Real transportation fuels, such as gasoline and diesel, are mixtures of thousands of different hydrocarbons. For multidimensional engine applications, numerical simulations of combustion of real fuels with all of the hydrocarbon species included exceeds present computational capabilities. Consequently, surrogate fuel models are normally utilized. A good surrogate fuel model should approximate the essential physical and chemical properties of the real fuel. In this work, we present a novel methodology for the formulation of surrogate fuel models based on local optimization and sensitivity analysis technologies. Within the proposed approach, several important fuel properties are considered. Under the physical properties, we focus on volatility, density, lower heating value (LHV), and viscosity, while the chemical properties relate to the chemical composition, hydrogen to carbon (H/C) ratio, and ignition behavior. An error tolerance is assigned to each property for convergence checking. In addition, a weighting factor is given to each property indicating its individual importance among all properties considered; the overall quality of the surrogate fuel model is controlled by a weighted error tolerance. It is observed that the solver can find an accurate surrogate fuel model for a low-cetane diesel fuel with 11 iterations. Finally, to further check the fidelity of the approach, the proposed surrogate fuel model is validated using a multi-dimensional engine simulation operated under a low temperature combustion (LTC) condition against the available experimental data. The results are also compared with a conventional single component model, viz., n-tetradecane representing physical properties and n-heptane representing chemistry. The results show that the proposed surrogate fuel model can accurately predict the overall combustion process and emissions, simultaneously; while the single component model is unable to predict the combustion process and emissions in the LTC condition for the low cetane diesel fuel.