Partitioning of indoor airspace for multi-zone thermal modelling using hierarchical cluster analysis

The article proposes a hypothetico-inductive approach to the formulation of suitable thermal zones, for subsequent multi-zone modelling and spatial control of microclimatic variables in buildings. Here, model structures are initially identified from data, thus avoiding undue reliance on prior hypotheses and ensuring that the resulting models are fully identifiable from the available temperature measurements. More specifically, an agglomerative hierarchical clustering approach is used to quantitatively distinguish and group thermal zones within an open airspace for any given ventilation and heating combination. To evaluate the new approach, the article utilises a previously developed Hammerstein type model for temperature, which is extended in this article to address the multi-zone modelling case. Experimental results are presented for a laboratory forced ventilation chamber, instrumented with 30 thermocouples, and recommendations are given for future application to a closed-environment agricultural grow cell being developed by the authors and industrial partners.