Computations of Laminar Premixed Flames using Extended FPI Modeling

INSA Rouen - CORIA, CNRS UMR 6614, 76801 Saint Etienne du Rouvray, FranceAbstractA new method to allow reacting flow computations using industrial type CFD solvers and detailed chemistry (throughthe FPI tabulation model) is introduced in this paper. Whereas FPI has already been extensively used in conjunctionwith laboratory CFD codes, its implementation in industrial solvers appears to be very challenging. In particular,species balance equations need to be solved with the corresponding chemical source terms being retrieved within theFPI databases. Among the difficulties arising from this weak coupling, it is noticeably shown that the interpolatedreaction rates may not be coherent with the locally computed flame structure. A specific modeling of the chemicalsource terms is hence required to prevent species concentrations from drifting from their tabulated states. To validatethis approach, freely propagating laminar premixed flames are successfully computed. Finally, the extension of themodel to non-adiabatic combustion thanks to the addition of an extra dimension (h) is discussed.IntroductionPredictingpollutantformationinindustrialdevicessuchas gas turbines or aeronautical combustion chambers hasbecome in recent years a really concerning issue due tothe increasingly restricting ecological norms. Within thiscontext,ComputationalFluidDynamics(CFD)hasraisedas a very promising tool to help industrials develop moreenvironment friendly technologies. However, predictingwith accuracy the behavior of turbulent reacting flows incomplex geometries remains a very challenging issue. Inparticular, the evaluation of minor species formation (es-pecially NOx and CO) requires a precise description ofthe chemical processes, generally involving hundreds ofintermediate species. To make such computations fea-sible within a limited CPU time, kinetic reduction tech-niquesarenecessarytodecreasethenumberofparametringvariables defining the problem. In that way, different au-tomatic strategies were developed, such as ILDM [1] orFPI [2, 3] (very similar to the FGM model [4]), based onthe fact that low dimensional attracting chemical man-ifolds can be identified during the evolution of simpli-fied reacting systems [1]. The existence of those mani-folds allows the construction of look-up tables where thechemical responses of elementary flames are stored ac-cordingtoalimitednumberofprogressvariables(usuallyless than 3). Therefore, the CPU time required for com-plex reacting flow simulations including detailed chem-istrywithaCFDcodeisdramaticallyreducedasthereso-lution of the whole reacting system is replaced by simpleinterpolations in chemical databases. Those approachesseem very promising but computations of complex 3Dgeometries with industrial type CFD solvers and tabu-lated chemistry models are still very tedious.Specific objectivesFew simulations of complex geometries using tabu-lated chemistry approaches were reported in the littera-