National Academies Press: OpenBook

Fatigue Evaluation of Steel Bridges (2012)

Chapter: Chapter 2 - Research Approach

« Previous: Chapter 1 - Background
Page 7
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 7
Page 8
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 8
Page 9
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 9
Page 10
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 10
Page 11
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 11
Page 12
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 12
Page 13
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 13
Page 14
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 14
Page 15
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 15
Page 16
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 16
Page 17
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 17
Page 18
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 18
Page 19
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 20
Page 21
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 21
Page 22
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 22
Page 23
Suggested Citation:"Chapter 2 - Research Approach." National Academies of Sciences, Engineering, and Medicine. 2012. Fatigue Evaluation of Steel Bridges. Washington, DC: The National Academies Press. doi: 10.17226/22774.
×
Page 23

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7 A review of the history of provisions for fatigue design and evaluation of bridges was performed in order to identify spe- cific provisions that can be improved, as well as to identify items that need to be researched further so that new provi- sions handling such items can be included. An effort was also spent in reviewing fatigue evaluation provisions of other countries in order to compare and contrast with existing pro- visions. Also, a survey was sent out to state DOTs, agencies, Canadian Provinces, and selected consultants in order to gain insight into the various issues related to bridge loadings and the effect they have on the assessment of the fatigue strength of bridge structures. History of AASHO/AASHTO Provisions for Fatigue Design and Evaluation of Bridges The first fatigue design provisions of the AASHTO Stan- dard Specifications for Highway Bridges appeared in 1965. In the 10th edition (1969) in Section 1.7.3 Fatigue Stresses, an allowable fatigue stress range was determined as a func- tion of loading, highway classification, detail type, strength of steel, and ‘R’ ratio (i.e., the algebraic ratio of the mini- mum stress to the maximum stress, which was not neces- sarily a live load stress range, as the minimum stress could have been produced by dead load). Eleven detail categories were defined, completely different than in use today, and are known as categories A through K. In the 11th Edition (1973), the fatigue provisions remained essentially unchanged with little modifications. The 12th Edition (1977) contained a completely revised approach for fatigue design of highway bridges. These changes were essentially a direct result of the NCHRP research con- ducted by Dr. John Fisher and colleagues. Section 1.7.2, titled “Repetitive Loading and Toughness Considerations,” was added and addressed both fatigue and fracture. This section of the Standard Specifications contained guidance related to the fatigue design of common steel details found in highway bridges. The Specifications contained illustrations of common bolted and welded details that are known to be fatigue sensi- tive, along with their associated fatigue resistance. These illus- trations remain essentially unchanged to this day. The details were grouped into categories of similar fatigue resistance that were labeled ‘A’ through ‘F,’ with ‘A’ being the highest. The most significant change was the introduction of the stress range concept for fatigue design. The results of NCHRP studies confirmed that for welded details, fatigue life was primarily a function of stress range, detail category, and the number of applied cycles. The other parameters previ- ously included in the earlier specifications, such as material strength and ‘R’ ratio, had no significant effect on the fatigue life of welded details commonly used in bridge construction. The allowable stress range at a specific number of loading cycles was provided for each of the new categories in Table 1.7.2A1. These effectively define the allowable fatigue resis- tance based upon the stress range concept. The fatigue endur- ance limit was also defined. Section 1.7.2 of the 12th Edition of the Standard Specifications also provided guidance on the number of cycles for which a given bridge member should be designed (indirectly this is the design life). Similar to the previous versions, the required design life is a function of member type, highway classification, and ADTT. This also includes infinite-life design. Infinite-life design is required in cases where a very high number of cycles are expected and/or no fatigue cracking can be tolerated. The fatigue provisions changed very little between the 14th and 16th editions of the Standard Specifications. The AASHTO LRFD Bridge Design Specifications, introduced in 1994, incorporated a reliability-based approach to all aspects of design related to highway bridges. The fatigue provisions were substantially revised, with the most significant changes being made to the load model used for fatigue design. The illus- trative examples and detail resistance (i.e., CAFL [constant amplitude fatigue limit]) essentially remained unchanged. C h a p t e r 2 Research Approach

