Accurate estimates of mean and peak pressure distributions on buildings facades are essential to assess wind hazards. CFD simulations with quantified confidence intervals can provide a detailed and accurate analysis for the wind load on structures that would otherwise only be evaluated with a building codes. They can also supplement wind tunnel tests required for complex tall structures, to achieve more complete results at a lower cost.
Our research aims to quantify, and improve, the accuracy of CFD results for pressure loads on buildings. The uncertainty in the CFD prediction of wind pressures on buildings arises primarily from two sources:
- uncertainty in the inflow boundary conditions representing the incoming atmospheric boundary layer, and
- uncertainty in the CFD model related to model choices such as the turbulence or subgrid model and wall model.
The specific objectives of our work are to establish methods to quantify both types of uncertainties in Reynolds-averaged Navier-Stokes (RANS) and large-eddy simulations (LES), and to validate the method with available test data for two different test cases: a low-rise and a high-rise rectangular building.
The inflow uncertainty quantification study accounts for the fact that the atmospheric boundary layer (ABL) measurement in the wind tunnel tests has an uncertainty associated to it. Based on the available measurements the uncertainty in the CFD model's inflow parameters, i.e. the roughness length, the velocity magnitude and the direction, was characterized. The figure below plots the range of velocity profiles that will be considered, and shows the resulting velocity field from one of the RANS simulations. The uncertainties in the inflow parameters will be propagated using a non-intrusive polynomial chaos method to quantify the uncertainty in the RANS prediction and validate the results against the experiments.
We established a turbulent inflow condition that enables investigating the influence of the incoming ABL turbulence characteristics on mean and peak pressure loads. The method is based on a divergence-free digital filter inflow condition . When performing simulations for conditions that mimic wind tunnel measurements of a neutral ABL, a decay in the turbulence intensity is usually observed downstream of the inflow. To overcome this problem, and ensure we achieve the correct turbulence statistics at the downstream location of interest, we implemented a gradient-based optimization algorithm. The algorithm adjusts the normal components of the Reynolds stress tensor specified at the inflow, until the components at the location of interest correspond to the wind tunnel measurements. The results demonstrate the expected monotonic behavior: higher Reynolds stresses at the inflow location result in higher Reynolds stresses at the future building location. The optimized result for the Reynolds stress profiles compares considerably better to the measurements.
This material is based upon work supported by the National Science Foundation under Grant Number 1635137. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.