A Second Law Approach to Dynamic Optimization of Enumerated Aircraft Thermal Management System Architectures

This article presents a novel framework for the automated enumeration and dynamic optimization of thermofluid system architectures, demonstrated on air cycle machine (ACM) thermal management systems (TMSs). The approach addresses the challenge of exploring large design spaces by combining graph-based modeling, symbolic equation generation, and dynamic optimization to automatically derive and evaluate the governing equations for each feasible architecture. A case study optimizes 164 ACM architectures over a 4000-s mission profile, targeting two objectives: total entropy generation and time-averaged bleed air mass flow rate ( $L_{2}$ norm). Results show that optimal architectures differ across metrics, revealing architecture-level tradeoffs. The top-performing designs align with known practical systems. The methodology is general and scalable, with potential extensions to two-phase thermal loops, vapor compression systems, and other regenerative energy systems.