Verification and validation framework for computational fluid dynamics simulation of wind loads on tall buildings

© 2019 Yousef Abu-ZidanWind loading is significant in tall buildings and typically governs their structural design. Controlling wind effects in tall buildings often requires major design modification to the building geometry and dynamic properties. Hence, there is a need for an accurate and computationally efficient wind analysis tool that would facilitate assessment of aerodynamic building performance during different stages of design. Computational fluid dynamics (CFD) is a numerical wind analysis tool that overcomes limitations of traditional wind tunnel testing including higher cost and time requirements. However, CFD is scarcely utilised in the building industry as a quantitative wind analysis tool due to a lack of confidence in the reliability of its predictions. Therefore, this thesis aims to develop a methodology for evaluating the performance of CFD models in predicting wind loads on tall buildings, so that confidence in CFD as a wind design tool can be established. To achieve this aim, the thesis first presents a review and synthesis of the theoretical knowledge required to adopt the generic verification and validation (V&V) framework from the field of Computational Science and Engineering (CS&E) to the specific context of CFD for tall building design. To the author's knowledge, no comprehensive work was dedicated to holistically implementing the V&V framework to CFD simulations of wind loading on tall buildings. A critical distinction between verification and validation activities is noted, where verification assesses whether the numerical model accurately reflects the modeller's conceptualisation of the problem, while validation evaluates the performance of the numerical model in predicting physical reality. Verification and validation activities relating to CFD simulations of tall buildings are described. Next, sensitivity studies are performed to investigate the nature of various numerical errors in CFD simulations of tall buildings, including ABL inhomogeneity, domain-size, and spatial discretisation errors. Notably, the study finds that errors due to ABL inhomogeneity and short upstream domain lengths can be especially significant and should therefore be controlled. Moreover, peak negative pressures that occurred in the first computational elements directly following flow separation edges are found to be highly mesh dependent and therefore unreliable. Next, the thesis implements the proposed methodology to Large-eddy simulations (LES) of a simplified tall building case. A validation experiment is conducted in a boundary layer wind tunnel and the results are used to validate LES predictions. The study reports on the accuracy of LES predictions of surface pressures, base reactions, and floor acceleration. The study discusses merits and limitations of the proposed methodology, as well as practical challenges to its implementation. These challenges include encouraging tall building industry projects to invest in validation-grade experiments and working with researchers to develop efficient strategies for the sharing of validation results