| authors |
Loveday, D.L., Virk, G.S., Cheung, J.Y.M. and Azzi, D. |
| year |
1997 |
| title |
Intelligence in buildings: the potential of advanced modelling |
| source |
Automation in Construction 6 (5-6) (1997) pp. 447-461 |
| summary |
Intelligence in buildings usually implies facilities management via building automation systems (BAS). However, present-day commercial BAS adopt a rudimentary approach to data handling, control and fault detection, and there is much scope for improvement. This paper describes a model-based technique for raising the level of sophistication at which BAS currently operate. Using stochastic multivariable identification, models are derived which describe the behaviour of air temperature and relative humidity in a full-scale office zone equipped with a dedicated heating, ventilating and air-conditioning (HVAC) plant. The models are of good quality, giving prediction accuracies of ± 0.25°C in 19.2°C and of ± 0.6% rh in 53% rh when forecasting up to 15 minutes ahead. For forecasts up to 3 days ahead, accuracies are ± 0.65°C and ± 1.25% rh, respectively. The utility of the models for facilities management is investigated. The "temperature model" was employed within a predictive on/off control strategy for the office zone, and was shown to substantially improve temperature regulation and to reduce energy consumption in comparison with conventional on/off control. Comparison of prediction accuracies for two different situations, that is, the office with and without furniture plus carpet, showed that some level of furnishing is essential during the commissioning phase if model-based control of relative humidity is contemplated. The prospects are assessed for wide-scale replication of the model-based technique, and it is shown that deterministic simulation has potential to be used as a means of initialising a model structure and hence of selecting the sensors for a BAS for any building at the design stage. It is concluded that advanced model-based methods offer significant promise for improving BAS performance, and that proving trials in full-scale everyday situations are now needed prior to commercial development and installation. |
| series |
journal paper |
| more |
http://www.elsevier.com/locate/autcon |
| references |
Content-type: text/plain
|
| last changed |
2003/05/15 21:22 |
|
