Validation of CFD Predictions of Urban Wind - Developing a Methodology
“Good mental health in a fluid or CFD modeller is always indicated by the presence of a suspicious nature, cynicism and a ‘show me’ attitude. These are not necessarily the best traits for a life mate or a best friend, but they are essential if the integrity of the modelling process is to be maintained.” (Meroney, 2004) Over the past 50 years, Computational Fluid Dynamics (CFD) computer simulation programs have offered a new method of calculating the wind comfort and safety data for use in pedestrian wind studies. CFD models claim to have some important advantages over wind tunnels; which remain the most common method of wind calculation. While wind tunnels provide measurements of selected points, CFD simulations provide whole-flow field data for the entire area under investigation (Blocken, 2014; Blocken, Stathopoulos, & van Beeck, 2016). Similarly, wind tunnel measurements must consider the similarity requirements involved with testing a model at small scale, while CFD simulations can avoid this as they are conducted at full scale (Ramponi & Blocken, 2012a). However, CFD simulations can also often be misleading; and they should only be trusted once they can be proven to be accurate. To appease the requirements for this cynical view- referenced in the above quote- proper verification and validation of a model is imperative. This thesis investigated and tested the current best practice guidelines around CFD model validation, using existing wind tunnel measurements of generic urban arrays. The goal of the research was to determine whether the existing data and guidance around the validation process was sufficient for a consultant user to trust that a CFD model they created was sufficiently accurate to base design decisions from. The CFD code Autodesk CFD was used to simulate two configurations first tested as wind tunnel models by the Architectural Institute of Japan, and Opus labs in Wellington. The Wellington City Council wind speed criteria were used to determine whether the CFD simulations met the required accuracy criteria for council consent. Results from the study found that the CFD models could not meet the accuracy criteria. It concluded that while the validation process provided sufficient guidance, there is a lack of available data which is relevant to CFD validation for urban flows. It was recommended that at least one improved dataset was required, to build a system by which a consultant can identify what the requirements of a CFD model are to provide accurate CFD analysis of the site under investigation. To accommodate the range of sites likely to be present in urban wind studies, it was recommended that the new dataset provided data for a variety of wind flows likely to be found in cities.