U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
FacebookYouTubeTwitterFlickrLinkedIn

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-17-020    Date:  February 2017
Publication Number: FHWA-HRT-17-020
Date: February 2017

 

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

INTRODUCTION

BACKGROUND

Many steel orthotropic bridge decks have been built over the last 60 years in Europe, the United States, Japan, and many other countries. The origin of this bridge deck type dates back to the “orthotropic plate” patent issued in Germany in 1948.(1) The major advantages of steel orthotropic decks include their light weight, rapid erection, and easy assembly. The original patent claimed that the steel consumption could be reduced by half. With these advantages, orthotropic bridge decks have been widely used on long-span highway, movable, cable-stayed, and suspension bridges because of their light weight. They have also been used on other types of bridges where fast construction is desired, such as temporary bridges and bridges in high population density areas. Orthotropic steel decks are also a common solution for redecking old bridges because of their easy assembly.

A steel orthotropic deck typically consists of a steel deck plate with welded stiffeners or ribs parallel to each other in the longitudinal direction. Transverse cross beams are typically used to support the ribs and provide stiffness in the transverse direction. The transverse cross beams typically serve as floor beams transferring the deck loads to the main structure. These floor beams are often integrated with the deck structure where the top flanges of the floor beams are often formed by the deck plate itself. The stiffening ribs can be open shapes, such as plates, inverted T-sections, angles, and channels or closed box-type ribs with different geometric shapes; trapezoidal closed ribs are the most common. Figure 1 is an illustration of a typical trapezoidal close-rib steel orthotropic deck panel where the large flat surface is the deck plate (with the stripes indicating typical lane markings) and the small trapezoids under the deck plate are the closed ribs. The first orthotropic steel deck with closed ribs was constructed in Germany in 1954. Compared to open stiffeners, the closed ribs have many advantages. First, closed ribs can transfer the traffic load much more efficiently in the transverse direction. As a result, closed ribs can have wider spacing than open ribs. This results in fewer ribs and therefore lighter weight compared to open-rib systems. Second, closed ribs can provide much higher flexural and torsional rigidity in the longitudinal direction, allowing longer spans to be achieved. In other words, fewer cross-beams are required, thereby reducing the deck self-weight and the number of welds associated with the cross-beams. Finally, because single-sided welds are used to attach the closed ribs to the deck versus double-sided welds for open ribs, the number of rib-to-deck welds is reduced by half. However, the one-sided welds cause quality control and inspection issues that can be a disadvantage for closed ribs.

Click for description

Figure 1. Illustration. Typical closed-rib steel orthotropic deck panel.

To overcome the disadvantages of one-sided welding and prevent premature fatigue failure, more careful consideration is needed to design rib-to-deck welds. Many of the earlier vintage orthotropic decks with closed ribs experienced fatigue cracking problems. There was a lack of knowledge about fatigue and a lack of guidance in the structural design codes. The complex stress state present at the rib-to-deck welds makes fatigue design even more difficult. Many orthotropic decks before the late 1970s were constructed under this state of knowledge. The quest for lighter self-weight led to relatively thin deck plates in the structural design. However, many of the designs with thinner deck plates were vulnerable to high local stress effects from wheel loads. The contribution of the wheel-load effect was not fully considered in early deck designs, and many bridges experienced fatigue cracking problems. Compared to main structural members, orthotropic steel decks tend to have a higher incidence of fatigue problems because of the local effects of wheel loads. Wheel loads cause large local stress variations, stress reversals, and an increased number of stress cycles that must be considered in fatigue design.

Steel orthotropic decks have been part of engineering practice and extensively studied in Europe for decades. Partially to the result of the use of relatively thin deck plates, premature fatigue cracking was observed in many European countries. (See references 2 through 6.) Steel orthotropic decks have also been widely constructed in the United States with mixed experiences relating to fatigue behavior. The situation has been steadily improving as more knowledge becomes available on how to improve fatigue resistance.

