Generally the wind loads on large span structures with a dome roof or free-standing canopy roof have been investigated by using wind tunnel experiments. However, in the case where wind pressures on isolated thin structures such as flat plate roofs are measured by using tubes, it is necessary to install the parts for storing tubes in the experimental model.
On the other hand, Computational Fluid Dynamics (CFD) is expected to be used as one of the methods except for wind tunnel experiments. CFD has enabled us to simulate without any physical restrictions on the experimental model. For wind-resistant design of buildings, it is required to predict accurately the peak wind loads, and the Large Eddy Simulation (LES) effective in obtaining approximately equivalent evaluation of wind loads on the buildings in urban areas to the results obtained in wind tunnel experiments. However, there are few studies on the evaluation of wind loads for large span structures using LES. In the case where wind loads on large span structures are estimated by using LES, it is required to reproduce appropriately the unique phenomenon such as conical vortices near the corner edge of the roof. In the previous study on LES for the flat plate roof model, although the present authors presented that it is possible to estimate the wind loads equivalent to the experimental results by using LES, the computed and experimental results of the minimum peak pressure near the corner edge of the roof did not coincide with each other, where the wind blew diagonally for the model.
The validation of the results of CFD against experimental results has been carried out by comparing statistics of wind pressure coefficients for each sampling point in general. On the other hand, wind pressures on each part of structures fluctuate with some correlation mutually. It is important for CFD to simulate the fluctuating pressure field equivalent to that of the wind tunnel experiment. Because it involves improving credibility of consideration on fluid fields, comparison of fluctuating wind pressure fields is significant and is one of the validations of the CFD results.
In this study, first, focusing on the experimental case that strong wind pressure occurs on the roof, examined are the computational conditions for reproducing this pressure by using LES. Validation of LES results is performed in terms of both statistics of wind pressure coefficients at each measuring point and fluctuating wind pressure fields obtained by using complex Proper Orthogonal Decomposition (POD) analysis. As a result, shown are the appropriate computational conditions and boundary conditions, which enable accurate evaluation of the conical vortices and related minimum peak pressure near the corner edge of the flat plate roof. Moreover, also proposed, is the technique which is valid as a method of evaluating the coincidence quantitatively for the correlation of fluctuating pressure fields.
Second, based on the computational conditions obtained the most accurate results in the above examination, wind pressure distribution on the flat plate roof is evaluated in detail by using LES for the original structure model excepted the parts for storing tubes. As a result, it is confirmed that the wind pressure which has the peak wind pressure coefficient of -14.0 and the peak wind force coefficient of -15.5 acts near the corner edge of the roof due to the conical vortices. Furthermore, coherent structures in the fluctuating wind pressure fields are examined by using complex POD analysis, and the relationship between the behaviors of the conical vortices and pressure fluctuation is discussed by using the visualization of computational results.
