D P Jenkins et al Probabilistic climate projections with dynamic building simulation: Predicting overheating in dwellings

 Energy and Buildings 43 (2011) 1723–1731

This study, as part of the LowCarbon Futures project, proposes amethodology to incorporate probabilistic climate projections into dynamic building simulation analyses of overheating in dwellings. Using a large climate projection database, suitable building software and statistical techniques (focussing on principal component analysis), output is presented that demonstrates the future overheating risk of a building inthe formof a probability curve. Such output could be used by building engineers and architects to design a building to an acceptable future overheating risk level, i.e. providing evidence that the building, with specific adaptation measures to prevent overheating, should achieve thermal comfort for the majority of future climate projections. This methodology is overviewed and the use of the algorithm proposed in relation to existing building practices.While themethodology is being applied to a range of buildings and scenarios, this study concentrates on night-time overheating in UK dwellingswith simple and achievable adaptation measures investigated.

Conculsions
A methodology for using probabilistic climate projections with
dynamic building simulation has been developed and shown to
have potential as an alternative to multiple, and time-consuming,
iterations of building simulation software. Generating this surro-
gate procedure from an initial dynamic simulation means that the
detail required for an overheating analysis is maintained while
allowing the user to account for many other climate projections
for that specific building. The result is that, rather than relying
on single design climates to estimate future overheating, a broad
array of climate projections can be used to capture the uncertainty
inherent in all future climate projections. Incorporating this into
existing risk analysis practices within building design is proposed
as being a route for integrating complex climate descriptions into
real building projects.
The proposed regression approach was able to predict between
78 and 86% of hourly internal temperatures when compared to the
ESP-r simulation software, with the validation process covering a
large selection of hourly climate files and a range of adaptation sce-
narios. The regression equation was calibrated based on a range
of locations, emission scenarios and timelines. Converting these
hourly predictions into a general overheating metric (of number
of hours above a threshold) showed that regression and simulation
values agreed within an error of 5% across all identified climates.
While specific software and interfaces are not presented here, a
selection of possible outputs of any future overheating tool are sug-
gested as being suitable for demonstrating the effects of different
adaptation scenarios on a current building that, although providing
adequate levels of thermal comfort in a current climate, might be
at risk of overheating in the future due to climate warming.

This study shows that probabilistic climate projections  can be used to avoid running building simulation models many times. A broad array of climate proections can be used instead of a single design climate, allowing a better understanding of possible future climate effects