technical

Influence of Planet Carrier Misalignments

on the Operational Behavior of Planetary
Gearboxes
Prof. Dr.-Ing. Christian Brecher, Dr.-Ing. Dipl.-Wirt.-Ing.
Christoph Löpenhaus and Julian Theling M.Sc.

Introduction and Motivation and misalignments in the gearbox. The loads resulting out of the radial, axial
The measurement results as well as input forces and bending moments can be and tangential components of the gear
the analysis of the present paper have simulated aerodynamically and applied on mesh forces. Due to the trend towards
been generated in the context of a proj- the input shaft by means of the NTL. higher power densities, the internal loads
ect funded by the Federal Ministry of In the test gearbox consisting out of a increase, while at the same time thinner
Economic Affairs and Energy (BMWi) planetary gear stage and two cylindrical and thus generally less stiff surrounding
with the title “Loads on the drive compo- gear stages approximately 150 sensors are structures are used (Ref. 9). As a result,
nents of wind turbines” (FKZ 0325799). applied, in order to measure for example higher deformations and thus also higher
The main objective of the project is the the misalignment and the effect of the shaft misalignments are to be expected in
further improvement of the validity of misalignment on the component load- the future.
today’s simulation models by means of ing. Therefore, position sensors as well as Shaft misalignments always occur
the first large-scale validation by means strain gauges in the tooth root of the sun when the rotary axes of one or more
of a wind turbine. In the long term, this and ring gear are applied. Furthermore, gears deviate from the ideal position e.g.
should enable the systematic optimiza- the rotational position of every shaft as due to deformation or assembly toler-
tion of the function and reliability of the well as the deviation of the whole housing ances, cf. (Fig. 2). A distinction can be
electromechanical drive-train of wind are measured. made between parallel displacements of
turbines (Ref. 2). the axes, which result in a center distance
The test object is a 2.7 MW wind tur- Shaft Misalignments in deviation and angular misalignments.
bine modified in cooperation with the Cylindrical and Planetary Regarding involute gears, the effect of
project partners, cf. (Fig. 1). It is driven Gearboxes angular misalignments is typically more
via a 4 MW direct drive, which provides a Similar to cylindrical gears, shaft mis- important. Angular misalignments can
maximum torque of 3.4 MNm starting at a alignments in planetary gearboxes result be distinguished in terms of inclination
rotational speed of 4 rpm, thus eliminating amongst other influences from manu- and skew. An axis skew corresponds to
the need for an additional gearbox on the facturing tolerances and load-dependent a rotation of the axis in relation to each
input side. A non-torque load unit (NTL) deformations of the shaft-bearing sys- other about the axes connecting line. An
is installed next to the drive motor. This tem. The loads that lead to this defor- axis inclination exists when the axes of
enables the test nacelle to be loaded in the mation can be either external loads the gears are rotated in the plane defined
remaining five degrees of freedom. Thus, from surrounding elements or internal by those axes in misaligned condition
it is possible to simulate bending moments
and forces of external wind loads and to
apply them to the test drive train even in
dynamic conditions. The gearbox has the
same design and macro geometry as a
Winergy wind turbine gearbox. The micro
geometry of the gears was redesigned so
that all geometry parameters of the gears
are known (Ref. 3).
One sub objective of the project is to
identify the influence of external loads
caused by the wind-field and electrical
grid on the loading of drive train com-
ponents. External loads caused by differ-
ent wind conditions result into bending
moments and forces on the input shaft of
the nacelle, which lead to deformations Figure 1 Test setup and objective.

This paper was first presented at the 8th WZL Gear Conference in the USA July 23/24, 2019.

46 GEAR TECHNOLOGY | July 2020

[www.geartechnology.com]
and thus intersect each other. According
to Wittke, both components of angu-
lar misalignments can be summed up
into a resulting lead angle deviation fHβ
(Ref. 16). The lead angle deviation can be
calculated by means of the pressure angle
at the operating pitch diameter αwt as well
as the inclination or skew angle (Ref. 16).
The misalignments occurring with cylin-
drical gears can in principle be transferred
to planetary gears. Due to the existing kine-
matics and the loads of the individual ele-
ments, other types of misalignment can
occur which can also have different local
effects on the gear mesh (Ref. 11). Figure 2 Shaft misalignments in cylindrical gears.
Simple planetary gearboxes consist of
the central elements sun gear, ring gear
and planet carrier, as well as usually
three or more planets (Fig. 2). Due to
the arrangement, each gear has several
meshes. The load on the surroundings,
i.e. — the shafts, bearings and carriers,
results from the superposition of the gear
loads from the individual meshes. The
sun gear is often mounted in an adjust-
able manner via a spline-tooth coupling
or spherical roller bearing in order to
find an ideal position for itself on the
basis of the individual force components.
The aim is to transfer the load as evenly
as possible to all planets. A displacement
is therefore intended and desired at the Figure 3 Misalignments in planetary gear stages.
sun wheel (Ref. 1). Ring gears are mainly
designed as housing elements; misalign-
ment of the ring gear is therefore strongly
dependent on the misalignment of the
housing (Ref. 13).
The planet gear is in contact with the
ring and sun gear. The tangential tooth
forces of the planet point in the same
direction due to the flank change and
have the same magnitude due to the
torque compensation and the equal base
diameter. In the case of helical gears, in
addition to radial and tangential forces,
axial forces also occur. In the ring gear
mesh and sun gear mesh of the planet,
the axial forces are directed in the oppo- Figure 4 Effect of misalignments on the gear mesh conditions.
site direction and balance each other
out. The force acting points of the two forces of the individual planetary gears connection. The effect is a twisting of the
meshes are located in different areas of are supported at the carrier and thus gen- two carrier walls in relation to each other
the planet, as a result of which an axial erate the carrier torque (Ref. 15). Due and thus a misalignment of the planet
bending moment is applied. This must to the one-sided torque transmission at shaft mounted in the carrier walls. Like
be balanced by the planet bearing and the the carrier shaft, the torsional moment the planets, the carrier is free of radial
planet pin. varies in magnitude over the face width. and axial forces. An additional displace-
The planet carrier has a significant Furthermore, the carrier wall facing away ment of the carrier caused by mesh forces
influence on the deformation within from the carrier shaft has a lower tor- is therefore unlikely. External loads such
planetary gearboxes. The tangential sional stiffness due to the bar-shaped as the weight force or forces acting from

