Real-Time Optimization of CO2 Heat Pumps for Industrial Dishwasher Applications
Transcritical CO2 heat pumps, due to their non-isothermal heat-rejection characteristics, are well suited for water-heating applications where high temperature lifts are required. However, maximizing the cycle performance while satisfying the thermal demands of real applications remains challenging, since the optimal high-side pressure depends on the sink temperature and operating conditions. In this work, a transcritical CO2 heat pump serving a professional dishwasher is modelled using a dynamic lumped parameters model. The dynamic model is used to verify that the required outlet water temperatures for the washing (55 °C) and rinsing (85 °C) phases are met, and to generate steady-state performance maps of the COP as a function of discharge pressure for different sink-side conditions. Analysis of these maps shows that, for a given outlet temperature, the optimal discharge pressure maximizing the COP is largely independent of the inlet sink temperature, making the high-side pressure an effective manipulated variable for supervisory optimization. A dither-free Least-Squares Extremum Seeking Control (LS-ESC) scheme is then implemented in MATLAB/Simulink and applied directly to the steady-state COP maps, operating on a 1-D lookup table without requiring an analytical model of the cycle. For each operating condition, the LS-ESC iteratively updates the discharge pressure and converges to the COP-maximizing value starting from non-optimal initial guesses. The results demonstrate that LS-ESC can reliably recover the optimum of realistic performance maps obtained from a detailed thermodynamic model, confirming its suitability as a supervisory optimizer for CO2 heat pumps during operation, adapting the operating conditions to achieve the best performance while satisfying the process constraints. The integration of the LS-ESC layer with the full dynamic dishwasher model for true realtime optimization is identified as promising future work.