Development of compact empirical models for variable-speed compressors for the prediction of energy consumption, mass flow and discharge temperature

In the context of heat pumps, developing prediction models for compressor performance holds significant importance for design, selection, control, and fault detection. Nevertheless, there is not currently an established standard for characterizing variable-speed compressors. Various published models offer different advantages and trade-offs in terms of accuracy and simplicity however, none of them strike the right balance between complexity and data fidelity, which is paramount, especially when working with limited or low-fidelity data.This study conducts a detailed analysis on how the compressor performance evolves within the compressor envelope focusing on the speed effect and, from this analysis, novel compact empirical models are obtained to predict mass flow rate and energy consumption. A study of discharge temperature is also presented in this work due to its relevance in quantifying heating capacity and power losses. To model it, a correlation based on electro-mechanical efficiency is proposed, which can be generalized for different suction conditions without introducing additional coefficients. The proposed models are data-driven by high-fidelity calorimetric tests and aim to maintain simplicity, using no more than six coefficients, while delivering acceptable accuracy with average prediction errors lower than 5%. Finally, a robustness analysis is carried out, which analyzes the required training data needed to fit the model and concludes that the proposed models can compete in accuracy with models that use much more coefficients and less training data.