8The fatigue life estimate also accounts for the growth in truck traffic volume. In the LRFR Manual, the correlation is made to the average lifetime daily truck traffic for a single lane. A figure is provided to correlate the present [(ADTT)SL]PRESENT with the lifetime average daily truck traffic (ADTT)SL. A section is also provided in the LRFR Manual with a number of differ- ent strategies to increase the remaining fatigue life, should it be deemed undesirable. The commentary indicates that retro- fit or load-restriction decisions should be based upon use of the Evaluation Life rather than the minimum life. Use of the mean life (with a 50% probability of failure) for remaining life estimates is also permitted if the estimate from the Evalu- ation Life is still unacceptable. Additional options that are prescribed include the recalculation of the fatigue life using more accurate data as input to the fatigue life estimate. These include improvements in the effective stress range, effective truck weight, the average daily truck traffic, or the number of cycles per truck passage. The fatigue life can also be improved by retrofitting the critical detail to improve the detail category rating and thereby increase the fatigue life. A section is provided in the LRFR Manual that deals with distortion-induced fatigue evaluation. The section is quite brief, and it indicates that distortion-induced fatigue is typi- cally a low-cycle fatigue phenomenon, with few stress range cycles needed to initiate cracking at distortion-induced prone details. The provisions state that “distortion-induced fatigue is a stiffness problem (more precisely the lack thereof) ver- sus a load problem.” No provisions are provided for how to address distortion-induced cracking when it occurs. Another addition to the LRFR Manual requirements is the prescrip- tion of a required fracture mechanics analysis if fatigue cracks have been visually detected. Alternatively, retrofitting mea- sures are recommended once fatigue cracking is detected. Section 7 titled “Fatigue Evaluation of Steel Bridges” of AASHTO’s The MBE was first issued in 2008 with a second edition in 2011 that combined the Manual for Condition Evaluation of Bridges, Second Edition (2000) and its 2001 and 2003 Interim Revisions with the Guide Manual for Condition Evaluation and LRFR of Highway Bridges, First Edition and its 2005 Interim Revisions. With this, all the previous bridge evaluation titles were archived by AASHTO. Section 7 of the MBE has been directly adopted from the LRFR Manual with minor changes in referencing. Review of Fatigue Specifications in Other Countries An effort was spent on collecting and reviewing selected fatigue design and evaluation specifications of other coun- tries representing the state of the practice in the world. A review of Eurocode and Australian specification codes is pre- sented in the following sections. The specification utilized the concept of an “effective fatigue truck” of prescribed loading and axle spacing. The Guide Specifications for Fatigue Evaluation of Exist- ing Steel Bridges (1990)—or Guide—was a significant devel- opment that introduced a comprehensive method to evaluate the fatigue life of steel bridges. The fatigue evaluation proce- dures in the Guide Specifications were developed in NCHRP Project 21-83 and presented in NCHRP Report 299 by Moses et al. (1987). The procedures provided an alternative to the design specification requirements, which were not well suited to the evaluation of existing bridges. Section 7 of the AASHTO Manual for Condition Evaluation and LRFR of Highway Bridges (2003) (or LRFR Manual) repre- sents a notable update of the AASHTO Guide Specifications that were issued in 1990. Language was added denoting the difference between load-induced fatigue versus distortion- induced fatigue. For load-induced fatigue damage evalua- tion, the fatigue requirements in the LRFR Manual utilize the equations and categories in the AASHTO LRFD Bridge Design Specification rather than the comparable values in the AASHTO Standard Specifications that were used by the Guide Specifications. Nevertheless, the same calibration was used to establish the values for the fatigue S-N curves for both the Guide Specification and the LRFR Manual. The LRFR Manual further indicates that the effective stress range shall be estimated as either the measured stress range or a calculated stress range value determined by using a fatigue truck as specified in the LRFD Bridge Design Specification, or a fatigue truck determined by a truck survey or weigh- in-motion study. The LRFR Manual makes use of partial load factors that adjust the stress range as uncertainty in the estimate is reduced as a result of improved analysis or site- specific information. The lowest possible partial load factor is associated with the use of measured strains to obtain the stress range values. Once the effective stress range has been determined, then a check is made to determine whether or not the details are prone to load-induced fatigue damage. In the LRFR Manual, a resistance factor, RR, was used in the fatigue life expression to account for one of the three types of life estimates being determined: a minimum expected fatigue life that would be conservative and be equal to the design fatigue life, an evaluation fatigue life that would give a con- servative estimate of fatigue life, and the mean fatigue life, which is the most likely fatigue life estimate. A table is given for RR to account for the appropriate type of fatigue life value being estimated for each of the various AASHTO detail cat- egories. According to the Commentary of the LRFR Man- ual, the probability of failure associated with the fatigue life approaches 2%, 16%, and 50% for the minimum, evaluation, and mean fatigue lives, respectively. This represents an offset of one and two standard deviations from the mean fatigue life for the evaluation and minimum fatigue lives, respectively.

9 Eurocode Specification The Eurocode (2005) has been revised since its first ver- sion reviewed in NCHRP Report 299 by Moses et al. (1987) when the Guide Specification was developed. It also uses the S-N curve concept for assuring an adequate life of bridges considering steel fatigue. Different from the AASHTO coun- terpart for normal stress ranges in members, the slope of the S-N curves is not a constant for the entire life range or stress range. Two slopes, 3 and 5, are used for different regions of the curves. In addition, 14 fatigue categories are used instead of 8 as in the AASHTO MBE (2011). Furthermore, for shear stress ranges, another set of S-N curves are provided with a constant slope of 5. The S-N curves for normal stress ranges have different infinite-life limits for constant and variable amplitude cases, whereas those for shear have only one limit for both cases. The Eurocode specification also allows the use of the hot-spot stress method. Lastly, the provisions in the Eurocode for non-welded details or stress-relieved welded details allow the mean stress influence on fatigue strength to be taken into account by determining a reduced effective stress range DsE,2 (corresponding to 2 million cycles) in the fatigue assessment when part or all of the stress cycle is com- pressive. The effective stress range may be calculated in the Eurocode by adding the tensile portion of the stress range and 60% of the magnitude of the compressive portion of the stress range. Australian Specification The Australian bridge design standard (Council of Stan- dards Australia, 2004a) and rating standard (Council of Stan- dards Australia, 2004b) both contain provisions regarding the fatigue failure mode in steel bridge components. The S-N curves in the Australian specifications are similar to those in the Eurocode as discussed above. Nevertheless, one more fatigue strength category is adopted in the Aus- tralian standards, which is at the high-strength end of the spectrum whose failure (cracking) perhaps more likely will not be observed very often. Two different slopes of the S-N curves are included, as in the Eurocode, as well as two lev- els of limit stress for infinite life, one for constant, and the other for variable amplitude stress variation. The Australian standards also include S-N curves for shear stress ranges appearing to be identical to those for the Eurocode. The Australian standards also explicitly include bolts and shear studs into the fatigue strength categorization, being differ- ent from the Eurocode. Note also that Australia is the only country found in this effort of literature review to have a separate bridge rating standard covering fatigue, although the rating procedure refers to the design S-N curves without many additional provisions directly dealing with the variety of situations possibly encountered in rating. Some examples of such situations may include, but are not limited to, how to use truck load records when an inadequate life is found, or what needs to be done when a fatigue crack is observed in the bridge. Survey Surveys were received from 30 of the more than 80 sent out, with 26 received from DOTs and other agencies and 4 from fatigue experts. A blank copy of the DOT survey, a sum- mary of the DOT survey responses, and a blank copy of the fatigue expert survey is included in Appendix A. Summary of DOT Survey Results The results of this survey can be summarized as follows: 1. It is extremely rare for states to perform regularly sched- uled fatigue evaluations. While states acknowledge that many common types of bridges require fatigue evalua- tions, they are only performed when cracks are discov- ered during inspections. Several states do not perform any fatigue analyses whatsoever. The reasons given are low traffic volumes and the absence of fatigue issues in the past. 2. When considering requirements utilized for fatigue eval- uation, the Guide Specification is more popular among states, and those states that utilize Section 7 of the Man- ual indicate that they also use the Guide. The majority of states feel that these two documents meet their cur- rent needs, but some states who are content feel that the evaluations are often too conservative. The MBE had just recently been published when this survey was car- ried out. 3. The majority of agencies make use of field measurements in their evaluations. More states reported using external consultants, but many also make use of in-house pro- grams. Roughly one-third of states reported collecting and using WIM data. 4. Almost all states have observed distortion-induced fatigue cracking and the majority indicated that it is more com- mon than load-induced fatigue. Two types of retrofit/ repair details are commonly used to address distortion- induced fatigue cracking. Most states “soften” the detail by removing rivets or bolts and drilling holes to relieve stresses and arrest cracks. A smaller group of states will stiffen the detail by welding or bolting the vertical stiffener to the flanges of the girder. 5. Nearly all states retrofit details that have cracked from load-induced fatigue. The method used by all states is almost identical. Holes are drilled to arrest the crack, and the section is stiffened by adding bolted splice plates.

