Prediction of water formation temperature in natural gas dehydrators using radial basis function (RBF) neural networks

详细信息    查看全文

• 作者：Tatar Afshin^a ; ^{Afshin.Tatar@gmail.com" class="auth_mail" title="E-mail the corresponding author} ; Barati-Harooni Ali^b ; Moslehi Hossein^c ; Naseri Saeid^b ; Bahadori Meysam^d ; Lee Moonyong^e ; Bahadori Alireza^f ; ^{alireza.bahadori@scu.edu.au" class="auth_mail" title="E-mail the corresponding author} ; Najafi-Marghmaleki Adel^b
• 关键词：Gas dehydration ; Triethylene glycol ; Water dew point ; MLP ; RBF
• 刊名：Natural Gas Industry B
• 出版年：2016
• 出版时间：March 2016
• 年：2016
• 卷：3
• 期：2
• 页码：173-180
• 全文大小：5192 K

文摘

Raw natural gases usually contain water. It is very important to remove the water from these gases through dehydration processes due to economic reasons and safety considerations. One of the most important methods for water removal from these gases is using dehydration units which use Triethylene glycol (TEG). The TEG concentration at which all water is removed and dew point characteristics of mixture are two important parameters, which should be taken into account in TEG dehydration system. Hence, developing a reliable and accurate model to predict the performance of such a system seems to be very important in gas engineering operations. This study highlights the use of intelligent modeling techniques such as Multilayer perceptron (MLP) and Radial Basis Function Neural Network (RBF-ANN) to predict the equilibrium water dew point in a stream of natural gas based on the TEG concentration of stream and contractor temperature. Literature data set used in this study covers temperatures from 10 °C to 80 °C and TEG concentrations from 90.000% to 99.999%. Results showed that both models are accurate in prediction of experimental data and the MLP model gives more accurate predictions compared to RBF model.