Speaker
Description
Further investigation is required into the operational performance of a trans-critical CO2 heat pump when integrated with various energy sources, such as shallow geothermal and solar thermal collectors. However, conventional physics-based models of this system demand substantial computational resources when employed for operational optimization. To allow faster system simulations, a surrogate model of the CO2 heat pump has been developed using an artificial neural network (ANN). The ANN model takes in six (6) inputs: evaporator water-side mass flow and fluid temperature, gas cooler water-side mass flow and fluid temperature, set-point output temperature, and high-side heat pump pressure. Among these, some are parameters that connect with the other system components, while others are parameters that can be controlled for optimization. The model’s outputs comprise the electrical energy needed to run the heat pump, the heat from the gas coolers, and the temperature of the heat-pump heated fluid. Data used for training, validating, and testing the ANN model were generated by running a calibrated Modelica (TIL suite) model of the CO2 heat pump for 10000 combinations of input parameters obtained from Latin hypercube sampling. The training process involved the exploration of various hyperparameters, such as the number of layers, the number of neurons per layer, the activation function, the learning algorithm, the learning rate, and the number of training epochs, and was evaluated with its mean square error and root mean square error. The resulting surrogate ANN model can be integrated into the system model as a functional mock-up unit within Modelica to facilitate faster simulations for operational optimization.
Conference Topic Areas | Track9: Smart Energy Storage, Integration and Utilization |
---|