Optimization and Systems Theory Seminar
Friday March 1 at 11.00 Lindstedtsvägen 25, room 3721
Saab Aeronautics and Linköping University.
Stress and fatigue constrained topology optimization
The presentation concerns structural optimization in conceptual design stages, for which we have developed stress and fatigue constraints for topology optimization problems. Both stress and fatigue are local phenomena, but, in order to ease the computational cost related to a high number of constraints, we use a clustered approach which means that several stress evaluation points are clustered and one stress constraint is applied to each such cluster. As the number of clusters, and thus constraints, can be varied, the clustered approach allows for a trade-off between computational cost and how well the local stresses are constrained.
A clustered stress measure is created such that it approximates the local stresses in an adequate way. Different approaches for how to sort stress evaluation points into clusters and how to update the clusters, such that the results are sufficiently accurate for conceptual designs, are developed and evaluated. Two-dimensional examples are shown which confirm the theoretical discussions and the designs that are obtained have managed to avoid large stress concentrations, even for design domains with an initial stress singularity.
Using loads described by a variable load spectrum and material data from fatigue tests, the tensile principal stresses are constrained by a limit that is determined such that fatigue failure will not occur. In the examples, where the mass is minimized subjected to fatigue and static stress constraints, simple topologies are obtained and the structural parts are sized with respect to the critical fatigue stress and the yield limit.
Compared to the traditional stiffness based designs, the stress and fatigue constrained designs are considered to be closer to a final design, which will decrease the total product development time.
Calendar of seminars
Last update: February 11, 2013.