Figure 5

Many weldments are composed of material cut from sheet, plate, shapes or hollow box sections. In other words, most of the metals we use come from planar products, or in the case of shapes, essentially planar products that intersect at right angles. The welds that join components cut from these products, then, are typically linear and orthogonally oriented to each other.

Much of this changes when hollow, round tube sections are used instead. The welds are often circular and nonplanar; the typical orthogonal orientation between various welds usually disappears.

These changes have obvious consequences for manufacturing, requiring different methods of cutting materials and depositing weld metal. Less obvious, however, are the changes in analysis and design that should take place.

Consider the connection of a cantilevered beam made from a wide flange shape to a column as shown in Figure 1. Welds are applied to the flanges and the web. All the welds are linear, and the web welds are perpendicular to the flange welds. Typically, it is assumed that the flanges (and flange welds) will resist the moment, while the web (and web welds) will resist the shear. For most beam configurations, the moment will require the application of a much larger weld than will be required to resist the shear. Also, for most sections that would be used for beams, the width of the flanges is smaller than the depth of the section, permitting longer welds to be placed on the web versus the flanges. Therefore, the required weld size for the flange welds is significantly larger than that needed for the web weld. With this simple geometry, it is easy to visualize how the forces will be transferred through the connection.

Figure 2 shows a similar situation, but the wide flange rolled section has been replaced with a round tube section. If the loading conditions remain the same as in Figure 1, then the moment will still be larger than the shear. The bending loads are best resisted with the portions of the circular weld that are the greatest distance away from the neutral axis, whereas the weld near the neutral axis will offer no resistance to bending.

The upper and lower quadrants of the round tube, as well as the circular weld, resist 82 percent of the applied moment, as is shown in Figure 2b. It is therefore reasonable (and conservative) to assume that all moment is resisted by welds at these locations, then to size the weld accordingly. The remaining two quadrants with a generally vertical orientation can be assumed to transfer shear. In that the bending moment is likely to generate stresses greater than those created by the shear, the size of the weld could change correspondingly in size around the perimeter. This may not be aesthetically desirable, but structurally, the weld need not be of the same size around the full circumference.

A basic principle of connection design is that the force will ultimately enter into a member, or a part of a member, that lies parallel. Figure 3 illustrates possible member types to which lugs may be added, and serves to illustrate this principle. In the first example, connecting the lug directly under the web of the rolled section results in a direct load transfer. The load is transferred immediately into the web (shown with shading in Figure 3a) since it is the member that is parallel to the force that will be applied through the lug.

The load transfer path in Figure 3b is harder to envision. No segment of the circular member is truly parallel to the applied force, but in cross section, it is easy to see that the side quadrants are nearly parallel. Once again, 82 percent of the vertical force will be resisted by these two areas (shown with shading). However, there is no direct load path from the lug to this quadrant. The bracket and the connection-design must encourage load transfer to these quadrants.

One possibility would be to form the lug from a ring that surrounds the tube as shown in Figure 4a. If this were done, the vertical loads could be resisted by the side quadrants, and welds need only be applied at these locations. Of course, such an arrangement could pose an assembly problem, and tolerances on both the tube and the hole in the lug could pose intermittent problems.

Figure 4b represents what might be the best solution to this problem: The lug is attached to a pair of side brackets that can be attached directly to the side of the tube. Load transfer is efficient, and assembly is easy.

In this space, I often use examples of flawed designs to illustrate the principle being discussed, but this time, I'll end with an example of a good design. The guy bracket shown in Figure 5 is a real-life example of an optimized design. Many designers would have chosen to apply a single lug to the side of the tower (as in Figure 3b) and would have connected the cables to the lug. Instead, in this case, the designer applied the principles of "circular reasoning" and selected a yoke arrangement so the angle of the cables could vary. This transfers the applied load to the two quadrants where it can be effectively resisted.

Omer W. Blodgett, Sc.D., P.E., senior design consultant with The Lincoln Electric Co., struck his first arc on his grandfather's welder at the age of ten. He is the author of Design of Welded Structures and Design of Weldments, and an internationally recognized expert in the field of weld design. In 1999, Blodgett was named one of the "Top 125 People of the Past 125 Years" by Engineering News Record. Blodgett may be reached at (216) 383-2225.