Prediction of turbulent Reactive Flows
A project by Ghent University which aims to develop a flexible, open source Large-Eddy Simulations (LES) Computational Fluid Dynamics (CFD) code-base for multiscale modelling of several multidisciplinary applications.
Thanks to ever increasing computer hardware and continuous improvement of software (numerical algorithms and modelling), computer simulations play an increasingly important role in engineering applications and research. In this project, the focus is on turbulent reacting flows, one of the most challenging problems in fluid dynamics.
Turbulence increases mixing and friction, both of which are of great importance in practical engineering applications. Turbulent flows are irregular, random and chaotic. An analytical description of turbulence is often impossible due to the complexity of the flow, and numerical simulations using computational fluid dynamics (CFD) have become very popular in the last few decades. Simulations are more flexible and cost effective than experimental methods, but experimental data remain very important to confirm validity.
Turbulent flows contain a very wide range of time, length and velocity scales, covering typically 3 or more orders of magnitude. Consequently, it is often unaffordable (in terms of computing times) to perform ‘Direct Numerical Simulations’ (DNS) of flows in engineering applications. Therefore, turbulence is modelled. The most extreme way of modelling is the ‘Reynolds Averaged Navier Stokes’ (RANS) approach: all turbulence is modelled, only the mean flow is resolved. Whereas this is attractive from a computational point of view, (too) much accuracy is lost, hindering further progress at the level of fundamental understanding of physical phenomena or at the level of technological developments. This explains the increasing popularity of the ‘Large-Eddy Simulations’ (LES) approach: the largest turbulent scales (‘eddies’) are resolved in time and space and the smaller ones are modelled. This approach is adopted in the present proposal. The computational cost is still substantial, but recent developments of High Performance Computing (HPC) facilities open up new perspectives to the scientific community and make projects as presented here possible.
Beside turbulence, (bio-)chemical reactions are of key interest in the present proposal. Applications ranging from bio-systems to fire, over chemical reactors and combustion devices (furnaces and engines) will be studied. Whereas these end applications may seem strongly different at first sight, they all have in common that finite rate (bio-)chemical kinetics interplay with the underlying turbulent flow and that their behavior can be described by almost the same mathematical equations. This implies that in all these domains, common problems are encountered. Hence, a major strength of the present consortium and project outline, in that different research groups, all active in the broad domain of turbulent reacting flows and each developing knowledge and expertise in their specific research area and applied to a specific system or process, join forces in a multidisciplinary consortium to jointly tackle similar problems and seek for generic solutions. Indeed, there are high-potential opportunities in combining all available (but now scattered) expertise, such that problems can be solved at a faster rate (efficiency gain through knowledge exchange) and substantial new steps forward can be taken by each of the research groups in their specific domains separately, as well as by the multidisciplinary consortium as a whole.