The construction of low greenhouse gas emission buildings requires the consideration of both embodied and operational emissions. While embodied emissions mainly occur in the construction phase, the operational emissions occur during the entire building life and are influenced by future developments such as climate change, electricity grid decarbonization, and user behavior. Suitable methods which allow considering these uncertainties already during the design stage are robustness assessments.

This study proposes a scenario-based robustness assessment method to investigate the effects of future developments on buildings’ greenhouse gas emissions. We investigate the scenario generation, building energy simulation, calculation of embodied and operational emissions, and the statistical evaluation of the results according to different robustness metrics. Two simulation models with different time scales are compared with a total of 12 test cases. Finally, the robustness assessment is demonstrated on a case study, where a total of 40 configurations against 1260 scenarios are investigated using five different robustness metrics: minimax regret, Spread, Maximin, Laplace’s Principle of insufficient reasoning, and Starr’s Domain criterion.

For the case study, a multi-family house located in Zurich, Switzerland, we demonstrate how future developments, e.g., climate change and electricity grid decarbonization, influence the building's greenhouse gas emission performance. For example, with slow decarbonization, heat pump systems perform most robustly compared to the other analyzed heating and cooling systems. Depending on the chosen robustness metrics, it is further possible to rate the solutions according to optimistic, pessimistic, or optimal preference, making the proposed framework a valuable tool for decision-makers.

In a time of transition, we conclude that analyzing building performance without future-oriented data is not suitable, and we encourage building designers to assess the impact of possible future scenarios in the design phase instead of purely using norm-based standard values, entirely based on data from the past.