July 2020 | GEAR TECHNOLOGY 47

 technical
connecting components of the carrier inclination and skew curves are phase- tooth hunt is calculated as a sequence
shaft, such as rotors in wind turbines, shifted by exactly a quarter rotation of the of individual pitches. The resulting time
are therefore the main cause of misalign- carrier. Due to the different influence of signal is transferred to the frequency
ment. Other influencing factors are bear- inclination and skew on the gear mesh, domain and the effects are analyzed.
ing clearance and tolerances as well as the resulting lead angle deviation is not Subsequently, a method to calculate local
manufacturing and assembly deviations. constant. Planet carrier misalignments damage is developed, which considers the
In Figure 4 the effect of the different are comparable to a wobble in cylindri- variable mesh positions in combination
misalignments on the resulting lead angle cal gears. The amplitude of the lead angle with the angular positions of the elements
deviation and therefore meshing condi- deviation depends on the rotation angle in planetary gearboxes.
tions is presented. Therefore, tolerances of the carrier. For the carrier positions At the end, the tooth flank has to be
of the shaft-hub connections are ana- 90° and 270°, there are only inclination optimized in order to achieve less dam-
lyzed and transformed into a resulting effects — for 0° or 360° and 180° only age and a lower noise excitation with
lead angle deviation according to Wittke skew. the aid of derived characteristic values.
(Ref. 16). The selected displacements rep- Therefore, an approach is used which is
resent the characteristic displacements Objective and Approach based on characteristic values derived
of planetary gears: “S-Shape” due to the The objective of this report is to describe in the first work packages that are input
bending moment of the axial forces on the effects caused by misalignments in variables of a variant calculation carried
the planet; planetary displacements due planetary gearboxes and take them into out with the FE-Stirnradkette in com-
to different twisting of the carrier walls account during the optimization and bination with the statistical evaluation
and a misalignment of the carrier due design process. Therefore, measure- software µOpt to ensure a balanced
to external loads or misalignment of the ment data of the FVA nacelle has to be and robust design against additional
housing. evaluated with respect to misalignments misalignments.
The planetary axis misalignment due and their influence on the loading of The input stage of the wind turbine
to axial forces of the planet causes a pure the gears. In order to achieve a balanced gearbox of the FVA nacelle is selected
inclination of the planet, since the axis of design, both the load on the tooth flanks as the application case for investigating
rotation of the tilting is perpendicular to and the excitation behavior have to be the influence of misal- ignments in plan-
the axis connection line of sun and planet included in the design process. In order etary gear stages. Besides the fact that the
(Fig. 4); the carrier position shows no to achieve the objective, four sub-objec- geometry is known, a further reason for
influence. tives have to be fulfilled (Fig. 5.) the choice of the demonstrator is the exis-
A similar effect is caused by the plane- In order to regard the misalignments in tence of numerous test results. In particu-
tary displacement due to a different twist- the tooth contact analysis, the FE-based lar, the misalignments of the planet car-
ing of the carrier walls. The displacement tooth contact analysis FE-Stirnradkette rier in relation to the housing are known
takes place in the tangential direction. In is extended in order to depict a com- for numerous load situations with loads
contrast to the planetary axis misalign- plete tooth hunt of a planetary gear stage in all six degrees of freedom. Selected test
ment, a pure skew is present; there is also and take into account time-variable mis- points together with the corresponding
no dependence on the carrier position. alignments. This is necessary because the measured carrier misalignment represent
Due to the mathematical transformation effect of a carrier misalignment has dif- the investigated operating points.
of the angular misalignment on the gear ferent influences on the mesh, depending
mesh via the sine of the pressure angle on the current position. The extension Description of Contact Conditions
at inclination and the cosine at skew, the is validated afterwards by a comparison in Misaligned Planetary
skew has a bigger influence for the same between the calculated tooth bending Gearboxes
angular misalignment. stress and the measured stresses. The first sub-objective is the description
Carrier misalignments result in a com- To evaluate the excitation behavior, of the contact conditions for misaligned
ponent of inclination and skew. The the transmission error of a complete planetary gearboxes. First, a method
is developed which converts the shaft
misalignment to the mesh according
to the Wittke method as shown (Fig. 4)
(Ref. 16). A main challenge is the consid-
eration of the position-dependent carrier
misalignment. The mesh specification is
then used as input for the FE-based tooth
contact analysis. Subsequently, measure-
ment results of the FVA nacelle are pre-
sented and essential effects are described.
The measurement is then compared to
the simulation to validate the method.
Extension of the FE-Stirnradkette.
Figure 5 Objective and approach. In the following section, a method is

