Driven by legislative and community pressure, the energy, power and process industries are constantly being urged to improve the efficiency of their systems and significantly reduce the hazardous emissions from their facilities. To meet these challenges, but at the same time maintain a viable business, industries are seeking alternatives that enable them to comply with the emission constraints at minimum cost and with as few modifications as possible to their existing systems. For the last two decades, computational modelling has been emerging as a cost-effective, reliable and powerful tool for such applications. This paper demonstrates the capabilities of the Computational Fluid Dynamics (CFD) approach in full-scale complex industrial systems. A number of alternative modifications for an existing 8 MW landfill-gas flare were simulated, and an optimised design was selected. Various turbulence and combustion models were used. Although the k-epsilon turbulence model failed to predict the flow pattern in inert-swirl flows, and to capture the flame characteristics in the reacting flows, a Reynolds stress model succeeded. The limitations in using the Eddy Break-Up (EBU) combustion model and a single global chemical reaction to simulate ignition, extinction, blow-off and lift-off phenomena are discussed. The CFD results provided detailed information on the flow structure, eg flow reversal, recirculation and swirl patterns, and also on flame stability and characteristics. The results were then used to provide guidance for selecting the most appropriate design. The full-scale flare was then modified, and its performance compared with the original design. The results of the numerical simulations of the flow and flame behaviours were found to be in good agreement with experimental observations. Overall, the final design provided a stable and confined flame over a wide range of operational conditions. Relative to the original design, the measurements of temperature and combustion products at the exit plane of the flare showed lower gas temperatures and significant reductions in emissions of co (45%) and NO (60%).