Two novel resistance-capacitance network models to predict the dynamic thermal behavior of active pipe-embedded structures in buildings

Abstract

Active pipe-embedded structures (APES) are promising low-energy systems for reducing cooling and heating loads in buildings. These structures are usually multilayered, having a main concrete layer where pipes are located. Simplified heat transfer models of these systems are often required for building energy performance simulations. The majority of the simplified models available in the literature show limitations to capture accurately the dynamic thermal behavior of these systems, especially when they have large thermal mass. The goal of the present effort is to introduce two new Resistance-Capacitance (RC) network models, the Delta and the Umbrella, for the main concrete layer of a prototypical APES system. The parameters of the models are obtained through a genetic algorithm, which is dynamically coupled with the models. This algorithm minimizes the error between the RC networks and a baseline dataset that was generated through a frequency-domain finite-difference (FDFD) model of an APES system. To assess the performance of the proposed models, they are compared with two others from the literature for three different thicknesses of the main layer. The results show that the Delta RC network has, in general, similar performance to the literature models. The overall nRMSE (normalized root mean square error) values for these models is between 6% and 10%. They are able to predict accurately the thermal response of the system to some of the boundary conditions analyzed, but show limitations for massive concrete layers under high frequency thermal fluctuations. The Umbrella RC network, on the other hand, was shown to have remarkable accuracy at higher frequencies than the other models, even for the massive structures. The overall nRMSE value for this model was of 3%.