Portal de Periódicos da CAPES

Achieving minimum length scale in topology optimization using nodal design variables and projection functions

2004; Wiley; Volume: 61; Issue: 2 Linguagem: Inglês

10.1002/nme.1064

1097-0207

James K. Guest, Jean H. Prévost, Ted Belytschko,

Advanced Multi-Objective Optimization Algorithms

Abstract A methodology for imposing a minimum length scale on structural members in discretized topology optimization problems is described. Nodal variables are implemented as the design variables and are projected onto element space to determine the element volume fractions that traditionally define topology. The projection is made via mesh independent functions that are based upon the minimum length scale. A simple linear projection scheme and a non‐linear scheme using a regularized Heaviside step function to achieve nearly 0–1 solutions are examined. The new approach is demonstrated on the minimum compliance problem and the popular SIMP method is used to penalize the stiffness of intermediate volume fraction elements. Solutions are shown to meet user‐defined length scale criterion without additional constraints, penalty functions or sensitivity filters. No instances of mesh dependence or checkerboard patterns have been observed. Copyright © 2004 John Wiley & Sons, Ltd.