سال انتشار: ۱۳۸۹
محل انتشار: اولین همایش ملی توسعه تکنولوژی در صنایع نفت، گاز و پتروشیمی
تعداد صفحات: ۴
Ehsan Khamehchi – Amirkabir University of Technology, 424 Hafez Avenue, Tehran, IranAssistant Professor, Amirkabir University of Technology (Tehran Polytechnic)
Ebrahim Shamohammadi – MSc student of Gas Engineering, Amirkabir University of Technology (Tehran Polytechnic
Seyed Vahid Yasrebi – MSc student of Petroleum Engineering, Amirkabir University of Technology (Tehran Polytechnic
Ahmad Ebrahimi – PHD student of Petroleum Engineering, Amirkabir University of Technology (Tehran Polytechnic)
Gas hydrates are a costly problem when they plug oil and gas pipelines. The best way to determine the Hydrate Formation Temperature (HFT) and pressure is to measure these conditions experimentally for every gas system.Since this is not practical in terms of time and money, correlations are the other alternative tools. There are a few numbers of correlations for specific gravity method to predict the hydrate formation. As thehydrate formation temperature is a function of pressure and gas gravity, an empirical correlation is presented based on the Hammerschmidt correlation for predicting the hydrate formation temperature. In order to obtain a new proposed correlation, 357 experimental data points have been collected from gas-gravity curves. This correlation is programmed and assessed with respect to its capabilities to match experimental data published in the literature under varying system conditions (i.e. temperature, pressure, and composition). The LINGO software has been employed for statistical analysis of the data. Accuracy of our correlation is more accurate than the Hammerschmidt correlation. In order to establish a method to predict the hydrate formation temperature, a new neural network has also been developed with the BP (back propagation) method. This neural network (IPS) model enables the user to accurately predict hydrate formation conditions for a given gas mixture, without having to do costly experimental measurements. It is found that the IPS neural network and the AUT correlation have the same results and are more accurate than the empirical correlation.