The maintenance and safety of existing bridges is an important concern of all highway
agencies. To assure adequate safety and to assist in assessing maintenance needs. The
objective of the present study is to develop practical fatigue evaluation procedures that:
Realistically reflect the actual fatigue conditions in highway bridges. Give an accurate
estimate of the remaining fatigue life of a bridge and permit this estimate to be updated in the
future to reflect changes in traffic conditions. Provide consistent and reasonable levels of
reliability. Permit different levels of effort to reduce uncertainties and improve predictions of
remaining life. Apply consistently to both the evaluation of existing bridges and the design
of new bridges. Can be conveniently modified to reflect future [Link] suitable for
inclusion in the AASHTO maintenance manual and design specifications.
CONCEPTS OF FATIGUE SAFETY
The concept of safety as applied to repetitive loads that cause fatigue damage is quite different
from the concepts of safety that are applied in the normal rating or strength design of a
bridge with respect to maximum static (nonrepetitive) loads. An understanding of the differences
is needed in selecting suitable fatigue evaluation and design procedures for bridges. Therefore,
the differences are discussed here. A single occurrence of a loading exceeding the corresponding
static strength causes unacceptable permanent damage or collapse due to excessive yielding,
buckling, or fracture. The critical condition for static rating or design is the worst combination
of loads that can occur simultaneously during the life of the bridge; for example, heavy trucks
with full impact effect in the worst positions in all lanes at the same time. Only one occurrence
of this critical condition needs to be considered and must have a correspondingly small
probability of occurrence. For fatigue, many loading repetitions are required to produce
a failure at some time in the future, usually far in the future. Generally, all truck loading stresses,
whether above or below the allowable stress range value, cause fatigue damage that could
result in a failure in the far future. An exception to this is the case in which all stress range cycles
are below the fatigue limit. To achieve adequate safety, static loading must be kept below
the maximum rating or design loading. In contrast, it is not necessary to keep the fatigue loading
below any particular value to assure adequate safety. The only effect of increasing or decreasing
the fatigue loading is to shorten or lengthen the life of the bridge. Therefore, the effects of fatigue
loading on an existing bridge can best be defined in terms of the remaining safe fatigue
life of the bridge. Similarly, the effects of fatigue loading on a new bridge can best be defined in
terms of the total life of the bridge, although it may be convenient to use a permissible stress
range corresponding to a desired design life to facilitate the reproportioning of members that do
not have an adequate life. Safety factors can be applied in calculating the remaining or
total life to assure that the actual life will exceed the calculated life with a desired degree of
probability or reliability.
BACKGROUND
The fatigue-life (remaining life) approach can be applied in several ways to existing bridges
to achieve adequate safety. If an evaluation of an existing bridge reveals that the calculated
remaining fatigue life is less than desired, the Engineer has four options. First, he could
recalculate the remaining life using more accurate data. For example, he could use more
accurate calculations of lateral distributions. Also, he could make a traffic survey to obtain
site-specific data on the volume and weight distribution of trucks rather than using general
values. Second, he could restrict the weight and/or volume of trucks to increase the fatigue
life. Third, he could modify the bridge to improve its fatigue life. For example, if a particular
detail caused the short fatigue life, it could be retrofitted to improve its fatigue
characteristics. Alternatively, the stress level in the bridge could be reduced by adding cross
section or by other means; this would also increase the fatigue life. Fourth, he could institute
periodic inspections at appropriate intervals to assure that fatigue cracks could be detected
before components actually failed. Estimates of the remaining lives of various details would
be helpful in selecting appropriate inspection intervals and allocating
inspection efforts. With any of these four options, the bridge can be easily reevaluated at any
time in the future to reflect changes in traffic or other conditions. An accurate estimate of the
remaining fatigue life of a bridge also has other important uses. Such an estimate is needed
in bridge management systems that are used to make cost-effective decisions regarding
inspection, maintenance, repair, rehabilitation, and replacement of existing bridges.
Estimates of remaining fatigue life would also be very useful in assessing permit-vehicl
policy or determining the effects of permitting a certain class of overloaded vehicles to use
the highways. Similarly, remaining life estimates could be used in assessing legislative
policies such as permissible truck weights. The fatigue-life approach can also be applied in
the design of new bridges. Generally, any particular detail that does not provide the desired
life would be redesigned in the same way that it would be with allowable-stress procedures.
To facilitate there design, it may be convenient to use a permissible stress range
corresponding to a desired safe design life. Instead of redesigning, however, the designer
might alternatively choose to recalculate the life with more accurate data or decide the
calculate life is acceptable for that particular bridge. The latter may be a logical decision for,
say, a rehabilitation job. The estimated life of a new bridge must be based on assumptions
regarding future conditions, especially traffic loadings, which are likely to change
significantly over the 50- to 100-year life of a typical bridge. Since the bridge can be
periodically re-evaluated in the future on the basis of the actual conditions that have
occurred, it is not realistic to impose unnecessarily rigid requirements for the calculated total
fatigue life of a new bridge being designed. Defining the effects of fatigue in terms of fatigue
life is a much more useful approach than imposing rigid allowable fatigue stresses.
