Target-driven splitting SPH optimization of thermal conductivity distribution

• DOI: 10.1016/j.ijheatmasstransfer.2024.125476
• Publication Date: 2024-04-10T19:59:29Z
• Authors: Bo Zhang; Chi Zhang; Xiangyu Hu
• Keywords: Regularization; Heat equation

Efficiently enhancing heat conduction through optimized distribution of a limited quantity high thermal conductivity material is paramount in cooling electronic devices and numerous other applications. This paper introduces target-driven all-at-once approach for PDE-constrained optimization derives splitting smoothed particle hydrodynamics (SPH) method optimizing the problems. In this method, iteration system split into several easily addressed steps. A targeting step employed to progressively enforce direct target, which potentially leads increased PDE residuals. Then, these residuals are recovered an evolution design variable. After this, solution carried out further decrease residuals, ready next iteration. Unlike simulation-based approaches, present does not rely on adjoint state equation converged variable field each iteration, process significantly simplified accelerated. With utilization implicit SPH operator general numerical regularization formulation, information propagation accelerated stability greatly enhanced. Typical examples demonstrate that current yields optimal results comparable previous methods exhibits considerable computational efficiency. Moreover, feature more moderate extreme values, offers distinct advantages easier selection appropriate with conductivity.