Simulation and validation of a VAV system with an ANN fan model and a non-linear VAV box model

The results of modelling and simulating a laboratory variable air volume (VAV) test rig are presented. The modelling for each component, such as fan, duct and VAV terminal box, with their control systems, in the test rig has been achieved and the simulations via the logical connection of the component models have been performed by means of an HVACSIM+ platform. A polynomial curve fitting method and a ten neuron sigmoid artificial neural network (ANN) model are used to model the non-linear characteristic of the fan. The ANN is shown to be superior. A non-linear characteristic terminal box model based on the experimental modelling of a European box is developed. The ANN fan model and the terminal box model are inserted in the program of the HVACSIM+ platform as the new component subroutine. The accurate simulation results for the VAV test rig via HVACSIM+ are validated by the real system operation performance. (C) 2002 Elsevier Science Ltd. All rights reserved.