Accurate and Dynamic Accounting of Fuel Composition in Flame Propagation During Engine Simulations 2016-01-0597

A methodology has been implemented to calculate local turbulent flame speeds for spark ignition engines accurately and on-the-fly in 3-D CFD modeling. The approach dynamically captures fuel effects, based on detailed chemistry calculations of laminar flame speeds. Accurately modeling flame propagation is critical to predicting heat release rates and emissions. Fuels used in spark ignition engines are increasingly complex, which necessitates the use of multi-component fuels or fuel surrogates for predictive simulation. Flame speeds of the individual components in these multi-component fuels may vary substantially, making it difficult to define flame speed values, especially for stratified mixtures. In addition to fuel effects, a wide range of local conditions of temperature, pressure, equivalence ratio and EGR are expected in spark ignition engines. Flame speed variations with fuel composition and local operating conditions need to be captured well to predict combustion phasing and heat release rates.

To capture fuel effects, flame speeds of 44 potential surrogate fuel components have been considered. These 44 fuels included components representing the chemical families n-alkanes, isoalkanes, cyclo-alkanes, alkenes, cyclo-alkene, iso-alkene, aromatics, ethers, cyclo-ethers, alcohols and methyl esters. Laminar flame speeds, which are used to calculate turbulent flame speeds in ANSYS Forte CFD, were generated as tables using the ANSYS Chemkin 1-dimensional Flame-speed Calculator for each of the 44 surrogate fuels. Using pure-fuel flame speeds and local fuel composition in the CFD simulation, multi-component-fuel flame speeds were calculated on-the-fly using non-linear blending of the single-component values. The accuracy of this non-linear blending approach was verified using Chemkin simulations covering the range of operating conditions of interest. For each pure fuel flame speed table, about 1500 conditions of temperatures, pressures, equivalence ratios and EGR were evaluated, to cover most engine conditions.

These ANSYS Chemkin simulations used well validated chemistry from the Model Fuels Library, where a single master mechanism includes 9179 species and 38505 reactions, representing combustion and pyrolysis reactions of all 44 fuel components. From this super-set, targeted mechanism reduction was performed to assemble a skeletal but still very detailed mechanism for each fuel. Care was taken to verify the appropriate grid-resolution for the flame-speed calculations and to establish a method to address very high temperatures and pressures, where auto-ignition competes with flame propagation. The approach presented provides (a) simplicity in the user input required for the CFD simulation (only fuel composition is required), (b) a high degree of accuracy afforded by the Chemkingenerated flame-speed library for an extensive range of fuel components and (c) the automation of the blending without compute performance penalty. Example engine simulations show the impact on spark-ignited flame propagation, relative to conventional approaches.