MULTIPLE LINEAR REGRESSION MODEL OF A DESICCANT WHEEL BASED ON EXPERIMENTAL DATA

Desiccant cooling systems (DCS) nowadays represent a suitable alternative to conventional systems based on electric-drivenchillers, for air-conditioning of residential and commercial buildings, thanks to several benefits. These benefits should becorrectly assessed by means of dynamic simulations of DCS, taking into account the operating context, such as climaticconditions, building loads, tariffs of energy vectors, etc. Solid DCS are typically based on desiccant wheels (DW), for whichseveral models are available in literature, to predict their performance as a function of operating parameters, such as inlettemperature and humidity ratio, regeneration temperature, air flow rates, etc. Nevertheless, physical models are rather complex tobe implemented in tools for the dynamic simulation of building-integrated energy systems. Regression models can represent asuitable alternative, as they can provide high accuracy but with a lower modelling effort with respect to physical models. In thispaper, experimental data collected in a test facility equipped with a DCS, are used to develop regression models for predicting thedehumidification and thermal performance of a desiccant wheel. Statistical tools are used to investigate the effect of severaloperating parameters on the DW behaviour. Comparisons with data provided by manufacturers software are also carried out.