48 GEAR TECHNOLOGY | July 2020

[www.geartechnology.com]
discussed to consider misalignments of
the planet carrier and the planets in tooth
contact (Fig. 7). The calculation of the
displacements from internal or external
additional forces can be carried out in
advance analytically or with the aid of
finite element analyses. The method then
takes these misalignments into account
in the tooth contact analysis and can thus
be used to analyze the influence of the
respective deviations or for modification
design; the method consists of four essen-
tial elements.
First, the displacements of the individ-
ual shafts are determined in a preceding
Figure 6 Test setup and occurring effects.
calculation step. It is possible to use FE
calculations or analytical approaches, as
well as misalignments acquired by mea-
surement. The input is made via the dis-
placement of the bearing positions. From
these, the vectors of the shafts are deter-
mined in a common coordinate system
for all wheels.
Mathematical operations are used to
transform the vectors for each mesh posi-
tion of a complete tooth hunt on to the
tooth contact. For this purpose, the state
of the art methods for the transforma-
tion of inclination and skew are extended
by the kinematics of planetary gears
(Ref. 16). The results of this transforma-
tion are resultant fHβ curves for the sun-
Figure 7 Validation of the simulation method.
planet and ring gear-planet mesh.
The resulting curves serve as input
variables for the FE Spur Gear Chain → misalignment of the planet carrier with the same contact conditions always occur
FE-Stirnradkette. Deviations from the respect to the sun and ring gear. The at the fixed ring gear on one tooth, so
ideal tooth contact are considered as con- loads on the sun gear are measured in the that this tooth always experiences the
tact distances in the spring model; a new test gearbox via strain gauges in the tooth same load. On the sun and planet gear,
FE calculation is therefore not necessary, root. With the help of the rotation angle however, the angular position at which
which leads to an optimization of the cal- measuring systems, the current position a tooth comes into contact varies. This
culation time and quick evaluation of the of the gears can be determined in a post- means that every contact position occurs
results. Loads of the individual gears are processing and thereby at which carrier on every tooth during a complete tooth
determined for each meshing position of position the loads occur. The results show hunt and the tooth load is time-variable.
a complete tooth hunt. The results of the that the load distribution varies over the This finally results in internal load spec-
tooth contact analysis can be post-pro- carrier angle due to the misaligned posi- tra already at constant external loads and
cessed afterwards. tion. This can lead to significant over- displacements due to changed contact
Measurement results of the FVA loads in the tooth contact. conditions.
nacelle. The planet carrier is connected The highest stresses occur in the Validation of the extension. In the fol-
to the rotor of the wind turbine and sup- tooth root if the meshing takes place at lowing section, the calculation approach
ported by a main bearing (Fig. 6). The an upper 180° or lower position 0°/360°. for the consideration of the carrier mis-
gear housing, in which the remaining Between these two positions the location alignments in the tooth contact analysis
gears are mounted, is supported by two of the maximum stress also changes from is validated. Measurement results of the
flexible mounting points on the torque the right to the left flank side of the sun FVA nacelle are used and compared with
arm of the main frame. This mounting of gear. The lowest loads and the best load the results of the simulation method. The
the housing enables a tilting of the gear distribution are at the horizontal posi- tooth root stress of the sun gear and the
housing due to the weight force in com- tions at 90° and 270°, respectively. load distribution factor KFβ for each posi-
parison to the planet carrier (Ref. 15). Because of the dependence of the con- tion are used as evaluation variables. The
D ue to t his ef fe c t , t here is a tact conditions on the carrier position, load case corresponds to the nominal