10 Summary of Identified Fatigue Expert Survey Results A short questionnaire was sent out to identified fatigue experts in order to gain useful information and opinions. The first question inquired about recent test data that may be useful for revising Section 7 of the Manual. All four experts pro- vided some information on possibly useful experimental data. The second question solicited the expert’s opinion about which parts of the Manual (or other fatigue design and evalu- ation specifications) should be revised. The intention of the question was to identify areas of the Manual that need to be revised and/or updated considering the latest advancements of knowledge. One expert did not identify any areas in the current specifications. Another expert identified the need to address the issue of very few stress cycles above the CAFL known to produce fatigue failure, and this risk is not covered by CAFL. The third expert pointed to the same issue of very few but very high stress cycles, also related to an inconsistency on this subject between the design and evaluation provisions in the two sets of AASHTO specifications. The fourth expert pointed out that the Manual does not include non-destructive evaluation informa- tion as a tool to characterize fatigue damage. The third question in the questionnaire specifically solicited ideas to deal with possible negative remaining life. One expert expressed the importance of refined analysis using a 3-D model of the bridge. Another expert agreed with this, but also pointed to the impact factor of 15% as overestimating dynamic stress effects but expressed a belief that these two factors alone (the overestimated static load effect and impact factor) still are not fully responsible for the problem. A third expert believes that the stress estimation must be wrong when a negative remain- ing life results, and the fourth expert offered no opinion. The last question of the questionnaire was an open solici- tation for any suggestions or comments that the expert may wish to express regarding revisions to the fatigue evaluation requirements of existing bridges. Two of the experts indi- cated nothing, a third one expressed support of the needed revision of the Manual, and the fourth one suggested includ- ing methods for estimating the effective stress range. Identify Critical Issues and Needed Research A review was conducted to decide specific directions for revising and updating Section 7 of the MBE. The review involved identification and prioritization of issues and proce- dures needing revision based on existing knowledge. Where additional research is needed to revise the MBE, a number of items were individually identified for revision based on the literature review and survey results. These items and factors are noted herein and briefly discussed. The S-N Curve The S-N curve is the foundation underlying the fatigue evaluation of steel bridges in the MBE. The curves were devel- oped primarily based on fatigue test data under constant amplitude cyclic loading. These curves were the result of a sig- nificant effort as part of NCHRP Project 12-15(5) by Keating and Fisher (1986) published as NCHRP Report 286 to gather available fatigue data and establish a consistent design meth- odology. They drew heavily from experimental fatigue test data reported in NCHRP Report 102 (Fisher et al. 1970) and NCHRP Report 147 (Fisher et al. 1974) on steel bridge details. A new set of fatigue curves was established with a consistent slope of -3.0 to better model the behavior of welded bridge details. Since the development of the MBE, technical advancement has been made for understanding long-life fatigue behavior under low-magnitude and variable amplitude cyclic loading. The current use of the S-N curves with a linear extension below the CAFL to account for long-life behavior was exam- ined, and the behavior contrasted with that obtained by using S-N curves as in the Eurocode and the Australian code which utilize multiple slopes. The impact of changing the S-N curves from the present behavior to one that more closely follows the practice in foreign countries also was examined in terms of the accuracy of the predicted cyclic life. It also should be noted that S-N curve development based on constant ampli- tude stress range testing results is different from that using variable amplitude test data, because the latter involves a new dimension of uncertainty associated with the load effect. It is worth mentioning that in bridge applications, this uncer- tainty is often more pronounced than that associated with the strength that is treated as the only uncertainty in S-N curve regression using constant amplitude test data. Unnecessary Approximation Leading to Unreliable Estimates In Article 7.2.5, “Estimating Finite Fatigue Life,” of the Man- ual (AASHTO 2003), or equivalently in Article 3.2 of the Guide (AASHTO 1990) and Article 7.2.5 of the MBE (AASHTO 2011), the following equation is provided to compute the estimated finite fatigue life Y of a fatigue-prone detail. Y R A n ADTT f R SL eff = ( ) ( )( )365 13∆ ( ) As seen, ADTTSL in the denominator is required to find Y, as the average number of trucks per day in a single lane aver- aged over the fatigue life Y. Since the fatigue life Y is being sought, ADTTSL cannot be found directly. Conceptually, ADTTSL needs to be found by iteration, using this equation. Nevertheless, the commentary to Article 7.2.5 in the Manual

11 (and Article 3.5 in the Guide) recommends an approxima- tion using the chart in Fig. C7-1 in the Manual (Figure 3.5A in the Guide) to estimate ADTTSL, and then in turn Y. This chart, as seen in Figure 1, does not include the unknown Y but only present age A (a in the commentary) that is only part of Y. Therefore, this chart contains an approximation, which can be reduced to enhance the reliability of estimation. Negative Remaining Fatigue Life A negative fatigue life occurs when the bridge age exceeds the predicted cyclic life. This situation is conceptually depicted in Figure 2. The current Section 7 of the AASHTO MBE (2011) provides a couple of options for handling situations where the evaluation computation results in a negative remaining life (estimated) and field inspection has observed no fatigue cracking for the particular steel connection detail in the bridge. These options involve accepting a greater degree of risk, using more accurate data to compute fatigue life, or retrofitting the detail altogether. As illustrated in Figure 2, the real fatigue life of the detail is a random variable, expressed by the probability distribu- tion curve shown with little triangle symbols. The total life estimated according to the MBE is a deterministic value, also indicated on the abscissa. It is defined in the calibration pro- cess of the specifications (NCHRP Report 299 by Moses et al. [1987]) as a value, up to which the failure probability (the shaded area) is equal to the failure probability corresponding to the target reliability index. In other words, the probability of the real life being smaller than this value is controlled to be under the targeted (acceptable) risk level associated with the target reliability index. Due to conservatism required in the process and a significant amount of uncertainty involved, sometimes this estimated total life is so small that subtract- ing the present age of the detail (bridge) from it results in a Figure 1. Figure C7-1 in the manual for estimating ADTTSL. Figure 2. Negative remaining life resulting from uncertainty in fatigue life estimation (shaded area is equal to targeted failure probability).

