Early Building Design: Informed decision-making by exploring multidimensional design space using sensitivity analysis

[Display omitted] •Development of a design methodology that can handle the vast design space in early building design.•A global design space is modelled from extensive Monte Carlo simulations.•Sensitivity analysis methods applied to guide decision-makers.•Interactive visualizations help the multi-actor design team explore thousands of building performance simulations.•Metamodels are used to run additional simulations and demonstrate the holistic consequences of input changes. This paper describes a novel approach to explore a multidimensional design space and guide multi-actor decision making in the design of sustainable buildings. The aim is to provide proactive and holistic guidance of the design team. We propose to perform exhaustive Monte Carlo simulations in an iterative design approach that consists of two steps: 1) preparation by the modeler, and 2) a multi-collaborator meeting. In the preparation phase, the simulation modeler performs Morris sensitivity analysis to fixate insignificant model inputs and to identify non-linearity and interaction effects. Next, a representation of the global design space is obtained from thousands of simulations using low-discrepancy sequences (LPτ) for sampling. From these simulations, the modeler constructs fast metamodels and performs quantitative sensitivity analysis. During the meeting, the design team explores the global design space by filtering the thousands of simulations. Variable filter criteria are easily applied using an interactive parallel coordinate plot which provide immediate feedback on requirements and design choices. Sensitivity measures and metamodels show the combined effects of changing a single input and how to remedy unwanted output changes. The proposed methodology has been developed and tested through real building cases using a normative model to assess energy demand, thermal comfort, and daylight.