A new way to look at welding:
researchers at Battelle have developed a modeling method that can predict fatigue life in weld joints and thereby cut millions in development costs
This is not just an incremental improvement. This method can cut development cost and time by more than half." The speaker is Pingsha Dong, technical director, Center for Welded Structures Research at Battelle (www.battelle.org; Columbus, OH). The subject is Verity, a modeling method developed by Dong and his team that they claim can reliably predict welded joint fatigue life and make much of the physical testing now used redundant. This may not sound particularly revolutionary, but calculating weld fatigue life has remained one of the few areas that has stubbornly resisted yielding to a simulated solution. Engineers still engage in the laborious and somewhat arbitrary practice of consulting standards books which offer information based on the physical test results of various examples. Once they find an example that resembles their design they essentially crib the test data, plug it into their calculations and hope for the best. It's an inexact method that necessarily leads to a lot of extra hours and dollars expended on physical testing to confirm durability.
BACK TO BASICS. Verity is a post-processor designed to utilize the data from finite element analysis (FEA) software solutions like MSC. Nastran (www.mscsoftware.com; Santa Ana, CA) and ABAQUS (www.hks.com; Providence, RI). Dong explains that all FEA programs generate data on nodal forces acting on a structure, but that this information is largely disregarded by engineers who are interested in more sophisticated stress calculations. But the problem with using stress data is that it can vary wildly depending on the mesh size of the model, so to maintain uniformity all engineers must agree to the same strict guidelines concerning mesh size, which rarely makes anyone entirely happy. And when it comes to large models like an entire body-in-white, the complexity of the structure makes controlling mesh size impossible. But by using the more fundamental nodal forces as its calculation basis, Verity removes mesh sensitivity as a factor, providing uniform results across any mesh size. That opens the door to the application of standardization and quality tools like Six Sigma that are not possible under the old example-based method.