12 negative remaining life (also as an estimate). As indicated in Figure 2, the estimated remaining life ends at the left side of the origin, being negative. As symbolically indicated in Figure 2, the fatigue life distribution is widely “spread,” modeling a wide range of random variation or uncertainty. Therefore, with- out further information, it is difficult to insure the estimated remaining life to be positive. When it is known that no fatigue cracking has been iden- tified for the detail, this information should be taken into account in the evaluation process. In other words, it would be desirable if the information for suitable performance (no cracking) could be used to revise the life estimate with the same level of risk maintained. Multiple Presence Factor Truck loading is the primary cause of steel bridge fatigue damage. Heavy trucks may appear on a highway bridge span in one or more lanes simultaneously. Depending on the structure’s configuration, trucks in multiple lanes can induce much higher load effects to bridge components than those in one lane. Therefore, the governing load effects due to these loading configurations need to be addressed in bridge evalu- ation. However, the AASHTO MBE (2011) prescribes an approximate approach to counting truck loads for their effect in inducing fatigue failure. It specifies that only the truck load on the shoulder lane needs to be counted for fatigue evalua- tion. This decision was made without rigorous investigation (NCHRP Report 299 by Moses et al. (1987)) and also perhaps because there were almost no WIM data available at the time. Regarding multiple presence of trucks on a bridge span, it is obvious that ignoring trucks in other lanes can lead to under- estimation of the real load. This effect can be significant espe- cially when a two-girder, two-truss, or two-arch system with multiple lanes is concerned. In addition, when the shoulder lane carries relatively less traffic, i.e., when other lanes carry relatively more truck traffic, such under-estimation of stress can also be more pronounced. Note that the current AASHTO MBE (2011) prescribes a uniform 80% to be the percentage of the total traffic carried by the shoulder lane, which has been verified to be often an overestimate of the shoulder lane’s truck traffic. This subject has been investigated in this research project using a large amount of WIM data. Lack of Detailed Guidance for Field Measurement as an Option Field measurement of load effect (in strain, displacement, etc.) for the stress range histogram is a practical approach to reducing uncertainties associated with load effects in a fatigue evaluation. The MBE (AASHTO 2011) has listed this method as an alternative method to determine the effective stress range. However, there is a lack of sufficient details for the user to adequately employ this method for producing consistent results. This situation results in significantly dif- ferent, or sometimes inconsistent, results depending on the individual who performs the work. The method used to col- lect strain data is critically important. Clearly, the collection of incorrect or inadequate strain data could lead to signifi- cant errors if the fatigue life is predicted to be significantly greater than the true behavior. Consequently, it is anticipated that additional guidance is needed to describe the minimum procedures to be used when collecting strain data to assess the fatigue strength. For example, the use of detailed analysis and/or proper modeling should be considered as a tool to assist in determining proper gage locations. Lack of Guidance on Tack Welds and Riveted Members The MBE (AASHTO 2011) does not include any provi- sions or guidance on treating tack weld induced cracks. The MBE mentions riveted members but refers to the LRFD spec- ifications for details. The information provided in the latter is not adequate for evaluation, either. Tack welds are mostly in existence on many bridges having built up sections, such as trusses. They often have been used to temporarily hold members in place before they are riveted, bolted, or welded. Tack welds have many start/stop locations during placement and therefore run the risk of weld flaws at the weld termination. Often these welds do not have a par- ticularly high quality since they were not detailed or designed to carry any measurable load per se but are only placed to facilitate the fabrication. Cracking of the tack welds through the throat of the tack weld will probably not pose a great danger to the structure. In cases where the tack weld is only partially cracked through the throat, it may be prudent to grind off the tack weld before the crack has a chance to propagate into the base metal of the primary structural element. Although certainly not common, there are documented cases where a fatigue crack has devel- oped at the toe of a tack weld. Figure 3 shows some examples of typical tack welds used in bridges. Also shown in the figure is an example of a tack weld that has developed a fatigue crack through its throat. Tack welds are currently classified as Category E details with a corresponding very low fatigue strength. However, data are insufficient to support this classification. As a matter of fact, one of the reasons for this classification originally was to discourage the use of tack welds. This category likely underes- timates their fatigue strength. If the fatigue strength is actually higher than Category E, and the tack weld presents little risk to seriously damage the primary structural members, then con- siderable cost savings may be realized if the need to repair tack welds is delayed or avoided altogether. This can be significant

13 for some older riveted structures that may have literally hun- dreds of tack welds that may need to be ground off otherwise. The survey sent out to state Departments of Transportation contained two questions specifically addressing tack welds. The first question asked if fatigue cracks associated with tack welds had been observed in any bridges. A total of 13 states out of the 23 that answered the question, or slightly more than one-half, indicated that they had observed fatigue cracks at tack welds. Unfortunately it was not clear if the fatigue cracks reported were simply a severing of the tack weld through the throat, which is fairly benign, or if the fatigue crack was at a weld toe, which can be potentially serious. Nevertheless, numerous fatigue cracks were detected at tack weld details. The second question asked if the agency performed a fatigue evaluation for tack welds. In this case, only three states of the 23 who responded to the question answered in the affirma- tive. Hence, it appears that most states are not thoroughly evaluating the tack weld details at present. The literature search located very few data collected to assess tack welds. Therefore, laboratory testing was con- ducted to better quantify their fatigue strength and their influence on built-up members. (a) (b) (c) Figure 3. Examples of tack welds used in bridge structures: (a) tack weld left in place after riveting, (b) tack weld throat crack, and (c) transverse tack weld.

