Study On The Characteristics Of Oil Mixing In Multi-product Pipelines Based On Data Analysis

In the batch transportation of product oil pipelines,adjacent batches of oil products are mixed at the first station leading to a mixed oil section,which is developing to a large extent through the the pipeline and stations.This process is directly affected by many factors such as pipeline parameters,oil properties and operating parameters.Accurate prediction of the concentration distribution at the station can reduce the loss in the process of mixed oil cutting on the basis of ensuring the oil quality.In this paper,the data of the pipeline is extensively combined with numerical simulation and data statistics to study the development of oil mixing along the way and through the station.Firstly,this paper introduces and compares different contamination computation models,reached the conclusion that the one-dimensional contamination computation model has great potentialto calculate oil mixing in medium-long pipelines.Based on the one-dimensional model with updating the diffusion coefficient K,the improved model inclueds the difference in oil properties and couples factors such as the initial distribution,the change in throughput,the change in oil temperature along the pipe,the change in pipeline parameters,etc.To a certain extent,this method achieves accurate prediction of the concentration distribution of the product oil pipeline batch.Secondly,the computation model is accelerated with dynamic mesh technology,which improves the calculation efficiency in the real pipeline.The author collects,processes and analyzes the data from multiple domestic oil pipelines,and summarizes the main factors and laws influencing the oil mixing along the pipeline.This paper proposes two methods to quantify the concentration distribution.One is obtaining the abscissa of five concentration points through quadratic linear interpolation(The concentration quintile),the other is fitting the key parameters by the S-Weibull function within a given concentration range(Inflection point coordinates).Then,the paper dedicates to establish the prediction model of the concentration distribution with BP-Artificial Neural Network technology,based on the concentration distribution quintile and large dataset.The model reflects the relationship between the influencing factors of oil-mixing and the concentration distribution to a certain extent.The intelligent algorithm has smaller errors and stronger generalization,and shows certain engineering reference significance with smaller errors and stronger generalization,however,it relys on a large learning samples.Finally,this paper puts forward a method to study the development of passing-station oil mixture by observing of oil-mixing sound velocity curve,based on the ultrasonic flowmeters at the inlet and outlet of the Gang-Zao oil product pipeline.Through analysis,the corresponding relationships between increment of mix oil and factors such as the length of mix oil,the quantity of flow,the order of batch have been explored.These conclusions can not only provide guidance for the design and construction of the station,but also generate new ideas for future research and quantification of oil-mixing passing station.In addition,this paper analyzes and summarizes the unreasonable problems during installation and operation of the station in Gang-Zao pipeline,and raises specific suggestions for transformation.