Hybrid Surrogate-Based Multi-Objective Optimization of Graded BCC Lattices

Abstract

Functionally graded BCC lattice structures hold promise for applications requiring simultaneous impact absorption and thermal dissipation, yet existing optimization frameworks rely on raw geometric parameters that are spatially blind to features such as the orientation of design gradients. In this study, we optimize density-graded BCC lattices for concurrent crashworthiness and heat dissipation via surrogate-based goal programming, initially using raw truss diameter as design variables. The lattice is discretized into three zones to achieve optimal dimensionality, and Pareto-optimal designs were identified that improve specific energy absorption and peak stresses during collisions while also enhancing its Nusselt number and pressure drop under forced convection relative to a non-graded lattice. Key insights of the effect of material distribution along the gradation axis on the performance are discussed in detail. We then introduce Physics-Informed Geometric Operators (PIGOs), which are scalar quantities derived from both the diameter profile and the triangulated surface mesh as candidate surrogate design variables. Pearson correlation analysis reveals that the raw diameter variables remain competitive, with the porosity gradient emerging as the most broadly predictive PIGO, strongly capturing both peak stress and pressure drop. The Nusselt number resists prediction by all variables tested, confirming that heat transfer is irreducibly multidimensional in this configuration due to competing surface-area and flow-blockage effects. The PIGO framework is topology-agnostic and extends naturally beyond the three-zone parameterization to finer gradations and alternative lattice topologies.