- Email: [email protected]
Vinyl Acetate Monomer (VAM) Plant Model: A New Benchmark Problem for Control and Operation Study
11th IFAC Symposium on Dynamics and Control of 11th IFACSystems, Symposium on Dynamics and Control of Process including Biosystems 11th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems 11th IFAC Symposium on Dynamics and Controlonline of Available at www.sciencedirect.com June 6-8,Systems, 2016. NTNU, Trondheim, Norway Process including Biosystems June 6-8,Systems, 2016. NTNU, Trondheim, Norway Process including Biosystems June June 6-8, 6-8, 2016. 2016. NTNU, NTNU, Trondheim, Trondheim, Norway Norway
ScienceDirect
IFAC-PapersOnLine 49-7 (2016) 533–538
Vinyl Vinyl Acetate Acetate Monomer Monomer (VAM) (VAM) Plant Plant Model: Model: Vinyl Monomer (VAM) Model: Vinyl Acetate Acetate Monomer (VAM) Plant Plant Model: Study A New Benchmark Problem for Control and Operation A New Benchmark Problem for Control and Operation Study A A New New Benchmark Benchmark Problem Problem for for Control Control and and Operation Operation Study Study
Yuta Machida* Shigeki Ootakara** Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano† † † Yuta Machida* Shigeki Ootakara**§ Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano # # † ‡ § † Yuta Machida* Shigeki Ootakara** Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano Yasuhiro Miyake Naoto Anzai Masayoshi Sawai Takashi Katsuno Toshiaki Omata # # ‡ § § # Yuta Machida* Shigeki Ootakara** Seki*** Yoshihiro Hashimoto**** Manabu Kano ‡ Naoto § Takashi Yasuhiro Miyake Anzai§ Hiroya Masayoshi Sawai Katsuno## Toshiaki Omata # ‡ ‡ Naoto Anzai§ § Masayoshi Sawai§ § Takashi Katsuno# Toshiaki Omata# Yasuhiro Miyake Yasuhiro Miyake Naoto Anzai Masayoshi Sawai Takashi Katsuno Toshiaki Omata *Omega Simulation Co.,Ltd., Tokyo 169-0051, Japan (e-mail: [email protected]) *Omega Simulation Co.,Ltd., Tokyo 169-0051, Japan (e-mail: [email protected]) ** Mitsui Tokyo Chemicals Inc., Osaka 592-8501, Japan *Omega Co.,Ltd., 169-0051, Japan (e-mail: *Omega Simulation Simulation ** Co.,Ltd., 169-0051, Japan592-8501, (e-mail: [email protected]) [email protected]) Mitsui Tokyo Chemicals Inc., Osaka Japan *** Chemical Resources Laboratory, Tokyo Institute of Technology, Yokohama 226-8503, Japan ** Mitsui Chemicals Inc., Osaka 592-8501, **Laboratory, Mitsui Chemicals Osaka 592-8501, Japan Japan *** Chemical Resources Tokyo Inc., Institute Yokohama 226-8503, Japan of Technology, Technology, ****Dept. of Computer Science, Nagoya Institute of Nagoya 466-8555, Japan *** Chemical Resources Laboratory, Tokyo Yokohama 226-8503, Japan *******Dept. Chemical† Resources Tokyo Institute of Technology, Nagoya Yokohama 226-8503, Japan of ComputerLaboratory, Science, Nagoya 466-8555, Japan Dept. of Systems Science, Kyoto University, Kyoto 606-8501, Japan ****Dept. of Computer Science, Nagoya Institute of Technology, Nagoya 466-8555, Japan ****Dept. † ofDept. Computer Science, Nagoya Institute of Technology, Nagoya 466-8555, Japan † of Systems Science, Kyoto University, Kyoto 606-8501, Japan † †Dept. of‡ Systems Science, Kyoto University, Kyoto Ube Industries, Yamaguchi 755-8633, Japan Japan Dept. of‡ Science,Ltd., Kyoto University, Kyoto 606-8501, 606-8501, ‡Systems Ube Industries, Ltd., Yamaguchi 755-8633, Japan Japan ‡ § ‡Ube Industries, Ltd., Yamaguchi 755-8633, Corp., Tokyo 100-8246, Japan Japan § Ube Industries, Ltd., Yamaguchi 755-8633, §Zeon Zeon Corp., Tokyo 100-8246, Japan Japan # § §ZeonElectric Corp., 100-8246, Tokyo 169-0051 Japan ## Yokogawa Zeon Corp., Corp., Tokyo Tokyo 100-8246, Japan Japan ## Yokogawa Electric Corp., Tokyo 169-0051 Japan Yokogawa Electric Corp., Tokyo 169-0051 Japan Yokogawa Tokyoacetate 169-0051 Japan (VAM) production was Abstract: A rigorous dynamic plant Electric model Corp., of a vinyl monomer Abstract: A rigorous dynamic plant model of a vinyl acetate monomer (VAM) production was developed. This plant model enables the users to experience realistic plant operation, since it reflectswas the Abstract: A rigorous dynamic plant model of aa vinyl monomer (VAM) production Abstract: A rigorous dynamic plant model ofexperience vinyl acetate acetate monomer (VAM) production was developed. This plant model enables the users toon realistic plant operation, since it reflects the real plant characteristics and practical problems the basis of experienced practitioners’ opinions. More developed. This enables users realistic plant since it the developed. This plant plant model model enables the the users to toonexperience experience realistic plant operation, operation, since it reflects reflects the real plant characteristics and practical problems the basis of experienced practitioners’ opinions. More importantly, the plant model provides problems a new benchmark problem; the users practitioners’ can investigate start-up/shutreal plant characteristics and practical on the basis of experienced opinions. More real plant characteristics and practical problems on the basis of experienced practitioners’ opinions. More importantly, the plant model provides a new benchmark problem; the users can investigate start-up/shutdown operation, plant-wide control, detection and diagnosis, Multiple scenarios importantly, the model provides aa new benchmark problem; the can investigate start-up/shutimportantly, the plant plant modelprocess provides new fault benchmark problem; the users usersand canothers. investigate start-up/shutdown operation, plant-wide process control, fault detection and diagnosis, and others. Multiple scenarios prepared in the developed model cannot be simulated in conventional benchmark problems. The plant down operation, plant-wide process control, fault detection and diagnosis, and others. Multiple scenarios down operation, plant-widemodel processcannot control, detection and diagnosis, and others. Multiple The scenarios prepared in the developed be fault simulated in conventional benchmark problems. plant model can be used also for chemical engineering education. This advantageous plant model is released prepared in the developed model cannot be simulated in conventional benchmark problems. The plant prepared developed cannot be simulated in conventional benchmark problems. plant model caninbetheused also formodel chemical engineering education. This advantageous plant model isThe released from Omega Simulation Co., Ltd. with a free limited license of Visual Modeler, which is a commercial model can be used also for chemical engineering education. This advantageous plant model is released modelOmega can beSimulation used also for education. advantageous is released from Co.,chemical Ltd. withengineering a free limited license This of Visual Modeler, plant whichmodel is a commercial dynamic simulator and can linked MATLAB®. Thisof article to introduce VAM plant from Simulation Co., with aa free limited Visual Modeler, which commercial from Omega Omega Simulation Co.,be Ltd. withwith free limited license license Visualaims Modeler, which is isthe commercial dynamic simulator and can beLtd. linked with MATLAB®. Thisofarticle aims to introduce theaa VAM plant model, the steady-state balance, various disturbances and malfunctions, and operation scenarios. dynamic simulator and can be linked with MATLAB®. This article aims to introduce the VAM dynamicthesimulator and balance, can be linked MATLAB®. This article aims to introduce the VAM plant plant model, steady-state variouswith disturbances and malfunctions, and operation scenarios. model, the steady-state balance, various disturbances and and operation scenarios. model, steady-state balance, various and malfunctions, malfunctions, and operation scenarios. © 2016,the IFAC (International Federation ofdisturbances Automatic Hosting by Elsevier Ltd. All rights reserved. Keywords: Process simulators, process models, Control) dynamic models, benchmark examples, chemical Keywords: Process simulators, process models, dynamic models, benchmark examples, chemical industry, plant-wide Keywords: Process control. simulators, process process models, models, dynamic dynamic models, models, benchmark benchmark examples, examples, chemical chemical Keywords: Process simulators, industry, plant-wide control. industry, plant-wide control. industry, plant-wide control. simulator attempts to keep the fidelity of the rigorous simulator attempts to keep the fidelity of the rigorous 1. INTRODUCTION nonlinear as high as possible. The dynamic 1. INTRODUCTION simulator attempts to the of rigorous simulator model attempts to keep keep the fidelity fidelity of the the simulator rigorous nonlinear model as high as between possible. Themodel dynamic simulator 1. INTRODUCTION 1. INTRODUCTION makes a good compromise the accuracy and model as high as possible. The dynamic simulator A test problem which checks the industrial relevance of new nonlinear nonlinear model as high as possible. The dynamic simulator a good compromise between the model accuracy and A test problem which checks the industrial relevance of new makes solvability. We improved the VAM plant model developed aa good between model accuracy and ideas technical is of great importance in makes A problem which