non re-entry component shape. The
selection of entities to change the component drawing for this purpose should be made in such a way that each proposed
non re-entry shape is separate and unambiguous. The designer should try to propose those non re-entry shapes for the re-
entry shape that could form exactly two or more similar non re-entry shapes. Extra entities that are required to create the
proposed non re-entry shapes should be recorded carefully. These entities are extra and do not require frictional power
calculation. Care should also be taken that frictional power is calculated only once for entities of the re-entry shape.

2. The center of gravity (xc, yc) and the cross-sectional area (Aa) for each proposed non re-entry shapes is found using
Advanced Modelling Extension (AME) of AutoCAD.

3. Separate DXF file is created for each proposed non re-entry shape as obtained from step 1. Each of them is processed
separately to extract line and arc entity data.

4. Each of the proposed non re-entry shape from step 2 is treated separately and steps 3–9 are followed to calculate the
individual optimum power (ϕmin) and the optimum die length (Lopt) using the Golden Section Unimodel Optimisation
Algorithm.

5. Let there be number of non re-entry shapes proposed for a given re-entry shape, where . CG and cross
sectional area (Aa) for each proposed non re-entry shapes is known from step 2. For the required reduction (% red)
separate billet for each of the proposed non re-entry shape could be obtained. The diameter of billet is found as

, where and Ana are required billet diameter and the cross-sectional area of

proposed non re-entry shape. Let and be the optimal power and optimal die length for proposed non re-

entry shape. With increase in friction conditions, optimal die length reduces, whereas the corresponding optimal

power increases.

The minimum overall power ϕo and the minimum billet diameter do for the re-entry shape (step 1) can be found as
,

The average ram pressure can be converted to relative extrusion pressure (Rs) as .

6. The overall optimal die length (Lopt) for the re-entry shape is decided on the basis of power (ϕo ) using Golden Section
Optimisation technique for the proposed non re-entry shapes of the re-entry shape. The relative optimum die length is
evaluated as RL = (Lopt/Ro ). If all the proposed non re-entry shapes are of symmetrical in nature (similar profile and cross
sectional area), individual die length (L) is same for all the proposed non re-entry shapes and is equal to Lopt. The values
are different when the proposed non re-entry shapes are non-symmetrical in nature. For a non-symmetrical re-entry
component, individual value of L for each non re-entry shape is taken as L= (2 Lopt/do ). This kind of assignment of optimal

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 die length for the proposed non-symmetrical non re-entry shape simplifies the solution without much loss in accuracy of
the result. In case of symmetrical shapes, however, exact solution is obtained.

Figure 5.6 Segmentation of re-entry shape extrusion into proposed non re-entry shapes. With
permission from Santosh et al. [1]

The above procedure for re-entry shape using the proposed upper bound solution methodology is general in nature and can be
applied for any re-entry shape. This procedure, however, requires intelligent user involvement as different non re-entry shapes
for a given re-entry shape can be proposed.

5.2.9. Examples
To test the validity of the proposed upper bound model for cold extrusions of re-entry and non re-entry shapes, several
examples are selected and the results are compared with the experimental results obtained by earlier workers. These
examples are explained below.

Example 1 (non re-entry)

Yang et al. [12] conducted experiments for axisymmetric extrusion of circular sections, details of material specifications and other process
parameters taken from their paper are given below.

Materials: AISI 4041 Steel and Al 2024 (annealed).

Stress–strain relationships for these materials are

Billet radius ( R o) = 12.5 mm, Friction condition ( m) = 0.12

Die profile: A fourth order die profile is used to validate the proposed model. Let, ratio (initial die length to the billet radius be equal to 2.0,

ram speed = 1.0 mm/s, temperature be the room temperature (20 °C).

Geometrical Construction of Shape

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
The final component geometry is circular with product radius Rp corresponding to different reductions are given in Table 5.1.
These shapes are drawn accurately using AutoCAD. For the drawn shape, DXF file is created and preprocessed as given in the
feature recognition methodology [1]. Minimum number of pseudo vertices (Ns) corresponding to each arc entity. The power
elements found on this basis is minimum in number. This ensures that average error in incompressibility condition
does not exceed by ± 2 %, and total error in angles on billet side (ϕ) and on product side (ϕ) does not exceed by ± 1°.
Minimum number of power elements generated on this basis are used for overall power (ϕo ) calculations.

