Study On The Wax Deposition And Prediction Method Of Waxy Crude Oil

Wax deposition is one of the key factors that affect the safety operation of the waxy crude oil transportation pipeline. Wax sediment in the pipeline leads to reduced effective inside diameter of pipeline, increased transport pressure, even potential blockage, which impose serious threat on the safety of crude oil transportation. The cost caused by wax deposition is huge, especially for offshore oil field, the difficulty and the cost for cleaning work are increased significantly once the deposition thickness get up to some extent. Pigging is effective to prevent deposition layer from becoming too thick, so a reasonable pigging period is very important. To determine the pigging period, it is necessary to grasp the deposition thickness distribution regularity along pipeline. Therefore, exploring the deposition regularity of waxy crude oil, and establishing a method for predicting wax deposition rate of waxy crude oil pipeline have actual meaning for the making scientific pigging plan and safety operation of pipeline.Waxy crude oil sample from North China Oil Field and model oil were taken as research objects, the regularities of how temperature difference between oil and pipe wall, the content of wax and asphaltene affect wax deposition were analyzed experimentally by using cold figure equipment and associated equipments. By measuring the wax content and carbon number distribution of sediments, the influence of the temperature difference on wax content and carbon number distribution of sediments was analyzed. By measuring the crystal lattice parameters of the crude oil and oil wax with and without adding the pour point depressant (PPD), and crystal lattice parameters of sediments which deposited from the model oil with and without adding the PPD, the mechanism of pour point depression was analyzed and the impact of PPD on wax deposition were emphasized, and the reason for sediment amount change caused by adding PPD was discussed. According to the characteristics of Graetz problem, based on dimensionless process, energy balance equation was transformed into Kummer equation(F-K method), then temperature distribution of the test section of the loop system was solved, the results were compared with that calculated by Svendsen method(S method). On consideration of the complicate non-liner relationship among the factors that affect wax deposition, at the instance of loop experiment data, BP neutral network was applied to predict wax deposition rate, and its reliability and accuracy were proved. Wax deposition rate prediction model was established based on Least Squares Support Vector Machines, the influences of different initial value of regularity parameters and kernel bandwidth were analyzed. Considering the main influence factors of wax deposition, and the wax deposition was ragared as a dynamic balance process of the deposition and removal, the dynamic model of wax deposition rate was established base on molecular diffusion law, the calculation errors of different models were compared. Considering the impact of variation of deposition layer thickness on the physical properties of crude oil and the heat transfer characteristics of pipeline, software program for the prediction of deposition thickness along actual pipeline was established, and the differences of the results obtained from different deposition rate models were discussed. According to analysis of heat and mass transfer in radial direction, considering kinetics of crystallization phenomenon in the mass equation, deposition thickness distribution of actual pipeline with different operation time were calculated. On the basis of loop experiment data and the analysis of wax deposition process, the models that show the relationship of wax deposition thickness and time were developed, the consistency between calculated value and experimental value was analyzed, and the model is verified by the calculation data of actual pipeline.The results show that:with the temperature difference between oil and pipe wall increasing, the quality and wax content of the sediment increase, the proportion of the higher carbon number wax increase. When the temperature difference remains the same, sediment quality increase with the decrease of temperature range of oil and pipe wall, the proportion of low carbon number wax in the sediment is higher. The sediment quality increases while the wax content of the model oil increase, but the affect of asphaltene content on wax deposition has two sides. With the increase of asphaltene content, sediment quality increase at first, then decrease, and increase again at last. Before and after adding the PPD in the crude oil and oil wax, the structure of the wax crystals of both crude oil and oil wax were changed. There is no large molecular such as asphaltene and resins in the oil wax, so co-crystallization is the main mechanism of pour point depression. Adding the PPD into model oil can decrease the sediment quality, the reason can be explained by that the PPD increase the interplanar spacing of wax crystal in the sediment according to the result of XRD test. The higher the increasing amplitude of interplanar spacing, the larger sediment quality decrement rate was observed. The regularities of temperature distribution calculated by S method and F-K method remain the same. However, temperature results calculated by S method is generally higher than that obtained by F-K method. Maximum difference of the calculation results by these two methods is small when the temperature difference between oil and pipe wall and the axial distance are small. Maximum difference of the calculation results by these two methods method is large when the axial distance is large and flow rate is low. Temperature field distribution formula obtained by F-K method is convenient for application because it avoid the inconvenient brought by solving numerical integration and Bessel Function. Wax deposition rates obtained respectively by LS-SVM method and BP neutral network match well with experiment result, and the accuracy are higher than multiple linear regression method. Compared to BP neutral network and multiple linear regression method, LS-SVM method has its advantages because explicit formulation for wax deposition rate can be obtained and the non-liner relationship among the factors which affecting wax deposition can be depicted. As we can see from the calculation results of different wax deposition kinetic models, the logarithmic least square error is relatively low when the modified model of this article was used, and the model accuracy is within an acceptable range. The variation of wax deposition rate along the actual pipeline is complicate, the deposition rate change depend on which factors play major roll. Wax deposition thickness distribution along the pipeline predicted by different model varied notably, the comparability of models is poor. The calculation results obtained from modified Wang Jifeng model is two to three times of that calculated by Huang Qiyu model while their trends are the same. When the wax deposition thickness was the same, the increasing of wax mass fraction in the deposition layer leads to decreasing surface temperature of deposition layer, and wax deposition rate increased gradually. In the condition of the wax content was the same, with the increase of deposition thickness, the surface temperature of deposition layer increased and the tube wall temperature decreased gradually. Under different operation time, the deposition thickness along the pipeline increases with deposition time increased, the increase trend is not linear growth, but a trend of growing fast at first then get slow. As can be seen from the model of wax deposit thickness changes with time, the coincidence degree of logarithmic model and the experimental results was the best, while the coincidence degree of exponential model and dynamic balance model were much the same. In the initial stage of deposition, the coincidence degree of exponential model was higher than dynamic balance model, but when the deposition time is closer to the experimental set time, the coincidence degree of dynamic balance model was higher than exponential model.