checks relevance of good compromise between the model accuracy and solvability. Wecompromise improved the VAM the plant model developed A test test and problem whichdevelopments checks the the industrial industrial relevance of new new ideas and technical developments is of great importance in makes by Yumoto We et al. (2010)the with reference to practitioners’ solvability. improved VAM plant model developed the study of process control and process data analysis. A ideas and technical developments is of great importance in solvability. We improved the VAM plant model developed Yumoto et al. (2010) with reference to practitioners’ ideasstudy and technical of great importance the of processdevelopments control and is process data analysis. in A by opinions to et make operation scenarios. The new by al. (2010) reference to great number of researchers haveprocess utilizeddata the analysis. Tennessee the of control A by Yumoto Yumoto al. realistic (2010) with with reference to practitioners’ practitioners’ to et make realistic operation scenarios. The new the study study of process process control and and A opinions great number of researchers haveprocess utilizeddata the analysis. Tennessee VAM plant model has the following features, which the opinions to make realistic operation scenarios. The Eastman challenge problem in various fields including plantgreat number of researchers have utilized the Tennessee opinions to make realistic operation scenarios. The new new VAM plant model has the following features, which the great number of researchers have utilized the Tennessee Eastman challenge problem in various fields including plant- VAM Tennessee Eastman process model cannot cope with: plant model has the following features, which the wide control, production efficiency optimization, soft-sensor Eastman challenge problem in various fields including plantVAM plant model has the following features, which the Tennessee Eastman process model cannot cope with: Eastman challenge problem in various fields including plantwide control, production efficiency optimization, soft-sensor Tennessee Eastman process model cannot cope with: design, fault detection and diagnosis, etc., since it was wide control, control, production production efficiency efficiency optimization, optimization, soft-sensor soft-sensor -Tennessee Eastman process model cannot cope with: start-up It can simulate transient operations including wide design, fault detection and diagnosis, since it1993). was - It can simulate transient operations including start-up introduced more than 20 years ago (Downsetc., and Vogel, design, fault detection and since was andcan shut-down. design, fault detection and diagnosis, diagnosis, etc., since it it1993). was -- It simulate introduced more than 20 years ago (Downsetc., and Vogel, It simulate transient transient operations operations including including start-up start-up andcan shut-down. Even now, the Tennessee Eastman problem is the most introduced more than 20 years ago (Downs and Vogel, 1993). It can generate noises and disturbances with such introduced 20 yearsEastman ago (Downs and Vogel, shut-down. Even now,more the than Tennessee problem is the1993). most -- and and shut-down. It can generate noises and disturbances with such popular problem. Later, Luyben and Tyreus (1998) Even the Eastman problem is the realistic sources asnoises measurement noise of sensors, nonEven now, now, the Tennessee Tennessee Eastman problem is developed the most most -- It can generate and disturbances with such popular problem. Later, Luyben and Tyreus (1998) developed It can generate noises and disturbances with such realistic sources as measurement noise of sensors, nonapopular model problem. of a vinylLater, acetate monomer (VAM)(1998) plant, which is a popular Luyben and developed ideality in actuators such as valve dead band, daily Luyben and Tyreus Tyreus developed realistic sources as measurement noise of sensors, nona model problem. of a vinylLater, acetate monomer (VAM)(1998) plant, which is a realistic sources as measurement noise of sensors, nonideality in actuators such as valve dead band, daily containing unit operations alarger of acetate monomer (VAM) which temperature fluctuations, etc. a model modelsystem of aa vinyl vinyl acetate standard monomerchemical (VAM) plant, plant, which is is aa ideality in actuators such as valve dead band, daily larger system containing standard chemical unit operations ideality in actuators such as valve dead band, daily temperature fluctuations, etc. for chemical components. The process design - temperature larger system containing standard unit It can simulate various failure which are set on largerreal system containing standard chemical chemical unit operations operations fluctuations, etc. for real chemical components. The process design - temperature fluctuations, etc. modes, It can simulate various failure modes, which are set on background is well-established in the The original paper. Several for real chemical components. process design the basis of the advice of experienced engineers for real chemical components. The process design It can can simulate simulate various various failure failure modes, modes, which which are set setand on background is well-established in the original paper. Several -- It are on the basis of the advice of experienced engineers and studies on control system design fororiginal this plant have been background is well-established in the paper. Several plant operating personnel. background is well-established in the original paper. Several the basis of the advice of experienced engineers and studies on control system design for this plant have been the basis of the advice of experienced engineers and operating personnel. conducted andsystem McAvoy, 2003; al., 2005; Seki - plant studies on control design for this plant have It provides an operator interface, which mimics a control studies on (Chen control design forOlsen this et plant have been been operating personnel. conducted (Chen andsystem McAvoy, 2003; Olsen et al., 2005; Seki - plant plant operating personnel. It provides an operator interface, which mimics a control et al., 2010; Tu et al., 2013; Psaltis et al., 2014). conducted (Chen and McAvoy, 2003; Olsen et al., 2005; Seki room of real plants, so that thewhich usersmimics can experience conducted 2003; Olsen et al., 2005; Seki -- It provides an operator interface, aa control et al., 2010;(Chen Tu etand al., McAvoy, 2013; Psaltis et al., 2014). It provides an operator interface, control room of real plants, so that thewhich usersmimics can experience et al., 2010; Tu et al., 2013; Psaltis et al., 2014). “real” plant operations. et al., 2010; Tu et al., 2013; Psaltis et al., 2014). room of real plants, so that the users can experience room of real plants, so that the users can experience This paper introduces the advanced VAM plant model, which “real” plant operations. This paper introduces the advanced VAM plant model, which In the “real” operations. next plant section, the VAM plant is described. In section 3, “real” operations. is implemented on the the commercial simulator, This paper advanced VAM model, which the next plant section, the VAM plant is described. In section 3, This paper introduces introduces advanceddynamic VAM plant plant model, Visual which In is implemented on the the commercial dynamic simulator, Visual details of the dynamic model Visual In the next section, the plant is 3, Modeler (Omegaon Simulation Co., Ltd.), as asimulator, new benchmark is the dynamic Visual In the next the VAM VAM plant implemented is described. described. In Inon section 3, of section, the dynamic model implemented onsection Visual is implemented implemented the commercial commercial dynamic Visual details Modeler (OmegaonSimulation Co., Ltd.), as asimulator, new benchmark Modeler are explained, including static balance and process details of the dynamic model implemented on Visual problem in lieu of the Tennessee Eastman problem. Visual Modeler (Omega Simulation Co., Ltd.), as a new benchmark details of the dynamic model implemented on Visual Modeler are explained, including static balance and process Modeler (Omega Co., Ltd.), as a problem. new benchmark problem in lieu ofSimulation the Tennessee Eastman Visual Modeler diagram. Section 4including shows operation scenarios also are static and process Modeler haslieu found applications in problem. industry Visual as an flow problem of Tennessee Eastman Modeler are explained, explained, static balance balance andand process diagram. Section 4including shows operation scenarios and also problem in of the themany Tennessee Eastman Modeler in haslieu found many applications in problem. industry Visual as an flow demonstrates an example of start-up operation. In section 5, flow diagram. Section 4 shows operation scenarios and also operator training simulator. It is capable of handling various Modeler has found many applications in industry as an flow diagram. Section 4 shows operation scenarios and also demonstrates an example of start-up operation. In section 5, Modeler has found many applications in industry as an operator training simulator. It is capable of handling various demonstrates summary of supporting environment for the users is an example of start-up operation. In section 5, modes of plant operations including start-up/shut-down as a operator training simulator. It is capable of handling various demonstrates an example of start-up operation. In section 5, summary of supporting environment for the users is operator training simulator. It is capable of handling various modes of plant operations including start-up/shut-down as a described. conclusions are drawn. for summary Finally of supporting supporting environment for the the users users is is prerequisite of the operator training system, whereas the modes of plant operations including start-up/shut-down as a summary of environment modes of plant including Finally conclusions are drawn. prerequisite of operations the operator trainingstart-up/shut-down system, whereas asthea described. described. Finally conclusions are drawn. prerequisite of the operator training system, whereas the described. Finally conclusions are drawn. prerequisite of the operator training system, whereas the