Table 5.1 Product radius (Rp) at different reductions

Billet radius ( R o) (mm) Reductions (% red) Output radius (R p) (mm) Area (Aa) (mm 2 )

12.50 40 9.682 294.36

12.50 50 8.84 245.49

12.50 60 7.9056 196.56

12.50 70 6.846 147.42

12.50 80 5.5902 98.164

With permission from Santosh et al. [3] publisher.

Results obtained at different reduction ratio (Table 5.1) are compared in Fig. 5.7 with the experimental results obtained by
Yang et al. [12]. It can be seen from Fig. 5.7 that the results obtained from the proposed model are in very close agreement
with the experimental results.

Figure 5.7 Comparison of results with Yang et al. [12]. With permission from Yang et al. [12] and
Santosh et al. [3] publishers

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Example 2 (non re-entry)

Yang et al. [7] have conducted experiments for non-axisymmetric products like rounded square Fig. 5.8 (a) ], ellipse square [Fig. 5.8 (b)], clover
[Fig. 5.8 (c)] and trocoidal gear [ Fig. 5.8 (d)] sections. The details of material specifications and other process parameters of the experiment
taken from Yang et al. [7] are given below.

Figure 5.8 Examples for validation of the model (non-re-entry). With permission from Santosh et al. [3] publisher

Materials: Al 2024 with stress–strain relationship as .

Billet radius ( R o) = 14.75 mm, die profile: Yang used a fourth order die profile. Equation ( 5.12) is taken as the die profile to validate the proposed
upper bound model for the given non re-entry shapes.

, Ram speed = 1.0 mm/s and temperature is the room tempe-rature (20 °C).

Geometrical Construction of the Shape

The final geometrical details of rounded square and ellipse are taken fromYanget al. [8], and clover and trocoidal gear are
taken from Yang et al. [7]. Their cross-sectional profiles are drawn accurately on true scale using AutoCAD. For these shapes,
a separate DXF file is created and preprocessed to separate the geometrical data. Minimum number of pseudo vertices (Ns),
i.e., minimum power elements for ellipse and square are generated, insuring that the average error in incompressibility not to
exceed by ± 2% and total error in angles on billet (ϕ) and product side (ϕ) do not exceed by ± 1o. Minimum number of power
elements generated on this basis are used for overall power calculations.

Experimental results obtained by Yang et al. [7] at various reductions are compared in Table 5.2. It can be seen from the table
that the results obtained using the proposed model are in close agreement with the experiments.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Table 5.2 Comparison of extrusion pressure with proposed model and experiment of Yang et al. [7]

S. Section shape % red (γ* ) Minimum number of power Extrusion pressure (Pave ) at optimal, Difference with
No. elements MPa experiment

Experimental Proposed

1 Rounded 60 1.0 210 317.3 376.63 + 18 %

square

2 Ellipse 60 1.5 142 306.7 319.64 +4%

3 Clover 60 1.5 288 349.9 398.70 + 13 .9 %

4 Trocoidal 45.15 1.5 368 317.92 344.89 + 8.0 %

gear

The shape factor as proposed by Yang et al. [7] With Permission from Santosh et al. [3] publisher.

Example 3 (re-entry)

Yang et al. [7] conducted experiments for non-axisymmetric extrusion of rounded square and clover sections as shown in Example 2. These
examples are solved by the procedure (Sec. 5.2.8.2) considering the shape as a re-entry. Details of material specifications and other process
parameters are taken the same as given in Example 2.

Geometrical Construction of the Shape

Initial geometrical details of rounded square and clover section are taken from Example 2. Their cross-sectional profiles are
drawn accurately on true scale using AutoCAD. Following the steps given in Sec. 5.2.8.2 for re-entry shape procedures,
rounded square and clover section are broken into two and five non re-entry shapes, respectively (Table 5.3). For all these
shapes, separate DXF files are created and preprocessed to separate the geometrical data. Minimum number of pseudo
vertices (Ns), i.e., minimum power elements for individual sub-shapes are generated. The results obtained for these two cases
are compared in Table 5.4 with the experimental result of Yang et al. [12]. The results clearly show that Example 4 (Re-entry)
extending the non re-entry solution methodology for re-entry shapes does not compromise on accuracy and there is no
difference in splitting the re-entry shapes into two or more non re-entry shapes. Results obtained clearly show that the same
results are obtained for all symmetric re-entry shapes. For non symmetric re-entry shapes, however, accuracy will get
affected.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Table 5.3 Examples to validate re-entry shapes of extrusion