The fatigue-life approach provides the following advantages: ( 1 ) it defines the actual effects of
fatigue loadings on a bridge, (2) it permits the Engineer to choose from among
many suitable options to assure adequate safety, (3) it permits the Engineer to react in a rational
way to future changes in fatigue loadings or other pertinent factors, (4) it replaces the
rigid go/no-go approach imposed by allowable-stress procedures with a flexible approach that is
more appropriate for the uncertain and changing conditions that affect fatigue, and (5)
it helps the practicing engineer to recognize that fatigue loadings affect safety in a far different
way than do static (nonrepetitive) loadings.
PROPOSED FATIGUE DESIGN OR EVALUATION
PROCEDURES
Two comprehensive European fatigue specifications have been adopted in recent years the ECCS
fatigue specifications and the British fatigue code . The fatigue provisions of the present
AASHTO design specifications (209) and maintenance-inspection manual (132) were
discussed earlier in this chapter and will not be discussed further here. ECCS For several years
the European Convention for Constructional Steelwork (ECCS) has been preparing
recommendations (172) for the fatigue design of steel structures, which
are intended to apply to highway and railway bridges, crane and machinery supports, and other
structures. The ECCS recommendations follow the AASHTO approach of classifying
structural details according to their fatigue strength; however, ECCS uses 14 different detail
categories instead of the 7 presently used by AASHTO, and includes some details not covered
by AASHTO. Furthermore, ECCS gives a complete design SN curve (stress range vs. life) for
each category instead of allowable stresses corresponding to specific life categories. These curves
are similar to those used to develop the AASHTO allowable stress range values.
The ECCS uses- an effective stress range concept similar to that developed in NCHRP Project
12-12 (198) to define variable- amplitude spectra. This effective stress range is based on
Miner's Law (139). In the fatigue check, the effective stress range corresponding to the design
loading is used with the appropriate SN curve. For highway bridges, ECCS also uses the fatigue-
truck concept developed in the NCHRP project (196, 198); in this concept the variety of trucks in
typical traffic is represented by standard fatigue vehicles. The fatigue check can be made in terms
of either stress or life and rigid allowable stresses are not imposed. A typical ECCS design SN
curve is shown in Figure 2. The dashed horizontal line represents the constant-amplitude fatigue
limit for the detail. If all of the stress cycles in a variable amplitude spectrum are below this limit
the life is assumed to be infinite and no further fatigue check is required. The level of the
horizontal line is set at the point where the sloping finite life portion of the curve intersects a life
of 5 million cycles. It is assumed that the break between the sloping and horizontal portions of an
SN curve occurs at this same life of 5 million cycles for all details. However, this contrasts with
data that suggest the break life increases with the severity of the detail (119). For example, the
break life for AASHTO category E' details is about 22 million cycles (119). If some of the cycles
in a variable-amplitude spectrum are above the constant-amplitude fatigue limit while others are
below, the fatigue life can be determined from the solid curve shown in Figure 2. The sloping
portion that is below the constant- amplitude fatigue limit has a slope of 5. The lower horizontal
line is drawn at the level where this sloping portion intersects 100 million cycles. This lower
cutoff line is at a level approximately 55 percent of the constant-amplitude fatigue limit
for each detail. Alternatively, a slope of 3 may be used between the constant-amplitude fatigue
limit and the cutoff line; this is a simpler and more conservative approach. All stress cycles
below the lower cut-off line may be ignored. For welded members, the ECCS recommendations
treat compression cycles, in which the applied stress varies in magnitude
but is always compressive, the same as tension cycles, in which a portion or all of the applied
cycle is in tension. The rationale for this provision is that tensile residual stresses in
welded members shift the applied compressive stress cycles into the tension region.
British. The recent British fatigue code (212), which applies to both highway and railwa bridges,
gives three different procedures of varying complexity for highway bridges. All three
procedures use the same design SN curves to define the fatigue strengths of various detail
categories. The classification of details into categories is similar to that used in the ECCS
recommendations (172). Also, the finite-life portions of the SN curves for the various categories
are generally about the same as those used by ECCS (172). In the British code, however, the
constant amplitude fatigue limit for each category is taken as the stress range at a life of 10
million cycles and the SN curve is projected below this stress range at a slope of 5. There is no
cutoff leve below which stress cycles have no effect. For welded members,
the British code treats compression cycles the same as tension cycles.
In the simplest of the three procedures, an effective applied stress range for each detail is
calculated by loading the bridge with standard fatigue trucks that represent typical truck traffic.
This calculated effective stress range must be below an allowable stress range corresponding to a
life of 120 years and to a truck volume that is defined for various types of highway. Each truck
passage is assumed to cause one stress cycle. The highest of the specified truck volumes is 2
million per year in one lane; this corresponds to 240 million stress cycles over the expected 120-
year life of the bridge. The effective stress range and fatigue truck concepts used in this
procedure are based on concepts developed in a previous NCHRP study (198) and are consistent
with Miner's Law (139). The other two procedures are much more complex and permit the direct
calculation of the fatigue damage caused by each truck passage. Miner's Law (139) is used to
assess the cumulative fatigue damage caused by such passages.
