A Modelica/Simulink Co-Simulation Framework with Improved Particle Swarm Optimization for the Optimal Chiller Loading Problem

Optimizing chiller load (OCL) distribution in multi-chiller HVAC systems is critical for energy efficiency, yet existing algorithms often struggle with accuracy and convergence. This challenge is compounded by the fact that existing research predominantly focuses on chiller-centric optimization, often neglecting the significant energy consumption of auxiliary components. To address this gap, this study proposes a novel method utilizing Modelica/Simulink co-simulation to accurately model the entire refrigeration system, including chillers, pumps and cooling towers, thereby eliminating complex mathematical derivations and enhancing real-world applicability. To solve this holistic optimization problem, an Improved Particle Swarm Optimization (IPSO) algorithm is developed, which integrates a Phased Adaptive Decreasing Inertia Weight (PADIW) strategy, adaptive learning factors, and a mutation operator to enhance its global search capability and robustness. A case study of a shopping mall demonstrates the approach’s efficacy: over a six-month period, the optimization method reduces total refrigeration system consumption by 25.5% compared to the strategy of distributing the load equally and 15.5% compared to the human experience strategy. Notably, this case revealed that the water pumps, while accounting for less than 20% of total consumption, held a disproportionately large energy-saving potential of over 25%. Comparative experiments and Monte Carlo simulations further confirm the proposed IPSO’s superior convergence and robustness over standard PSO and other common metaheuristics. This study demonstrates that the synergy of Modelica/Simulink co-simulation and the IPSO algorithm is crucial for realizing the full energy-saving potential of the entire system, particularly from previously overlooked components like the water pumps.