PULLING IT ALL TOGETHER: FUEL CHEMISTRY MODELING ACROSS THE SCALES FROM INDIVIDUAL MOLECULES TO ENGINE SIMULATIONS

摘要

The need for accurate predictive models for fuel chemistry has been apparent for many decades: it is very difficult to design a new fuel or new combustion device if one cannot predict with any confidence whether or not the new system will work better than existing technology. By combining modern quantum chemistry with advanced kinetic modeling techniques, and by combining methods developed over several decades by a large number of researchers on several continents, it is now becoming possible to accurately predict combustion chemistry in detail for fuels containing molecules in the gasoline range. However, these simulations are quite complex, usually involving more than 10~3 species and 10~4 reactions, raising a host of practical and numerical issues at many different scales. Certainly practical fuel / combustor designers will not be able to construct and use such complex simulations without considerable help from easy-to-use software. Even for a fuel chemistry expert, it can be very challenging to identify and diagnose problems with these large simulations. Ideally the simulations should be presented in a way which makes it easy for the entire fuel chemistry / combustion community to identify problems and to add fixes / improvements, so that the fuel or device designers will be working from the best possible predictive models. Many members of the combustion chemistry community have recognized this need and are now addressing various aspects of the overall problem, ranging from database and validation issues through the development of better methods for building and solving large simulations. Here some recent progress in addressing several of the many technical issues that arise is outlined, with a focus on methods for estimating chemically- activated reaction rates, and methods for facilitating broader involvement of the chemistry community in making predictive fuel chemistry modeling a reality. Simulations of several recent experiments performed at MIT and at Yale are presented to illustrate some of the issues.