Optimisation of energy consumption and daylighting using building performance surrogate model

Today, the architects are expected to identify solutions that provide best trade-offs among an excessively large number of possible design alternatives. Within this context, computational intelligence techniques prove to be valuable decision support tools. In parallel to this agenda, the current study aimed to present a novel approach towards identifying non-dominated design solutions that minimize annual building energy consumption and improve indoor daylight conditions. We applied the method to an L plan shaped office design. In this hypothetical building design, parameters of footprint area, number of levels, fenestration, shading, U-Values of building elements and HVAC system selection were set as variables; whereas total floor area and floor height were kept as constants in order to facilitate further practical relevance. A total of 105 simulations were performed for various values of the parameters. The resulting dataset was used to obtain two approximation models, for each of the objective functions. A Multi-Objective Evolutionary Algorithm was subsequently used to obtain the set of non-dominated solutions for the problem. Our results indicated the applicability of the proposed approach for decision-making practices at the conceptual design phase of relevant cases.