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.
146 CHAPTER 5. CONCLUSIONS AND RECOMMENDATIONS This research project primarily consisted of the experimental testing of 12 full-scale gusset plate connections. These 12 connections were used to investigate shear and buckling limit-states, as well as to consider the effects of simulated section loss, multi-layered gusset plates, and edge stiffening. The experimental matrix was limited to a single geometry: a five member connection with collinear chords, two diagonals entering into the connection at 45 degrees to the chords, and one vertical entering the connection perpendicular to the chords. However, to account for the variety of connection geometries and load combinations found in the bridge inventory, a finite element parametric study was used to appropriately extend the experimental results. A robust finite element modeling methodology was established by comparing the finite element simulations to the experimental results. These comparisons established high confidence in both the modeling method used and the interpretation of the simulations, ensuring that the 201 connections modeled as part of the analytical parametric study produced trustworthy results. To draw conclusions, the results of both the experimental and analytical studies were grouped by failure mode to establish professional factors that could be used in an LRFD/LRFR calibration of predictive resistance equations. Also, because bridge owners have relied on the FHWA Guide to load rate gusset plates in response to the recommendations made in Technical Advisory T 5140.29, Load-carrying Capacity Considerations of Gusset Plates in Non-load-path-redundant Steel Truss Bridges, an attempt was made to maintain the existing predictive resistance equations of the FHWA Guide during this research. The following list of observations and recommendations highlights the major findings of this project. 1. It was recognized that the influence of the developed resistance equations on the design of new gusset plates would be minimal, at most requiring thicker gusset plates. The largest implications of the developed resistance equations would be in the rating of the existing inventory where repair and retrofitting is difficult. In keeping within the scope of the project, this research only focused on an LRFD/LRFR philosophy. 2. The limit-states for design were calibrated to a target reliability index of 4.5, and ï¦- factors were selected at a DL/LL of 6.0. For rating, a lower target reliability index of 3.5 was deemed tolerable, and ï¦-factors were selected at a DL/LL ratio of 1.0. The justification of using a lower reliability index for rating is predicated on AASHTO mandating the use of system factors in rating gusset plates, where currently they are considered optional. In addition, the resistance factors must be further reduced for only rating provided the DL/LL ratio is greater than 1.0. The additional reduction decreases linearly from 1.00 to a minimum of 0.90 as the DL/LL ratio changes from 1.0 to 6.0. 3. A minimum plate thickness for use in the design of new gusset plates is recommended. This research found that plates thinner than 0.375 inches contributed greatly to the scatter of some limit-states, particularly gusset plate buckling. The selected ï¦-factors for design
147 assume this limit is adopted. If not adopted by AASHTO, some ï¦-factors may have to be reduced for design. A minimum plate thickness limit is not needed for rating, though the recommended ï¦-factors for rating account for the greater scatter associated with plates thinner than 0.375 inches thick. For this reason, owners who are trying to prioritize rating their gusset plate inventory should consider focusing on those gusset plates 0.375 inches thick or thinner. 4. The general formulation of shear yielding, 0.58FyAg, used in LRFD and the FHWA Guide was proven to be appropriate. On average the ï-factor was found to be 1.02, though a fixed value of 0.88 is assumed for both design and rating. A new ï¦-factor for gusset plate shear yielding is recommended for both the BDS and MBE. This ï¦-factor would be 0.80 for design and 1.00 for rating. 5. The current definition of ï needs to change, though this only applies to the existing FHWA Guide where it is used. It was found that gusset plates would plastify through the shear plane at the shear yielding limit-state. As a result, the present correlation of ï to a bending shear distribution was found to be weak as it has a much stronger association with a uniform shear distribution. 6. No data were specifically collected that either supported or questioned the formulation of the shear fracture resistance equation. However, in the seven experimental tests that resulted in shear failures, there were no shear fractures and two of the seven exceeded the predicted shear fracture strength before yielding in shear. Therefore, there is no recommendation to change this limit-state from what is published in the FHWA Guide. 7. Evaluating the capacity of gusset plate shear planes that pass over connected members was found to be inappropriate. Close inspection of the stress contours evaluated for each simulation revealed that yielding along gusset plate shear planes that also passed over a chord was dominated by shear away from the chord and by normal stress due to splice action near the chord. Therefore, shear yielding and fracture resistance equations need only be evaluated on shear planes that can mobilize without interacting with a connected member. While it was recognized that a plane cannot mobilize when it passes over a member, it is speculated that a shear plane could mobilize through the chord splice provided it is only lightly reinforced with alternate splice plates, although no evidence of this possibility developed during the research. 8. The existing Whitmore buckling resistance used in the FHWA Guide produces highly variable resistance predictions, especially based on the assumed effective column length factor (i.e., K-factor). A better prediction of buckling was found with a fixed value of K=0.5 used in combination with Lmid rather than Lavg used in the FHWA Guide. The calibrated ï¦-factor for the Whitmore buckling check was found to be 0.75 for design and 0.85 for rating. For compact connections where the members are clustered close together, Whitmore buckling was found to be a poor predictor of buckling resistance. Instead, the