2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2016, 2016 IFAC 533 Hosting by Elsevier Ltd. All rights reserved. Copyright 2016 responsibility IFAC 533Control. Peer review©under of International Federation of Automatic Copyright © 533 Copyright © 2016 2016 IFAC IFAC 533 10.1016/j.ifacol.2016.07.397
IFAC DYCOPS-CAB, 2016 534 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
2. VAM PLANT Figure 1 shows the overall process flow diagram of the VAM plant, which firstly generates VAM product in the reactor, then separates the product from the unreacted materials and by-products by using the distillation column and other units. 2.1 Process Overview Eight materials appear in the VAM plant as shown in Table 1. Three raw materials are introduced to the process. Ethylene (C2H4) and oxygen (O2) are fed in the gas phase; acetic acid (AcOH) is fed in the liquid phase and vaporized with superheated steam at the vaporizer. These three materials are mixed and introduced to the reactor, in which the following gas phase reactions are taking place. [Main reaction] C2H4 + CH3COOH + 1/2O2
→
Part of VAM removed from the top of the absorber is recycled to the inlet of the process. The remaining part of gas is introduced to the CO2 remover and the gas purge system, which keeps concentration of CO2 around 5~10mol% and C2H6 around 5mol% in the gas recycle line. The VAM crude at the intermediate buffer tank is fed to the azeotropic distillation column. VAM-H2O mixture discharged from the top of the column is condensed at the condenser and separated at the decanter. VAM forms the organic phase and H2O goes to the aqueous phase; VAM product is discharged as organic product from the decanter. Unreacted AcOH is discharged from the bottom and recycled to both the vaporizer and the absorber. 2.2 Process constraints The VAM process must be operated under the following constraints, which come from safety, equipment protection, and product quality requirements.
[Side reaction] C2H4 + 3 O2
-
→
-
The main reaction generates VAM (CH2=CHOCOCH3) product and by-product water (H2O) from C2H4, AcOH (CH3COOH), and O2. The side reaction generates byproduct carbon dioxide (CO2) and H2O from C2H4 and O2. Both reactions are exothermic, thus reaction heat is removed by boiler feed water (BFW) circulation and steam is generated at the shell side of the reactor. The reactor outlet gas containing about 5mol% VAM product is cooled down to 37degC with two coolers. Unreacted AcOH, H2O, and VAM are condensed as liquid VAM crude at the separator. On the other hand, separated gas leaving from the separator includes unreacted C2H4, O2, by-product CO2, inert ethane (C2H6), and a small amount of uncondensed VAM. This separated gas is compressed by the compressor to circulate recycle gas flow, then introduced to the absorber. The uncondensed VAM is absorbed by the cold AcOH which is fed from the top of the absorber. The mixture of VAM and AcOH is discharged from the bottom of the absorber, and mixed with the VAM crude at the intermediate buffer tank.
-
3. IMPLEMENTATION OF THE VAM PLANT MODEL The VAM plant model was implemented with Visual Modeler (Omega Simulation Co., Ltd.), which is the commercial dynamic simulation software package, with reference to the process configuration and the static balance proposed by Luyben and Tyreus (1998). In addition, the VAM plant model was arranged with reference to actual issues and opinions from the fields of Japanese chemical industry to make the model and operation scenarios more realistic than the original as follows.
Table 1. Materials in the VAM plant Material Ethylene (C2H4) Oxygen (O2) Acetic Acid (AcOH) VAM Water (H2O) Carbon Dioxide (CO2) Ethane (C2H6) Nitrogen (N2)
The O2 concentration must not exceed 8mol% anywhere in the gas pipeline to avoid an explosion. Pressure in the gas pipeline must not exceed 862kPaG because of the mechanical construction limit. The peak reactor temperature must remain below 200degC to prevent damage to the catalyst. The reactor inlet temperature must remain above 130degC to avoid dew point of AcOH. The AcOH concentration in the VAM product must remain below 150ppm as a product specification. The VAM concentration in the bottom of the distillation column must remain below 100ppm to prevent polymerization of VAM. The compressor must run while O2 is remaining in the gas pipeline to avoid forming of O2 hot spot and explosion.
Description Raw material of VAM Raw material of VAM Raw material of VAM Product By-product By-product Accompanying gas of Ethylene Inert gas
-
534
Tuning the steady-state balance Adding pipelines and units for start-up operation Adding disturbances and malfunctions for abnormal situation operation
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
535
Fig. 1. Process flow diagram of the VAM plant Table 2. Representative steady-state stream data. New balance / Original balance (Luyben and Tyreus, 1998) Stream number Flow Temperature Pressure O2 CO2 C2H4 C2H6 VAM AcOH H2O N2
Stream number Flow Temperature Pressure O2 CO2 C2H4 C2H6 VAM AcOH H2O N2
[kmol/h] [degC] [kPaG] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%]
C2H4 feed 1 59.4 / 49.9 30.0 / 30.0 933.0 / 931.0 0.000 / 0.000 0.000 / 0.000 0.999 / 0.999 0.001 / 0.001 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
AcOH feed 2 55.5 / 47.1 30.0 / 30.0 1299 / 931.0 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000
O2 feed 3 35.4 / 31.2 30.0 / 30.0 933.0 / 931.0 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
Reactor inlet 4 1143 / 1159 148.5 / 148.0 783.1 / 783.1 0.060 / 0.074 0.058 / 0.007 0.709 / 0.584 0.051 / 0.214 0.001 / 0.002 0.114 / 0.110 0.007 / 0.009 0.000 / 0.000
Reactor outlet 5 1116 / 1137 156.8 / 154.9 525.5 / 518.2 0.029 / 0.051 0.064 / 0.011 0.674 / 0.553 0.052 / 0.218 0.052 / 0.042 0.066 / 0.072 0.062 / 0.053 0.000 / 0.000
[kmol/h] [degC] [kPaG] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%]
CO2 purge 10 4.78 / 5.10 38.0 / 40.0 815.4 / 781.3 0.000 / 0.000 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
C2H6 purge 11 0.93 / 0.18 38.0 / 40.0 815.4 / 781.3 0.036 / 0.059 0.072 / 68ppm 0.821 / 0.667 0.064 / 0.266 0.002 / 0.002 0.004 / 0.005 0.001 / 0.001 0.000 / 0.000
Column feed 13 240.7 / 229.2 39.3 / 42.5 59.1 / 477.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.234 / 0.206 0.466 / 0.513 0.300 / 0.281 0.000 / 0.000
VAM product 15 58.5 / 49.6 41.0 / 40.0 22.7 / 22.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.963 / 0.950 23.0ppm/370 ppm 0.037 / 0.050 0.000 / 0.000
H2O product 16 58.7 / 49.9 41.0 / 40.0 22.7 / 22.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.002 / 0.002 16.1ppm/370ppm 0.998 / 0.998 0.000 / 0.000
535
IFAC DYCOPS-CAB, 2016 536 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
Table 3. Modification of the steady-state balance Changing points Reduce C2H6 concentration
Affected streams 4, 5, 6, 7, 9, 11, 12
Increase CO2 concentration
4, 5, 6, 7, 9, 10, 12
Increase C2H4 concentration; Decrease O2 concentration
1, 2, 3, 4, 5, 6, 7, 9, 10, 12, 13, 14, 15, 16, 17
Tighten specification of VAM product
15
Backgrounds C2H6 concentration is remained 2 to 5mol% in the actual process. In the past model, C2H6 concentration was more than 21% and it was decreasing efficiency of reaction. C2H6 concentration is remained around 5mol% in the new balance. CO2 concentration is remained around 5 to 10mol% in the actual process. In the past model, CO2 was removed perfectly with the CO2 remover, but it was not reasonable balance. CO2 concentration is remained around 6.0mol% in the new balance. VAM reaction efficiency is improved due to changing C2H6 and CO2 concentration as above. It enables to produce more VAM product with lower O2 feed concentration. As a result of modification, VAM production is increased 20% and raw material feed flow balance is changed. This improvement also affect to feed composition and operation condition of the distillation column. AcOH concentration in VAM product is tightened from 600ppm to 150ppm by reference to actual product specification.
