Corrosion in the petroleum industry is one the major concerns of the managers and engineers. Both upstream and downstream parts face this problem and the companies spend billions of dollars in this case. To remain competitive with the world market and win the game, this cost must be kept to a minimum and here the large demands for a reliable corrosion prediction model do exist which we can use in our decision making. On the other hand in many engineering problems the available information is vague and sometimes measured data or export knowledge is too imprecise to justify the use numbers. Fuzzy logic is a good solution here and helps us to compute with words. But the problem is that you can not train the fuzzy systems so neural networks may be useful here to add learning capability to fuzzy systems.

Using fuzzy logic as a powerful tool, 3-D corrosion models have been generated in the previous work. In this paper the problem of corrosion modeling will be discussed form the hybrid neurofuzzy approach. Teaching the fuzzy system with neural networks, and using pressure, temperature, oil and gas production rate, CO2 and H2S mole percent of the flow as the model inputs and the corrosion rate as the output, 3-D surfaces for the corrosion have been obtained and the prediction results of the model which is based on the pattern recognition between input and output parameters is tested. These graphs can be used as a reliable tool in corrosion prediction by managers and engineers.

You can access this article if you purchase or spend a download.