This research presents a general optimization workflow and a geometrical parametric study using computational fluid dynamics (CFD) analyses in conjunction with a Kriging-based active learning strategy. Motivated by the concept of a wind-resistant “cyclonic home”, the present work conducts a systematic numerical wind study for a typical gable-roof elevated house structure across a parameter space consisting of varying incident wind directions and roof pitch angles. The house model is fully parametrized such that geometrical and mesh generation are automatically performed to facilitate the deployment of active learning for establishing a surrogate model of wind-induced forces. A wind force map is produced which reveals the relationship between the incident wind angle and roof pitch. From this, statistical analysis exposes which roof pitch angles offer greater wind-resistance, i.e., a roof pitch of θ=32.9° is found to provide the lowest mean force across all possible wind angles, whereas roof pitches of θ=12.7° and θ=45.0° yield the largest forces. The influence of the overhanging eave, a typical feature of such houses, is varied to understand its role on overall roof performance. Quartering winds are found to provide a greater contribution to overall roof forces as the eave length is extended. In contrast, for winds perpendicular to the roof ridge, its extension provides a relatively stable total force. These findings can assist a variety of stakeholders from engineering and risk assessment disciplines in pursuit of buildings resilient to high-winds due to tropical cyclones.
Read full abstract