July 2020 | GEAR TECHNOLOGY 49

 technical
torque of the system with Mx = 1,650 kNm b)
and in dependence of the carrier angle. meshing planes of each single pitch
at an average speed of n in = 11 rpm. The amplitudes of the stresses are in the within one complete tooth hunt using the
Further loads applied to the shaft simu- same range as in the measurement, and Wittke method as a function of angular
late, on the one hand, the rotor weight also the location of the maximum load carrier position (Ref. 16).
of the turbine in the direction of Fz and, is identical. When comparing the width The result of the tooth contact analysis
on the other hand, bending moments load factor KFβ between simulation and is the transmission error for one pitch
and transverse forces from inclined wind measurement, identical characteristics for each angular position. The calcula-
flows on the rotor. can be determined. Both graphs show tion is carried out separately for the sun
The misalignment of the planet car- maxima at 0° and 180°. The KFβ of the and ring gear mesh without consider-
rier measured by distance sensors is used maximum at 180° in the simulation is at ation of cross influences between those.
as the input variable of the method. The KFβ = 1.35 and represents the load occur- The results of the individual single pitch
effects from the planetary axis are not ring in the measurement well. The sec- calculations are combined into an over-
considered in this example, since the ond maximum is displayed higher in the all signal in chronologically correct
influence of these displacements is small simulation. The reason for this is on the sequence. A consideration of differences
compared to the carrier displacement. one hand, the extended evaluation range in the length of a pitch is that prema-
The maximum values of the resulting fHβ in the simulation, because in the mea- ture and post-mature gear meshes or gear
from the measured misalignment in the surement, the results cannot be resolved clamps do not take place. The evaluation
sun mesh are ∆fHβ = ± 80 µm; Figure 7 up to the edge area, since no strain of the signal is done in frequency domain
shows the comparison of measurement gauges were applied there. On the other on the basis of the speed-dependent
and simulation. hand, the differences can be explained by orders. For this a fast Fourier transform
The measurement results in the load the lack of consideration of the flexible (FFT) of the time signal is carried out.
case evaluated here show a clear depen- planet shaft, which can reduce the tilting The evaluation takes place individually
dence of the load distribution and the due to the present clearance. for each load case.
locally occurring tooth root stresses on Figure 8 shows the results of the analy-
the current carrier position. As explained Calculation of the Transmission sis for a load case at nominal load with a
previously, the maxima of the load dis- Error of Misaligned Planetary carrier misalignment resulting in an fHβ
tribution are located at approximately 0° Gearboxes deviation of ∆fHβ = 144 µm, compared to
or 360° and 180°. This corresponds to The approach for calculating the trans- a simulation without carrier displace-
a mesh position in the upper and lower mission error of misaligned planetary ment. Considering a variable displace-
area of the ring gear. The highest load gears uses the results provided by the pre- ment, it can be seen that the rotational
occurs in the tooth mesh in the 0° range. vious developed and validated method. orders of the carrier are excited in the
Between the two maxima, a change of the As input variable of the tooth contact sun as well as in the ring gear. On the
more heavily loaded flank side can also analysis, one load case with measured other hand, sidebands appear around the
be seen. In the measurement, the width misalignments is used. In the further gear mesh orders. In the ring gear mesh
load distribution factor in the tooth root investigations only planet carrier mis- a reduction of the amplitude of the first
KFβ shows values between KFβ = 1.03 and alignments are taken into account dur- gear mesh order in comparison to the
K Fβ = 1.38, depending on the meshing ing the investigations, because all other non-displaced variant is noticeable. The
position. The maximum at approx. 180° misalignments have a position invari- amplitude remains almost unchanged in
has a KFβ = 1.25. able influence and could be optimized the sun mesh.
The progression of the tooth root afterwards by lead angle modifications Due to the low speeds of the carrier,
stresses can be correctly predicted with (Fig. 4). The misalignments of the car- the resulting excitations caused by the
the method — both in width direction rier are transformed to the individual carrier rotation orders lie in the range
below a frequency of f = 1 Hz and thus
are not in the audible range of the human
being. An excitation of these frequencies
is nevertheless to be reduced, since tower
natural frequencies are excited by the
low-frequency vibrations of the transmis-
sion, which can lead to a swinging up of
the turbine on the tower (Ref. 12).
The changed characteristics of the gear
mesh orders can be explained mainly
by modulation effects. Low-frequency,
superimposed oscillations of the rota-
tional orders lead to modulations and
thus to the development of sidebands
of higher-frequency oscillations of the
Figure 8 Results of the excitation behavior calculation with and without misalignments. gear meshes (Ref. 8). The formation of

50 GEAR TECHNOLOGY | July 2020

[www.geartechnology.com]
sidebands is often aimed for in order to
reduce tonalities in the system. In this
case, the amplitudes of the sidebands in
the sun mesh exceed both, the first order
of gear mesh and the amplitude of the
undisplaced variant, so that the excita-
tions of the system are amplified.