14 Lack of Guidance for the Evaluation of Distortion-Induced Fatigue Cracks The AASHTO MBE (2011) indicates that distortion- induced fatigue is more of a stiffness problem (or lack thereof) than a loading problem. The frequency of distortion-induced fatigue cracking is quite extensive. In fact, Connor and Fisher (2006) estimate that 90% of all fatigue cracking is the result of out-of-plane distortion at fatigue sensitive details. Common distortion-induced fatigue cracking sites include the web gap region at the end of vertical stiffeners or connection plates for floor beams, or in the web gap region of lateral gusset plates that intersect vertical connection plates. An example of distortion-induced fatigue cracking at the ends of vertical connection plates is shown in Figure 4. An experimental examination of distortion-induced fatigue damage was reported by Fisher et al. (1979) in NCHRP Report 206. In this study, several types of fatigue damage at various fatigue susceptible details were investigated, includ- ing welded partial length cover plates, vertical stiffeners, and web penetrations by a flange. The fatigue testing demon- strated that for cyclic out-of-plane displacement of web gap regions, the fatigue strength increased as the web gap length increased. The relationship between web gap length and the initiation of fatigue cracking, however, was found not to be directly proportional. Small web gaps (less than five times the web thickness) were found to have very erratic behavior under cyclic out-of-plane displacement. Fatigue cracks at the end of the stiffeners for the normal beam testing were found to be satisfactorily retrofitted by using drilled holes. A comprehensive study was conducted by Fisher et al. (1990) in NCHRP Report 336 to examine various types of distortion-induced fatigue cracking. The goal of this NCHRP study was to develop recommended criteria for designing steel girders so that distortion-induced fatigue cracking problems are minimized. Both transverse connection plate and lateral gusset plate details were studied. Experimental tests of speci- mens that simulated details used in practice were conducted to evaluate the structural behavior. It was found that drilled retrofit holes were effective in arresting fatigue crack growth when they conformed to the following relationship: ∆K y y ρ σ σ< ( )4 2for in ksi units ( ) where r is the retrofit hole radius, sy is the yield strength of the steel, and DK is the stress intensity range for a given nomi- nal stress range level, Sr. This relationship was developed by Fisher et al. (1980) in NCHRP Report 227 in a study of the fatigue behavior of welded bridge attachments. The retrofit used for positive attachment of the lateral connection detail consisted of a WT section that was bolted to both the flange of the beam and the transverse connec- tion plate. No retrofit hole was used along with the WT section that was bolted to the flange and transverse plate to evaluate the effectiveness of the bolted attachment. It was found that little additional crack growth occurred and that the positive attachment was effective in repairing the transverse connection plate detail. Of course, the small additional crack growth that did occur could be completely eliminated if a retrofit hole were used together with the bolted positive attachment. A number of observations were made from the cyclic testing program. First, the stress gradient for the web gap region varied significantly and produced larger stresses at (a) (b) Figure 4. Distortion-induced fatigue cracking at (a) lower end of vertical connection plate and (b) upper end of connection plate.

15 the transverse connection plate weld toe than at the lateral gusset plate weld toe. Second, for an in-plane stress range of 41.4 MPa (6 ksi), no cracking was detected for details A and B after 10,000,000 cycles of loading. Third, fatigue cracks did develop at the weld toe of the transverse stiffener opposite the gusset plate for detail C, where there was not a positive attachment of the lateral gusset plate to the transverse stiff- ener. Fourth, when the data were plotted on an S-N curve, with the stress plotted on the basis of the sum of the in-plane flexural stress plus the estimated out-of-plane bending stress, then all of the data exceeded the AASHTO category C level. Fifth, drilled holes at the crack tips were found to arrest the crack growth when the stress range levels did not exceed 138 MPa (20 ksi); higher stress levels require a positive attach- ment between the gusset plate and the transverse stiffener, even for large web gaps. A couple of additional comments in this study are worth noting. They mention that fitting gusset plates around a transverse stiffener without a direct attach- ment does not appear to be desirable. Moreover, even with a positive attachment, the web gap between the weld toes should be at least four times the web thickness. Large stress gradients can occur in the gap region, even for very small gaps. Lastly, it is stated that the intersection of the gusset longitudinal weld and the transverse stiffener weld must be avoided. Weld shrinkage strains will produce severe restraint and contribute to the possibility of weld discontinuities and inclusions at the weld intersection. Additional test data on the repair of distortion-induced web cracking at a vertical connection plate for variable ampli- tude, long-life cyclic loading is also provided in NCHRP Report 354 (Fisher et al. 1993). The report states that: The test results suggest that fatigue cracks are not likely to develop at transverse stiffener details in actual bridge structures unless out-of-plane distortion develops. The studies on out-of- plane distortion of transverse connection plates confirmed the findings given in NCHRP 336. Rigid connections of the plate to the top and bottom flanges by bolted or welded connections are needed to prevent fatigue cracks from out-of-plane deformation. Proper retrofit procedures to address and mitigate distortion- induced cracking are critically important. General methods for evaluating and retrofitting bridges, as well as a case study involving out-of-plane distortion cracking at the web gap, are presented by Connor and Fisher (2006). Web gap details are generally classified as Category C for fatigue strength evaluation. While this may be adequate for in-plane bending stresses, the out-of-plane distortions that occur make this a fatigue-prone detail. Methods of instrumentation discussed include gage placement and type as well as monitoring times and methods. A case study is reviewed of a three-span contin- uous haunched plate girder bridge which had been retrofitted once the fatigue cracks had been discovered at the ends of the vertical stiffeners and at the gusset plate to stiffener details. Angles were attached to the top flange of the girder and to the transverse connection plates. They were also installed between the lateral gusset plates and the intersected trans- verse stiffeners. The results of this instrumentation and mon- itoring were used to determine that several existing details required retrofitting and that some previously installed ret- rofits were not effective. Specifically, the 3 × 3 × 3⁄8 angles attached with only two bolts at the gusset plate to the stiffener web gap were not capable of providing sufficient rigidity to fully mitigate fatigue cracking. The web gap at the top of the transverse connection plate detail was retrofitted with an 8 × 8 × 5⁄8 angle connected with four bolts that resulted in a sig- nificantly stiffer connection. Measurements obtained at the top web gap at three separate locations where heavier angles and more bolts had been installed indicated that the angle retrofit was sufficiently stiff to decrease distortional displace- ments and subsequent stresses to a level that was effective in mitigating further fatigue cracking. As a general guideline, Connor and Fisher conclude that for retrofitting transverse connection plates, heavy back-to-back angles (19 mm mini- mum thickness) or comparable WT sections be used with four high-strength bolts in each leg. The main point is that retrofits for out-of-plane distortion must provide suffi- cient stiffness to prevent relative deflection and distortion between adjacent components. The survey that was sent to all of the state Departments of Transportation confirmed that the most commonly observed fatigue cracks in steel bridges are those due to distortion- induced localized stresses. In practice, there have been two general approaches for repairing distortion-induced fatigue cracks: stiffening or softening. Either approach may work satisfactorily if designed and installed properly. On the other hand, neither approach will work if not done properly in design and/or installation. The stiffening approach is to stiffen the distorted web and other elements of the detail. The key to the stiffening approach is to properly assess the required connection forces and provide a retrofit that has not only sufficient strength but, more importantly, very high stiffness to eliminate localized deformation. On the other hand, the key to the success of the softening approach is to properly analyze the load paths as well as the overall behav- ior of the structural system so that the softened structure will not cause adverse effects by allowing excessive local defor- mations. If done properly, the softened structure is flexible enough to deform under the required distortion so that the induced stresses do not cause significant fatigue damage. The MBE does not include proscriptive provisions as to how distortion-induced fatigue cracking can be evaluated or treated. Section 7 of the MBE contains a separate subsection for distortion-induced fatigue. The section indicates that distortion- induced fatigue is a stiffness problem rather than a load problem.

