Hybrid Probabilistic–Possibilistic Treatment of Uncertainty in Building Energy Models: A Case Study of Sizing Peak Cooling Loads

Abstract

18Optimal sizing of peak loads has proven to be an important factor affecting the overall energy consumption of 19 HVAC systems. Uncertainty quantification of peak loads enables optimal configuration of the system by opting 20 for a suitable size factor. However, the representation of uncertainty in HVAC sizing has been limited to 21 probabilistic analysis and scenario-based cases, which may limit and bias the results. This study provides a 22 framework for uncertainty representation in building energy modeling, due to both random factors and imprecise 23 knowledge. The framework is shown by a numerical case study of sizing cooling loads, in which uncertain climatic 24 data is represented by probability distributions and human-driven activities are described by possibility 25 distributions. Cooling loads obtained from the hybrid probabilistic-possibilistic propagation of uncertainty are 26 compared to those obtained by pure probabilistic and pure possibilistic approaches. Results indicate that a pure 27 possibilistic representation may not provide detailed information on the peak cooling loads, whereas a pure 28 probabilistic approach may underestimate the effect of uncertain human behavior. The proposed hybrid 29 representation and propagation of uncertainty in this paper can overcome these issues by proper handling of both 30 random and limited data. 31 Introduction 32Energy efficient building design merits special attention as the construction sector holds the largest share of energy 33 consumption in most countries (Birol 2010). The magnitude of energy consumption by the building sector has 34 resulted in governmental concerns that has led to implementing global and national regulations for promoting 35 energy efficiency in buildings (Guillén-Lambea, Rodríguez-Soria, and Marín 2016, Allouhi et al. 2015). To 36 comply with these regulations, new buildings are designed with special attention to both indoor comfort and 37 energy efficiency, while existing buildings undergo retrofits at envelope and systems levels. In either case, this 38 practice is associated with careful (re)design of the Heating Ventilation and Air-Conditioning (HVAC) systems. 39Indeed, HVAC design is very sensitive to the implementation of optimal temperature and humidity control, which 40 may account for up to 60% of the total electric energy consumption of a building (Pérez-Lombard, Ortiz, and Pout 41 2008, Zhao et al. 2013, Vakiloroaya et al. 2014). Studies show that cooling loads dominate the majority of HVAC 42 energy consumption in office buildings (Wan Mohd Nazi et al. 2017) and optimal configuration of chillers can 43 result in substantial energy saving (Salari and Askarzadeh 2015). 44The first necessary step for optimal design of HVAC system (that eventually results in the optimal configuration 45 of chillers/boilers) is to quantify the peak load on the heating/cooling system, which is commonly known as the 46 sizing process. Sizing the cooling system is frequently conducted according to the ASHRAE (Mitchell and Braun 47 2013) proce...

Uncertainty of solar radiation in urban canyons propagates to indoor thermo-visual comfort

“…A proper representation of uncertainty based on the nature and availability of data is key to obtaining a realistic perception from the environment (Khayatian et al, 2018). This study resorts to the possibilistic representation of the uncertain parameter "tree transmittance", as little to no information was available from the literature (Table 1).…”

“…The application of hybrid uncertainty treatment in building performance simulations has been previously discussed (Khayatian et al, 2018). This study opts for a similar approach to treat the uncertainties within urban canyons.…”

“…This study opts for a similar approach to treat the uncertainties within urban canyons. For the sake of brevity, details about the adopting framework in (Khayatian et al, 2018) for hybrid approach and possibilistic distribution on epistemic parameters are offered in Appendix A.…”