عدم قطعیت و تجزیه و تحلیل حساسیت از ایجاد عملکرد با استفاده از پیش بینی های آب و هوایی احتمالی: مورد مطالعه بریتانیا

This study explores the uncertainties and sensitivities in the prediction of the thermal performance of buildings under climate change. This type of analysis is key to the assessment of the adaptability and resilience of buildings to changing climate conditions. The paper presents a comprehensive overview of the key methodological steps needed for a probabilistic prediction of building performance in the long term future (50 to 80 years). The approach propagates uncertainties in climate change predictions as well as the uncertainties related to interventions in building fabric and systems. A case study focussing on an air-conditioned university building at the campus of the authors is presented in order to demonstrate the methodology. This employs the most recent probabilistic climate change projections for the United Kingdom (UKCP09 dataset) and takes into account facility management uncertainties when exploring uncertainties in the prediction of heating energy, cooling energy, and carbon emissions.

In the UK, buildings account directly or indirectly for approximately 40% of national carbon emissions [1]. This is a major constituent of the human-driven greenhouse gas emissions, which are increasingly tied to global warming [2]. Reciprocally, building thermal performance is directly affected by changing weather conditions. Hence it is important to make sure buildings remain thermally comfortable, while being energy efficient and employing low carbon technologies. The impact of climate change on building thermal performance and the adaptation of buildings to changing environmental conditions have become active research areas [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19] and [20]. CIBSE TM 36 provides a comprehensive analysis of energy use and overheating in different types of buildings by using climate projections for the UK released in the 2002 (UKCIP02) [4]. Research by Guan [5], studying the impact of climate change on air-conditioned buildings in Australia, indicates that under the 2070 high emissions scenario a 28–59% increase of cooling capacity will be needed to maintain thermal comfort conditions. De Wilde and Tian [6] have implemented a probabilistic method and sensitivity analysis to identify the key variables affecting the thermal performance of a mixed-mode office building in Birmingham, UK. Coley and Kershaw [7] propose “climate change amplification coefficients” to correlate indoor air temperature to predicted weather conditions. More research has been carried out based on different climate change scenarios in different countries, such as Canada [8], the Netherlands [9], New Zealand [10], Portugal [11], Slovenia [12], Switzerland [13] and [14], the United Arab Emirates [15], and the UK [16] and [17]. Some research takes a wider view and compares the trends in building behaviour in a range of different climate zones [18], [19] and [20]. However, most of this existing work on the impact of climate change on building thermal performance is deterministic in nature. Furthermore, any application of sensitivity analysis in building performance simulation (the main methodology for predicting future building behaviour) is mostly focussed on local sensitivity analysis (one-factor-at-a-time). At the same time meteorological research progresses quickly. In the UK, a new set of climate change projections (UKCP09) was released in June 2009 [2]. This new dataset is the first to attach probabilities to different levels of future climate change. However, this new data is also a challenge to the building science disciple due to its complexity. At the same time it also provides an opportunity to further analyse building behaviour under climate change by using (sampling-based) Monte Carlo approaches. This is a commonly used method for estimating the impact of uncertainty in inputs on a corresponding uncertainty in outputs [21] and [22]. There are many sources of uncertainty in building energy simulation, such as weather conditions, physical properties of building materials, internal heat gains [23], [24], [25] and [26], and accordingly the sampling-based methods are very suitable for research on impact and adaptation to climate change in the built environment [6]. This study explores the uncertainties and sensitivities in the prediction of the thermal performance of buildings under climate change. It presents a comprehensive overview of the key methodological steps needed for a probabilistic prediction of building performance in the long term future (50 to 80 years). The approach propagates uncertainties in climate change predictions as well as the uncertainties related to interventions in building fabric and systems. The methodology is demonstrated by means of a case study that quantifies the likely impacts of climate change on energy performance and carbon emissions of a real, complex case study building using sampling-based uncertainty and sensitivity analysis. The case selected for this work is the Roland Levinsky Building at the University of Plymouth. The study considers the uncertainties in building simulation due to weather conditions (UKCP09) and other inputs in building performance simulation and identifies the dominant factors affecting thermal performance using global sensitivity analysis. The research employs the UKCP09 climate change predictions and incorporates the uncertainties in other inputs into the building energy simulation. Sensitivity analysis is used to identify the most influential factors that affect three performance indicators: annual heating energy, annual cooling energy, and carbon emissions. Uncertainty analysis provides more information for lifecycle building performance assessment [27], taking into account that the weather conditions will change within the life expectancy of buildings. The structure of this article is as follows. First, the probabilistic climate projections from UKCP09 are briefly presented, and the method of creating weather files for building simulation according to Finkelstein–Schafer statistics is described. Then the article reports on the thermal model which is used for transient thermal simulation, using the EnergyPlus programme. Finally, the predicted energy performance and carbon emissions of the case study building are explored, employing uncertainty and sensitivity analysis.

