Non-linear identification of de-butanizer separator tower system located in a gas refinery in South Pars using experimental data.

Number of pages: 76 File Format: word File Code: 32246
Year: Not Specified University Degree: Master's degree Category: Electrical Engineering
  • Part of the Content
  • Contents & Resources
  • Summary of Non-linear identification of de-butanizer separator tower system located in a gas refinery in South Pars using experimental data.

    Master's Degree in Electrical Engineering with Control Orientation

    Electrical Engineering with Control Orientation

    Abstract

    Nowadays, the science of chemistry and oil and gas extraction and refining is one of the most important points of economic and production reliance of countries. One of the most common processes in refineries is the separation of chemicals. This is done using separator towers. The debutanizer tower is one of the towers that is used in the gas refinery to separate butane from sour gas in the debutanization unit.

    If a proper and good control system governs this system, we can produce high quality products with minimal energy and minimal cost. It will be possible.

    Various methods have been presented to identify process dynamic systems. In this treatise, we will examine the existing methods for identifying the system that governs this process and we will examine two groups of identification methods: 1- linear 2- non-linear and we will identify and simulate the system with the studied methods.

    At the end, after examining the fuzzy-neural method from the category of non-linear system identification methods, we will present a method to improve the identification performance with this method and compare the simulation results. performed by different methods in system identification, using experimental data obtained by sampling a gas refinery in operation, we will discuss how the identified systems are in the process of a debutanizer separation tower. Linear

    Keywords: Distillation column- System Identification-Debutanizer-Nonli

    Introduction to debutanizer separator tower, input and output description, data collection

    1-1-Introduction

    This tower consists of a feed input line from the middle of the tower and two output lines of the produced materials at the bottom and top of the tower. The governing system of such a tower has been described in different sources as a multi-input-multiple-output system, which we can control several inputs such as tower inlet feed flow rate, tower pressure, etc. and to have several outputs such as the lower and upper temperature of the tower that have a direct impact on the quality of the product produced from the process. If a proper and good control system governs this system, we can produce high quality products with minimum energy consumption and minimum cost. The design of this control system will be possible by reliable identification of the system governing the process by using the available information from the sampled input and output of the system.

    Identification and mathematical modeling of the system means obtaining a relationship between the input and the output of the system so that if the same input signal is applied to the system and the simulated model, the outputs of the original and modeled system are almost the same.

    Identification into three categories a) white box or analytical modeling b) black box or experimental modeling c) gray box or mixed division model 2[.

    a) White box or analytical modeling

    Analytical modeling means obtaining the mathematical relationship between system variables based on physical laws. What is problematic in the analytical modeling method is determining the parameters. Due to the diversity of the parameters of different systems, there is no specific method to determine the parameters in analytical modeling]2[.

    b) Black box or experimental modeling

    Experimental modeling is obtaining a mathematical model for the system using the results of an experiment. In experimental modeling, we do not know what is inside the system and what components it consists of. In this method, by sampling input and output signals by means of a sampler, numerical values ??are provided to the computer in order to obtain a model of the relationship between inputs and outputs, so that if another input is applied to the system, we can estimate the output by the obtained model, and the output estimated by the model is close to the output response of the system itself [2].

    c) Gray box or hybrid model

    In this method, the structure of the model is obtained using physical laws, but instead of obtaining individual parameters through separate tests, the input and output signals are sampled and the output and input vectors are calculated, which are the same sampled numbers, and by using these two vectors, we try to estimate the values ??of the parameters of the structure [2]. Botanizer has (it will be discussed in detail in the following) we have considered and selected the black box identification method to continue the work.

    In order to identify such systems by the black box method, various methods have been presented, and some of these methods are introduced in this treatise. Botanizer as a type of separating towers, looking at the past works in this field, we will do a detailed review of each of the identification methods mentioned, and we will identify the studied system with these methods. It is used separately. Nowadays, these towers are used in refineries and petrochemicals to separate chemicals and produce products]3,4[.

    In ancient times, separation was used to concentrate alcohol in drinks[5]. Figure (1-1) Separation was first used in Mesopotamia, but the Greeks were the ones who transferred this science to Europe. More or less, the effects of using this science in African and American tribes have also been observed in the discoveries made. 5

    Abstract

    Todays, chemistry, oil and gas extraction and refining, is one of the most important economic dependence and production of the countries. One of the most common procedures that are used in refineries is the separation of materials; This is done using distillation columns. Debutanizer column in butane gas separation unit on the refinery sour gas is used to separate butane.