Table 5.4 Proposed results from the model for re-entry shapes

Kiuchi et al. [13] have demonstrated an analytical method to calculate power, extrusion pressure at optimum die length for
extrusion of various kind of re entry (non-symmetric) cross-sections such as square, rectangle, hexagon, L, T, and H-sections,
etc. Various results and trends obtained by him have been claimed to be of acceptable range. However, these results have not
been validated with experiments. A re-entry shape extrusion of H-section (Fig. 5.9) is chosen to validate the trend obtained by
Kiuchi et al. [13]. This shape is taken to study the relationships between relative die length and relative stress using a conical
die. Details of material specifications and other process parameters for the study are given below. Since the billet material
used by Kiuchi et al. [13] is not specified, AISI 4140 steel has been taken as the material. Because of this only trend in the
result can be compared.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.9 An example (H-section). With permission from Santosh et al. [3] publisher

Materials: Steel AISI 4140; (MPa) for Billet radius (Ro ) = 40.8 mm, Die profile: Kiuchi et al.
[13] used a conical die profile. Equation (5.12) is used as the die profile to validate the proposed upper bound model for the
re-entry shape.

Ram speed = 1.0 mm/s, Temperature: Room temperature (20°C).

Geometrical Construction of the Shape

Initial geometrical details of H-section (Fig. 5.10) are taken from Kiuchi et al. [13]. Their cross-sectional profiles are drawn
accurately on the true scale using AutoCAD. Following the steps given in Sec. 5.2.8.2 for re-entry shape procedures, the shape
is broken into two sub-shapes (Fig. 5.10). For these shapes, separate DXF files are created and preprocessed. Minimum
number of pseudo vertices (Ns), i.e., minimum power elements for individual sub-shapes are generated. Minimum number of
power elements generated are used for overall power calculations.

Figure 5.10 Segmentation for example Fig. 5.9. With permission from Santosh et al. [3] publisher

Results obtained at different reductions are compared in Fig. 5.11 with the analytical results obtained by Kiuchi et al. [13]. It
can be seen that the trend obtained by the proposed methodology for re-entry shapes is similar to that obtained by Kiuchi et
al. [13].

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.11 Variation of relative die length vs. relative stress in extrusion from round billet to H-type
bar. With permission from Santosh et al. [3] publisher

Example 5 (re-entry)

Table 5.5 lists eight commonly used re-entry shapes along with the details of their cross-sectional area and the initial billet diameter required for
their extrusion. These shapes are chosen to demonstrate the result obtained by the proposed upper bound solution methodology for re-entry
shapes. Geometrical details of the shapes have been taken from various sources. Details of material specifications and other process
parameters for the analytical experiment are given below.

5.3. UPPER BOUND ANALYSIS FOR HOLLOW EXTRUSION

To calculate the extrusion power in cold case, the upper bound model presented in the previous sections can be extended for
hollow extrusion analysis. Any geometry, which is made by lines, polyline, arcs, circle, or ellipse, is the assembly of points. So
geometry is divided into a number of points for geometry recognition and further processing of point co-ordinates of the
geometry. A Bridge Die has been designed and manufactured for extrusion of commercially available Lead alloy (30% Sn +
70% Pb) as a working material for a circular profile. Results obtained by the experiments compared to theoretical results [7]
shows a good agreement. Various parameters such as product profile (product shape factor) extrusion die and mandrel
shapes, area reduction, friction at die, billet and mandrel interface, ram velocity, die length, etc., affect the extrusion power
and average ram pressure have been taken for parametric study. A comparative study has been done by choosing two profiles
(one is circular tube, second is a complex trocoidal shape tube with clover shape inner hole).

Area Division for Equal Velocity of Extruded Product

The data obtain in DATA.O and DATA.I is the large number of x and y coordinate of line at start and end position which after
drawing in a continuous manner makes the actual product OD and ID profile. One needs the further division of these data using
the following process:

a. During initial division of geometry it is not necessary that points obtained for both OD and ID are started from same
position.

b. It is not necessary that data written in DATA.O is equal to data in DATA.I. It is also possible that both coordinates will not
coincide at the line passing through the C.G. of final product shape and outer profile point.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
c. For the calculation of power, we divide whole geometry in large number of small elements having equal area of each small
entity in a pseudo points on inner and outer periphery of product. So that all elements have same velocity during extrusion.

