Improvement of different methods of predicting the vapor pressure of different materials

Thesis of Master's degree in chemical engineering

Vapor pressure is an important thermodynamic property in the design of process equipment and operation of chemical engineering unit. Therefore, experimental vapor pressure data that cover the entire range of vapor pressure are very valuable, but due to the lack of accurate measurements for the vapor pressure of some materials near the triple and critical point, equations that are able to predict the vapor pressure in such conditions are very important. in chemical engineering calculations, the calculation method is based on neural networks. In this research, a model was presented for predicting the vapor pressure of different substances using artificial neural networks. In this work, four groups of substances were used, including aromatic hydrocarbons, alkanes and alkenes, alcohols, and alkylcyclohexanes. The most suitable type of artificial neural network for predicting the vapor pressure of these substances is a three-layer feedforward network with backpropagation algorithm, in which the function The sigmoid tangent transfer is used in the hidden layer and the linear transfer function is used in the output layer. The input parameters of the network are: temperature, critical temperature, critical pressure and eccentricity coefficient. The data required for training and testing the network were collected from valid laboratory values. The error rate of the neural network method was compared with the error values ??obtained from different vapor pressure estimation methods. The simulation results show that the neural network method has been able to provide an accurate prediction of the vapor pressure of materials and is more accurate than other methods. Vapor pressure is an important thermodynamic property in many different chemical engineering processes such as chemical equilibrium, distillation, evaporation, and the like. Determining this characteristic can lead to the calculation of other important characteristics such as enthalpy of evaporation. In the oil and gas industry, the determination of steam pressure is of particular importance. In this area, steam pressure is dealt with in two main cases; One of these cases is the steam pressure in the tanks. One of the important methods for dividing the types of fluid storage tanks in the oil and gas industry is their division based on the vapor pressure of the desired fluid. For each range of material vapor pressure (low pressure, medium pressure, high pressure) special types of tanks are used. For example, for fluids with low vapor pressure, fixed roof tanks are used. For fluids with medium vapor pressure, floating roof tanks are used.

The vapor pressure of liquid products is another thing that is measured in the industry. One of the indirect ways of measuring the rate of evaporation of volatile petroleum solvents is their vapor pressure. Refinery products must also have special specifications and meet standards in order to be able to be present and sell well in global markets. Accordingly, the amount of steam pressure is one of the most important features, which, in addition to quality and price, is also very important from the point of view of safety during transportation and storage, and is always tested and controlled. Therefore, the importance of accurately determining the vapor pressure of liquids in the field of petroleum industry is not hidden. When the moment of equilibrium is approached, the number of vaporized molecules equals the number of condensed molecules. In fact, the rate of evaporation is equal to the rate of compression, and the specified pressure in this case is the vapor pressure of that liquid at that temperature.] 1 [

1-3-Factors affecting vapor pressure

In general, the vapor pressure of a liquid depends on the following two factors:

The nature and nature of the liquid

The temperature of the liquid

1-3-1- The nature of the liquid

Liquids that have intermolecular forces They are weak, more volatile and have a higher vapor pressure. For example, the vapor pressure of ethyl alcohol is higher than the vapor pressure of water.

1-3-2-liquid temperature

The vapor pressure increases with increasing temperature. This problem is due to the fact that with the increase in temperature, the rate of evaporation increases.

1-4-Problem statement

The vapor pressure of different materials is one of the properties required to perform chemical engineering calculations, such as equilibrium calculations and chemical engineering unit operations, which many researches are being conducted. The most common methods for determining vapor pressure include laboratory measurements, equations of state, empirical relationships, and relationships based on the law of corresponding states. The necessity of calculating vapor pressure using mathematical relationships increases when laboratory data is not available. Since the conditions of many chemical processes are such that it is not practically possible to determine the vapor pressure of materials experimentally and its measurement at certain pressures and temperatures is difficult and the obtained values ??are not very reliable, therefore providing models and methods for predicting vapor pressure will have a major contribution in determining this thermodynamic property. Therefore, numerous methods have been presented to predict this characteristic, and newer methods are introduced or old methods are corrected every year. ] 5-2 [

1-5-Justification of the need to conduct research

Since many experimental relations or relations of corresponding states have limitations in determining steam pressure and cannot be used to determine steam pressure in all required temperature ranges and do not have acceptable accuracy, therefore, it is recommended to use new methods that are far from these limitations. One of the modeling methods that has attracted the attention of many researchers in various sciences in recent years is artificial neural network modeling. Artificial neural networks are used as a subset of artificial intelligence methods, with a structure and function similar to the human brain in a wide range to solve many problems including evaluation, optimization, prediction, diagnosis and control. [6] One of the advantages of using artificial neural networks compared to old models is that it does not require determining a specific function to express the relationship between input and output data.  The relationship between the input and output data is obtained through the training process. ] 7 [

1-6-Research Objectives

Study different methods for calculating vapor pressure

Check the level of accuracy and how to apply existing methods to calculate the vapor pressure of different materials

Improve vapor pressure prediction methods by applying the new neural network method and compare it with old methods

Evaluate the results of modeling by neural network with experimental data and check the accuracy It

1-7-Stages of conducting research

In the first part of this thesis, various methods for calculating vapor pressure were comprehensively and completely examined and the parameters of each relationship, the accuracy of vapor pressure estimation and the optimal temperature range of each method were mentioned.

In the second part, the neural network method was introduced and examined as an accurate calculation method for predicting the vapor pressure of materials. In this stage of the thesis, the artificial neural network model was examined as a research method and a comprehensive explanation of this method and its uses and characteristics and how to apply this method to predict the vapor pressure of materials was presented. In the third part of the thesis, the new neural network method was applied to more accurately predict the vapor pressure of several groups of materials, and the optimal neural network was designed and the results of modeling by the artificial neural network method for different groups of materials were presented. In this section, using reliable sources and reliable experimental data, the level of accuracy and how to apply different conventional models to estimate vapor pressure was investigated. In the following, it was tried to investigate the effectiveness of different methods of predicting the vapor pressure of materials in different conditions and for different materials. In conclusion, the results obtained from the current research and suggestions for future research are presented. 1-8-Research Structure The most useful and famous vapor pressure calculation relationships were mentioned in the second chapter of the thesis. In this chapter, eight relationships were introduced and examined. Next, the background of using the neural network method to determine the thermodynamic properties of materials, including the vapor pressure of various compounds, was explained. One of the most important features of artificial neural networks is that they do not depend on the initial hypotheses about the input data; In the sense that the input data can have any desired statistical distribution [8] This important feature of neural networks is their special advantage in front of statistical methods and gives them the ability to equally use different types of input data with any desired distribution.