Evaluation and validation of CFD for predicting dilution from roof top stack exhausts

Evaluation and validation of CFD for predicting near field dilution from roof top stack exhausts

Flow around bluff bodies such as buildings has proven to be a difficult problem for Computation Fluid Dynamics (CFD) to simulate (Stathopoulos 2002), in large part because of the common occurrence of flow separation about the building corners. Because of these difficulties, it is crucial that CFD practitioners show validation for a given application (for example, the determination of mean or peak values of concentration, pressure, or wind speed) when modelling flow around buildings.

A previous study conducted by the authors (Banks et al. 2003) compared CFD simulation results using FLUENT with wind tunnel test data for a simple isolated rectangular building. Separate CFD simulations were conducted at two different institutions. In some cases, the predictions were quite accurate. In others, they differed substantially from the experimentally measured values, even though the conditions had only varied slightly (a doubling of the stack height, for example). In addition, apparently minor differences in modelling technique between the two CFD operators led to substantially different predictions.

The study concluded that in order for CFD simulations to be used as a design tool, both the CFD code and the manner in which the code is used need to be validated against experimental data for a similar situation. This combined operator/code validation exercise continues at CPP and CSU, and this continued effort is the subject of this paper.

Several aspects of CFD simulation stack flow are examined in this paper. The previous study used the k-omega Reynold?s Averaged Navier Stokes (RANS) turbulence model in a steady state fashion, and this paper examines the merits of a quasi steady technique, as well as the merits of other RANS turbulence models. Also at issue are the boundary conditions near the stack exit (isothermal jet in a cross flow), and the use of an approach flow which is in equilibrium with the surface turbulence generation. Grid independence and Lagrangian frame of reference plume trajectories are also considered.

Banks, D., Meroney, R. N., Petersen, R. L., and Carter, J. J. "Evaluation of FLUENT for predicting concentrations on buildings." A&WMA 2003 Annual Conference, San Diego, CA, paper # 70223.

Stathopoulos, T. (2002). "The Numerical Wind Tunnel for Industrial Aerodynamics: Real or Virtual in the New Millenium?" Wind and Structures, 5(2-4), 193-208.