Application of neural networks to predict the steady state performance of a Run-Around Membrane Energy Exchanger

Modeling the performance characteristics of thermal systems has been a research interest for many decades with moisture transfer systems experiencing a resurgence over the last decade, especially in heating, ventilating, and air conditioning (HVAC) applications. In this study, a neural network (NN) model is developed to predict the heat and moisture transfer performances (i.e., the sensible and latent effectivenesses) of a novel HVAC energy exchanger called the Run-Around Membrane Energy Exchanger (RAMEE) which is able to transfer both heat and moisture between exhaust and supply air streams. The training data set for the NN model covers a wide range of design and operating parameters and is produced using an experimentally validated finite difference (FD) model. Two separate NNs (one for sensible and one for latent energy transfer) each with five inputs and one output, are selected to represent the RAMEE. The results from NN models are numerically and experimentally validated. The root mean squared error (RMSE) between the FD and NN models are 0.05 °C and 2 × 10−5 kgv/kga, indicating satisfactory agreement for energy exchange calculations. The paper reports the weights and biases to make the results of this study reproducible. These NN models are very fast and easy to use therefore, they might be used for design and for estimating the annual energy savings in different buildings which use the RAMEE in their HVAC system. Additionally, the NN models can be used with optimization algorithms to maximize energy savings and minimize life-cycle costs for a given system.