Generation of Equal Number of Data on Inner and Outer Profile

This is a necessary step for the generation of data file, which gives coordinates of elements in equal area. First of all C.G. of
each profile is calculated by using equation

(5.32)

Due to above mentioned reasons; we need further division of area by considering points of OD and ID together.

or

(5.33)

where xi and yi is C.G. of small line for i = 1 to n, and are C.G. of product geometry, li is length of each line. By subtracting
C.G. from each coordinate a new data file is generated with respect to C.G. of profile. Now the data are arranged in increasing
tangent with respect to C.G. of each profile by using relation

(5.34)

Now take new arranged data file of both outer and inner profile together, and divide each profile by generating pseudo vertex
in such a way that each profile has equal number of data, which represents points on the periphery of the profile. In addition,
each vertex of both profiles coincide to the line joining C.G. of product and point at outer profile, i.e., each new generated point
on each profile has same tangent. Tangent of line made by two points on periphery of any profile can be calculated by using
relation and tangent of new vertex with respect to C.G. can be given by Eq. (5.34). So new point generated on either of profile

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
will have to satisfy both Eqs (5.34) and (5.35) simultaneously. Figure 5.13 shows the generation of equal number of data on
inner and outer profile by creating pseudo points using Eqs. (5.34) and (5.35).

Figure 5.12 Flow chart for reading DXF file. With permission from Malpani and Kumar [4] publisher

Figure 5.13 Division of equal elements on outer and inner profile. With permission from Malpani and
Kumar [4] publisher

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
The flow chart of the process is shown in Fig. 5.14. In Fig. 5.13, newly generated pseudo vertexes on outer profile are 5, 6, 7
and on inner profile are e, f, g.

(5.35)

Figure 5.14 Flow chart for equal data generation. With permission from Malpani and Kumar [4]
publisher

Generation of Pseudo Vertexes to Satisfy Continuity Equation

For the extrusion of straight hollow product, it is necessary that velocity of each extruded small quadrilateral element will be
equal. This means area of each element is selected in such a way that it is equal so that by using law of continuity velocity of
each element is also equal as shown by Fig. 5.15. Thus, if each small element is extruded with equal velocity, than there is no
chance of twisting or bending of product. Figure 5.16 shows how data are further divided to get equal pseudo elemental
extruded area. These are final data of vertexes of outer and inner profile used for calculating power in extrusion. Finally the
whole procedure of data preprocessing and data generation shown in Fig. 5.16 can be summarized in following steps.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.15 Product profile before and after division in equal area elements. With permission from
Malpani and Kumar [4] publisher

Figure 5.16 Flow chart for generation of pseudo vertexes satisfying continuity equation. With
permission from Malpani and Kumar [4] publisher

Step (1) Obtain the dimensions of the geometry of outer and inner profile of product from user and draw the profile in Auto
CAD using suitable entities. Care should be taken that the inner profile area should not be greater than outer profile area.

Step (2) Divide the both profile in more then 100 point entities. Delete original geometry and save these points geometry of
both profiles in separate files in Auto Cad R12 DXF format. For higher accuracy and good results, we have to divide
geometry in larger number of points.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.17 Flow chart showing complete plans for data preprocessing and data generation. With
permission from Malpani and Kumar [4] publisher

Step (3) Read the coordinates from Entity Section of DXF file generate in step 2 for both profiles. Calculate their C.G. and
subtract it from vertexes to get coordinates of each profile with respect to origin.

Step (4) Multiply coordinates to the zoom factor to get vertexes of original product. Divide the outer and inner profile in
equal number of elementals by using the concept of pseudo coordinate generation.

Step (5) Use these pseudo vertexes and again divide the whole geometry in equal elemental area of hollow product by
comparing it with required small elemental area and generating pseudo vertexes by satisfying Eqs (5.34) and (5.35). Save
the data in DATA.FF file.

Step (6) Reopen the DATA.FF file and calculate product area, outer profile periphery, minimum inscribing circle diameter,
shape complexity factor and total number of elements generated.

Step (7) Get the percentage reduction and velocity of extruded billet to get final product shape by user and calculate billet
and mandrel radius.

Step (8) Print the data calculated in step 6 and step 7 in file ARPERI.F. This completes the process for generating
coordinate having equal elemental area. These file are further used to calculate the power consumed during extrusion.

