Minimum length scale control in a NURBS-based SIMP method

Abstract

Controlling the minimum size (or length scale) of geometric features in topology optimisation analyses is of paramount importance. Particularly, the minimum length scale must always be considered according to the smallest manufacturable size. The goal of this work is to provide an intuitive methodology for controlling the minimum length scale in a NURBS density-based algorithm for topology optimisation. In this framework, the minimum length scale can be properly tailored by acting on some parameters tuning the shape of the NURBS geometric entity, i.e. degrees, number of control points and knot vectors components. The main consequence is that the proposed method does not need the introduction of an explicit optimisation constraint into the problem formulation to take into account for the minimum length scale requirement. The effectiveness of the proposed method is proven on meaningful benchmarks for both 2D and 3D applications. Results show that the minimum length scale requirement is properly met on the reassembled geometry at the end of the optimisation process.