The adoption of Bayesian causal models for rough sizing of buildings can be critical every time that innovative technologies are due to be adopted. Indeed, the shortage of proper simulation programs relating to recently developed technologies prevents their application in the contemporary construction market. Object Oriented Bayesian Networks are used to develop simulation models capable of coping with incomplete or uncertain information, and working in a way that is very similar to the one practiced by human designers.
The application to Roofponds is particularly meaningful, thanks to the highly complex behaviour of these components and the many contrasting variables involved. The models developed show the high potentials of Bayesian Networks for prediction and multi-criteria decision making.
This research have been carried on in collaboration with Prof. Alfredo Fernandez-Gonzalez and the team from NEAT Lab (University of Nevada – USA).
Alessandro Carbonari, Berardo Naticchia, Alfredo Fernandez-Gonzalez (University of Nevada – USA)