Upper bound model

The model is similar to the previous section. Figure 5.18 is taken to establish the model. The domain of deformation Ω is the
plastic deformation zone, τ is the shear stress on velocity discontinuity surfaces S i (Fig. 5.18).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.18 Deformation zone typical streamline surfaces in extrusion. With permission from
Malpani and Kumar [4] publisher

Under similar treatment of the analysis, with reference to Fig. 5.13, and in addition, at the product side coordinates are given
by

(5.36)

where n′ is given as

Putting values of n′ we find

(5.37)

where Ro is the billet radius ne is the distance from the axis formed by joining the C.G. of the product and the billet shape (i.e.,
O-O′) to an arbitrary point C at the die entry. ϕ and ψ are the angles between the plane of symmetry and the stream surface at
entry and exit of the die, respectively, and L is the die length, Ri is the mandrel radius at entry.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.19 Geometry of extrusion die and streamlines having generalized section. With permission
from Malpani and Kumar [4] publisher

Substituting these boundary conditions from Eqs (5.9) and (5.10) into Eq. (5.8) gives

(5.38)

Equation (5.9) can be rewritten in the following form

(5.39)

(5.40)

In the present analysis, function f is represented by the following fourth order curve.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.41)

Equation (5.38) describes the coordinates inside the plastically deforming region but the relationship between the Cartesian
and n, ϕ, z coordinate systems also. Although the present analysis employs a fourth order curve represented by Eq. (5.14) for
the description of die and mandrel profile and the assumed streamlines of particles, it is to be noted that function f in Eq.
(5.41) can be any general function of z provided the function satisfies the given boundary conditions as given in Eq. (5.40).

In most of cases mandrel is used as flat shaped for which function x and y coordinates are not the function of axial coordinate
z. Hence Eq. (5.41) for flat mandrel or die profile is given as

(5.42)

by putting the entry and exit boundary condition given by Eqs (5.9) and (5.10) into Eq. (5.39) gives

(5.43)

From Eq. (5.41) it is cleared that function f (z) = 0 for a flat shaped profile. So this equation can also be written as

(5.44)

(5.45)

In the present analysis, function f is represented by the following relation.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.46)

where C1 is given as C1 = 0.

The Jacobian of Eq. (5.38) can be found as

(5.47)

Here

(5.48)

where

(5.49)

Hence

(5.50)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Determinant of the Jacobian is therefore written as

Det

(5.51)

where

(5.52)

Hence,

(5.53)

Strain rate components are represented by,

(5.54)

where vi and vj represent the velocity components along xi and xj axes in Cartesian coordinate system, respectively. The partial
derivatives of the above equations are obtained with the aid of coordinate transformation, i.e.,

(5.55)

The expanded form of Eq. (5.55) can be seen in the work of Malpani and Kumar [4].

Subscripts of g (ϕ, z) ϕ and g (ϕ, z) z indicate partial differentiation of function g (ϕ, z) with respect to ϕ or z, respectively. First
and second primes on symbols, i.e., f ′, f ′ ′, etc., indicate the first and second derivatives of the symbol with respect to the
given coordinate system. It has been found analytically from the above that εxx + εyy + εzz = 0 and the proposed strain rate field
satisfies the incompressibility condition.

Assuming that the plastic zone is bounded by entry and exit shear surfaces, the velocity field components can be obtained.
Because of volume continuity, the velocity component along z-direction in the Cartesian coordinate system (vz) should be vo
at the entrance and vp at the exit of die. Vo and vp are the speeds of the billet and the outgoing product, respectively. vp can be
described in terms of vo as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.56)

These requirements are satisfied using Eqs (5.40) and (5.41). Therefore, other velocity components for incompressible
material are determined as

(5.57)

Using the velocity fields given by the above equations it has been found that the velocity boundaries conditions are satisfied.
Since the proposed velocity and strain rate satisfy the incompressibility condition, i.e., = 0. Hence,

the proposed velocity field model fulfils the stringent requirement for the construction of a kinematically admissible condition.

Estimation of Extrusion Power and Average Ram Pressure

The total power consumed within the element is the sum of power losses due to plastic deformation (ϕi), the velocity
discontinuities at entry (ϕe) and outlet (ϕo ) of the die and mandrel, and the friction power loss along the interface between the
material and the die and mandrel (ϕf). The above powers are consumed due to metal extrusion in die and to obtain total power
consumed in extrusion process, container and die land frictional powers are added in it. The predicted total power obtained
through the present velocity field would be higher (if no other redundant power losses are present) than the power actually
consumed.

a. Internal Power of Deformation : the incremental power of deformation (dϕi) can be expressed as

(5.58)

b. Frictional Power : If the frictional resistance of the material along the total-work interface is τ and the slip velocity (or the
tangential velocity discontinuity) along the interface is vt, then the incremental frictional power at the interface can be
expressed as

(5.59)

Based on Eq. (5.59), frictional power for the considered element is obtained from the above as

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 (5.60)

where , when ne = Ro

c. Shear Power : Thus, the shear power can be expressed as

(5.61)

where vt is the tangential velocity discontinuity along the dS and S is the area over which integration has to be carried out.
The shear power at entry (ϕe) and exit (ϕo ) side are obtained as:

(5.62)

(5.63)

where

d. Container Wall Friction Power : This includes the power consumed in friction between container and billet contact due to
relative velocity between them.

