Modeling HVAC degradation due to climate shocks and stresses using dynamic Bayesian networks

The impact of climate conditions on infrastructure is a major concern for the sustainability of built environment. Two main issues that add uncertainty and complexity in climate-change impact are of interest: multiple hazard types and non-stationarity of climate actions. This paper proposes an approach using dynamic Bayesian networks to assess the reliability of a building system considering both gradual and extreme climate factors over the service life of the asset. The methodology is illustrated on a case study that examine an HVAC system, considering overheating fault and degradation risk. Compared to conventional Markov model, the results show stochastic dependence in the degradation process at different time instants and hence affect the variability of degradation. The proposed approach includes economic-based impact analysis to determine costs and payoffs accrued as the consequences. By integrating climate stress and shock and accounting for dynamic changes of the hazard, this method helps decision-makers in identifying and prioritizing adaptation strategies for building system under climate change.