Evaporator modeling - A hybrid approach

In this paper, a hybrid modeling approach is proposed to model two-phase flow evaporators. The main procedures for hybrid modeling includes: (1) Based on the energy and material balance, and thermodynamic principles to formulate the process fundamental governing equations; (2) Select input/output (I/O) variables responsible to the system performance which can be measured and controlled; (3) Represent those variables existing in the original equations but are not measurable as simple functions of selected I/Os or constants; (4) Obtaining a single equation which can correlate system inputs and outputs; and (5) Identify unknown parameters by linear or nonlinear least-squares methods. The method takes advantages of both physical and empirical modeling approaches and can accurately predict performance in wide operating range and in real-time, which can significantly reduce the computational burden and increase the prediction accuracy. The model is verified with the experimental data taken from a testing system. The testing results show that the proposed model can predict accurately the performance of the real-time operating evaporator with the maximum error of ±8%. The developed models will have wide applications in operational optimization, performance assessment, fault detection and diagnosis.