Page 1 of 14 2013-01-1086 Large Eddy Simulation of Premixed Combustion in Spark Ignited Engines Using a Dynamic Flame Surface Density Model Ranasinghe, C, Malalasekera, W & Clarke, A Loughborough University, UK Copyright © 2012 SAE International ABSTRACT In this work, cyclic combustion simulations of a spark ignition engine were performed using the Large Eddy Simulation techniques. The KIVA-4 RANS code was modified to incorporate the LES capability. The flame surface density approach was implemented to model the combustion process. Ignition and flame kernel models were also developed to simulate the early stage of flame propagation. A dynamic procedure was formulated where all model coefficients were locally evaluated using the resolved and test filtered flow properties during the fully developed phase of combustion. A test filtering technique was adopted to use in wall bounded systems. The developed methodology was then applied to simulate the combustion and associated unsteady effects in a spark ignition engine. The implementation was validated using the experimental data taken from the same engine. Results show that, even with relatively coarser meshes used in this work, present LES implementation has been able to resolve the evolution of a large number of in-cylinder flow structures, which are more influential for engine performance. Predicted combustion rate and pressure rise is also in good agreement with the measurements. The limits of cyclic variations are well within the experimentally observed range. It has also been able to demonstrate the limits of cyclic fluctuations to a reasonable degree even with a fewer number of simulation cycles. A significant variation of flame propagation has also been predicted by the simulations. INTRODUCTION Use of the Large Eddy Simulation (LES) techniques has become increasingly popular for engineering simulations. In Reynolds Averaged Navier Stokes (RANS) method, the entire flow spectrum has to be modelled so that the predictions are model dependent to a larger extent. Whereas in LES, most of the flow field is resolved and only the sub-grid scale (SGS) part is modelled. Hence, LES results are more accurate, less model dependent and provide detailed information. In addition, LES is very much appropriate for engine simulations as it provides the opportunity to simulate inherent unsteady phenomena such as cyclic variations and combustion instabilities. Therefore, the use of LES in engine modelling provides a more reliable way of investigating operational and geometric refinements. LES techniques have been widely used in non-reacting flow modelling, but its application to combustion simulation is still in a preliminary stage. In particular, studies devoted to internal combustion (IC) engine simulations using LES is very limited. Studies [1-5] may be identified as some of the promising cold flow simulations which reveal the potential of LES predictions. Remarkably, compared to RANS, overall predictions of these simulations have found to be much better in agreement with experimental results. Application of LES in reacting flow modelling has largely been limited to non- premixed combustion [6, 7]. Premixed combustion modelling with LES is particularly a challenging task due to a number of difficulties detailed below. Usually, the premixed laminar flame thickness   , is so thin and it cannot be resolved using the classical LES mesh sizes. For example the flame thickness in SI engine applications is about 0.1 mm [8]. In order to adequately resolve the flame front, a minimum of 5-10 grid points are needed with a typical finite volume based CFD code [9]. Therefore, this resolution requirement is prohibitively expensive compared to the mesh resolution possible with current computing power, which is typically about 0.5mm. A possible solution is to neglect this physical consideration and model the combustion process using an eddy break-up (EBU) type formulation. In such an approach, any modelling discrepancy may be absorbed by an adjustable model constant. The EBU model has two major short comings: negligence of the chemistry interaction with combustion and poor predictions near highly strained regions such as near walls. Moreover, the EBU model constant seems to be strongly dependent on the flow condition and mesh configuration. However, there is only very little work carried out to investigate the applicability of this type of formulations [10] and some initial applications of the LES-EBU model can be found in [11, 12].