3.1 Tuning steady-state balance 3.3 Adding disturbances and malfunctions for abnormal situation operation
Table 2 shows the steady-state material balance of both the new VAM plant model and the original balance proposed by Luyben and Tyreus (1998). Table 3 shows representative modifications of the steady-state, which is the important target profile for the start-up operation.
Tables 4 and 5 show disturbances and malfunctions, which were implemented on the new VAM plant model. The users can activate any single or multiple events at any time, not only during steady-state operation but also during startup/shut-down operation. The users can flexibly change parameters of disturbances and malfunctions such as magnitude, time constant, and timing of recovering. The disturbances are coming from outside of the VAM process, such as changes in temperature and pressure of utilities. Thus, the control system should be designed to suppress these disturbances. On the other hand, process failure caused by malfunctions such as pump trip cannot be suppressed by control, so fault detection and diagnosis are needed. The users are able to investigate various issues related to process control, process monitoring, and others by using these malfunctions and disturbances.
3.2 Adding pipelines and units for start-up operations To make operation scenarios more realistic, the following pipelines and units are added to the plant model. -
[Configuration of the AcOH vaporizer] The AcOH vaporizer was expressed with two unit models, i.e., a heat exchange unit and a vapor-liquid separation unit, in the original model (Yumoto et al., 2010). In the new plant model, the vaporizer is expressed with a single unit model that can calculate heat exchange and separation simultaneously.
-
[Start-up/shut-down lines] Several start-up/shut-down lines were implemented to realize realistic start-up/shut-down operation procedures. In the distillation column, for example, an initial product charge line and an emergency deliquoring line were added to the decanter vessel.
-
[Intermediate buffer tank before the distillation column] An intermediate buffer tank is equipped between the reaction section and the separation section in many chemical processes to alleviate disturbances from the reaction section to the separation section. A new intermediate buffer tank was added before the distillation column as the same manner of the real process. This intermediate tank can be bypassed by switching lines around the vessel on demand.
4. OPERATION SCENARIOS The VAM plant model can provide realistic operation scenarios, which cannot be investigated in conventional benchmark problems. The users can experience various real operation of the VAM plant as if they are operators of the real plant. 4.1 Initial state The VAM plant model has initial state files which include information of all units and streams. The users need to load following initial state files before starting operation.
536
-
[Before start-up] This initial state is prepared for start-up operation. All process units remain shut-down condition at this state.
-
[Steady-state] This initial state is the steady-state described in Table 2.
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
Table 4. Disturbance list No 1 2 3 4 5 6 7 8 9 10 11
Description Change C2H4 feed composition Change AcOH feed composition Change C2H4 feed pressure Change O2 feed pressure Heavy rain Change cooling water temperature Change cooling water feed pressure Change 1.5MPa steam feed pressure Change 5.0MPa steam feed pressure Change 1.5MPa steam temperature Change 5.0MPa steam temperature
Pattern Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp
Table 5. Malfunction list No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Description Reduce reaction activity Reduce heat transmission in vaporiser Accident error of temperature sensors Stop absorber circulation pump Stop column feed pump Stop column bottom pump Stop column reflux pump Stop compressor Stick separator bottom valve Stick absorber bottom valve Stop analysing sensors Failure of gas feed pressure sensor Failure of column reflux flow sensor Failure of column bottom level sensor Crack weir of decanter Increase differential pressure of column Decrease efficiency of CO2 removal Measurement error of column bottom level sensor Measurement error of vaporiser level sensor
Pattern Step/Ramp Step/Ramp Step/Ramp Step Step Step Step Step Step Step Step Step Step Step Step Step/Ramp Step/Ramp Step/Ramp
Fig. 2. Start-up operation procedure of the VAM plant
Step/Ramp
Fig. 3. Time-series data of start-up operation 537
537
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538 538
ACKNOWLEDGMENTS
4.2 Start-up operation
This research has been conducted as a part of the projects of the Workshop No. 31 (Process Control Technology) of the Japan Society for the Promotion of Science (JSPS) 143rd committee on Process Systems Engineering. Discussions with the workshop members are greatly appreciated.
Figure 2 shows the outline of start-up operation procedure of the VAM plant model. Total time to finish this operation procedure is almost 20 hours, but the users can accelerate the speed of simulation calculations up to 200 fold in the case of time-consuming procedure, such as liquid filling of vessels and unit warming up. This operation procedure must be executed under process constraints described in section 2. Figure 3 shows representative time-series process data during the start-up operation.
REFERENCES Alexandre, C. D. and Costin, S. B. (2011). Chemical Process Design, chapter 10, Wiley. Chen, R. and McAvoy, T. J. (2003). Plantwide control system design: Methodology and application to a vinyl acetate process. Ind. Eng. Chem. Res., 42, 4753-4771. Downs, J. and Vogel, E. (1993). A plantwide industrial process control problems. Computers Chem. Engng, 17, 245-255. Luyben, M. L. and Tyreus, B. D. (1998). An industrial design/control study for the vinyl acetate monomer process. Computers Chem. Engng, 22, 867-877. Luyben. M. L. et al. (1998). An industrial design/control study for the vinyl acetate monomer process, Computers and Chemical Engineering. Vol. 22, No.7-8, pp.867-877. Machida, Y,. Nakaya, M., Omata, T. (2014). Process Education and Development Infrastructure Building by Virtual Plant in DCS Manufacturing Company, The International Federation of Automatic Control, Cape Town (IFAC 2014), pp.12237-12242. Olsen, D. G., Svrcek, W. Y., and Young, B. R. (2005). Plantwide control study of a vinyl acetate monomer process design. Chemical Engineering Communication, 192, 1243-1257. Psaltis, A., Kookos, I. K., and Kravaris, C. (2014). Plantwide control structure selection methodology for the benchmark vinyl acetate monomer plant. Computers Chem. Engng, 62, 108-116. Rangaiah, G. P. and Kariwala, V. (2012). Plantwide control, chapter 9, Wiley. Seki, H., Ogawa, M., Itoh, T., Ootakara, S., Murata, H., Hashimoto, Y., Kano, M. (2010). Plantwide control system design of the benchmark vinyl acetate monomer production plant, Computers Chem. Engng, 34, pp.12821295. Tu, T., Matthew, E., and Chistofides, P. D. (2013). Model predictive control of a nonlinear large-scale process network used in the production of vinyl acetate. Ind. Eng. Chem. Res., 52, 12463-12481. Yumoto, T., Ootakara, S., Seki, H., Hashimoto, Y., Murata, H., Kano, M., Yamashita, Y. (2010). Rigorous Dynamic Simulator for Control Study of the Large-scale Benchmark Chemical Plant, Dynamics and Control of Process Systems (DYCOPS 2010), pp.49-54.
