Date of Graduation

12-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level

Graduate

Department

Civil Engineering

Advisor/Mentor

Selvam, R. Panneer

Committee Member

Arnold, Mark E.

Second Committee Member

Murray, Cameron D.

Third Committee Member

Heymsfield, Ernest

Keywords

Computational fluid dynamics; Large eddy simulation; Peak pressure; Random Fourier Generation; Synthetic Eddy Method; Synthetic inflow turbulence

Abstract

Structural failures by extreme winds often leads to economic losses and even deaths. Proper design of buildings requires accurate estimation of wind loads to prevent structural damage. Using routine guidelines generally leads to the underestimation of buildings’ peak pressures, in which most failures occur. To estimate wind loads more accurately three different methods (i.e., field measurement, wind tunnel (WT), and computational fluid dynamics (CFD)) are in use. A well-validated CFD provides more flow field details at lower costs compared to field measurements and WT. As strong winds are extremely turbulent, the effects of turbulence in wind should be incorporated into CFD using various turbulence modeling methods. Among these turbulence modeling methods, Large Eddy Simulation (LES) is more reliable and applicable in the industry compared to other methods. However, in LES simulations, a proper turbulent flow field at the inlet as an inflow boundary condition is required to apply and predict peak pressure correctly. Otherwise, CFD underestimates peak pressure coefficients. The turbulent flow behavior in the computational domain is extremely dependent on the type of inflow generators. Inflow turbulence generation methods are categorized to (a) precursor database, (b) recycling method, and (c) synthetic turbulence methods. Synthetic inflow turbulence is a more applicable approach, as it does not require expensive prior flow simulations. In this study, different types of synthetic turbulence generator methods are considered to investigate their performance in wind engineering applications by plotting pressures and wind spectrums. The velocity spectrum at the inlet and building location is compared to the Von Karman spectrum for different inflow methods to determine how well the inflow field is representative of the real wind flow and how well the energy is carried from the inlet to the building location. Furthermore, spurious pressures are introduced and different methods are evaluated to see whether they produce spurious pressures in the domain. It is concluded that spurious pressure exists in all the considered methods except Synthetic Eddy Method (SEM) method with the Gaussian shape function (SEM-G). In addition, SEM-G is found to be a suitable method for peak pressure prediction on buildings with the upmost 30% error. Furthermore, for the considered Random Fourier Generation (RFG) Method (i.e., Consistent Discrete Random Flow Generator (CDRFG)) an approach is suggested to control spurious pressure and improve computed peak pressures on buildings.

Download

Share

COinS