16 However, no guidance is provided on how to evaluate or retrofit details that have developed distortion-induced fatigue cracks. Additional information is needed on how to evaluate details susceptible to distortion-induced fatigue cracking. Consider- able information is already available in the literature on use of the hole drilling and softening method to retrofit cracked details. However, more proscriptive recommendations on methods to design effective retrofit details for stiffening the web gaps of ver- tical connection plates would be helpful to increase the service life of details that have experienced distortion-induced fatigue cracking. Consequently, a series of distortion-induced fatigue tests were conducted to provide additional information on the behavior of the connection. Fatigue Serviceability Index The fatigue serviceability index (FSI), Q, is one of the meth- ods envisioned to assess the fatigue serviceability limit state. The FSI provides a relative measure of the performance of a struc- tural detail, at a particular location in the structure, with respect to the overall fatigue resistance of the member. Although the remaining fatigue life is used to determine the FSI, the final value for the fatigue serviceability index is dimensionless. Rela- tive values of this coefficient are intended to be used to charac- terize the overall serviceability relative to the fatigue limit state. Based on a combination of the quantitative value of the FSI and the overall qualitative assessment, engineers can make planning decisions regarding the viability of given bridges in their bridge inventory. Additional Factors Two additional factors that will be considered further for incorporation into Section 7 of the Manual are the use of fracture mechanics and hot-spot methods to assist the user in estimating the remaining cyclic life, which will then in turn be used to evaluate the FSI. Experimental Setup and Test Procedures Two of the factors identified previously require experi- mental testing. They are tack weld details and retrofits for distortion-induced fatigue cracks. The test setups and proce- dures that were implemented for the testing of these types of details are described in the following sections. Tack Weld Tests The tack weld tests involve a simple lap connection loaded in tension. A pair of plate members is attached to a test central plate using tack welds on the sides of the plate members. A typical tack weld specimen is shown in Figure 5. Various vari- ables were examined in the tack weld program. These include the number of tack welds along the side of the bolted plates, the length of the tack welds, the stress range applied, the position of the tack welds relative to the adjacent bolt hole, longitudinal vs. transverse position of the tack weld, and the effect of the bolt clamping force. All steel plates are ASTM A36. The purpose in examining different numbers of tack welds along the side of the bolted plates is to examine the degree to which the overall stiffness of the connection is influenced by the number of tack welds. It is well known that there is a shear lag effect that can occur in a long weld. The stresses in a long weld, or even a long bolted joint, tend to be greater at the ends than near the middle. If this influence is valid here, then cracking at the ends may be more likely for the three tack weld configuration than for the two tack weld geometry. The tack weld length is also examined. Most of the tack welds for the test program are about one inch long. If the “tack welds” are too long and placed in series, then they would be better classified as intermittent welds, rather than as tack welds. To evaluate this difference, a few tests were conducted with weld lengths of about 1.5 inches to see if there is a discernable difference in the fatigue strength. The tack welds were deposited on the plates at room tempera- ture with no significant pre-heat. Since the weld cooling rate is related to the hardness of the material in the heat- affected zone adjacent to the weld toe, the short tack welds will probably cool more rapidly than longer tack welds, and thus may be more susceptible to cracking in the heat- affected zone. Most of the tack welded connections were tested with tack welds parallel to the lines of fasteners in the joint. However, a few specimens were tested with the tack welds along the end Figure 5. Typical tack weld test specimen.

17 of the connection in a line perpendicular to the fasteners. It is well known that welds are significantly more ductile when they are oriented parallel to the direction of the loading rather than perpendicular to the load. It is expected that the tack welds along the end of the connection will crack through the weld throat rather quickly and not play a significant role in the fatigue life of the connection. Nevertheless, this factor was examined. Lastly, the role of nominal stress range on the “lightly” bolted joint is examined. The fatigue resistance of a connec- tion with tack welds is often anticipated to be Category E. Hence, to evaluate overall cyclic performance, two different stress range levels are examined: one at about 20 ksi, which is likely in the finite-life region, and one at about 12 ksi, which is just above the endurance limit for a Category C detail. This provided a range of values from which the fatigue resistance could be evaluated and compared. Table 1 shows the test matrix for the tack weld test program. Here ‘MP’ denotes a modified position tack weld specimen where the leading line of tack welds is shifted such that the tack weld toes are in line with the center of the adjacent bolt holes. ‘FT’ indicates a specimen where the bolts are fully tightened along with the welds being in the modified position. An R ratio of 0.1 was used to calculate loads to be applied for the different stress ranges. The R ratio is simply the ratio of the minimum stress to the maximum stress, and a posi- tive value indicates that the stresses are of the same sign. In the test specimens being tested, both the minimum and maximum stresses are tensile. The tack welds are about a quarter of an inch in size. Normal length welds are about an inch in length, while longer welds are about 1.5 inches in length. A 4-pole MTS servo-hydraulic test fixture is used to cycli- cally load the specimens. Two grip plates and two lap plates are used to grip the tack weld specimen. The grip plates fit into the grips of the MTS controller. The tack weld assembly, along with the dimensions of specimen plates and grip plates, is shown in Figure 6. All bolts on the assembly are fully tightened, except for the ones on the tack weld specimen. When replacing a specimen, only the bolts holding the specimen plates at the top and bottom in the assembly need to be loosened. The old specimen can then be slid out and replaced with a new speci- men. A Campbell CR5000 Data Acquisition System was used to record the strains from the strain gages attached to some of the specimens, including the initial specimen. The strain gage readings were utilized to measure the stress distribution across the plates and the stress range near the weld toe. A photograph Table 1. Test matrix for tack weld test program. No. of Tack Welds Tack Weld Position Tack Weld Length No. of Specimens Tested at S r Value 20 ksi 12 ksi 12 ksi 2 L <1-in. 2 3 L <1-in. 3 3 2 (FT) 2 L <1-in. 3 (MP) 2 T <1-in. 2 3 L >1-in. 2 Figure 6. Tack weld test assembly (specimen shown is circled).