(5.64)

where Lt=Lo in case of flat die, Lo is optimum billet length and for streamline die

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 t o o

t is the time taken as half of the 80% time of passing billet through die and mc is friction factor between container and
billet.

e. Die Land Friction Power : The power consumed in die land friction is given by

(5.65)

where Ll is length of die land which is taken as one seventh of optimum die length,Pf is outer profile periphery, Vdl is
velocity of product inside die land and is the flow stress of material after strain hardening.

f. Total Power and Average Extrusion Ram Pressure : The total power (ϕT) is the sum of its constituents including the
redundant powers. Therefore, the total power (ϕT) is given as

(5.66)

Once the total power (ϕT) consumed during extrusion is obtained, the upper limit to the average ram pressure (Pavg) for
extrusion is found as

(5.67)

The average ram pressure (Pavg) can be converted to relative forming stress (Rs) as

(5.68)

where σo is the effective stress at zero hardening condition. The relative die length is given as

(5.69)

Case study: (I)

To perform the extrusion of a circular tube a conical die with flat mandrel profile is manufactured. The product shape is
selected as circular for ease of manufacturing. The specifications of die and mandrel are given below.

Specifications of Die

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 • Profile - Conical profile with two cone angle 45° and 70.8°.

• Type - Welding chamber bridge die having four webs on the attached mandrel.

• Materials - Commercial grade die steel.

• Reduction - 80.94%

• Inlet diameter - 26.00 mm

• Outlet diameter - 13.00 mm

• Taper length - 7.00 mm and 6.00 mm, respectively.

• Bearing - 1.00 mm

• Total length - 13.00 mm

• Outer diameter - 38.00 mm

Mandrel Specifications

• Material - Commercial grade die steel.

• Length - 8.00 mm

• Web length - 7.00 mm

• Taper angle - 22.50° from vertical axis.

Experimental details

The extrusion of a circular tube is performed on pre fabricated extrusion set up using commercial grade Lead (70Pb, 30Sn)
billet at 80oC temperature. Billets are machined to 25.10 mm diameter and 45.00 mm length. The extrusion is done vertically
on a 2000 kN Hydraulic Press having constant strain rate facility. Lubricants used on ram and pressure pad is fine graphite
power (10–20% by weight) mixed in Mobil oil. Attention is given on displacement and load, and is noted at fixed time interval.
Figure 5.22 shows the extruded product.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.20 Assembled die and mandrel

Figure 5.21 Assembled die and from inlet side mandrel from outlet side. With permission from
Malpani and Kumar [4] publisher

Figure 5.22 Extruded circular tube attached with die assembly and left billet. With permission from
Malpani and Kumar [4] publisher

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
As the lead is hot worked at room temperature and it is independent of strain hardening, the flow stress and other properties
of the alloy is obtained experimentally using ring compression [9] test as:

Melting temperature = 160 °C

Extrusion temperature = 80 °C

Lubricant is used as a mixture of graphite powder (10–20% weight) in mobil oil on the ram and pressure pad. Friction factorm
for Lead is found to be 0.486 from the ring compression test at 80°C. Other details of experiment are given below.

Mandrel diameter = 7.00 mm

Product cross sectional area = 94.25 mm

Product outer diameter = 13.00 mm

Product inner hole diameter = 7.0 mm

Load vs. displacement and corresponding pressure obtain by experiment is given in Table 5.5.

Table 5.5 Experimental results of lead tube extrusion

S. No. Load in kN Displacement in mm/10s Average pressure (P ave ) in MPa

1 0 0 0

2 1 1.18 2.192

3 2 2.36 4.383

4 4 3.53 8.766

5 8 4.68 17.533

6 21 5.88 46.020

7 45 7.10 98.614

8 52 8.391 113.954

9 52 9.48 113.954

10 54 10.68 118.337

11 58 11.86 127.103

12 58 13.05 127.103

With permission from Malpani and Kumar [4] publisher.

The average velocity Vo is found to be 0.1239 mm/s and average pressure (Pave) is found to be 64.83 MPa. A comparison has
been done between experimental results and the results obtained by the proposed upper bound model for the circular tube as
shown in Table 5.6.

