Fluor Daniel Lowers Client's Maintenance Costs

HYPERGEN HELPS FLUOR DANIEL LOWER CLIENT'S MAINTENANCE COSTS

This large model contains 13,156 nodes and just under 60,000 elements. It was constructed and analyzed by Fluor Daniel Design Engineer Mike Holden.

A cold rolling aluminum mill is no place for components that provide less than the maximum possible amount of strength and the longest service life. That's why companies in this industry are constantly looking for improved designs that can reduce maintenance requirements and lower costs. Recently, a leading aluminum processing company turned to Fluor Daniel, themselves a leader in design engineering, to make improvements to a series of components in the expanding mandrels that hold the unwind and rewind reels used in cold aluminum sheet processing.

Mike Holden, Design Engineer in the Equipment Group of Fluor Daniel's Mechanical Systems Department in South Carolina, was given the responsibility. Mr. Holden decided to create an Algor FEA model of the existing parts, find the high stress areas, compare these to parts that had cracked in use, and then perform analyses on several revised designs until he found one that offered the needed improvements.

Critical Component

The most critical component was the "star cone". This part engages in the internal "T" slots of four separate segments, which expand to hold the roll of aluminum. The highest stress concentrations of the assembly were in this piece. Like the other components, Mr. Holden could not redesign the star cone without paying attention to several design constraints. In his words: "The part had to be manufacturable. It also had to work with existing parts, so there were some areas of the design that could not be changed. The way I saw it was, I kept what worked and what didn't work I redesigned."

This model is a quarter section of the star cone. Because of its use, it receives very high levels of stress. The new design is currently in production.

3-D Required

When Mr. Holden began modeling the existing parts, it became apparent that, to simulate both the components and the loadings, a 3-D solid model would be needed. "The star cone could only be roughly approximated using brick elements because of the geometry," says Mr. Holden, "and even then, the modeling effort would be tremendous. I had heard about Hypergen and decided to give it a try. I figured I might as well, I was literally at a dead end without it."

Began with 2-D

Mr. Holden began by creating a 2-D model of the star cone, using Superdraw II and Algor's Supergen 2-D automatic mesh generator. "This approach allowed me to quickly see what the stress distribution looked like prior to creating the 3-D model. Using Supergen I could easily refine the mesh in critical areas and get precision to a high level. Beginning in this fashion provided an insight that was inexpensive in terms of modeling time," says Mr. Holden.

The 3-D models were created with Supersurf, Superdraw II and, of course, Hypergen. In Mr. Holden's words: "Supersurf is super powerful. It was easy to create the surface mesh. I was able to make the mesh coarse where I could and, once I knew where the high stress areas were, refine the mesh."

Meeting Customer Needs

Once the existing components had been modeled, Mr. Holden began making design changes and running analyses to determine which design best met the customer's needs. "Some of the models were large," says Mr. Holden, "But Fluor Daniel's PCC Group helped me use our network to access a two-gigabyte hard drive. I also had access to an HP Paintjet, so I created a series of color printouts which were useful in showing the design changes to our client."

The new designs are currently in production. As to his initial experience with Hypergen, Mr. Holden says: "I couldn't have done it without Hypergen. I was at a dead end. Using Hypergen with Supersurf made it easy to create the surface mesh, and then change the mesh as needed. The use of tetrahedral elements also allowed me to slice through my 3-D models in Superview to analyze the internal stress distributions."