First Building Simulation and Optimization Conference Loughborough, UK 10-11 September 2012 © Copyright IBPSA-England, 2012 BSO12 MODEL FOR RETROFIT CONFIGURATION SELECTION USING MULTIPLE DECISION DIAGRAMS Niall P. Dunphy 1 , James Little 2 , and Roman van der Krogt 2 1 Cleaner Production Promotion Unit, Dept. of Civil & Environmental Engineering, University College Cork, Ireland 2 ThinkSmart Technologies Ltd., Cork, Ireland ABSTRACT The paper demonstrates the use of Multiple Decision Diagrams (MDDs) in consideration of building energy retrofit options. Candidate retrofit alternatives including associated key performance indicators (KPIs) (e.g. cost, energy, embodied carbon) can be compiled into MDDs where various performance implications can be effectively illustrated and KPI trade-offs explored. We argue MDDs are flexible supporting a wide range of computations around the decision process. Significantly, we show KPIs can also be used as constraints in the search for satisfactory retrofit configurations and conclude that the MDD approach complements existing methods of optimising building energy retrofit options by providing a reduced initial search space. INTRODUCTION There is widespread consensus that climate change is occurring and that anthropogenic greenhouse gases (GHG) are contributing significantly (Solomon et al. 2007). Consequently, there is international agreement that GHG emissions should be reduced to stablise atmospheric conditions and prevent dangerous interference with the climate (UNFCC 1992). The largest single source of greenhouse gases is the built environment, which accounts for up to 40% of global energy consumption and associated emissions (Cheng et al. 2005). As a consequence, improving building energy efficiency is seen as offering greater potential for energy savings and greenhouse gas reduction than any other single domain (EC 2010). The lifespan of buildings is such that 80% of European buildings in use today will still be used in 2030 (EC DG-Research and Innovation 2010). Consequently constructing new buildings to high energy-efficient standards on their own will not provide the required building stock energy performance improvements. Accordingly, substantial retrofitting for energy conservation and improved efficiency to existing buildings is needed. Choosing the right set of components, systems or interventions in the design of a new build or retrofitting energy efficiency measures to an existing building, is a complex problem. It involves multiple constraints in the context of the existing building and the specific requirements / preferences of clients. These assessment criteria include energy performance, cost, aesthetics, environmental impact, structural limitations, planning restrictions etc. CONTEXT No single configuration will maximise/minimise all the requirements above. There are too many trade- offs between the various building energy retrofit assessment criteria for designers to manually track. Typically, the objective of a building energy retrofit project is to maximise energy efficiency, thereby minimising operating costs while minimising capital expenditure (now and in the future). At the same time the existing building’s context must be considered and specific preferences of the client expressed. Significantly a life cycle perspective is increasingly being adopted with respect to these assessment criteria, particularly so with whole life carbon considerations and embodied carbon (Lane 2007, Sturgis and Roberts 2010, Jones 2011). In such types of decision-making, the modelling and articulating of complex trade-offs is of great importance for any system chosen to support the designer. Finding the optimal solution in such a subjective area (trade-offs are often difficult to put a value on) requires innovative solutions (algorithmically and visually) with respect to showing the designer what is available as he/she introduces constraints and expresses preferences while the building design is undertaking. Generally, this type of problem is a multi-criteria decision one and traditionally two approaches have been adopted in an effort to consider more than one criterion simultaneously (Triantaphyllou 2002, Kahraman 2008, Wang et al. 2009). The first is where all but one criterion is handled as a constraint and the final one made the objective. The second approach is where a weighted sum of each criterion is added to the objective function, reflecting the perceived importance of each. Both approaches require a priori information from designers: boundary conditions for the constraints and/or weights for the performance criteria. With little knowledge about the performance space of solutions in advance, designers may find it difficult to set appropriate values for those required inputs. Furthermore, only one optimal solution is obtained - 309 -