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
Table 5.6 Comparison of experimental and theoretical results for circular lead tube

S. No. Comparison properties Experimental results Proposed results Optimum results Difference (%) [(B/A) – 1]*100

1 Average pressure (Pave ) 64.83 MPa 67.75 MPa 49.69 MPa +4.51 %

2 Average power 3.67 KW 3.83 KW 2.808 KW +4.36

3 Die length 13.05 mm 13.05 mm 17.75 mm

Permission from Malpani and Kumar [4] publisher.

From Table 5.6, it is clear that theoretical results obtained by the proposed upper bound model shows a good agreement with
experimental result for the circular tube of lead alloy. Figure 5.23 shows the variation of displacement rate with average
pressure and load required for extrusion. It is clear from the graph that initially load increase at low rate and then after a
certain displacement it increase sharply to reach at a maximum value and after that it become constant. It is because, as the
extrusion start; upsetting of billet continues up to point (a) and billet accumulate itself in the container inner size, so load
increases slowly. After point (a) piercing and shearing of billet starts and billet is sheared in four parts which is continued up
to point (b) and after point (b), billet is sheared completely and material starts to flow over the mandrel hence again decreases
in the load appears. After point (c) welding of the four sheared section takes place due to smaller cross sectional area
available and hence more temperature rise takes place near the bearing which helps in welding to tubular shape. Finally
material passes through bearing and deforms completely to obtain the final cross section with proper ID and OD of the
circular shape extrusion. Thus extrusion is continuously done keeping the load constant after point (d).

Figure 5.23 Variation of average pressure and load with displacement rate for circular tube extrusion
of Pb (Vo = 0.1239 mm/s, Ro =12.55 mm, %r = 80.94). With permission from Malpani and Kumar [4]
publisher

II. Yang et al. [12] conducted experiment for three arbitrarily shaped tubular sections as: ellipse [Fig. 5.24(a)], clover [Fig.
5.24(b)], and trocoidal gear [Fig. 5.24(c)] with circular inner hole. Using Al2024 alloy as the working material and a forth order
streamline die profile with a flat mandrel profile. The experimental conditions are taken as:

Material: Al2024 with Stress–Strain relationship as: (MPa).

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Billet diameter = 29.5 mm Billet length = 80 mm

Friction factor m = 0.12 Ratio L/R o = 2.0

Ram speed = 1.0 mm/s. Temperature = Room temperature (20°C)

Figure 5.24 Examples used for validation of the proposed model. With permission from Malpani and
Kumar [4] publisher

The final geometrical details of elliptical, clover, and trocoidal gear are taken from Yanget al. [12]. The cross section of profile
is drawn on true scale using AutoCAD Minimum number of power elements is generated for equal pseudo small elemental
extruded area in such a way so that final geometry area calculated should be with in ± 1% and total error in angles on billet (ϕ)
and product side (ψ) do not exceed by ± 1°.

Experimental results obtained by Yang et al. [12] at various reductions are compared in Table 5.7. It can be seen that the
results obtained using the proposed model agrees well with experiment.

Table 5.7 Comparison of extrusion pressure with experiment [12] and the proposed model (r: reduction)

Extrusion Pressure at optimal ( Pave ) in

MPa

S.No Section r (%) Area Shape Element Results obtained by Yang Proposed % Difference with
shape factor number et al. [12] results experiments

Experi- Theore-
mental tical

1. Ellipse 67.79 220.15 1.5 587 773.5 843.1 843.38 +9.03

2. Clover 67.79 220.15 1.5 590 834.7 865.9 848.88 +1.70

3. Trocoidal 51.01 334.84 1.5 825 617.9 632.4 639.49 +3.49

gear

With permission from Malpani and Kumar [4] publisher.

5.4. FINITE ELEMENT METHOD FOR EXTRUSION

SIMULATION
Slip-line and upper bound method are useful in making good predictions for forming loads, the overall geometry changes with
the deforming work piece, the qualitative modes of metal flow and the optimal process conditions. However, very accurate
determination of various process parameters on metal flow became possible only after the finite element method was

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
introduced. In the finite element method, the domain of interest is divided into a number of suitable elements and the
governing differential equations are reduced to algebraic equations by using appropriate approximations for the field
variables over the elements.

Almost all metal forming processes have been analyzed using finite element method due to its ability to handle a wide class
of boundary conditions without restriction on workpiece geometry by selection of a proper mesh. In this section a 3D rigid
plastic FEM formulation is presented.