    If the control system is adequate on this process system, we can produce high quality products at the lowest cost and energy spending; that this matter will be easy if we design a reliable control system detected the process system utilizes the information of input and output sample from the system.

    Various methods are proposed for identification of dynamical process systems. In this paper we review existing methods for identifying the process system. Two groups of methods: 1-linear methods 2-nonlinear method, we will study, examine the systems and will identify and simulate system with these methods.

  • Contents & References of Non-linear identification of de-butanizer separator tower system located in a gas refinery in South Pars using experimental data.

    List:

    Chapter 1: Getting to know the debutanizer separator tower, description of input and output, data collection

    1

    1-1- Introduction

    1

    1-2- Separation process and looking back

    1-3- Review of past works

    3

    4

    1-4- Description of a separator tower

    6

    1-5- Debutanizer tower in South Pars Gas Refinery

    8

     

     

    Chapter Two: Identification by linear and non-linear methods

    2-1- Introduction

    11

    11

    2-2- Identification by linear methods

    11

    2-2-1- Linear identification by parametric method

    11

    2-2-1-1- Identification by ARX method

    12

    2-2-1-2- Identification by OE method

    14

    2-2-1-3- Identification by BJ method

    15

    2-2-2- Linear identification based on subspace analysis

    16

    2-3- Identification by non-linear methods

    20

    2-3-1- Identification by non-linear ARX (NLARX) method

    20

    2-3-2- Identification by method Hammerstein-Wiener (NLHW)

    21

    2-3-3- Recognition by Neural Networks

    23

    (MLP) 2-3-3-1- Multilayer Perceptron

    23

    2-3-3-2- Learning by Lunberg-Marquardt

    25

    2-3-4- identification by fuzzy-neural method

    26

    2-3-4-1- classification or clustering

    27

    2-3-4-2- subtractive classification

    28

    2-3-4-3- membership function

    29

    2-4- aggregation Classification

    Title

    30

    Page

    Chapter 3: Implementation of linear and non-linear identification methods on dibutanizer system

    31

    3-1- Introduction

    31

    3-2- Collecting data to identify the dynamic system of dibutanizer

    3-2-1- Variables Physical and precision instruments

    32

    32

    3-2-2- Sampling and charting the variables of de-butanizer tower

    34

    3-3- Implementation of ARX identification method

    37

    3-4- Implementation of OE identification method

    39

    3-5- Implementation of identification method BJ

    41

    3-6- Implementation of N4SID identification method

    44

    3-7- Implementation of NLARX identification method

    47

    3-8- Implementation of NLHW identification method

    48

    3-9- Implementation of neural network identification method

    50

    3-10-Implementation of recognition by neuro-fuzzy method

    3-11- Summary

     

    53

    56

    Chapter four: Expanded neuro-fuzzy method

    4-1-Introduction

    57

    57

    4-2- Type fuzzy systems- 2

    57

    4-3- Categorization by reduction method, fuzzy type-2

    58

    4-4- Determining the neighborhood radius of mutual influence of membership functions

    61

    4-5- Implementation of identification by extended fuzzy-neural method

    4-6- Summary

    65

    68

    Chapter five: discussion, conclusions and suggestions

    69

    Resources

    71

     

     

    Source:

    [1] Chang, R. General chemistry. McGraw-Hill, 2006

    [2] M. Karrari, System identification. Tehran: Amirkabir industrial university, 1388

    [3] Abdullah, Z., N. Aziz and Z. Ahmad. Nonlinear modeling application in distillation column. Chemical Product and Process Modeling 2(3-12), 2007.

    [4] Skogestad, S. Dynamics and control of distillation columns - A critical survey. In IFAC-SymposiumDYCORD, Maryland (April-1992).

    [5] Forbes, R. Short History of the Art of Distillation from the Beginnings Up to the Death of Cellier Blumenthal (Second Edition ed.). Brill Academic Publishers, 1970. [6] Kister, H. Distillation Design. McGraw-Hill, Inc, 1992.

