Simulation is an informative tool to plan the design and renovation of engineering systems, but its use requires accurate inputs from the users. Inputs such as the unknown future operating conditions might be expected in a system’s lifetime, e.g., the future climate for which we design buildings today. Inaccurate representations of such unknowns reduce the effectiveness of planning. In this work, we discuss a planning method that works well with reasonably accurate predictions of the changing climate. The method relies on simulations with diverse plausible operating conditions. We demonstrate its use in designing robust systems with imprecise long-term predictions from climate models. We simulate the effect of the weather (outdoor environment) on environmental conditions inside a building, and estimate the energy required to make the indoor environment livable. Current practice is to select a small set of ‘representative’ future climates (typical weather) and use this small sample of outputs to make decisions. We show that, instead, simulating energy performance for a large sample of possible future climates can enable robust building designs. While this does increase computational load during the planning process, we have previously shown that fast and accurate approximations of simulations make it feasible to test large samples of operating conditions.