4.3 Production load-up/load-down operation The users can execute load-up/load-down operation from the steady-state. Production load of the VAM process can be increased by raising the O2 feed concentration and the reactor temperature. Load-down operation is totally opposite. After changing operation condition of the reactor, the distillation column condition should be adjusted in conformity to reactor load. This operation scenario will be a good example to investigate the plant-wide control loop configuration. 4.4 Abnormal situation operation The users can start a simulation of the VAM plant model from the steady-state initial condition and experience abnormal situation operation by triggering disturbances and malfunctions at anytime. The users can study dynamic behavior caused by such disturbances and malfunctions, and also they can configure control logic and fault detection logic on the custom calculation blocks which are supported on the VAM plant model. 5. SUPPORTING ENVIRONMENT FOR THE USERS The following elements are available on the Omega Simulation Co., web site for free. -
The VAM plant model with a free limited license of Visual Modeler. - Time-series data sets for data analysis study including fault detection and diagnosis. - Explanation documents to use the VAM plant model, which include process description, operation procedure, system manual, and so on. We are also developing interface which links the VAM plant model and MATLAB®. This interface will be helpful to test control logics developed on MATLAB® as if it is applied to the real plant. 6. CONCLUSIONS A rigorous plant model of vinyl acetate monomer (VAM) process was developed. The users can get feelings of the real plant operation through this model. It can also be used as a test environment for various studies in the process control field, such as plant-wide control, process monitoring, optimal operation, and so on. The developed VAM plant model with a free limited license of Visual Modeler is already available through the web site of Omega Simulation Co., Ltd.: http://www.omegasim.co.jp/contents_e/product/vm/cnt4/ (Official site : http://www.omegasim.co.jp/index_e.htm) 538
ScienceDirect
IFAC-PapersOnLine 49-7 (2016) 533–538
Vinyl Vinyl Acetate Acetate Monomer Monomer (VAM) (VAM) Plant Plant Model: Model: Vinyl Monomer (VAM) Model: Vinyl Acetate Acetate Monomer (VAM) Plant Plant Model: Study A New Benchmark Problem for Control and Operation A New Benchmark Problem for Control and Operation Study A A New New Benchmark Benchmark Problem Problem for for Control Control and and Operation Operation Study Study
Yuta Machida* Shigeki Ootakara** Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano† † † Yuta Machida* Shigeki Ootakara**§ Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano # # † ‡ § † Yuta Machida* Shigeki Ootakara** Hiroya Seki*** Yoshihiro Hashimoto**** Manabu Kano Yasuhiro Miyake Naoto Anzai Masayoshi Sawai Takashi Katsuno Toshiaki Omata # # ‡ § § # Yuta Machida* Shigeki Ootakara** Seki*** Yoshihiro Hashimoto**** Manabu Kano ‡ Naoto § Takashi Yasuhiro Miyake Anzai§ Hiroya Masayoshi Sawai Katsuno## Toshiaki Omata # ‡ ‡ Naoto Anzai§ § Masayoshi Sawai§ § Takashi Katsuno# Toshiaki Omata# Yasuhiro Miyake Yasuhiro Miyake Naoto Anzai Masayoshi Sawai Takashi Katsuno Toshiaki Omata *Omega Simulation Co.,Ltd., Tokyo 169-0051, Japan (e-mail: [email protected]) *Omega Simulation Co.,Ltd., Tokyo 169-0051, Japan (e-mail: [email protected]) ** Mitsui Tokyo Chemicals Inc., Osaka 592-8501, Japan *Omega Co.,Ltd., 169-0051, Japan (e-mail: *Omega Simulation Simulation ** Co.,Ltd., 169-0051, Japan592-8501, (e-mail: [email protected]) [email protected]) Mitsui Tokyo Chemicals Inc., Osaka Japan *** Chemical Resources Laboratory, Tokyo Institute of Technology, Yokohama 226-8503, Japan ** Mitsui Chemicals Inc., Osaka 592-8501, **Laboratory, Mitsui Chemicals Osaka 592-8501, Japan Japan *** Chemical Resources Tokyo Inc., Institute Yokohama 226-8503, Japan of Technology, Technology, ****Dept. of Computer Science, Nagoya Institute of Nagoya 466-8555, Japan *** Chemical Resources Laboratory, Tokyo Yokohama 226-8503, Japan *******Dept. Chemical† Resources Tokyo Institute of Technology, Nagoya Yokohama 226-8503, Japan of ComputerLaboratory, Science, Nagoya 466-8555, Japan Dept. of Systems Science, Kyoto University, Kyoto 606-8501, Japan ****Dept. of Computer Science, Nagoya Institute of Technology, Nagoya 466-8555, Japan ****Dept. † ofDept. Computer Science, Nagoya Institute of Technology, Nagoya 466-8555, Japan † of Systems Science, Kyoto University, Kyoto 606-8501, Japan † †Dept. of‡ Systems Science, Kyoto University, Kyoto Ube Industries, Yamaguchi 755-8633, Japan Japan Dept. of‡ Science,Ltd., Kyoto University, Kyoto 606-8501, 606-8501, ‡Systems Ube Industries, Ltd., Yamaguchi 755-8633, Japan Japan ‡ § ‡Ube Industries, Ltd., Yamaguchi 755-8633, Corp., Tokyo 100-8246, Japan Japan § Ube Industries, Ltd., Yamaguchi 755-8633, §Zeon Zeon Corp., Tokyo 100-8246, Japan Japan # § §ZeonElectric Corp., 100-8246, Tokyo 169-0051 Japan ## Yokogawa Zeon Corp., Corp., Tokyo Tokyo 100-8246, Japan Japan ## Yokogawa Electric Corp., Tokyo 169-0051 Japan Yokogawa Electric Corp., Tokyo 169-0051 Japan Yokogawa Tokyoacetate 169-0051 Japan (VAM) production was Abstract: A rigorous dynamic plant Electric model Corp., of a vinyl monomer Abstract: A rigorous dynamic plant model of a vinyl acetate monomer (VAM) production was developed. This plant model enables the users to experience realistic plant operation, since it reflectswas the Abstract: A rigorous dynamic plant model of aa vinyl monomer (VAM) production Abstract: A rigorous dynamic plant model ofexperience vinyl acetate acetate monomer (VAM) production was developed. This plant model enables the users toon realistic plant operation, since it reflects the real plant characteristics and practical problems the basis of experienced practitioners’ opinions. More developed. This enables users realistic plant since it the developed. This plant plant model model enables the the users to toonexperience experience realistic plant operation, operation, since it reflects reflects the real plant characteristics and practical problems the basis of experienced practitioners’ opinions. More importantly, the plant model provides problems a new benchmark problem; the users practitioners’ can investigate start-up/shutreal plant characteristics and practical on the basis of experienced opinions. More real plant characteristics and practical problems on the basis of experienced practitioners’ opinions. More importantly, the plant model provides a new benchmark problem; the users can investigate start-up/shutdown operation, plant-wide control, detection and diagnosis, Multiple scenarios importantly, the model provides aa new benchmark problem; the can investigate start-up/shutimportantly, the plant plant modelprocess provides new fault benchmark problem; the users usersand canothers. investigate start-up/shutdown operation, plant-wide process control, fault detection and diagnosis, and others. Multiple scenarios prepared in the developed model cannot be simulated in conventional benchmark problems. The plant down operation, plant-wide process control, fault detection and diagnosis, and others. Multiple scenarios down operation, plant-widemodel processcannot control, detection and diagnosis, and others. Multiple The scenarios prepared in the developed be fault simulated in conventional benchmark problems. plant model can be used also for chemical engineering education. This advantageous plant model is released prepared in the developed model cannot be simulated in conventional benchmark problems. The plant prepared developed cannot be simulated in conventional benchmark problems. plant model caninbetheused also formodel chemical engineering education. This advantageous plant model isThe released from Omega Simulation Co., Ltd. with a free limited license of Visual Modeler, which is a commercial model can be used also for chemical engineering education. This advantageous plant model is released modelOmega can beSimulation used also for education. advantageous is released from Co.,chemical Ltd. withengineering a free limited license This of Visual Modeler, plant whichmodel is a commercial dynamic simulator and can linked MATLAB®. Thisof article to introduce VAM plant from Simulation Co., with aa free limited Visual Modeler, which commercial from Omega Omega Simulation Co.,be Ltd. withwith free