Influence of Planet Carrier

Misalignments on the Load
Carrying Capacity
A local accumulation of damage is used
to assess the influence of moving contact
patterns due to misaligned central ele-
ments in planetary gears. The method is Figure 9 Approach for local damage accumulation.
similar to the Sfar and Ziegler method,
but does not only consider the lower sin- Table 1: Overview of regarded load bins
gle contact point, but also the complete Nr. Mx [kNm] My [kNm] Mz [kNm] Carrier angle [°] Load Cycles [%]
tooth flank (Refs. 13 and 17). The main 1 304 0 0 0.0277 10
difference to the existing methods is the 2 639 0 0 0.0157 5
consideration of internal load spectra 3 973 0 0 0.0187 20
resulting from shifted central elements 4 1312 0 0 0.0170 30
with the help of the calculation of a com- 5 1650 0 0 0.0204 20
plete tooth hunt. For the load calculation, 6 1650 1000 0 0.0065 7
a method for the analysis of the tooth 7 1650 -1000 0 0.0157 8
contact with misaligned central elements
developed and validated in the previ- in relation to the sun gear. (Fig. 9, bottom-center).
ous section is used. The load capacity is The contact conditions as well as the For each load bin “v”, a complete tooth
represented by an S/N curve, and sliding torque are used individually for each hunt is simulated so that all angular posi-
velocities are currently not taken into pitch as input data of the FE-based tooth tions of the gear elements relative to each
account. In the second part of this sec- contact analysis FE-Stirnradkette. As a other are regarded. This means that sev-
tion, a comparison of the load carrying result of the tooth contact analysis, the eral pitches are rolled over and all teeth
capacity of the tooth flanks is carried tooth flank pressures according to Hertz are in contact several times. A partial
out with variable, constant and none dis- are used for the further calculation. The damage by an additional contact is added
placed contact patterns for a load spec- tooth root load capacity is not examined to the already existing damage, so that
tra in order to identify the influence of in this report. For this purpose the tooth finally a partial damage of all teeth and
variable misalignments on local dam- root stresses would be evaluated instead all flank elements for a complete tooth
age. Measured displacements of the FVA of the Hertzian pressure and a local accu- hunt of one load bin exists. This value is
nacelle are used. mulation of damage in the tooth root then divided by the number of pitches
Approach to determine local dam- would be carried out. calculated for a complete tooth hunt and
age. The previously developed method In order to assess the local damage, multiplied by the absolute number of
is extended in the present section so the tooth in contact z corresponding to cycles Nv of the load bin. The result is a
that the determined data can be post- the angular position of the gear elements total damage caused by the individual
processed in order to calculate the local is identified at each gear for each sin- load bin. It should be noted here that the
damage of the tooth flank. In addition, gle pitch. Each tooth is additionally ras- number of load cycles at the sun gear and
the possibility to consider different exter- tered in i rectangular elements along the ring gear must be multiplied by the num-
nal drive torques and misalignments height and j elements in width direction. ber of planets, since the individual flanks
via a load spectra specification is made For each area of the grid, the maximum are over-rolled several times. The results
possible. pressure is determined when one pitch is of the individual load bins are summed
In a first step, operating points “v” or rolled over. The occurring flank pressure up so that a total damage of the indi-
load bins are specified (Fig. 9, bottom- σi,j,z of each element is the input variable vidual flank elements results, taking into
left). In addition to the drive torque and for the evaluation of the individual dam- account the load spectra and correspond-
the percentage of a load bin in the total age. An S/N curve is used to determine ing measured misalignments.
number of load cycles, the displacement the bearable number of load cycles Ni,j,z Results of the evaluation of local dam-
of individual gear elements is another for the determined pressure σi,j,z. The par- age. Selected test points together with
input variable. Possible specifications tial damage si,j,z of an element due to one the corresponding measured carrier mis-
are, for example, misalignments of the load cycle is defined as the reciprocal alignment represent the load bins of the
planet carrier, the ring gear or the planet of the bearable number of cycles Ni,j,z, investigated load spectrum. This leads

July 2020 | GEAR TECHNOLOGY 51

 technical
to plausible combinations of torques and of the ring gear. The durable permissi- the calculated local damage between
misalignments. The internal load spectra ble flank pressure of the case-hardened individual flank areas or designs is legiti-
result from the carrier displacement. The gears is σH = 1,300 MPa, of the ring gear mate. Therefore, a normalized damage
—
distribution of the number of load cycles σH = 520 MPa (Fig. 10, top-center). The value SZ is used. This value is defined as
among the respective collective classes consideration of carrier misalignment in the ratio of the maximum damage per
does not correspond to reality, but is dis- the damage accumulation on the flank is tooth smax, lok, z to the maximum damage
tributed in such a way that a large part carried out in three different ways. In the of a gear smax, lok, var in the calculation with
of the operation takes place at nominal first variant, as described previously, the variable displacement.
torque. An overview of the input data misalignment is calculated for each gear The upper-right part of Figure 10
used can be found in Table 1. All operat- mesh on the basis of the angular position shows the results of the local damage
ing points show a constant weight force of the elements (variable misalignment). accumulation for the planet gear in the
of the rotor in z direction of Fz = - 488 kN, In the second variant the misalignment sun mesh. It can be seen that the nor-
a loading with axial and bending forces Fx of the carrier is not considered (no mis- malized damage assumes a constant
—
and Fy does not occur. alignment), and in the third variant the value of SZ = 1 for all teeth on the planet
In addition to the loads resulting from maximum occurring value is assumed for gear despite variable misalignment is
the operating conditions, the strength all positions (maximum misalignment). regarded. The same behavior is observed
of the material is a necessary input vari- The pressures on the tooth flank cal- at the sun gear. At the ring gear, however,
able for describing the local damage. The culated from the tooth contact analysis different damages occur, depending on
material behavior is described by an S/N do not correspond to the standard value the tooth position. Here the tooth at the
curve (Fig. 10). The corresponding values of the flank pressure σH, since applica- seventh position is most damaged. The
for the sun and planetary gears are deter- tion factors are not taken into account. graph shows a maximum at this point, a
mined for case-hardened gears from ISO In addition, the load spectrum does not second local maximum with lower ampli-
6336 Part 5 for the material quality MQ correspond to reality, so that a calculation tude occurs with 180° offset at tooth
(Ref. 5). Values for wrought alloyed mate- of the absolute damage value is not con- 51. The results for the planet gear flank,
rial are used for the material behavior ducive. However, a relative comparison of which is in contact with the ring gear, are
not displayed. The pressures occurring
are all below the fatigue strength level
because of the concave-convex contact,
so that no damage can be determined.
The constant damage to the sun and
planet gears can be explained by the rota-
tion of the two gears. Each tooth of the
sun gear, as well as of the planet gear,
does not have a fixed global mesh posi-
tion; the global position depends on the
actual carrier position. The deviations
of the gear mesh resulting from the mis-
alignment of the carrier are also depen-
dent on the carrier position; therefore,
temporally variable deviations occur at
one tooth of the sun and planet gear.
Figure 10 Results of the local damage accumulation. After a complete tooth hunt of the plan-
etary gear stage, each tooth of the sun
and planet gear are exposed to the same
positions and thus to the same loads.
Due to the fixed positioning in a load
bin, a tooth on the ring gear always has
identical mesh conditions. Therefore, the
degree of damage changes depending on
the tooth position. The second local max-
imum is smaller, since areas of the tooth
flank are stressed in which the support-
ing effect of the helix angle is missing. As
a result, the flank can be deformed better
and the pressure values are lower.
The diagrams also show the results of
the local accumulation of damage with-
Figure 11 Analysis of the location of local damages. out and with maximum displacement