OBJECTIVES

The first objective of this study was to determine the effect of weld process and geometry on the fatigue resistance of the rib-deck weld. This was accomplished by fatigue testing a series of 159 specimens with different welding processes and different levels of weld penetration under two different loading regimes. A statistical analysis of the data was used to determine the effect each variable has on fatigue resistance.

The second objective of this research was to optimize the size and shape of rib-to-deck welds as the basis of the detailing requirements. At the time the project began (2008), the fourth edition of the American Association of State Highway and Transportation Officials (AASHTO) Load-and-Resistance Factor Design (LRFD) Bridge Design Specification was the current version.(7) Article 9.8.3.7.2 stated that “Eighty percent partial penetration welds between the webs of a closed rib and the deck plate should be permitted,” and the commentary states that “Such welds, which require careful choice of automatic welding processes and a tight fit, are less susceptible to fatigue failure than full penetration groove welds requiring backup bars.”(7)(pp. 9–23) This provision was commonly perceived as rib-to-deck welds requiring a minimum penetration of 80 percent with a tight fit not to exceed 0.01 inch. An upper bound on penetration is typically applied such that the welds cannot have blow-through or melt-through that creates defects inside the ribs. In reality, because of the thin rib plate thickness that is commonly used in steel orthotropic decks, 80 percent penetration without melt-through or blow-through is difficult to achieve consistently, and, because of to the natural waviness of hot-rolled plate, the tight fit-up was also difficult to consistently achieve. A more tolerant penetration requirement is needed to reduce the need for weld repairs and increase fabrication efficiency. However, any relaxation of the 80 percent penetration must not increase the potential for fatigue cracking.

The last objective of this research was to validate the level 3 design approach currently published in the seventh edition of the AASHTO LRFD Bridge Design Specification (here forth referred to as AASHTO BDS in the remainder of this document) for the design of rib-to-deck welds.(8) This approach adopts a local structural stress (LSS) methodology for fatigue design where three-dimensional finite element models are used to characterize the stress state near weld toes. Stress is interrogated at two points away from the weld toe and then extrapolated to a theoretical stress at the weld toe that is used in the fatigue assessment. Fatigue life data collected in this project will be compared to the extrapolated LSSs based on finite element analysis of the specimens.

APPROACH

Fatigue resistance is typically characterized into S-N curves where fatigue test data is plotted based on the stress range (S) and the number of cycles to failure (N) on a logarithmic scale. Most previous orthotropic deck research used full-scale components to conduct fatigue testing. While this provides an accurate representation of a real structure in terms of boundary conditions and stress fields, the cost of specimens and testing equipment is prohibitively expensive. This research took a different approach where full scale was maintained but through a small test specimen. In this case, a full-thickness deck plate and full-scale rib are welded together into a panel that varied between 3 and 6 ft in length. This creates a full-scale geometry, and the residual welding stresses should be commensurate with a real structure. However, instead of testing the panel to only acquire one fatigue data point, the panel was sectioned into 4-inch-wide specimens, and loads were applied to the rib while supporting the deck, which caused out-of-plane (from the viewpoint of an entire deck) flexing of the rib, placing the weld in a high demand. While this loading pattern does not create a realistic stress pattern in the specimen as compared to a real deck, the concept does allow for rapid and cheap comparison of many specimens testing a variety of different variables on an equal playing ground.

It can be argued that sectioning of the panel reduces the residual stress and that the weak-link concept is lost. The weak-link concept can be illustrated when placing an entire panel under a fatigue test. Likely, that panel will develop a single fatigue crack at some localized point where there is an internal discontinuity within the weld. However, if that panel were sectioned into many smaller specimens, then that localized discontinuity is only within one of them, and that specimen will likely have the lowest fatigue resistance. The argument is that the distribution of fatigue life attained from many specimens extracted from one panel may be different than the distribution produced from testing many single panels. The researchers believe that it was more important to generate a large population of data rather than a small population.

 

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101