ALGOR FEA HELPS DESIGN WHEEL HUB WITH MILLION-CYCLE FATIGUE LIFE AND MINIMUM WEIGHT

Photograph of truck drive hub Webb designed using FEA.

Finite element analysis (FEA) on a personal computer has helped a manufacturer of commercial vehicle wheel hubs design a cast steel hub with a long, million-cycle fatigue life and minimum weight. The FEA program, compared with an SAE fatigue test, proved itself highly accurate, matching the levels and areas of high stress the SAE fatigue test had determined.

Now Webb Wheel Products, Inc., of Cullman, Alabama, a member of the Chicago-based Marmon Group of companies, will begin regular use of FEA in the design process, according to Vice President of Engineering Paul Levering. He says one of the important benefits is that the program's onscreen analysis is so much faster than physical testing. That will dramatically reduce the prototype cycle, Levering says. "We'll never eliminate the prototypes and fatigue testing, but we now expect to come much closer to the optimal design the first time around."

Webb uses the Algor Finite Element Analysis (FEA) System on a Compaq 386 with a color monitor and EGA graphics. The package integrates analysis with graphics, modeling and meshing. After Webb began using the CADKEY 3-D CAD system on a personal computer, the firm began looking for computer-aided engineering tools, notably FEA. (Of course, its not necessary to purchase a separate CAD system to prepare models for FEA analysis. The Algor system contains all the CAD you need for FEA.) Levering, who had performed FEA on mainframes, says he liked Algor's features, such as its ease of meshing and the availability of several yield criteria to define tri-axial stresses.

"Typically, it takes around six months and costs $5,000 to build and test a prototype and redesign it," states Levering. Webb normally builds about ten pieces in the prototype stage, fatigue testing at least three, and submitting the rest to the customer. "Our goal in using computer-aided design and the finite element analysis software," he adds, "is to reduce the amount of time from when the drawing is finished to when we have parts ready for sale in the field."

A large company entered into an agreement to purchase trucks from a major truck manufacturer. For Webb, which has been given the front and rear hub program for these trucks, that means a long production run. Additionally, there will be other applications for this design.

Levering says this was an appropriate time for additional testing of their design. "If you have a failure in the product, you're faced with a product liability suit and you're faced with a possible recall, which is very expensive - and damaging to your reputation, so it's important to make sure the design is strong enough to do the job," he explains.

But it was also important to make the design as light as possible. "Since production quantities will be quite high, we would realize a substantial cost savings for every ounce of metal we could do without. The trucking industry is very weight conscious anyway."

The 45-pound hub, made from a cast steel grade that is roughly equivalent to SAE 1030, has an eight-bolt pattern of 22 mm studs to hold the wheel in place and six 0.75-inch bosses for rotor mounting on the back side and eight 0.56 bosses for drive flange mounting on the front side.

"We suspected the critical area on the hub was going to be around the junction of the eight-stud pattern and the six-stud pattern," relates Levering. "So we modeled that area, constructing ¼ of the hub on CADKEY, placing five elements of 18 degrees per element into it. We took two section views and meshed them in and translated them through the Algor link program into the finite element analysis program."

To prepare for meshing the model, Levering used Algor's MSHGEN and RADGENBR, a general purpose 2-D and 3-D brick cylindrical mesh generator, respectively. He made a pie section of the six-stud pattern and a pie section of the eight-stud pattern in RADGENBR to make a 90 degree section, then joined the two using Algor's Substruct. He says Substruct features an automatic "push" capability to connect the two shapes; all Levering had to do was define three noncolinear nodes in the two models to position the two parts for joining.

With the mesh completed, Levering used Algor's AEdit to establish the boundary conditions on the model and to apply the loads, which Levering chose to be those prescribed in the SAE fatigue test for hubs. The SAE test, a rotary bending fatigue test based on the offset of the wheel, the axle loading and a 1.4 test acceleration factor, called for a bending load of 13,481 foot-pounds on Webb's hub or 5000 pounds-per-stud. Levering also restrained two sets of bearings, entering the Y and Z coordinates of the nodes he wanted to have restrained. The program selected all the nodes that fit the parameters.

The von Mises Criterion predicted exactly what the fatigue test had shown - that the high stress area would be at the junction of the outside diameter of the rotor mounting boss and at the inside diameter of the wheel mounting boss. It also predicted a stress level that was going to be at a safe million-cycle fatigue life. Levering says that matched Webb's fatigue test results; of the prototypes that Webb tested, two pieces showed cracks at a million cycles, but still sustained load. These cracks initiated in the fatigue tests right where Algor predicted, at the high stress levels.

After this analysis, Levering tested the front hub in the very same manner, with Algor's prediction again matching Webb's. Good news for Webb, the front hub had a lower stress level than the rear's and zero failures for a million cycles.

Because of the success of verifying the hub design, Levering says he will regularly use FEA for upcoming designs. "Algor is just coming up with transient heat transfer analysis," says Levering. "A goal of mine will be using that in the future for castings. This will be of great help, since we have to design the castings for directional solidification."

Copyright © 1988 Algor, Inc. All rights reserved.