52 GEAR TECHNOLOGY | July 2020

[www.geartechnology.com]
(Fig. 10). The locally determined max-
imum damage with variable displace-
ment was kept as the reference value. It
can be determined that no damage to
the tooth flanks occurs in a calculation
without misalignment. All pressures are
below the fatigue strength of the material
and are not relevant for damage. If the
maximum misalignment is used to assess
the damage, the same maximum value is
obtained for the ring gear as for the local
evaluation. The damage to the sun and
planet gear is six times higher than in
the local calculation. A dimensioning of
the gears with the maximum value of the Figure 12 Approach of the design method.
misalignment would thus be safe, but at
the same time clearly over-dimensioned. contact areas are relieved due to the tip of the load increase of the tooth flank
A dimensioning without consideration relief provided on both flanks. This is over one rotation of the carrier and ∆ϕγ
of the influence of the variable effects why the maximum damage occurs in the as the deviation of the transmission error
over the circumference, however, leads to double-contact area. over one rotation of the carrier for each
under-dimensioned gears. nominal design and each load step.
In Figure 11, the damage values are Optimization of the Operational For this purpose, the maximum value
plotted on the maximum loaded tooth Behavior of Planetary Gearboxes of the rotational error ∆ϕmax and the flank
flanks. It can be seen that the damaged The procedure explained here focuses pressure σHmax is determined from all val-
areas on all gears are extremely localized. on the design of the flank corrections of ues occurring over one rotation of the
On the planet and sun gear, damaged the planet. The optimization goal is to carrier and divided by the value occur-
–
areas occur on the left and right side of increase the robustness of the load car- ring for the average f Hβ. An additional
each tooth flank. Here, the left side of the rying capacity against misalignments. In boundary condition, which is added in
tooth is stressed more than the right side addition, the overall excitation should the evaluation, is the symmetrical design
of the flank. The damage to the ring gear be reduced — especially in the low-fre- of the flank corrections on both tooth
depends on the position of the tooth on quency range. As input variables for the flanks of the planet.
the circumference; the highest damage is design, the averaged lead angle devia- The maximum Hertzian pressure, the
–
determined at tooth 7. This occurs locally tion fHβ over one carrier rotation as well as transmission error at the average mis-
in the tip area on the left side of the flank. the amplitude of the variation of the lead alignment as well as the two newly
At the second local maximum at tooth 51, angle deviation f̂ Hβ from maximum to introduced load increase factors K H,γ
the right area of the tooth flank is sub- minimum value are taken into account. and ∆ ϕγ are included as parameters in
jected to greater stress. Here, too, the high- The design is carried out with a vari- the evaluation of the variants. All values
est damage is found in the tip area, but it is ant calculation in the FE-Stirnradkette as are determined for each nominal vari-
significantly lower than on tip seven. well as a post-processing and a statistical ant for the sun and ring gear mesh and
Due to positive slip, the head areas are evaluation of the results with the help of then evaluated together. The evaluation
generally less relevant to flank damage the µOpt program (Fig. 12). takes place depending on the load case
than high loads in the negative slip area, The variant calculation is carried out and the parameter with the help of the
but nevertheless the local limitation of separately for the individual meshes µOpt software. Here the load cases are
the damage shows that the flank is used between sun and planetary gear, as well weighted more strongly with increasing
sub-optimally. A uniform damage over as ring gear and planetary gear. For this rotor torque and the pressure is evaluated
the entire flank would correspond to an purpose the lead angle deviation is finely higher than the transmission error. The
ideal geometry where all points of the resolved over a wide variation range. overview of the individual variables is
flank fail at the same time. The areas with Using the mean lead angle deviation f̂Hβ shown (Fig. 13).
the highest damage result from edge con- and the amplitude of the lead angle devia- The optimized flank geometry essen-
tact that cannot be robustly compensated tion f̂ Hβ all possible deviations for one tially differs from the nominal design by
by the gear geometry. In the ring gear nominal design are determined in a post- an applied profile angle correction and a
contact, the maximum stress is located processing. The individual results of the higher lead crowning. Due to the positive
in the area of the upper single contact, transmission error and the flank pressure profile angle correction at the planet, the
since here the highest loads have to be are determined for each lead angle devia- tooth tip of the ring gear is relieved and
transmitted by a pair of teeth. On the tion occurring in the misalignment range the area of maximum load moves in the
planetary flank there is a tip relief, which in one load bin. direction of the tooth center. The effect
reduces the load on the second single From the sum of the individual results, of the changed microgeometry becomes
meshing area. In sun contact, both single the value KH,γ is determined as the factor clear when comparing the results of local

