Low Carbon Futures Project Decision support for building

The risk of buildings overheating in a
low-carbon climate change future
a
Banfill ,
a
Jenkins ,
b
Gibson ,
a
Menzies ,
Prof PFG
Dr DP
Prof G
Dr G
Dr S Patidarb, Dr M Gula
aUrban Energy Research Group, School of Built Environment
bSchool of Mathematical and Computer Sciences
The project
LCF regression tool
The aim of the project was to design a method that
enabled probabilistic climate projections to be incorporated
into building design. The focus was on using such a
method to identify the potential risks of buildings “failing” in
the future due to climate and inappropriate design. The
main failure criteria concerned thermal comfort and HVAC
systems. The work combines statistical and building
sciences with various user feedback techniques to
adequately convey the opinions of practitioners.
The LCF tool uses Principal Component Analysis to
identify a correlation between hourly climate variables and
building simulation outputs. With a regression relationship
established, the tool is applied to hundreds of climate files
(such as those from the Weather Generator) to produce
hourly internal temperatures, heating loads and cooling
loads for each climate file. This information can then be
collated into useful output, as shown below:
Current climate
Med emission, 2030
Med emission, 2050
Med emission, 2080
100%
Probability of occurence
90%
80%
70%
60%
50%
40%
30%
20%
Overheating
threshold
10%
0%
0
5
10
15
20
% of occupied hours > 28°C
2080, High
2080, Medium
2080, Low
2050, High
2050, Medium
2050, Low
2030, High
2030, Medium
2030, Low
Current climate
% chance of failure
81-100
61-80
41-60
21-40
0-20
No Adaptation
With adaptations
Practitioners
Applying UKCP09
The UK Climate Projections 2009, through the Weather
Generator, can provide thousands of equally probable
climate descriptions across several emission scenarios
and timelines to 2099.
Manually processing all this climate information through
detailed building simulation is impractical. The LCF project
applied statistical techniques to detailed dynamic building
simulation as a form of model emulation. This allows the
full spectrum of climate projections to be processed
through a given building design.
To understand the requirements of building designers, a
series of focus groups, questionnaires and interviews were
carried out . This also provided feedback for the application
of the tool, influencing the form of output provided where
complex probabilistic information is generated alongside
more immediately intuitive colour-coded risk analysis.
The feedback suggested that the risk of future overheating
was highest in the non-domestic sector, although dwellings
in the south of the UK might also cause concern. There
was a general agreement that a clear and concise
mechanism for quantifying this risk would be welcome.
Distinctly Ambitious
Prof Phil Banfill
[email protected]
www.hw.ac.uk
@HWUrbanEnergy