5.4.1. A Rigid-Plastic FEM Formulation for 3D Extrusion

If the plastic strains are very large as compared to the elastic strains and can be considered to be insignificant as far as the
deformation phenomena are concerned. Under the assumption of negligible elastic strains, the material can be idealized as
rigid-plastic rigid-viscoplastic. Deformation of such materials can be conveniently analyzed by Eulerian or flow formulation for
3D extrusion that this section is concerned. The constitutive equations for rigid-plastic material used here are the same as
described. Initially the domain (Fig. 5.5) is divided into many small slices (Figs 5.25 and 5.26) for simplification and easy
solution.

Figure 5.25 Division of one slice into finite elements for the analysis. With permission from S. Kumar
[5]

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
 Figure 5.26 One element from the power slices. With permission from S. Kumar [5]

5.4.2. Governing Equations

In general, the material behavior of the deforming metal is governed by the equations of conservation of mass, conservation
of momentum, and conservation of energy. Present formulation deals with the steady-state 3D cold extrusion process where
the forming speed is low and therefore the temperature rise is small. Thus, the process is considered to be isothermal.
Therefore, the equation of conservation of energy is not required for the analysis. The momentum equation for steady state
extrusion process (neglecting the body forces) is given as:

(5.70)

Since volume remain constant in case of plastic deformation. The continuity equation is given as:

(5.71)

5.4.2.1. Weak formulation

In the present work, mixed pressure–velocity formulation is used. Let vx, vy, vz be velocities and p (pressure) be is the function
that satisfy a weak solution if the following integral is satisfied:

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.72)

where Wp, Wx, Wy, and Wz be the appropriate weighting function. Since hydrostatic pressure is compressive and hence
negative, weighting function Wp is also taken negative. Performing integration by parts:

(5.73)

Where

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.74)

(5.75)

In finite element formulation, it is convenient to express integrals in matrix form as:

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.76)

where

5.4.2.2. Finite element approximation

In the present work a brick element is used for finite element approximation with 20 nodes in each corner. However, since the
essential boundary condition is to satisfy the continuity equation, the primary variable velocity is taken at each node to be
continuous inside the element. But pressure variable is taken constant inside the element. Another possibility may be taking
the shape function for velocity as quadratic and the same for pressure linear at each corner nodes. For nodal velocity, 20
noded and for pressure, eight noded elements are selected.

Shape function

The square shape function as 20 noded elements with a corresponding master element is taken for velocity and eight noded
(having corner nodes) for pressure is assumed constant over an element.

The natural coordinates for the 20 and eight noded elements as used are given as below:

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
20 noded:

1 (– 1, – 1, – 1) 11 (1, – 1, 0)

2 (– 1, – 1, 1) 12 (0, – 1, – 1)

3 (1, – 1, 1) 13 (– 1, 0, – 1)

4 (1, – 1, – 1) 14 (– 1, 0, 1)

5 (– 1, 1, – 1) 15 (1, 0, 1)

6 (– 1, 1, 1) 16 (1, 0, – 1)

7 (1, 1, 1) 17 (– 1, 1, 0)

8 (1, 1, – 1) 18 (0, 1, 1)

9 (– 1, – 1, 0) 19 (1, 1, 0)

10 (0, – 1, 1) 20 (0, 1, – 1)

where,

(5.77)

Eight noded:

1 (– 1, – 1, – 1) 5 (– 1, 1, – 1)

2 (– 1, – 1, 1) 6 (– 1, 1, 1)

3 (1, – 1, 1) 7 (1, 1, 1)

4(1, –1, –1) 8 (1, 1, – 1)

The shape function for die billet interface is chosen as eight noded as given below:

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.78)

(5.79)

where,

(5.80)

det

(5.81)

For boundary nodes (die billet interface):

det

(5.82)

5.4.2.3. FE formulation for ε and ε (W)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
5.4.2.3. FE formulation for ε and ε (W)

(5.83)

The present formation an isoperimetric, representation is used, hence,

(5.84)

Now taking an arbitrary function f which is a function of x, y and z. so,

(5.85)

Taking differential

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.86)

or,

where [J] = Jacobian matrix

Hence

(5.87)

where JI is inverse of J. Now, if f is considered as vx, vy or vz

Then,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.88)

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
(5.89)

(5.90)

(5.91)

(5.92)

where,

© McGraw-Hill Education. All rights reserved. Any use is subject to the Terms of Use, Privacy Notice and copyright information.
or,

Similarly,

Therefore,

(5.93)

Secondary ele