Vahab Montazeri - Department of Chemical Engineering, Razi University, Kermanshah, Iran
Atefeh Ghazi - Department of Chemical Engineering, Razi University, Kermanshah, Iran

AbstractIt is essential to separate two immiscible liquids from gas to produce the light liquid, heavy liquid, and vapor phases. The separation of water from hydrocarbons is a practical example in the oil industry. For such separation in industry, three phase separator is used. In this study, different parameter and the weight of the three-phase separator was optimized with the genetic algorithm (G.A.) and finally, the total cost of manufacturing the separator was decreased. Different types of three-phase separators are vertical, horizontal, and spherical. The separator works in the operating condition of ۱۷۲ kPa and ۴۴۵ K, respectively and the real weight of the separator is ۸۱۳۱ kg. For the optimization target, the flow of vapor, light liquid, and heavy liquid was considered constant during the optimization process. The objective function (O.F.)  is obtained from the weight of the separator and ۳ multiparameter equations. Also, ۷ parameters which include: separator aspect ratio (L/D), the height of heavy liquid (HHL), height of light liquid (HLL), hold-up time (TH), surge time (TS), low liquid level (HLLL) and vapor level (Hv) are used in G.A. as constraints. The weight of the optimized separator was calculated ۶۰۰۱ kg approximately. So, with this method, the total weight of the separator decreases by about ۲۶.۲ % as compared to the real weight of the separator. On the other hand, the maximum difference between the answers was ۳.۳%, which is acceptable.  Also, error analysis of the predicted results is calculated by mean absolute percentage error (MAPE) for ۷ design parameters of the three-phase separator and separator weight, which are in an acceptable level of accuracy. The presented approach can have potential application for the development of low-cost manufacturing of three-phase separators in the petroleum industry.

optimization, Three-phase separator, Genetic Algorithm, Petroleum, Objective Function