A novel building design — the lift-up design — has shown promise in removing obstacles and facilitating wind

circulation at lower heights in built-up areas, yet little is understood about how their design parameters can

influence the surrounding wind environment. This study develops a framework to study these parameters, and, using the knowledge, to modify the lift-up design to improve both the wind and thermal environments for pedestrians. The framework combines an Artificial Neural Network (ANN)-based surrogate model, an optimization algorithm (Genetic Algorithm), and Computational Fluid Dynamics (CFD) simulation to find the best lift-up design that maximizes either pedestrian wind comfort or thermal comfort or both. The optimization is done for two diametrically different climates: a hot climate with calm wind conditions ('hot-calm'), and a cold climate with windy conditions ('cold-windy). By adjusting eight parameters, the proposed framework enlarges, by more than 46% and 37% for öhot-calm’ and öcold-windy’ climates respectively, the area near a lift-up building where there is pedestrian wind comfort, and by 18% and 10% respectively for the two climates, the area where there is thermal comfort. These results indicate that optimum lift-up designs strongly depend on how the objective function of the optimization is set: e.g., whether to maximize area with pedestrian wind comfort or with thermal comfort or both.