    [7] R. GANI, C. A. Ruiz and I. T. CAMERONS: A GENERALIZED MODEL FOR DISTILLATION COLUMNS-MODEL DESCRIPTION AND APPLICATIONS, Computers & Chemical Engineering, Vol. 10, No. 3, pp. 181-198, 1986

     

    [8] S. Ismail KIRBASHLAR, Ahmet AYDIN, Umur DRAMUR: Identi_cation of a Binary Distillation Column Using Pulse Testing, Tr. J. of Engineering and Environmental Science 22 (1998), 255-of Engineering and Environmental Science 22 (1998), 255-267.

    [9] Shiraz Amjad, Closed Loop Identification For Model Predictive Control: A Case Study, MS Thesis, KING FAHD UNIVERSITY OF PETROLEUM & MINERALS October 2003

    [10] K.Ramesh, N.Aziz, and S.R. Abd Shukor: Development of NARX model for distillation column ans studies on effect of regressors, Journal of Applied Sciences 8(7):1214-1220, 2008 ISSN 1812-5654 [11] H. Guidi: Open and Closed loop model identification and validation, MS Thesis, University of Pretoria, December 2008 [12] J. R. C Barroso: Modeling and Intelligent Control of a Distillation Column. Applications (0975 – 8887) Volume 29– No.7, September 2011

     

    [15] S.Meenakshi, G.Saravanakumar: Subspace identification and controller design for Distillation Column, International Conference on Computing and Control Engineering (ICCCE 2012), 12 & 13 April, 2012

    [16] PetroVietnam Gas Company, Condensate processing plant project—process description, Tech. Rep. 82036-02BM-01, PetroVietnam, Washington, DC, USA, 1999.

    [17] Skogestad, S. The dos and don'ts of distillation column control. Chemical Engineering Research and Design (Trans IChemE, Part A) 85, 2007.

    [18] Lundstr ¨om, P. and S. Skogestad. Opportunities and difficulties with 5x5 distillation columns. Journal of Process Control 5(4). IFAC Symposium: Advanced Control of Chemical Processes, 1995.

    [19] Cammarata, L., Fichera, A., and Pagano, A., Neural prediction of combustion instability, Applied Energy, 72:513-528, 2002.

    [20] Kister, H. Distillation Operation. McGraw-Hill, Inc, 1990.

    [21] POGC (Pars Oil and Gas Company), Oil and Gas Processing Plant Design and Operation Training Course Gas Sweetening Processes, < www.pogc.ir>, 2002

     

    [22] POGC (Pars Oil and Gas Company), Document number: SPP-910-107-C-102-03, DB-910-107-P332-5000-Rev0,2009

     

    [23] Li, T. and Georgakis, C. Dynamic input signal design for the identification on constrained systems. Journal of Process Control, 18, 332-346.2008

    [34] LJUNG, L., System Identification: Theory for the User, 2nd edn, Prentice Hall, Upper Saddle River, N.J. ISBN 0-13-656695-2, 1999.

     

    [25] T.S. Soderstorm, Petre G. Stonica, System Identification, Prentice hall, 1989.

     

    [26] Y. zhu, Multivariable system Identification for process control, Elsevier Science, 2001.

     

    [27] J. Kon, Y. Yamashita, T. Tanaka, A. Tashiro and M. Daiguji, Practical Application of Model Identification based on ARX Models with Transfer Functions, Control Engineering Practice, on-line available, 2012

     

    [28] So¨derstrom, T., Fan, H., Carlsson, B., and Bigi, S.: Least squares parameter estimation of continuous-time ARX models from discrete-time data, IEEE Trans. Autom. Control, 42, (5), pp. 659-673, 1997. [29] Michel Verhaegen, "Filtering and System Identification, A Least Squares Approach", Vincent Verdult, 2007. [30] U. Forsseland L. Ljung, Identification of unstable systems using output error and box-jenkins model structures, IEEE Transactions on Automatic Control, vol. 45, pp. 137-141, 2000. [31] Mats Viberg. Subspace-based methods for the identification of linear time-invariant systems. Automatica, 31(12):1835-1851, 1995. [32] M. Verhaegen. Identification of the deterministic part of MIMO state space models given in innovations form from input-output data. Automatica, 30:61-74, 1994.

Non-linear identification of de-butanizer separator tower system located in a gas refinery in South Pars using experimental data.