A real-time algorithm for the determination of R744 systems optimal high pressure

• DOI: 10.1016/j.ijrefrig.2012.01.005
• Publication Date: 2012-01-28T20:15:32Z
• Authors: Luca Cecchinato; Marco Corradi; Giovanni Cosi; Silvia Minetto; Mirco Rampazzo
• Keywords: CO₂; Heat pump; Artificial neural network (ANN); Transcritical cycle; 0202 electrical engineering, electronic engineering, information engineering; CO2; Optimisation; 02 engineering and technology; 7. Clean energy

In this paper the optimal energy efficiency and high pressure problem in single stage transcritical carbon dioxide vapour compression units is addressed. A real-time model-based optimisation algorithm for the optimal (or quasi-optimal, close to the optimal) pressure determination is developed as a more efficient and robust solution than literature approximated ones. The problem is solved by a model estimating the system performance and by measuring its boundary conditions. The model is obtained by an on-line artificial neural network identification technique and the optimisation problem is worked out by a particle swarm technique. The proposed algorithm is dynamically tested by simulation, considering the performance of a supply water temperature controlled carbon dioxide heat pump. It appears to successfully approximate the optimal cycle discharge pressure showing an average daily pressure deviation of 0.9 × 105 Pa over a two years simulation period, corresponding to 1.1% increase in the cumulated energy consumption, compared to the optimal pressure solution.