The Study Of The Oretical Models On Wax Disapperance Temperature In Waxy Crude Oil

When the temperature and pressure condition get changed,there is wax deposition in the production,processing and transportation in the crude oil and finally decreasing the rate of production.The problem of wax deposition will cause negative effect in the oilfield and the important factor about wax deposition is the Wax Disappearance Temperature(WDT).Grey correlation analysis,Pearson correlation and Spearman correlation analysis as the basis of modeling process,and combining the Support Vector Machine(SVM),Adaptive Neuro-Fuzzy Inference System(ANFIS)and Grey Wolf Optimization GWO),Particle Swarm Optimization(PSO),Ganetic Algorithms(GA),the six models(GWO-SVM,LS-SVM,GA-SVM,PSO-SVM,GA-ANFIS,PSO-ANFIS)were built successfully.The input parameters are pressure and molar mass,and the output is WDT at every point for the prediction of WDT in binary,ternary and multicomponent system(272 date sets in all)in the range of 0.1-100MPa and 87-282 g/mol.And the proportion of train and test set is 7:3.And six models are evaluated and validated.Finally a evaluation and optimization developed at different pressure(0.1,20,40,60,80,100MPa).For the binary,ternary and multicomponent system and all data sets,the AARDs is 0.1878%,0.3794%,0.0345%,0.7128%for GWO-SVM respectly;the AARDs is 0.2235%,0.4288%,0.0918%,0.7128%for GA-ANFIS;the AARDs is 0.2347%,0.4097%,0.7610%,0.7518%for PSO-ANFIS;the AARDs is 0.3414%,0.6420%,0.3730%,0.8969%for GA-SVM;the AARDs is 0.4527,0.6799%,0.3832%,0.7218%for PSO-SVM;the AARDs is 0.5897%,1.0046%,0.0976%,0.8818%for LS-SVM.For the binary,ternary and multicomponent system and all data sets,there is 0,0,0,6 doubtful data points in GWO-SVM respectly;there is 4,0,1,5 doubtful data points in GA-ANFIS respectly;there is 2,0,1,9 doubtful data points in PSO-ANFIS respectly;there is 0,0,1,6 doubtful data points in GA-SVM respectly;there is 0,0,0,6 doubtful data points in PSO-SVM respectly;there is 0,0,0,8 doubtful data points in LS-SVM respectly;A evaluation and optimization developed at different pressure.GWO-SVM is suitable to predict WDT in the range of 0.1-40MPa and 60-100 MPa,and GA-ANFIS is suitable to predict WDT in the range of 40-60MPa.After the comparison of developed models,GWO-SVM is the best controller,which presented satisfactory performances no matter which system it is in and meets the industrial requirements.