limited license license Visualaims Modeler, which is isthe commercial dynamic simulator and can beLtd. linked with MATLAB®. Thisofarticle aims to introduce theaa VAM plant model, the steady-state balance, various disturbances and malfunctions, and operation scenarios. dynamic simulator and can be linked with MATLAB®. This article aims to introduce the VAM dynamicthesimulator and balance, can be linked MATLAB®. This article aims to introduce the VAM plant plant model, steady-state variouswith disturbances and malfunctions, and operation scenarios. model, the steady-state balance, various disturbances and and operation scenarios. model, steady-state balance, various and malfunctions, malfunctions, and operation scenarios. © 2016,the IFAC (International Federation ofdisturbances Automatic Hosting by Elsevier Ltd. All rights reserved. Keywords: Process simulators, process models, Control) dynamic models, benchmark examples, chemical Keywords: Process simulators, process models, dynamic models, benchmark examples, chemical industry, plant-wide Keywords: Process control. simulators, process process models, models, dynamic dynamic models, models, benchmark benchmark examples, examples, chemical chemical Keywords: Process simulators, industry, plant-wide control. industry, plant-wide control. industry, plant-wide control. simulator attempts to keep the fidelity of the rigorous simulator attempts to keep the fidelity of the rigorous 1. INTRODUCTION nonlinear as high as possible. The dynamic 1. INTRODUCTION simulator attempts to the of rigorous simulator model attempts to keep keep the fidelity fidelity of the the simulator rigorous nonlinear model as high as between possible. Themodel dynamic simulator 1. INTRODUCTION 1. INTRODUCTION makes a good compromise the accuracy and model as high as possible. The dynamic simulator A test problem which checks the industrial relevance of new nonlinear nonlinear model as high as possible. The dynamic simulator a good compromise between the model accuracy and A test problem which checks the industrial relevance of new makes solvability. We improved the VAM plant model developed aa good between model accuracy and ideas technical is of great importance in makes A problem which checks relevance of good compromise between the model accuracy and solvability. Wecompromise improved the VAM the plant model developed A test test and problem whichdevelopments checks the the industrial industrial relevance of new new ideas and technical developments is of great importance in makes by Yumoto We et al. (2010)the with reference to practitioners’ solvability. improved VAM plant model developed the study of process control and process data analysis. A ideas and technical developments is of great importance in solvability. We improved the VAM plant model developed Yumoto et al. (2010) with reference to practitioners’ ideasstudy and technical of great importance the of processdevelopments control and is process data analysis. in A by opinions to et make operation scenarios. The new by al. (2010) reference to great number of researchers haveprocess utilizeddata the analysis. Tennessee the of control A by Yumoto Yumoto al. realistic (2010) with with reference to practitioners’ practitioners’ to et make realistic operation scenarios. The new the study study of process process control and and A opinions great number of researchers haveprocess utilizeddata the analysis. Tennessee VAM plant model has the following features, which the opinions to make realistic operation scenarios. The Eastman challenge problem in various fields including plantgreat number of researchers have utilized the Tennessee opinions to make realistic operation scenarios. The new new VAM plant model has the following features, which the great number of researchers have utilized the Tennessee Eastman challenge problem in various fields including plant- VAM Tennessee Eastman process model cannot cope with: plant model has the following features, which the wide control, production efficiency optimization, soft-sensor Eastman challenge problem in various fields including plantVAM plant model has the following features, which the Tennessee Eastman process model cannot cope with: Eastman challenge problem in various fields including plantwide control, production efficiency optimization, soft-sensor Tennessee Eastman process model cannot cope with: design, fault detection and diagnosis, etc., since it was wide control, control, production production efficiency efficiency optimization, optimization, soft-sensor soft-sensor -Tennessee Eastman process model cannot cope with: start-up It can simulate transient operations including wide design, fault detection and diagnosis, since it1993). was - It can simulate transient operations including start-up introduced more than 20 years ago (Downsetc., and Vogel, design, fault detection and since was andcan shut-down. design, fault detection and diagnosis, diagnosis, etc., since it it1993). was -- It simulate introduced more than 20 years ago (Downsetc., and Vogel, It simulate transient transient operations operations including including start-up start-up andcan shut-down. Even now, the Tennessee Eastman problem is the most introduced more than 20 years ago (Downs and Vogel, 1993). It can generate noises and disturbances with such introduced 20 yearsEastman ago (Downs and Vogel, shut-down. Even now,more the than Tennessee problem is the1993). most -- and and shut-down. It can generate noises and disturbances with such popular problem. Later, Luyben and Tyreus (1998) Even the Eastman problem is the realistic sources asnoises measurement noise of sensors, nonEven now, now, the Tennessee Tennessee Eastman problem is developed the most most -- It can generate and disturbances with such popular problem. Later, Luyben and Tyreus (1998) developed It can generate noises and disturbances with such realistic sources as measurement noise of sensors, nonapopular model problem. of a vinylLater, acetate monomer (VAM)(1998) plant, which is a popular Luyben and developed ideality in actuators such as valve dead band, daily Luyben and Tyreus Tyreus developed realistic sources as measurement noise of sensors, nona model problem. of a vinylLater, acetate monomer (VAM)(1998) plant, which is a realistic sources as measurement noise of sensors, nonideality in actuators such as valve dead band, daily containing unit operations alarger of acetate monomer (VAM) which temperature fluctuations, etc. a model modelsystem of aa vinyl vinyl acetate standard monomerchemical (VAM) plant, plant, which is is aa ideality in actuators such as valve dead band, daily larger system containing standard chemical unit operations ideality in actuators such as valve dead band, daily temperature fluctuations, etc. for chemical components. The process design - temperature larger system containing standard unit It can simulate various failure which are set on largerreal system containing standard chemical chemical unit operations operations fluctuations, etc. for real chemical components. The process design - temperature fluctuations, etc. modes, It can simulate various failure modes, which are set on background is well-established in the The original paper. Several for real chemical components. process design the basis of the advice of experienced engineers for real chemical components. The process design It can can simulate simulate various various failure failure modes, modes, which which are set setand on background is well-established in the original paper. Several -- It are on the basis of the advice of experienced engineers and studies on control system design fororiginal this plant have been background is well-established in the paper. Several plant operating personnel. background is well-established in the original paper. Several the basis of the advice of experienced engineers and studies on control system design for this plant have been the basis of the advice of experienced engineers and operating personnel. conducted andsystem McAvoy, 2003; al., 2005; Seki - plant studies on control design for this plant have It provides an operator interface, which mimics a control studies on (Chen control design forOlsen this et plant have been been operating personnel. conducted (Chen andsystem McAvoy, 2003; Olsen et al., 2005; Seki - plant plant operating personnel. It provides an operator interface, which mimics a control et al., 2010; Tu et al., 2013; Psaltis et al., 2014). conducted (Chen