July 2020 | GEAR TECHNOLOGY 53

 technical
damage accumulation before and after
optimization. The damage values of the
new design are determined and divided
by the reference value. The reference
value of the normalized damage is the
value smax,lok,origin of the original variant.
In the sun mesh, a reduction of the
damage can be achieved to a value of
X = 0.75 — both at the planet and at the
sun gear. The flank damage of the ring
gear can be reduced to a value close to
Sz = 0. The difference between the level
of damage depending on the tooth posi-
tion at the ring gear is also reduced. The
reason for this is, on the one hand, the Figure 13 Influence of the optimization on the local damage.
avoidance of edge contacts which have
led to high loads in the edge area of the excitation behavior are determined. In a nominal designs are compared with the
gear teeth. This resulted in high dam- post-processing of the previously devel- help of the µOpt software. The recalcula-
age values in the original design. On the oped method, the transmission error of tion of the best variant shows a reduced
other hand, the contact pattern in the the single meshes, as well as the dam- damage of all flanks, despite a higher
ring gear mesh moves in the direction of age locally resolved on the tooth flank crowning. The effect of the varying dam-
the flank center due to the profile angle of each tooth of each gear, is calculated. age over the ring gear circumference can
correction, whereby locally high pres- The transmission error of misaligned almost be eliminated.
sures in the tip area can be avoided. and aligned planetary gearboxes dif- As a next step, slippage effects will be
fers. In misaligned conditions, sidebands regarded during the local damage cal-
Summary and Outlook of the tooth mesh frequencies occur. culation to better predict the localiza-
This study deals with the modeling and Furthermore, the carrier rotational fre- tion of failure. Also measurements will
consideration of misalignments in plan- quency is excited. be done on a smaller test rig to validate
etary gearboxes in the optimization and The results of the local damage calcula- the effects observed in the evaluation of
design process. Procedures for taking tion show that due to the rotation of the the excitation behavior. Therefore, dif-
into account misalignments in cylindri- sun and planet gear, changing meshing ferent distinct misaligned conditions can
cal gearboxes are standardized and estab- conditions are present on a tooth. In sum, be set and the transmission error can be
lished in industry. Misalignments of cen- each tooth on these wheels sees the same measured.
tral elements like carrier, sun gear or ring loads, resulting in homogeneous damage Power Transmission Engineering
gear in planetary gearboxes, cause-vary- over the wheel. One tooth is always sub- (FVA) for the software FE-Stirnradkette.
ing contact positions and variable loads, jected to the same load on the fixed ring The authors gratefully acknowledge sup-
depending on the angular position of the gear. This results in variable damage over port by the WZL Gear Research Circle
central elements. This load, which is vari- the circumference, with some teeth being for the software µOpt. The authors grate-
able over the circumference, is not taken much more damaged than others. If the fully acknowledge financial support
into account in the standardized proce- misalignment of the carrier is not taken by the Federal Ministry for Economic
dures, despite its effects on the loads on into account, the wheels will not be dam- Affairs and Energy (FKZ 0325799) for
the gears. aged at all. If the maximum misalignment the achievement of the project results.
Within the scope of this report, a is assumed in all positions, too high dam- For more information.
method is developed in order to regard age is calculated, except for the ring gear. Questions or comments regarding this paper?
carrier misalignments in the tooth con- The optimization is carried out via the Contact Julian Theling at [email protected]
aachen.de.
tact analysis FE-Stirnradkette. Because of flank corrections on the planetary gear.
carrier position-dependent mesh condi- For this purpose the deviation range, as
tions, a complete tooth hunt is calculated well as the mean value of the resulting
and post-processed afterwards. By means lead angle deviation, are included in the
of a comparison between simulated and calculation in order to take the misalign-
measured tooth root stresses of the FVA ment of the carrier into account. New
nacelle, a validation of the method is suc- characteristic values are generated for
cessfully carried out. the consideration of the load increase
By means of the validated method, over the circumference KHγ as well as the
the influence of carrier misalignments deviation range of the transmission error
on the load carrying capacity and over the circumference ∆ϕγ, and different

