How does the fatigue constraint influence the topology optimization result?

Jenny_Koeckritz

Hi,

I was wondering how the fatigue constraint, which can be set in the DTPL card for a topology optimization, influences the topology optimization result. In the out. file, I can tell that no sensitivities are calculated for the fatigue constraint, only a check is included. However, the final result with/without fatigue constraint shows a difference in the element distribution.

Therefore I wonder: How does the fatigue constraint constribute to the change of the design variable? Does somebody know?

This is a more general question, therefore I did not attach any model files.

Thank you in advance,

Jenny

SouravDas0306

Hi Jenny,

The fatigue constraint in the DTPL card does not usually act like a fully sensitivity driven response in the same way as compliance or stress constraints, which is why you do not see explicit fatigue sensitivities written in the .out file. In many topology optimization implementations, fatigue is treated more as a response check or indirect constraint evaluation rather than a directly optimized gradient-based quantity. However, it can still influence the optimization because the optimizer modifies the design variables to satisfy all active constraints simultaneously during the iterative update process. Even if fatigue sensitivities are not explicitly reported, the fatigue response can still affect element retention/removal through internal approximations, response screening, scaling, or by influencing which regions are considered feasible during the optimization cycle. Since fatigue behavior is strongly dependent on local stress amplitudes and stress redistribution, introducing a fatigue constraint typically pushes the optimizer toward smoother load paths, reduces sharp stress concentrations, and retains more material in dynamically critical regions compared to a compliance-only solution. This is why you observe different element distributions with and without the fatigue constraint, even though the .out file mainly shows a fatigue “check.” In practice, the fatigue constraint acts as an additional feasibility requirement that biases the density update process toward designs with lower cyclic stress amplitudes, even if the solver does not expose full analytical sensitivities in the output.

Regards,

Sourav