148 research showed that the load determined to cause a partial shear plane to yield was also a much better predictor of buckling. 9. Although it may seem odd to use a shear formulation to predict buckling, the research showed that once a partial shear plane in the gusset plate along a compression member yields, its elastic modulus decreases and thus reduces the out-of-plane rotational restraint the plate can provide to the idealized column. The factored resistance formulation described in Recommendation #4 applied to the partial shear plane was sufficient to predict this effect; different ï¦-factors are not needed for the partial plane shear yielding criteria. 10. The proposed two-folded approach to buckling may find gusset plates will have a reduction in rating values for compression resistance despite the FHWA Guide producing favorable ratings. In particular, the Whitmore buckling criteria overestimates buckling strength of compact joints and the partial shear yielding check will control and produce lower resistances. The connections most susceptible are those with prior ï¬avg values less than 1.0 using the FHWA Guide. On the contrary, the Whitmore buckling criteria in the FHWA Guide, especially assuming K=1.2, produced overly conservative resistance predictions when ï¬avg was greater than 1.5. The new approach with a fixed K=0.5 and using Lmid will predict higher resistance for these joints. In the transition region between 1.0<ï ï¬avg <1.5 it could go either way. 11. Only three analytical simulations produced tension failures and none of the experimental specimens experienced tensile failures. Therefore, there was insufficient data to either support or question the Whitmore gross and net section checks for tension members used in the FHWA Guide. As a result, it is recommended to keep these existing resistance equations and associated ï¦-factors. 12. A calibrated ï¦-factor is presented for block shear based on the current block shear equations included in the BDS. The required ï¦-factor for design was found to be 0.80 and for rating it was 1.00. 13. The research found that the Whitmore criteria should no longer be used to check the chord splice because it does not characterize the true stress distribution well. Rather, a new chord splice procedure was developed that equates the gusset plate stress in the chord splice with the stress distribution produced assuming a linear bending gradient resulting from the eccentric loading. The ï¦-factor for design was found to be 0.65 and for rating it was found to be 0.85. 14. For gusset plate connections built-up from multiple layers of individual plates, the research found that treating the individual plates as uncoupled and summing their individual resistances produced results within the scatter band of the associated resistance calibration.
149 15. The consideration of section loss in gusset plates in determining resistance was evaluated as part of this research. A âsmearedâ section loss model was considered and found to be conservative. This approach determines an equivalent plate thickness on the relevant failure plane depending on whether a shear or buckling failure mode is being evaluated. It was found in shear failure scenarios with section loss that failure occurred after plastification of the entire failure plane. It did not matter if the section loss was non- uniform in the plate itself or unbalanced between two gusset plates in the same connection. 16. An attempt was made to collect all historical rivet shear strength data from the existing literature. Multiple sources were found dating back to 1882. All of the rivet shear strength data identified was evaluated statistically and an alternate approach of considering the connection length effect originally proposed by Tide was assessed. Using the Tide inequalities to characterize the proportioning of the connection, it was found that the connection length factor can be defined by two fixed values. It was also found that the population of rivets could be broken down into three strength categories: A141/A502 Grade 1 rivets, A195/A502 Grade 2 rivets, and all other unknown rivets (rivets not from one of those four grades). The statistics for the unknown and A195/A502 Grade 2 rivets were biased from lack of data and the factored shear strengths for them were considered unreliable. The factored resistances for the A141/A502 Grade 1 rivets were developed from a large pool of data and were very similar to those values currently published in the 2011 MBE Interims. Therefore, no changes in the values were considered necessary at this time, without a better pool of strength data for unknown or A195/A502 Grade 2 rivets. 17. The research found no correlation to the slenderness of a gusset plate free edge and the associated compression buckling resistance. Therefore, it is recommended that AASHTO revise their specification language to clarify the intent of the slenderness limit in the BDS to say it represents good detailing practice and reduces initial imperfections, but is not meant to enhance buckling resistance. It is also recommend that the MBE not require rating based on the slenderness of the gusset plate free edge. However, if properly implemented, edge stiffening can be used to enhance buckling resistance if it adds out-of- plane stiffness to the compression member relative to the adjoining members. 18. It was recognized given the number of different connection geometries used to define the resistance equations that the equations will be overly conservative in some situations to ensure that all connections would be conservative. Therefore, if bridge owners do have gusset plates with unfavorable load ratings based on the resistance equations developed from this research, they also have the option to conduct a refined analysis. However, a refined analysis would not address the inherent variability of material properties and the fabrication process that a ï¦-factor is meant to account for. Therefore, an analysis factor was developed to account for these variabilities when applied to the results of a refined analysis. The analysis factor was found to be 0.70 for design, and 0.90 for rating.