and McAvoy, 2003; Olsen et al., 2005; Seki room of real plants, so that thewhich usersmimics can experience conducted 2003; Olsen et al., 2005; Seki -- It provides an operator interface, aa control et al., 2010;(Chen Tu etand al., McAvoy, 2013; Psaltis et al., 2014). It provides an operator interface, control room of real plants, so that thewhich usersmimics can experience et al., 2010; Tu et al., 2013; Psaltis et al., 2014). “real” plant operations. et al., 2010; Tu et al., 2013; Psaltis et al., 2014). room of real plants, so that the users can experience room of real plants, so that the users can experience This paper introduces the advanced VAM plant model, which “real” plant operations. This paper introduces the advanced VAM plant model, which In the “real” operations. next plant section, the VAM plant is described. In section 3, “real” operations. is implemented on the the commercial simulator, This paper advanced VAM model, which the next plant section, the VAM plant is described. In section 3, This paper introduces introduces advanceddynamic VAM plant plant model, Visual which In is implemented on the the commercial dynamic simulator, Visual details of the dynamic model Visual In the next section, the plant is 3, Modeler (Omegaon Simulation Co., Ltd.), as asimulator, new benchmark is the dynamic Visual In the next the VAM VAM plant implemented is described. described. In Inon section 3, of section, the dynamic model implemented onsection Visual is implemented implemented the commercial commercial dynamic Visual details Modeler (OmegaonSimulation Co., Ltd.), as asimulator, new benchmark Modeler are explained, including static balance and process details of the dynamic model implemented on Visual problem in lieu of the Tennessee Eastman problem. Visual Modeler (Omega Simulation Co., Ltd.), as a new benchmark details of the dynamic model implemented on Visual Modeler are explained, including static balance and process Modeler (Omega Co., Ltd.), as a problem. new benchmark problem in lieu ofSimulation the Tennessee Eastman Visual Modeler diagram. Section 4including shows operation scenarios also are static and process Modeler haslieu found applications in problem. industry Visual as an flow problem of Tennessee Eastman Modeler are explained, explained, static balance balance andand process diagram. Section 4including shows operation scenarios and also problem in of the themany Tennessee Eastman Modeler in haslieu found many applications in problem. industry Visual as an flow demonstrates an example of start-up operation. In section 5, flow diagram. Section 4 shows operation scenarios and also operator training simulator. It is capable of handling various Modeler has found many applications in industry as an flow diagram. Section 4 shows operation scenarios and also demonstrates an example of start-up operation. In section 5, Modeler has found many applications in industry as an operator training simulator. It is capable of handling various demonstrates summary of supporting environment for the users is an example of start-up operation. In section 5, modes of plant operations including start-up/shut-down as a operator training simulator. It is capable of handling various demonstrates an example of start-up operation. In section 5, summary of supporting environment for the users is operator training simulator. It is capable of handling various modes of plant operations including start-up/shut-down as a described. conclusions are drawn. for summary Finally of supporting supporting environment for the the users users is is prerequisite of the operator training system, whereas the modes of plant operations including start-up/shut-down as a summary of environment modes of plant including Finally conclusions are drawn. prerequisite of operations the operator trainingstart-up/shut-down system, whereas asthea described. described. Finally conclusions are drawn. prerequisite of the operator training system, whereas the described. Finally conclusions are drawn. prerequisite of the operator training system, whereas the
2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2016, 2016 IFAC 533 Hosting by Elsevier Ltd. All rights reserved. Copyright 2016 responsibility IFAC 533Control. Peer review©under of International Federation of Automatic Copyright © 533 Copyright © 2016 2016 IFAC IFAC 533 10.1016/j.ifacol.2016.07.397
IFAC DYCOPS-CAB, 2016 534 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
2. VAM PLANT Figure 1 shows the overall process flow diagram of the VAM plant, which firstly generates VAM product in the reactor, then separates the product from the unreacted materials and by-products by using the distillation column and other units. 2.1 Process Overview Eight materials appear in the VAM plant as shown in Table 1. Three raw materials are introduced to the process. Ethylene (C2H4) and oxygen (O2) are fed in the gas phase; acetic acid (AcOH) is fed in the liquid phase and vaporized with superheated steam at the vaporizer. These three materials are mixed and introduced to the reactor, in which the following gas phase reactions are taking place. [Main reaction] C2H4 + CH3COOH + 1/2O2
→
Part of VAM removed from the top of the absorber is recycled to the inlet of the process. The remaining part of gas is introduced to the CO2 remover and the gas purge system, which keeps concentration of CO2 around 5~10mol% and C2H6 around 5mol% in the gas recycle line. The VAM crude at the intermediate buffer tank is fed to the azeotropic distillation column. VAM-H2O mixture discharged from the top of the column is condensed at the condenser and separated at the decanter. VAM forms the organic phase and H2O goes to the aqueous phase; VAM product is discharged as organic product from the decanter. Unreacted AcOH is discharged from the bottom and recycled to both the vaporizer and the absorber. 2.2 Process constraints The VAM process must be operated under the following constraints, which come from safety, equipment protection, and product quality requirements.
[Side reaction] C2H4 + 3 O2
-
→
-
The main reaction generates VAM (CH2=CHOCOCH3) product and by-product water (H2O) from C2H4, AcOH (CH3COOH), and O2. The side reaction generates byproduct carbon dioxide (CO2) and H2O from C2H4 and O2. Both reactions are exothermic, thus reaction heat is removed by boiler feed water (BFW) circulation and steam is generated at the shell side of the reactor. The reactor outlet gas containing about 5mol% VAM product is cooled down to 37degC with two coolers. Unreacted AcOH, H2O, and VAM are condensed as liquid VAM crude at the separator. On the other hand, separated gas leaving from the separator includes unreacted C2H4, O2, by-product CO2, inert ethane (C2H6), and a small amount of uncondensed VAM. This separated gas is compressed by the compressor to circulate recycle gas flow, then introduced to the absorber. The uncondensed VAM is absorbed by the cold AcOH which is fed from the top of the absorber. The mixture of VAM and AcOH is discharged from the bottom of the absorber, and mixed with the VAM crude at the intermediate buffer tank.
-
3. IMPLEMENTATION OF THE VAM PLANT MODEL The VAM plant model was implemented with Visual Modeler (Omega Simulation Co., Ltd.), which is the commercial dynamic simulation software package, with reference to the process configuration and the static balance proposed by Luyben and Tyreus (1998). In addition, the VAM plant model was arranged with reference to actual issues and opinions from the fields of Japanese chemical industry to make the model and operation scenarios more realistic than the original as follows.
Table 1. Materials in the VAM plant Material Ethylene (C2H4) Oxygen (O2) Acetic Acid (AcOH) VAM Water (H2O) Carbon Dioxide (CO2) Ethane (C2H6) Nitrogen (N2)
The O2 concentration must not exceed 8mol% anywhere in the gas pipeline to avoid an explosion. Pressure in the gas pipeline must not exceed 862kPaG because of the mechanical construction limit. The peak reactor temperature must remain below 200degC to prevent damage to the catalyst. The reactor inlet temperature must remain above 130degC to avoid dew point of AcOH. The AcOH concentration in the VAM product must remain below 150ppm as a product specification. The VAM concentration in the bottom of the distillation column must remain below 100ppm to prevent polymerization of VAM. The compressor must run while O2 is remaining in the gas pipeline to avoid forming of O2 hot spot and explosion.
