In this work, a topology optimization parallel-computing framework is developed to design support structures for minimizing deflections in Laser Powder-bed Fusion produced parts. The parallel-computing framework consists of a topology optimization model and an Inherent Strain Method (ISM) model. The proposed framework is used to design stiffer support structures to reduce the before and after-cutting deflections in printed cantilevers. Gravity load and residual stresses calculated from ISM are applied in the topology optimization model. The optimized results were printed and analyzed for validating the effectiveness of the proposed model. Experimental results show that the optimized supports can achieve over 60% reduction in part deflection as well as over 50% material usage reduction compared to the default support structure. In addition, ISM also was used to predict the part deflections and shows good agreement (average error of 6%) between the experimental and simulated results. Lastly, the multi-node parallelization of the proposed framework showed ~ 5 times speedup compared to a single-node implementation.

在这项工作中，开发了一种拓扑优化并行计算框架，以设计支撑结构，以最大程度地减少激光粉末床聚变生产部件中的挠度。并行计算框架由拓扑优化模型和固有应变方法（ISM）模型组成。所提出的框架用于设计较硬的支撑结构，以减少印刷悬臂梁在切割前和切割后的变形。由ISM计算的重力载荷和残余应力被应用到拓扑优化模型中。打印并分析了优化结果，以验证所提出模型的有效性。实验结果表明，与默认的支撑结构相比，优化的支撑可以将零件变形减少60％以上，并将材料使用量减少50％以上。此外，ISM还用于预测零件变形，并在实验结果和模拟结果之间显示出良好的一致性（平均误差为6％）。最后，与单节点实现相比，所提出框架的多节点并行化显示出约5倍的加速。