150 19. According to national policy, Load Factor Ratings (LFR) are still admissible for reporting to the National Bridge Inventory. According to this projectâs scope, LFR was not studied. However, in Appendix J recommended language was proposed to Section 6B of the MBE to provide LFR rating procedures for gusset plates since AASHTO is no longer maintaining the Standard Specifications. A reverse calibration from LRFR to LFR cannot be performed as the project did not attempt to look at the live load variability associated with short- and long-span trusses. Therefore, one cannot use the LRFR resistance factors in LFR because of the differences between the HS-20 and HL-93 live load models. Limit- states that changed from what was published in the FHWA Guide (like buckling and shear) were assigned an LFR resistance factor of 1.00. If nothing changed from that published in the FHWA Guide, the factor from the FHWA Guide was copied. For clarity, the major recommended changes to the BDS and MBE are represented in Table 39. This table compares the recommendations to those in the FHWA Guide. Table 39 Summary of Recommended Changes Compared to FHWA Guide Limit State Factor FHWA Guide Proposed LRFD ï¨ï¢=4.5 @ DL/LL=6) Proposed LRFR ï¨ï¢=3.5 @ DL/LL=1) a Buckling ï¦ï 0.90 0.75 0.95 K 0.65 to 2.0 0.5 0.5 Whitmore Section 30°, Lavg 30°, Lmid 30°, Lmid Shear Yielding ï¦ï 0.95 0.80 b 1.00 b ïï 0.74 or 1.00 0.88 b 0.88 b Shear Fracture No changes from existing FHWA Guide Block Shear ï¦ï 0.80 0.80 1.00 Chord Splices ï¦ 0.90 or 0.95 using Whitmore section for compression and tension respectively 0.65 0.85 Tension Members No changes from existing FHWA Guide Rivet Shear Strength No changes from existing 2011 MBE Interims Simulation Analysis ï¦ Not applicable 0.70 0.90 a â An additional reduction shall apply for DL/LL ratios greater than 1.0. The additional reduction decreases linearly from 1.00 to 0.90 as the DL/LL ratio changes from 1.0 to 6.0. The additional reduction shall not be less than 0.90. b â Will now apply to partial planes around compression members.
151 SUGGESTED RESEARCH 1. Additional research is needed to establish the ideal target reliability index for trusses and gusset plates. It was beyond the scope of this project to investigate this rigorously and an assumed value of 4.5 for design and 3.5 for rating was used in this work. Generally, reasonable but conservative choices of various parameters have been chosen when extracting proposed design and rating specification provisions from this research. In particular, the choices of statistical parameters such as the mean and coefficient of variation (COV) of loads as well as the assumption that the physical properties of two or more gusset plates making up a joint, and the loads applied to them, are totally uncorrelated are assumptions that could be reviewed and refined at a later date. The possibility that the COV of the lower tail of the distribution of steel yield strengths, both modern and legacy, is lower than the COV calculated for the full population of test data may be fertile ground for the study leading to improvement in the published calibration results of this report. 2. In two of the analytical simulations that failed around tension members, the three tension limit-state checks used by the FHWA Guide unconservatively predicted the failure load. In these three connections it appears that coupled mechanisms were formed that are not addressed by either the block shear or regular shear checks. The coupled mechanisms involved either multiple members pulling out from the gusset plate simultaneously, or having an entire corner of gusset plate pull out with the member. As a result, this is an area of needed additional research. 3. It is critical that further work be performed to determine whether or not the connection length effects exist and the Tide criteria are valid. In addition, it is necessary that further work be performed to gather rivet strength data for rivets of unknown origin or those made to the A195/A502 Grade 2 specification. The data from both exercises could be used to further refine the factored rivet shear strengths.
