Optimization of Rib-to-Deck Welds for Steel Orthotropic Bridge Decks

CONCLUSIONS AND RECOMMENDATIONS

This research was performed to develop design and detailing recommendations to make the rib-to-deck weld for orthotropic decks more friendly for fabrication. At the time the research began, the current standard for this detail was an 80-percent minimum weld penetration requirement with an initial tight fit and no tolerance for melt- or blow-through in the as-welded condition. As an example, the difference between 80- and 100-percent penetration given a ⁵/₁₆-inch rib wall thickness is quite small, and further coupling with a fit-up tolerance not to exceed 0.01 inch makes the weld prohibitively difficult to fabricate. Variations in the fit-up tolerance between the rib and deck can easily result in melt-through or blow-through when attempting to achieve nearly full penetration, especially if the fit-up between the plates is not consistent.

A number of variables were hypothesized to affect the fatigue resistance of this detail, and it was not possible to test a statistically significant population of full-size decks for all of the variables. Therefore, a full-scale, small-specimen fatigue testing protocol was developed to investigate the fatigue resistance of this detail by slicing an individual welded panel into numerous individual specimens. These tests are a cost-effective way to rapidly generate large amounts of data and study the significance of weld procedure variables.

Overall, the study determined that the primary factor in the fatigue resistance of the rib-to-deck weld is its geometry. Welds with larger leg dimensions along the deck plate perform better in fatigue, and, as this leg dimension increases, there is a commensurate decrease in the required penetration. Welds with penetration varying from 40 to 100 percent can be designed assuming a level 3 fatigue analysis with category C fatigue resistance provided the weld leg dimension meets minimum size requirements.

RECOMMENDATIONS

The stress analysis used throughout this report was consistent with a level-3 design procedure. The following recommendations are proposed for consideration and potential adoption into the Federal Highway Administration Manual for Design, Construction, and Maintenance of Orthotropic Steel Deck Bridges and the seventh edition of the AASHTO LRFD Bridge Design Specifications; however, no consideration of level-1 or level-2 design procedures was considered:^(10,8)

1. Weld geometry must be made to satisfy the inequality that penetration shall be greater than 0.222(d₁/d₄) ^−1.50, where d₁/d₄ is the normalized leg weld dimension along the deck plate. Therefore, two options exist: (1) pick a target penetration value and calculate minimum leg size or (2) select a leg size and calculate a minimum penetration. Either way will produce welds that exceed category C design resistance.
   
   
2. The ratio of weld leg dimensions, h‑to‑d₁, shall be between 0.6 and 2.0, and d₁/d₄ shall be between 0.40 and 1.30. This is to maintain consistency with the specimens tested and reported herein.
   
   
3. Qualification shall be performed to ensure welding procedures produce production welds with a closed root condition. The closed root condition is not defined as being fused; it is merely the notion that the inside rib corner is visibly in close contact with the deck plate. Provided a closed root condition is maintained, root failures should be suppressed.
   
   
4. The current pre-welded fit-up gap of 0.020 inch is considered sufficient. For the two panels fabricated with 0.020-inch fit-up gaps, the gap mostly closed across the whole length of panel from weld shrinkage, and there were no weld root failures from these populations of specimens^[1]. Some evidence was presented that ¹/₃₂-inch gaps can also close, though they were largely not fatigue tested, and the 0.020-inch recommendation is considered conservative.

FUTURE RESEARCH

The full-scale small-specimen test(s) should be developed into a standardized testing protocol that can evaluate the rib-to-deck weld prior to fabrication. Most bridge owners already require fabricators to produce full-scale mock-ups of rib-to-deck weldments as a qualification test prior to fabrication. It would be relatively inexpensive to section some fatigue test specimens from the mock-ups, and it takes about 1 to 2 d to perform a fatigue test. Such a qualification test can be useful to allow fabricators leeway for efficient design of this weldment.

Currently, the LSS method is limited to weld toes only. The reason is the LSS method assumes that the disparity between the predicted structural stress and the actual stress is uniform throughout all weld toes. Hence, the fatigue resistance for all weld toes follows the same S-N curve. However, this assumption no longer holds for weld roots because the notch effect is much more severe. This is especially true for open root gaps because they create crack-like defects whose fatigue resistance highly depends on the local geometry and is thus difficult to capture by a single S-N curve. A more comprehensive methodology to determine the fatigue resistance from the root of partial penetration welds will be a useful advance in practice.

The use of the AASHTO category C resistance curve has been shown to work well with the LSS method in the finite life regime where all stress cycles exceed the fatigue threshold. However, it is unknown if the 10-ksi fatigue threshold for this category applies to the rib-to-deck weld. Additional testing is recommended at lower stress ranges to determine the proper threshold for this weldment. This will enable design of orthotropic decks for infinite fatigue life.

More work is needed to correlate the full-scale small-specimen tests to full-scale testing and in-service performance of orthotropic decks. The full-scale, small specimen tests should be considered to supplement any future tests of orthotropic decks.

There is one exception to this statement. Specimen UB-2 was reported to have a closed root condition and did fail in the weld root under reversal loading. However, a macroetch of that weld was not provided, and it is possible that this may have been reported as a closed condition because that was the intent of the UB panel, but the actual welded condition may have been open.