: Numerical simulation of fuel in diesel engine condition using reduced chemical mechanisms is presented in the present work. In order to do that new model Neuro-PaSR is offered. In the new model that is constructed based on partially stirred reactor model, a series of Artificial Neural Networks are used instead of solving differential equations for kinetic simulations. In order to generate the required data for constructing the network, a zero dimension Perfect Stirred Reactor is used. More than 100000 patterns are generated for the training phase of the Artificial Neural Network (ANN), using Perfect Stirred Reactor (PSR) model with detailed chemistry. The input layer consists of major species concentrations and temperature. The rates of production of species are considered as the output layer neurons. The new model is constructed and used for Methane and Heptane. For each fuel two versions of new models are presented one with 4 species(CO 2 , O 2 , H 2 O, Fuel) and the other with 5 species(CO, CO 2 , O 2 , H 2 O, Fuel). For each fuel, results are presented for a constant pressure chamber with new models, an extended mechanism Arhenius model and a one step Arhenius model. The new model performance in all cases( including initial temperature and initial concentration of fuel and oxidant) is better than one step Arhenius. In order to check the performance of new model, a constant volume chamber is used. Comparison between the results of the new model for the constant volume Aachen Bomb and other numerical simulations reveals the advantage of the present model, where very low computational cost and good predictions are observed. In the second part of the thesis blended fuel are considered. A new blended fuel including commercial diesel fuel and Isopropyl alcohol is proposed for diesel engine. Thermo physical properties of new fuel is measured experimentally. Cold behavior of new fuels spray is studied using simulation and compared against pure diesel fuel. Finally a 4 cylinder diesel engine fed with new blended fuel is tested and performance is compared against pure diesel. Key words: NeuroPaSR, Artificial Neural Networks, Non-premixed combustion, Diesel Engine