Turbulence modeling is among the most important challenges for the researchers. Over the past decades, different models have been proposed to explain the behavior of turbulent flows. An important feature of turbulent flows is the chaotic characteristics of the flow field. Such a feature has led researchers to adopt statistical approaches in tackling turbulence modeling challenge. The averaging method, in which the instantaneous quantities are decomposed into the mean and fluctuating components, is a well-known approach in turbulence modeling. As a result of this scheme, new unknowns are introduced into the governing equations that are subject to further modeling. Accuracy in predicting turbulent flow quantities will lead to precise prediction of the flow field and thus accuracy in all of the numerical simulation results. Modeling and simulation of turbulent reacting flows have been the focus of many researches. In order to simulate turbulent reacting flows, an appropriate combustion model is required. Due to separating chemical reactions calculations from the solution of turbulent flow field, the present flamelet model provides the possibility of considering chemical kinetic with any level of complexity. The purpose of this study is to evaluate the performance of various models of turbulent flux and Reynolds stress in turbulent reacting flows. The model considered in this study are the Reynolds stress model (RSM), Algebraic Reynolds Stress (ARSM) and turbulent flux models. In order to validate and choose the best model for turbulent flux, channel, Couette and backward facing step flows are studied in this study. Considered turbulent flux models include the standard model (Boussinesq analogy), the Daly and Harlow model and the Abe and Suga model. Simulation results show that the standard gradient model gives good predictions for normal heat flux, but fails to predict the stream-wise heat flux (errors up to 100%). It is also verified that the model of Abe and Suga accurately predicts the normal heat flux, and by a 25% enhancement, yields good predictions for the stream-wise heat flux. The model of Daly and Harlow also accurately predicts the normal heat flux profiles, but underestimates the stream-wise heat flux despite showing its correct trend. The Abe and Suga model predicts best results and is more consistent with the experimental data than the other two models. Subsequently, the reactive flow are studied for the case of a non-premixed bluff body flames using turbulent flux of Abe and Suga, Algebraic Reynolds Stress (ASM) and Reynolds Stress (RSM) models. Results of various turbulent flow models for velocity field, temperature field, species mass fraction and mixture fraction are compared with each other. The result reveals a good agreement with the experimental data. In this study, the results of reacting flow field in a swirl burner is also provided and axial velocity, swirl velocity, temperature distribution and oxygen concentration are presented at different cross sections. Results show that the combined model of Reynolds stress and Abe and Suga improves the prediction of turbulent flows. Keywords : Turbulent Flow, Turbulent Flux Model, Flamelet model, Algebraic Reynolds Stress Model, Non-Premixed Flame