This research presents a comprehensive overview of the key methodological steps needed for a probabilistic prediction of building performance in the long term future (50 to 80 years). The approach propagates uncertainties in climate change predictions as well as the uncertainties related to interventions in building fabric and systems. The methodology described in this paper is suitable for other type of buildings at different locations. Even if probabilistic climate projections are not available, the uncertainties in other input factors in building simulation can still be analysed. In general, global sensitivity analysis is a better choice to identify the key variables affecting building energy performance although the local sensitivity analysis is still very popular. To get more robust results from sensitivity analysis, it is necessary to apply two or more methods in the analysis. It should be emphasised that the relative importance from sensitivity analysis is dependent on the assumed range of the inputs. Defining these inputs requires great diligence, for instance where a correlation between specific climate change scenarios and actual HVAC system sizing needs to be taken into account in order to obtain proper results which are not skewed due to a one-size-fits-all oversizing approach. This study indicates that the use of multiple weather files is important in quantifying the uncertainties in the prediction of the future performance of buildings. At present this is a non-trivial task for building performance analysts, as one needs to establish multi-year weather datasets, with data from weather stations often containing missing values and meteorological predictions for the future having to be transformed into a format that is suitable for building simulations. The paper discusses the application of the approach to an actual case study, exploring the uncertainties in the prediction of the thermal performance of the Roland Levinsky Building at the authors' campus under climate change, taking into account the probabilistic nature of climate change predictions (using the UKCP09 data) as well as the uncertainties related to interventions in the building fabric and systems. Trends and uncertainties in annual heating energy, cooling energy, and carbon emissions have been studied for the baseline period (1961–1990) and the 2050s (2040–2069). The following conclusions pertain to the Roland Levinsky Building under climate change. Note that these conclusions will not apply to other buildings operated in a different climate. (1) Simulation results indicate that annual heating energy will decrease and annual cooling and carbon emissions will increase as the climate warms. By the 2050s under a medium emission scenario, the mean annual cooling energy increases by 122% and the mean annual heating energy decreases by 40% relative to the baseline period (1961–1990). The uncertainty in predicted annual cooling energy is significantly higher than the uncertainty in predicted heating energy. A changeover from natural gas to electricity as the main fuel will lead to an increase in carbon emissions as the external temperature rises. Mean annual greenhouse emission will increase by about 14% by the 2050s for a medium emission scenario compared to baseline period. The uncertainty in carbon emissions also increases over time. (2) The change in thermal performance for this building is strongly related to annual heating or cooling degree days and the relationship between them is approximately linear. Other weather variables, like solar radiation and relative humidity, have only small effects. (3) From the baseline to the 2050s, the uncertainty in heating energy remains quite constant, while the uncertainty in cooling energy shows a significant increase when more variability in input parameters is introduced in the simulation. There is little overlap of the distributions for three performance indicators between the baseline and the 2050s, which suggests that significant changes in the thermal performance for this building are to be expected. (4) The thermal characteristics of the windows of the building have a significant influence on all three performance indicators. Infiltration rate also has important effects on heating energy, and reduced lighting heat gains will lead to a substantial decrease in carbon emissions. The uncertainty associated with climate becomes increasingly important from the baseline to the 2050s.