Combustion process in flare systems is very important from protection of environment aspects. It is used heavily in chemical industries especially in refinery, gas and petrochemical plants. In this process poisonous, combustive and corrosive gases are converted into less hazardous materials. The length and amount of flame deviation due to cross flow (wind) are two major subjects in non-premixed flames of flares. The shape and length of flame play an important role in radiative heat transfer to equipments in site as well as calculation of flare height. In order to study different factors such as ratio of cross flow to the flow of discharged fuel jet, heating value of the fuel, and Wobbe number on shape and length of flame, simulation of combustion process in flare at jet velocities of 40 m/s and various cross flow velocities have been done. According to results, the flame length decreases initially by increasing of the ratio of air velocity to jet velocity and then it increases. It also can be understood that flames with higher heating value are longer and fuels with similar Wobbe number results in similar flame shapes. In order to prepare a comprehensive model for fast prediction of length and deviation of flame, an Artificial Neural Network (A.N.N.) model was designed and developed. For this purpose, all of required data for preparing A.N.N. model were obtained from simulation of combustion process in flare at jet velocities of 20, 30, 40, 60 and 80 m/s and various cross flow velocities. Another A.N.N model was designed and developed for identifying the combustion specification such as heating value, stoichiometric mass ratio of oxidant to fuel, and enthalpy of combustion product. The required data for using in A.N.N. model for various fuels is obtained from a pre-developed program based on chemical equilibrium. Also for easy generation of data for conventional design methods of flare and developed A.N.N. model, a software with user friendly graphical interface is designed and created.