In our fast-paced world, engineers may “find the answer” and solve a problem without ever asking the question “Why?”

This is too bad, because the correct answer will often yield an understanding of underlying principles.

This month, we'll look back on some old data, apply the “why” question, and in doing so, we'll learn something about the behavior of weldments in fatigue.

Back in the days when girders for bridges typically were riveted, it was common to take a rolled shape and add coverplates in the center of the span length where the bending moment was greatest — see Figure 1.

As welding gained acceptance, designers began to specify welded coverplates instead. When such girders were put into service, fatigue cracks sometimes occurred in the rolled beam, at the ends of the longitudinal welds or at the toe of the transverse weld toe, near the end of the coverplate.

We now know this configuration as a Category E or E' fatigue detail, and we understand that the stress range capability of such details is limited.

Today, most designers will select a thicker or wider flange in the areas of high moment, and use a welded butt splice with a 2.5:1 transition. When the weld is ground flush and inspected with radiographic or ultrasonic inspection, the configuration improves to a Category B detail.

While welded coverplates are not a common bridge detail today, many manufacturers of agricultural and earthmoving equipment use coverplate-like devices that increase the capacity of the member in a localized area. The ends of such reinforcements behave similarly to the ends of coverplates.

The heavy equipment industry uses a variety of transitional profiles to enhance fatigue behavior.

At the University of Illinois in the early 1960s, Munse and Stallmeyer evaluated six different end plate details. A summary of their data is presented in Table 1.

Stress ranges were reported for both 100,000 cycles and 2,000,000 cycles. A higher value meant that the configuration had greater capacity.

In fatigue, particularly of welded assemblies, there is always variation in the results. Theoretically, what behaves best at 100,000 cycles should also behave best at 2,000,000 cycles. For this data set, specimen 5 had the best behavior at 100,000 cycles, and specimen 3 had the best behavior at 2,000,000 cycles. Despite these anomalies, we still can observe trends.

Each specimen contained two welds that were generally longitudinal to the beam's axis, and some contained a weld that was generally in the transverse direction. Table 1 uses blue and yellow to illustrate the weld's orientation. The specimens with only longitudinal welds (specimens 3 and 4) by and large performed better than those with transverse welds.

Some of the specimens had square-cut ends (specimens 1 and 4) whereas the others had some type of transition. Specimens that used a transition end (specimens 2-3, 5-6) did better than those without a transition. Specimen 3 with a “concave” transition had the best 2,000,000 cycle performance of the six, even though it had a transverse weld.

At this point, it is worthwhile to ask some “why” questions.

Why is the transition helpful?

Table 1 includes a sketch of the section properties of the cross section (shown in green), along with the resultant stresses (shown in orange). The stress plot assumes a constant moment, a reasonable assumption for the region of the beam in the immediate area of the coverplate termination.

Since stresses “flow” though members, the actual stresses in the vicinity of the weld are more complicated than the diagrams imply, but the plots are useful nevertheless.

These plots explain the effect of the transition: The change in section properties is more gradual, permitting a more even flow of stress through the member where the coverplate terminates (see the orange plots in Table 1). This reduces the stress concentration created by the square cut coverplate (specimens 1 and 4).