The motivation for using virtual prototyping for durability at BMW is not just faster vehicle development - maintaining and improving the already high quality levels is equally important. Engineers within the vehicle development program want to improve the accuracy of the durability modeling process, and to be able to compare more design alternatives in virtual space before signing a prototype over to a physical test.
Development of a process for numerical durability analysis of body structures started in a joint project with LMS several years ago. Although BMW was already using LMS FALANCS for chassis components, virtual prototyping for durability of body structures required a separate development, in particular to account for spot welds and to accurately and efficiently perform stress analysis for large body structures.
The front section of a concept vehicle served as the application case for checking and improving the developed methods. BMW had three key requirements for validation: performing numerical analysis before physical test (‘true prediction’); checking intermediate results, in particular the stress analysis; and checking fatigue life - not just at a single critical location on the body structure, but also at locations where cracks started later. The applied loads were measured on BMW’s proving ground.
Results of the numerical stress analysis were checked against rosette strain gage measurements at 11 different locations. Several improvements of the finite element modeling were determined to be necessary. For example, accounting for the compliance of the test rig, and updating the finite element mesh to reflect the exact locations of spot welds and the actual sheet metal thickness of the physical prototype. With these improvements, results of the stress analysis were typically within 10% of the measured data, with the exception of two strain gages located in an area of rapidly changing stress.
Based on the improved stress analysis, the durability analysis with LMS FALANCS correctly determined the three critical locations in the sheet metal and the three critical locations at spot welds. FALANCS predictions correlated very well with test results: in the worst case the predicted crack initiation life was 25% of the test life. This, however, was due to detecting the crack in the test only at a crack length much larger than the 0.5mm used for the prediction. Accounting for this, the ratio of predicted to actual fatigue life was well within the typical test scatter.
Based on this validation of the process, BMW has started to use LMS FALANCS in its vehicle development programs. The first application was the optimization of the shock tower of the concept vehicle. Faced with a changed vehicle specification, which lead to much higher loads, BMW used FALANCS to come up with a new design just using virtual prototyping, which would otherwise have required an additional physical prototype.
As Horst Stamm, leader of the newly formed computer aided durability engineering group in the BMW body structures departments says, “We are convinced that our new virtual prototyping process for body durability will help us reduce development time while improving design quality. In particular, being able to detect all critical locations - including those which may not show up in a single physical test for statistical reasons or because the test is stopped after the first crack initiates - will reduce design loops.”
