Study On Prediction Model Of Scaling Trend In Oilfield Produced Water

With the development of oil and gas fields in our country entering the middle and late stage,the amount of produced water in the process of oil field production increases rapidly.Due to the high salinity of the produced water,the complex ion composition and the high scale amount,the calcium carbonate and other scales are easy to deposit in the gathering pipeline and underground pipe string,which causes the scaling and corrosion of the water injection pipeline to intensify,therefore,the prediction and judgment of the scaling type of produced water is of guiding significance for the establishment of targeted scale inhibition technology.Due to the complex causes of scaling in oil field water,it is gradually formed through the interaction of chemistry,crystallization kinetics and fluid mechanics,as a result,the current Ryznar Stability Index method,for example,has some deficiencies,which do not take into account the effect of fluid mechanics on scaling.In view of the various factors that affect the scaling of produced water,this paper is based on a large number of static experimental data in the early stage of laboratory and takes Scalechem Software as a template,by adopting the research method of combining the traditional mechanism prediction simulation with the neural network simulation in the big data era,a new prediction model of scaling trend is established,the main conclusions are as follows:(1)According to the analysis results of oilfield water quality and using Scale Chem software to analyze the scaling tendency;Based on the analysis of the scaling trend prediction curve,it is concluded that the scaling types in produced water of Shanbei oilfield are mainly calcium carbonate and calcium sulfate,and most of them are mixed scale Compared with the results of type and quantity analysis of scale formation in water,the agreement between the predicted results and the actual analysis results is more than 98%.(2)According to the prediction curve of scaling trend,the calcium sulfate scale in oilfield produced water is deposited on the carrier of the formed carbonate scale crystal,and when the temperature is raised from 25 ℃ to 100 ℃,the calcium sulfate scale in oilfield produced water is deposited on the carrier of the formed carbonate scale crystal,the scaling trend index of calcium carbonate increased from 2.23 to 12.14,and that of calcium sulfate increased from 0.83 to 2.49.When the pressure increased from 0.1 mpa to 6 mpa,the scaling trend index of calcium carbonate decreased from 5.56 to 4.88,the calcium sulfate trend index decreased from 1.35 to 1.22.(3)Using the Langlier saturation index method,Davis-Stiff saturation index method and other empirical formulas to predict the scaling trend of produced water has a high accuracy(the error range is less than 12%),however,the prediction accuracy of scale amount is relatively poor(the error is about 18%).(4)Through the analysis of error fluctuation mean,the best modeling method of neural network model is: using genetic algorithm to optimize better initial parameters in forward propagation,and then using back propagation to optimize parameters,the initial network parameters can be configured optimally,and the mean of error fluctuation is minimum 0.27.When the method is used to predict the scaling trend,the coincidence rates of single scaling trend and scale amount are 98.3% and 93.4%,respectively,and that of mixed scaling trend and scale amount are 96.8% and 90.8%,respectively.(5)Among the available data,the accuracy of synthetic prediction is up to 94.3%,and the more accurate and timely the data of water quality analysis is,the better the prediction result of neural network modeling is.