54 GEAR TECHNOLOGY | July 2020

[www.geartechnology.com]
References
1. Bastert, C.-C. “Die Verlagerung der Zentralräder
in Planetengetrieben.” In: Forsch Ing-Wes, 37. Jg.,
1971, Nr. 1, S. 21–29.
2. Brecher, C., C. Löpenhaus and D. Piel. “Auslegung
und Untersuchung eines Forschungsgetriebes
für Windkraftanlagen mit der FE-basierten
Zahnkontaktanalyse DMK,” Dresden: TUD press, Prof. Dr.-Ing. Christian Brecher has since
2015. January 2004 been Ordinary Professor for Machine
3. Brecher, C., C. Löpenhaus, J. Theling, M. Tools at the Laboratory for Machine Tools and
Schroers and D. Piel. “FE-Based Method for Production Engineering (WZL) of the RWTH Aachen,
Design of Robust Tooth Flank Modifications for as well as Director of the Department for Production
Cylindrical and Planetary Gear Stages Regarding Machines at the Fraunhofer Institute for Production
Manufacturing Tolerances,” AGMA Fall Technical Technology IPT. Upon finishing his academic studies
Meeting, 2017. in mechanical engineering, Brecher started his
4. Ingeli, J. “Erweiterung der FE-basierten professional career first as a research assistant
Zahnkontaktanalyse zur normkonformen and later as team leader in the department for
Grübchentragfähigkeitsberechnung symmetrischer machine investigation and evaluation at the WZL.
und asymmetrischer Verzahnungen,” Dissertation From 1999 to April 2001, he was responsible for
IIF - Institut für Industriekommunikation und the department of machine tools in his capacity as a Senior Engineer. After
Fachmedien GmbH, 2017.
a short spell as a consultant in the aviation industry, Professor Brecher was
5. ISO 6336 Teil 5. (Juli 2003) Calculation of load
appointed in August 2001 as the Director for Development at the DS Technologie
capacity of spur and helical gears, Strength and
Werkzeugmaschinenbau GmbH, Mönchengladbach, where he was responsible
Quality Materials.
6. ISO 6336 Teil 3. (September 2006) Calculation of for construction and development until December 2003. Brecher has received
load capacity of spur and helical gears, Calculation numerous honors and awards, including the Springorum Commemorative Coin; the
of Tooth Bending Strength. Borchers Medal of the RWTH Aachen; the Scholarship Award of the Association
7. ISO06b ISO 6336 Teil 2. (September 2006) of German Tool Manufacturers (Verein Deutscher Werkzeugmaschinenfabriken
Calculation of load capacity of spur and helical VDW); and the Otto Kienzle Memorial Coin of the Scientific Society for Production
gears, Calculation of Surface Durability (Pitting). Technology (Wissenschaftliche Gesellschaft für Produktionstechnik WGP).
8. Landvogt, A. “Einfluss der Hartfeinbearbeitung und Dr. -Ing. Dipl. -Wirt. -Ing. Christoph
der Flankentopographieauslegung auf das Lauf- und Löpenhaus has since 2014 served as Chief
Geräuschverhalten von Hypoidverzahnungen mit
Engineer in the Gear Department of WZL, RWTH
bogenförmiger Flankenlinie,” Diss. RWTH Aachen
Aachen / Laboratory of Machine Tools and Production
University, 2003.
9. Lubenow, K., F. Schuhmann and S. Schemmert. Engineering (WZL), RWTH Aachen. He previously held
Requirements for Wind Turbine Gearboxes with positions there as (2011–2014) Team Leader, Group
Increased Torque Density with Special Attention to a Gear Testing Gear Department Chair of Machine
Low-Noise Turbine Operation,Conference for Wind Tools Laboratory of Machine Tools and Production
Power Drives 2019, Norderstedt: Books on Demand, Engineering (WZL) RWTH Aachen; (2010–2011)
2019. Research Assistant, Group Gear Testing Gear
10. Miner, M. “Cumulative Damage in Fatigue,” Journal Department Chair of Machine Tools Laboratory of
of Applied Mechanics-Transactions of the ASME, 12. Machine Tools and Production Engineering (WZL)
Jg., 1945, Nr. 3, S. A159-A164. RWTH Aachen: (2007–2009) as Student Researcher,
11. Niemann, G. and H. Winter. Maschinenelemente. Group Gear Design and Manufacturing Calculation Gear Department Chair of
Band 2: Getriebe allgemein, Zahnradgetriebe - Machine Tools Laboratory of Machine Tools and Production Engineering (WZL)
Grundlagen, Stirnradgetriebe. Bd. Nr. 2, 2. Aufl. RWTH Aachen; and (2004–2009) as a student in Industrial Engineering RWTH
Berlin: Springer, 2003. Aachen.
12. Seidel, M. “Auslegung von Hybridtürmen für
Windenergieanlagen - Lastermittlung und Since June 2019 Julian Theling M.Sc. has
Nachweis der Ermüdungsfestigkeit am Beispiel been the leader of the Gear Acoustics group at
einer 3,6-MW-WEA mit 100 m Rotordurchmesser,” the Laboratory of Machine Tools and Production
Beton- und Stahlbetonbau, 97. Jg., 2002, Nr. 11, S. Engineering (WZL). He started his career as a
564–575. researcher at the WZL in 2016 after receiving his
13. Sfar, M. “Bestimmung von Verzahnungskorrekturen master’s degree in mechanical engineering from
und Lagerkräften in Planetengetrieben für RWTH Aachen University. Theling’s research focus
Lastkollektive,” Diss. Ruhr-Universität Bochum, is the influence of flexible surroundings on the
2011. application behavior of planetary gear stages.
14. Theling, J., C. Brecher and C. Löpenhaus. “Einfluss
von Planetenträgerabweichungen auf die Lokale
Lastverteilung in Planetengetrieben,” KISSsoft AG
(Hrsg.): SMK 2018. Dresden: TUD press, 2018.
15. Vriesen, J. “Berechnung der
Verzahnungskorrekturen von Planetenradgetrieben
unter Berücksichtigung der Steg- und
Hohlradverformungen,” Diss. Ruhr-Universität
Bochum, 2001.
16. Wittke, W. “Beanspruchungsgerechte
und Geräuschoptimierte Strinradgetriebe For Related Articles Search
Toleranzvorgaben und Flankenkorrekturen,” Diss.
RWTH Aachen University, 1994. planetary gears
17. Ziegler, A. “Zur verkürzten
at www.geartechnology.com
Systemlebensdauerprüfung von Zahnradgetrieben,”
1, Aufl. Aufl. München: Verl. Dr. Hut, 2011.

July 2020 | GEAR TECHNOLOGY 55