Fine-tuning FEA Software

Customization improves analysis speed and efficiency

A longstanding goal at many engineering organizations is to extend the benefits of finite-element analysis to product designers. Traditionally, designers have used CAD-integrated tools to perform FEA functions. Lately, though, there’s been a growing trend to retrofit high-end FEA software specifically for designers. This trend is called FEA process automation. Essentially, it’s a way for companies to develop customized packages of analysis knowledge for deployment by FEA experts and non-experts alike.

“Ideally, customization automates tasks for an expert user while also providing sophisticated functionality to infrequent or non-expert users,” says David Winkler, principal engineer at ABAQUS Inc. “In the latter case, the expert takes on two new roles, one of tool builder and one of mentor to infrequent users, in addition to methods development and high-end analysis roles. Customization can be anything from a script that automates x-y plots during post processing to a fully supported analysis application with a unique graphical user interface, database, and solver routine.”

For example, Procter & Gamble customized their FEA process to speed up analysis of package designs and to make the interpretation of results user-friendlier. The company married HyperMesh from Altair with the ABAQUS FEA solver to create a Virtual Packaging System that automatically simulates the performance of plastic containers subjected to loads typical of shipping, stacking, filling, and so on. Engineers now achieve geometry healing, meshing, FEA solving, and report generation in 24 hours, without having to set up original analysis models. Faster analysis results contribute to the design of structurally innovative packages.

United Technologies Research Center (UTRC) uses a proprietary reliability code in the design of gas turbine components. They wanted to integrate reliability analysis with structural analysis, which was performed separately in ABAQUS. By incorporating their code as a module within ABAQUS, UTRC eliminated a number of intermediate steps in data input and streamlined their process. A custom user interface now guides engineers through the reliability analysis. Among the results is a twenty-fold improvement in analysis efficiency. Additionally, UTRC can calculate reliability during structural analysis at loads less than maximum stress. 

The international metals company Corus, formerly British Steel, developed a simulation system that completely automated the analysis of hot-rolled steel sections. Their project included FEA solver customization and creation of a designer-friendly user interface that enables engineers to simulate hot rolling of steel sections and packages analysis results for quick design decision making. Roll-pass designers on the mill floor now run section designs through virtual hot rolling trials that analyze the behavior of the metal as it passes through each mill stand, whereas previously all analysis was performed off-site. 

“Automation makes the job of an experienced analyst easier and leaves more time for higher-level thinking about analysis techniques,” says Jon Carter, a development engineer at ABAQUS UK who helped to develop the interface Corus employs as a front-end for automated roll-pass design analysis. “In most of our projects, analysts package up their established methodologies to increase their own efficiency. Corus was an exception, in that the analysts packaged their methodology and delivered it to the designers.”

Other ABAQUS projects have included developing a material model for a new automotive exhaust system, creating a new FEA element that simulates the hysteretic behavior of piping system couplings and supports without adding excessive computation time, and providing push-button automatic assembly of an analysis model for a transmission with over 400 parts.

Winkler and Carter of ABAQUS suggest the following areas to investigate for FEA process automation:

          •Analysis tasks that are repetitive, tediously manual, or prone to input error
          •Any existing analysis methodology that yields reliable results
          •Usability workarounds that can be permanently written into an interface

For analysts interested in doing their own customization work, the company includes an embedded version of Python, an open source and well-supported command scripting and GUI extensibility tool.