18 of the specimen fitted into the hydraulic grips along with the entire test setup is shown in Figure 7. Distortion-Induced Fatigue Tests One problematic condition that has not been well studied is the retrofit details for connections that exhibit distortion- induced cracking. A majority of fatigue cracks detected in steel bridges belong to this category. Several different meth- ods have been used to repair these cracked details. In general, these include repairs that either stiffen the detail or make it more flexible. Distortion-induced cracks often form at the end of vertical connection plates that are attached to the web of a girder, but not attached to the girder tension flange. Out-of-plane forces are developed in the cross frame members when one bridge girder deflects a different amount than an adjacent girder. The cross frame members are typically attached to the vertical connection plate, introduce pumping of the web region, and tend to develop out-of-plane stresses that combine with in- plane bending stresses. These elevated stresses can often lead to a premature cracking at the end of the weld to the verti- cal connection plate and/or the toe of the longitudinal weld attaching the web to the flange plate. A common retrofit used to prevent cracking at the end of the vertical connection plate is to use an attachment that is bolted to the vertical connec- tion plate and to the flange of the girder. A variety of different attachment details are used to make this positive connection, including a WT section, a pair of angles, single angles, or a pair of plates that are welded together and then bolted to the girder flange and vertical connection plate. If the attachment is stiff enough, the out-of-plane stresses will be significantly reduced and the likelihood of distortion-induced cracking will be eliminated if it has not already occurred. However, the stiffness of the retrofit detail is critical. Connor and Fisher (2006) describe a situation where a retrofit detail with a small thickness was not fully effective in preventing further crack growth at a detail with distortion-induced fatigue cracking, while a thicker detail used elsewhere on the same bridge was effective in halting further crack growth. Clearly, the stiffness of the detail is quite important. But questions remain about determining the required stiffness to mitigate distortion- (a) (b) Figure 7. Tack weld specimen in a 4 pole MTS servo-hydraulic actuator.

19 induced fatigue cracking. Other than the experimental results reported by Fisher et al. (1990) in NCHRP Report 336: Distor- tion-Induced Fatigue Cracking in Steel Bridges, few data exist on the fatigue strength and performance of attachments used to repair distortion-induced cracking. This subcomponent test involves testing a portion of a welded girder with a welded connection plate attached to the web. The cross section of the welded girder involves a section with a 34-in. × 3⁄8-in. web plate and 12-in. × 1-in. flange plates. These dimensions were intentionally selected to be very similar to the cross section dimensions used in NCHRP Report 336. The performance of the web region is evaluated by introduc- ing a given displacement to the vertical connection plate. Since this type of a displacement occurs in an actual connection that experiences distortion-induced fatigue cracking, and because the large magnitude of the out-of-plane stresses developed when this transverse loading occurs, the subcomponent test is believed to be appropriate for studying the behavior of the web gap and the retrofit/repair details used in the web gap region. It should be noted, however, that a limitation of the subcom- ponent test is that it does not effectively model primary bending stress that occurs parallel to the primary axis of the girder mem- ber. As cracks developed from distortion propagate away from the connection plate and begin to turn, then the influence of the primary bending stress would come into play. Nevertheless, it is believed that out-of-plane cracking behavior can still be effec- tively studied with the simple subcomponent test specimen. The connection plate for an actual bridge structure would often be terminated short of the tension flange to prevent fatigue cracking at the plate to flange weld. This would result in a web gap between the end of the connection plate and the flange at one end only. However, for the subcomponent test, a web gap region exists at each end of the connection plate to obtain the greatest amount of possible information. Accord- ingly, two attachments are connected to the specimen, one at each end of the connection plate, which effectively allows for two tests to be conducted simultaneously (see Figure 8). The connection plate is loaded at two locations along the plate length. The dimensions for the test specimen subcomponent are shown in Figure 9. The critical factors that are studied to examine their role on the repair of distortion-induced fatigue cracking include: the web gap size, the type of attachment detail, attachment thicknesses, magnitude of differential distortion, and attach- ment geometry. The web gap for most of the WT specimens is 1.5 in., with two WT tests with a smaller web gap of 0.75 in. The 1.5-in. web gap corresponds to the web gap used in NCHRP Report 336, while a smaller value of 0.75 in. provides a more critical situation with greater out-of-plane stresses for the smaller gap. Two of the WT specimens (labeled with an RH) were repaired with a drilled retrofit hole prior to installing the bolted retrofit detail to verify that the detail will fully arrest further distortion-induced cracking. Two additional WT specimens (labeled as B) were tested with a reduced num- ber of bolts attaching the retrofit flange and web elements in order to study the influence of a reduced number of bolts on the effectiveness of the retrofit detail. Most of the retrofit details were tested with four bolts attached to each flange ele- ment and four bolts to the connection plate. For the “B” tests, however, only two bolts were used to attach the retrofit detail to the flange elements and to the connection plate. Variation of the number of bolts provides information on the impor- tance of the connection stiffness on the fatigue performance. The use of double-angle and single-angle retrofit speci- mens is also studied since there are situations where these details will be more convenient to use than a WT section due to geometric constraints, such as interference with a horizon- tal attachment plate or the elimination of the need to cut out part of the vertical connection plate to make the positive con- nection to the vertical plate. A smaller web gap is used here since it produces greater out-of-plane stresses and because this is an excellent example of where the angle(s) would sim- plify the retrofit since there will be no need to cut the vertical connection plate. Two different angle thicknesses were used to study the influence of detail stiffness. A total of 13 subassembly tests were conducted, with two specimens per subassembly. Hence, a total of 26 speci- mens were tested that includes 14 WT retrofit specimens, 6 double-angle specimens, and 6 single-angle specimens. Two specimen web gaps of 0.75 in. and 1.5 in. were used, along with three different differential distortions. WT retro- fit flange thicknesses tested were 0.5 in. and 0.75 in. Double angles of thicknesses of 5⁄8 in. and ¾ in. were tested. Single angles of thicknesses ¾ in. and 1 in. were tested. The test matrix for the distortion test program is given in Table 2. Pre-cracking the specimens was found to range from 1.5 to 3.5 million cycles at a loading frequency of 4 Hz. Hence, Figure 8. Subassembly unit with retrofit details installed.

