This paper discusses a method of significantly reducing geometric math model (GMM) passive radiator node count while maintaining accurate temperature responses from thermal math model (TMM) runs. Multiple GMM/TMMs are combined into larger models to predict the integrated system response. GMM reduction is often necessary because the central processing unit (CPU) time required to run large models typically increases exponentially with increasing node count. The result of using this technique is that GMM node count is greatly reduced while maintaining accurate temperature responses and saving CPU time.The methodology uses sink temperatures, computed during the TMM run, to serve as boundary conditions for detailed radiator grid thermal nodes. The methodology consists of defining nodes in the GMM/TMM, connecting simplified sink temperature nodes with high fidelity radiator thermal nodes, computing sink temperatures, and resetting external surface node temperatures so that the simplification has a minimal effect on other components.