A Nonlinear Model Order Reduction Framework for Dynamic Vapor Compression Cycles via Proper Orthogonal Decomposition
Dynamic modeling, Model order reduction, Vapor compression cycle, Proper orthogonal decomposition
A computationally efficient and accurate modeling approach is critically important for designing and evaluating controls and fault detection and diagnosis (FDD) algorithms. This paper proposes a reduced order modeling approach for vapor compression cycles (VCC) that involves application of nonlinear model order reduction (MOR) methods to dynamic heat exchanger (HX) models to generate reduced order HX models. A reformulated finite volume HX model was first developed that matches the baseline MOR model structure. Then, a nonlinear MOR framework based on Proper Orthogonal Decomposition (POD) and a Discrete Empirical Interpolation Method (DEIM) was developed for generating nonlinear reduced order HX models. The proposed approach was implemented within a comprehensive VCC model. Reduced order HX models were constructed for a centrifugal chiller system and coupled to quasistatic models of a compressor and expansion valve to complete the reduced order VCC model. The reduced cycle model was implemented within the Modelicabased platform and used to predict loadchange transients over a wide range of operating conditions for comparison with measurements. The proposed reduced order modeling approach is computationally efficient and accurately captures cycle dynamics.