The seal in this case was called a wiper, also known as an excluder. Designed to seal out dirt, the part originally had dual lips so that if the outer lip failed, the inner lip would continue to provide protection. An initial analysis showed that the outer lip lifted during installation, making the wiper ineffective. That problem was resolved by changing the design, but the solution caused excessive stresses and also caused the inner lip to lift. A better solution turned out to be using only one lip - a decision that was made months sooner thanks to computer simulation.

 "The problem with seals is that once they're installed, you can rarely see them work," explains Jay Nogan, an applications engineer at HPS. "With this wiper, we might have seen the problem with the outer lip when we tested a prototype but we never would have known the inner lip was lifting. We probably would have learned it didn't work, but not without spending nine months of testing, if not a year."

One-Stop Seal Shopping

One of the company's strengths is the fact that it doesn't manufacture seals but instead works with a wide variety of seal manufacturers as strategic partners. "That was a conscious decision, made years ago, so that we wouldn't be limited to products that just our own factory could make," explains Nogan. "We are free to propose the best seal for the application." HPS products include a full range of rubber and plastic compounds such as the HPS-designed rod and rubber seals, metal-encased rotary seals, custom-designed components and standard components such as O-rings, U-cups, and oil seals.

 The company's use of computer simulation is fairly recent. It came about because of a problem with a seal that performed well in prototype testing, and then developed problems in certain installations but not in others. In some situations, the seal's inside diameter increased instead of decreasing as it was expected to. The problem, lip rollout, baffled HPS engineers, their customer, and their supplier.

 The customer, who uses the ANSYS design analysis and optimization software from ANSYS, Inc. (Canonsburg, PA) to predict the performance of metal parts, suggested applying the technology to this problem. Since no one at HPS had experience with nonlinear finite-element analysis, the company hired a consultant to do the work.

 "This was a trial case to see what FEA (finite element analysis) would show versus what we were seeing in the field," says Nogan. "ANSYS predicted the rollout and helped us understand why we were getting it in some cases and not others."

 This experience demonstrated the value of the computer simulation for observing otherwise hidden seal behavior. Soon after, HPS purchased its own Windows NT version of ANSYS and a Pentium Pro PC to run it on. The use of the software has eliminated most design-related physical testing, which used to take up to three weeks per sample.

Excluder Simulation

 Nogan then entered other information about the problem. In this analysis, since the problem involved material nonlinearity, he needed to supply the material's stress-strain curve. That information was used by the software to generate the Mooney-Rivlin constants needed for analysis of a hyperelastic part. (HPS is now gearing up for a much more extensive use of analysis, having 24 additional materials tested at an independent materials testing lab to determine stress-strain data.) Nogan also defined the boundary conditions for the problem. These included: the press fit of the wiper (which was modeled as the movement of the outside diameter of the seal toward the centerline); the bottom of the counterbore (which involved restraining a portion of the seal in the direction of the shaft axis); and the use of general contact elements to simulate the shaft.

 This portion of the simulation, preparing the model for analysis, took between two and two and one half hours. The compute time for the simulation took several minutes. After viewing the results, Nogan collaborated with the supplier's engineers to revise the design. After each design change, he ran another analysis on the new shape. After several iterations, it became clear that the dual-lip design would not work and they went to a single lip. In all, it took a couple of weeks and six iterations to arrive at the final decision. "Without FEA, getting to that point would have taken closer to a year," says Nogan.

 In addition to decreasing the time needed to develop this product by eliminating months of testing, the use of simulation also reduced costs. "When we make molds to manufacture prototypes, there is a big expense for tooling," says Nogan. "We eliminated much of that with ANSYS."

 An additional benefit of simulation for HPS, for the excluder problem and for the seal design in general, is that it makes the design process more scientific. Says Nogan, "It has been my experience in seal design that much of it is art. But FEA takes some of the art out of the process and makes it more scientific. Now that we have this technology, we intend to use it whenever possible to improve our seal designs.