DEVELOPMENT OF AN ALGORITHM FOR DETERMINING THE SATURATED VAPOR PRESSURE OF STABLE CONDENSATE IN REAL TIME

Vollständiger Abstract

Worum geht es in dieser Arbeit?

At the gas condensate stabilization unit, product quality control is carried out by regular laboratory analysis of the saturated vapor pressure (VP) of stable condensate. In the current practice, condensate samples are taken twice a day for VP subsequent monitoring. This approach requires significant time and resources personnel participation, transportation of samples and waiting for analysis results. In addition, the obtained VP values are received with a delay, which limits the efficiency of making technological decisions. The goal of the research described in this article was to create a software algorithm capable for significant time and resources: personnel participation, transportation of samples and waiting for analysis results. In addition, the obtained VP values are received with a delay, which limits the efficiency of making technological decisions. The goal of the research described in this article was to create a software algorithm capable for predicting the value of the VP of stable condensate from the current technological data of the plant in real time. The required forecast accuracy is comparable to the accuracy of laboratory measurements, which will allow the predicted value to be used for operational product quality control and process control. Machine learning methods are used to implement this task. Development was carried out in the Jupyter Notebook environment using the Python language and a number of specialized libraries: Pandas for working with data, NumPy for operations with arrays, scikit-learn for creating and training various types of forecasting models. Using Matplotlib and Seaborn libraries, data and simulation results were visualized. The paper discusses the stages of data preparation, various approaches to forecasting, selection of the optimal model for subsequent integration of the developed algorithm into the circuit of an automated system for controlling technological parameters.