Load forecasting for the efficient energy management of HVAC systems

Abstract

In this paper, Artificial Neural Networks (ANNs) are used to achieve cooling load forecasting in HVAC (Heating, Ventilating, and Air Conditioning) systems. Load forecasting is crucial in plant configurations making use of thermal storage technologies, where, during the nighttime, part or most of the energy required during daytime is produced at lower cost by cooling or icing water. Load forecasting is then needed to quantify the energy to be stored for the following daytime and to set up strategies for its release during daytime. Although many algorithms have been presented in the literature for load forecasting, they often need as input a large data set, that is not always available in practical situations. In this paper, we present an algorithm based on ANNs that allows to obtain sufficiently accurate load predictions by exploiting a limited data set, obtained by measuring quantities that are typically available in standard HVAC installations. Furthermore, knowledge of the current thermal load (which is needed to setup the data set for ANN training) can be obtained by using a load estimation algorithm previously proposed by some of the authors, that only need basic knowledge of the system hydronics. Another distinctive feature of the algorithm is the use of the AHU schedule as a means for inferring information on the internal loads, which is in general not available in practice. Simulation results for both CAV and VAV HVAC systems confirm the viability of the approach.