20 Figure 9. Top and side view of subcomponent test specimen. Table 2. Test matrix for the distortion test program. Connection Type Detail Thickness (in.) Differential Distortion, 0.01 in. Differential Distortion, 0.02 in. Web Gap, ¾ in. Web Gap, 1-½ in. Web Gap, ¾ in. Web Gap, 1-½ in. WT 1/2 X X X X 3/4 X X X X X X X(RH) X(RH) X(B) X(B) Differential Distortion, 0.0075 Differential Distortion, 0.01 DA 5/8 X X X X 3/4 X X SA 3/4 X X X X 1 X X

21 in order to speed up the time required for testing, two test setups were used. A schematic of Test Setup 1 is shown in Figure 10. The top of the loading spreader beam is attached to a 220-kip capacity MTS servo-hydraulic actuator. The spreader beam is attached to a pair of angles that pulls on the specimen stiffener at two points while the specimen flanges are firmly bolted to the side supports. In between the flange and the side support is an extra plate which accommodates different bolt hole patterns for the retrofits to be tested with- out having to change the side supports. Test Setups 1 and 2 for the distortion tests are shown in Figure 11. Test Setup 2 is similar to Test Setup 1 except that concrete supports have been used to hold the specimen instead of steel supports. Also, a 55-kip capacity MTS servo-hydraulic actuator was used, rather than the 220-kip capacity actuator used for Test Setup 1. The distortion of the stiffener was measured with an MTS clip-on gage. An aluminum attachment was fabricated so that the clip gage can be attached and held in position. The attachment, as shown in Figure 12, consists of two parts. The primary part sits on top of the specimen stiffener, reaching out horizontally over the specimen flange. It has a small steel knife edge at this end. The secondary part sits on top of the specimen flange and also has a small knife edge. This part will hold the clip gage in place during the test. The clip gage can latch on to the two knife edges, which effectively measures the deformation between the flange and the stiffener edge. The actuator can then pull on the stiffener according to the Figure 10. Test specimen mounted in test jig attached to strong floor. (a) (b) Figure 11. Distortion test setups 1 and 2.

22 measured clip gage reading, and thereby control relative dis- tortion of the web gap between the stiffener and the flange. In order to evaluate the effectiveness of the clip gage in measuring the distortion of the stiffener, Linear Variable Differential Transformers (LVDTs) with a range of ±0.05 in. were used on the test specimen as shown in Figure 13. These LVDTs give a more accurate estimate of the actual distortion occurring in the stiffener-to-web weld toes at the two ends of the specimen. The tests were initially planned to be run in distortion con- trol with the help of the clip gage. This was achieved success- fully, but the maximum frequency at which the test could be run was unsatisfactory. Hence the tests were instead run in a pseudo-distortion control, where force control was primarily employed, but the force was adjusted based on the monitor- ing of the resultant distortion, such that a nearly constant distortion was always maintained. The tests were mostly run at a frequency of 4 Hz. The testing procedure involves running the test initially in pseudo-displacement control to initiate a distortion- induced fatigue crack that is about ½ to 1 in. long on each side of the vertical connection plate. At this point, holes are drilled through the specimen flange to install the retrofit at both ends of the vertical connection plate. It should be noted that the fatigue cracks in the web were not treated by drilling a stop hole to remove the fatigue crack tip. Although this would normally be done as recommended normal prac- tice, it was not done in these tests so that the effectiveness of the retrofit detail repair could be assessed, even if the crack tips were not first blunted. (Note: stop holes were drilled for two retrofit specimens—labeled as RH—to assess expected normal practice.) The required initial load for cracking the specimen is roughly doubled, and the loading is resumed in force control. Since the retrofits are rigidly attached to the ends of the stiffener, force control also corresponds to a fixed amount of distortion at the ends of the specimen. The Figure 12. MTS clip-on gage fitted in aluminium attachment. (a) (b) (c) (a) (b) Figure 13. Linear variable differential transformers (LVDTs).

23 loading is continued until failure or completion of 5,000,000 loading cycles. At this point, the specimen is removed in order to examine any fatigue cracks to determine if the ret- rofit was effective in significantly slowing or halting further fatigue crack growth. Strain gages were attached to the loading angles to mea- sure the stress flowing through the angles. This is necessary to ensure that the load is being equally distributed to both ends of the specimen and also to observe the change in the distri- bution of load as fatigue cracks form at the ends of the speci- men. The strains were recorded using a Campbell CR5000 Data Acquisition System. Strain gages were also attached to the initial specimen and some of the retrofits to measure the stress range at critical locations of interest. Figure 14 shows a typical bolt pattern for a WT retrofit. The holes used in all retrofits have a diameter of 15⁄16 in. Figure 15 shows the bolt pattern for a 2 bolt WT retrofit. Figure 16 shows the typical bolt pattern for double-angle or single-angle retrofits. Figure 14. Typical WT retrofit bolt pattern. Figure 15. Two bolt WT retrofit pattern. Figure 16. Typical double-angle/single-angle retrofit bolt pattern.

Next: Chapter 3 - Findings and Applications »
Fatigue Evaluation of Steel Bridges Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 721: Fatigue Evaluation of Steel Bridges provides proposed revisions to Section 7—Fatigue Evaluation of Steel Bridges of the American Association of State Highway and Transportation Officials Manual for Bridge Evaluation with detailed examples of the application of the proposed revisions.

Appendixes A-D to NCHRP Report 721 are only available electronically. The appendices, which are in one electronic document, are as follows:

• Appendix A - Survey Interview Forms

• Appendix B - AASHTO Fatigue Truck Validation Analysis Results

• Appendix C - Tack Weld Tests

• Appendix D - Distortion Induced Fatigue Tests

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!