Development and implementation of practical modelling tools to assess the dispersion and transport of stack exhaust is key to reducing the risk of any potential exhaust re-entrainment. In addition to potential health and odor risks, re-entrainment of hot or moist exhaust can adversely affect equipment efficiency and the effective operating range. The tools currently used include desktop numerical models, wind tunnel testing and Computational Fluid Dynamics (CFD). There is some controversy on whether these tools all perform adequately. The current study aims to assess the accuracy levels of CFD modelling approaches for exhaust dispersion through comparison with wind tunnel testing data in an actual industrial environment. The CFD model is set-up using the same-scale, geometry and testing conditions as the wind tunnel to avoid any scaling uncertainties. Comparison between the capabilities of different Reynolds Averaged Navier Stokes (RANS) and Detached Eddy Simulation (DES) models in predicting the impacts of the surrounding structures on the overall wind flows and the transport of the exhaust mass as well as the dispersion and dilution of the exhaust mass as it is transported downstream. The limited capabilities of RANS models to predict pollutant distribution and the location of the pollutant affected regions are highlighted. Implementation of DES is recommended to model exhaust dispersion because of the improved prediction of wind flows around buildings – especially the separation and reattachment zones – using DES.