Response mapping methodology for premixed systems.

摘要

Practical combustion systems like gas turbines and internal combustion engines involve non-equilibrium chemistry. Numerical solution for such equations is tedious and involves solution of conservation and chemistry equations concurrently. A simpler approach is sought. Response modeling is one such technique. It involves pre-integrating the chemistry information and making it available to the physical system. The method reduces the stiffness of the equations, while maintaining the non-equilibrium information.; Premixed system while simple to model captures the complexities of a real combustion systems. Developing a method for using a response model for such a system is the goal of the present work. Propane is an hydrocarbon which exhibits properties similar to heavier molecules like gasoline. It is also a commonly used fuel. Moreover the reaction set for propane is manageable for purposes of integration. Hence, propane-air combustion is chosen for simulations.; A zero dimensional model representing a fluid particle traveling in a premix space while receiving energy fluxes is developed. The zero-dimensional solution from the model captures all the essential physics of a 1D premix combustion. Modifying the two key parameters in the model gives us wide range of solutions to different initial conditions.; Fifty-six simulations with different initial conditions are run. The information is used for mapping. The mapping procedure calls for certain parameters be mapped directly with neural networks while parameterizing others into profile shape for implementation. All the key variables necessary for developing the response model are obtained from the pre-integrated set.; Modified one-dimensional premix equations without the species fluxes are solved with the response map. The equation set is reduced to two first order ordinary differential equations and integrated by a fourth order Runge-Kutta method. Results are compared for different cases.