Description Raw material of VAM Raw material of VAM Raw material of VAM Product By-product By-product Accompanying gas of Ethylene Inert gas
-
534
Tuning the steady-state balance Adding pipelines and units for start-up operation Adding disturbances and malfunctions for abnormal situation operation
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
535
Fig. 1. Process flow diagram of the VAM plant Table 2. Representative steady-state stream data. New balance / Original balance (Luyben and Tyreus, 1998) Stream number Flow Temperature Pressure O2 CO2 C2H4 C2H6 VAM AcOH H2O N2
Stream number Flow Temperature Pressure O2 CO2 C2H4 C2H6 VAM AcOH H2O N2
[kmol/h] [degC] [kPaG] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%]
C2H4 feed 1 59.4 / 49.9 30.0 / 30.0 933.0 / 931.0 0.000 / 0.000 0.000 / 0.000 0.999 / 0.999 0.001 / 0.001 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
AcOH feed 2 55.5 / 47.1 30.0 / 30.0 1299 / 931.0 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000
O2 feed 3 35.4 / 31.2 30.0 / 30.0 933.0 / 931.0 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
Reactor inlet 4 1143 / 1159 148.5 / 148.0 783.1 / 783.1 0.060 / 0.074 0.058 / 0.007 0.709 / 0.584 0.051 / 0.214 0.001 / 0.002 0.114 / 0.110 0.007 / 0.009 0.000 / 0.000
Reactor outlet 5 1116 / 1137 156.8 / 154.9 525.5 / 518.2 0.029 / 0.051 0.064 / 0.011 0.674 / 0.553 0.052 / 0.218 0.052 / 0.042 0.066 / 0.072 0.062 / 0.053 0.000 / 0.000
[kmol/h] [degC] [kPaG] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%] [mol%]
CO2 purge 10 4.78 / 5.10 38.0 / 40.0 815.4 / 781.3 0.000 / 0.000 1.000 / 1.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000
C2H6 purge 11 0.93 / 0.18 38.0 / 40.0 815.4 / 781.3 0.036 / 0.059 0.072 / 68ppm 0.821 / 0.667 0.064 / 0.266 0.002 / 0.002 0.004 / 0.005 0.001 / 0.001 0.000 / 0.000
Column feed 13 240.7 / 229.2 39.3 / 42.5 59.1 / 477.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.234 / 0.206 0.466 / 0.513 0.300 / 0.281 0.000 / 0.000
VAM product 15 58.5 / 49.6 41.0 / 40.0 22.7 / 22.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.963 / 0.950 23.0ppm/370 ppm 0.037 / 0.050 0.000 / 0.000
H2O product 16 58.7 / 49.9 41.0 / 40.0 22.7 / 22.8 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.000 / 0.000 0.002 / 0.002 16.1ppm/370ppm 0.998 / 0.998 0.000 / 0.000
535
IFAC DYCOPS-CAB, 2016 536 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
Table 3. Modification of the steady-state balance Changing points Reduce C2H6 concentration
Affected streams 4, 5, 6, 7, 9, 11, 12
Increase CO2 concentration
4, 5, 6, 7, 9, 10, 12
Increase C2H4 concentration; Decrease O2 concentration
1, 2, 3, 4, 5, 6, 7, 9, 10, 12, 13, 14, 15, 16, 17
Tighten specification of VAM product
15
Backgrounds C2H6 concentration is remained 2 to 5mol% in the actual process. In the past model, C2H6 concentration was more than 21% and it was decreasing efficiency of reaction. C2H6 concentration is remained around 5mol% in the new balance. CO2 concentration is remained around 5 to 10mol% in the actual process. In the past model, CO2 was removed perfectly with the CO2 remover, but it was not reasonable balance. CO2 concentration is remained around 6.0mol% in the new balance. VAM reaction efficiency is improved due to changing C2H6 and CO2 concentration as above. It enables to produce more VAM product with lower O2 feed concentration. As a result of modification, VAM production is increased 20% and raw material feed flow balance is changed. This improvement also affect to feed composition and operation condition of the distillation column. AcOH concentration in VAM product is tightened from 600ppm to 150ppm by reference to actual product specification.
3.1 Tuning steady-state balance 3.3 Adding disturbances and malfunctions for abnormal situation operation
Table 2 shows the steady-state material balance of both the new VAM plant model and the original balance proposed by Luyben and Tyreus (1998). Table 3 shows representative modifications of the steady-state, which is the important target profile for the start-up operation.
Tables 4 and 5 show disturbances and malfunctions, which were implemented on the new VAM plant model. The users can activate any single or multiple events at any time, not only during steady-state operation but also during startup/shut-down operation. The users can flexibly change parameters of disturbances and malfunctions such as magnitude, time constant, and timing of recovering. The disturbances are coming from outside of the VAM process, such as changes in temperature and pressure of utilities. Thus, the control system should be designed to suppress these disturbances. On the other hand, process failure caused by malfunctions such as pump trip cannot be suppressed by control, so fault detection and diagnosis are needed. The users are able to investigate various issues related to process control, process monitoring, and others by using these malfunctions and disturbances.
3.2 Adding pipelines and units for start-up operations To make operation scenarios more realistic, the following pipelines and units are added to the plant model. -
[Configuration of the AcOH vaporizer] The AcOH vaporizer was expressed with two unit models, i.e., a heat exchange unit and a vapor-liquid separation unit, in the original model (Yumoto et al., 2010). In the new plant model, the vaporizer is expressed with a single unit model that can calculate heat exchange and separation simultaneously.
-
[Start-up/shut-down lines] Several start-up/shut-down lines were implemented to realize realistic start-up/shut-down operation procedures. In the distillation column, for example, an initial product charge line and an emergency deliquoring line were added to the decanter vessel.
-
[Intermediate buffer tank before the distillation column] An intermediate buffer tank is equipped between the reaction section and the separation section in many chemical processes to alleviate disturbances from the reaction section to the separation section. A new intermediate buffer tank was added before the distillation column as the same manner of the real process. This intermediate tank can be bypassed by switching lines around the vessel on demand.
4. OPERATION SCENARIOS The VAM plant model can provide realistic operation scenarios, which cannot be investigated in conventional benchmark problems. The users can experience various real operation of the VAM plant as if they are operators of the real plant. 4.1 Initial state The VAM plant model has initial state files which include information of all units and streams. The users need to load following initial state files before starting operation.
536
-
[Before start-up] This initial state is prepared for start-up operation. All process units remain shut-down condition at this state.
-
[Steady-state] This initial state is the steady-state described in Table 2.
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538
Table 4. Disturbance list No 1 2 3 4 5 6 7 8 9 10 11
Description Change C2H4 feed composition Change AcOH feed composition Change C2H4 feed pressure Change O2 feed pressure Heavy rain Change cooling water temperature Change cooling water feed pressure Change 1.5MPa steam feed pressure Change 5.0MPa steam feed pressure Change 1.5MPa steam temperature Change 5.0MPa steam temperature
Pattern Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp Step/Ramp
Table 5. Malfunction list No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Description Reduce reaction activity Reduce heat transmission in vaporiser Accident error of temperature sensors Stop absorber circulation pump Stop column feed pump Stop column bottom pump Stop column reflux pump Stop compressor Stick separator bottom valve Stick absorber bottom valve Stop analysing sensors Failure of gas feed pressure sensor Failure of column reflux flow sensor Failure of column bottom level sensor Crack weir of decanter Increase differential pressure of column Decrease efficiency of CO2 removal Measurement error of column bottom level sensor Measurement error of vaporiser level sensor
Pattern Step/Ramp Step/Ramp Step/Ramp Step Step Step Step Step Step Step Step Step Step Step Step Step/Ramp Step/Ramp Step/Ramp
Fig. 2. Start-up operation procedure of the VAM plant
Step/Ramp
Fig. 3. Time-series data of start-up operation 537
537
IFAC DYCOPS-CAB, 2016 June 6-8, 2016. NTNU, Trondheim, Norway Yuta Machida et al. / IFAC-PapersOnLine 49-7 (2016) 533–538 538
ACKNOWLEDGMENTS
4.2 Start-up operation
This research has been conducted as a part of the projects of the Workshop No. 31 (Process Control Technology) of the Japan Society for the Promotion of Science (JSPS) 143rd committee on Process Systems Engineering. Discussions with the workshop members are greatly appreciated.
Figure 2 shows the outline of start-up operation procedure of the VAM plant model. Total time to finish this operation procedure is almost 20 hours, but the users can accelerate the speed of simulation calculations up to 200 fold in the case of time-consuming procedure, such as liquid filling of vessels and unit warming up. This operation procedure must be executed under process constraints described in section 2. Figure 3 shows representative time-series process data during the start-up operation.
REFERENCES Alexandre, C. D. and Costin, S. B. (2011). Chemical Process Design, chapter 10, Wiley. Chen, R. and McAvoy, T. J. (2003). Plantwide control system design: Methodology and application to a vinyl acetate process. Ind. Eng. Chem. Res., 42, 4753-4771. Downs, J. and Vogel, E. (1993). A plantwide industrial process control problems. Computers Chem. Engng, 17, 245-255. Luyben, M. L. and Tyreus, B. D. (1998). An industrial design/control study for the vinyl acetate monomer process. Computers Chem. Engng, 22, 867-877. Luyben. M. L. et al. (1998). An industrial design/control study for the vinyl acetate monomer process, Computers and Chemical Engineering. Vol. 22, No.7-8, pp.867-877. Machida, Y,. Nakaya, M., Omata, T. (2014). Process Education and Development Infrastructure Building by Virtual Plant in DCS Manufacturing Company, The International Federation of Automatic Control, Cape Town (IFAC 2014), pp.12237-12242. Olsen, D. G., Svrcek, W. Y., and Young, B. R. (2005). Plantwide control study of a vinyl acetate monomer process design. Chemical Engineering Communication, 192, 1243-1257. Psaltis, A., Kookos, I. K., and Kravaris, C. (2014). Plantwide control structure selection methodology for the benchmark vinyl acetate monomer plant. Computers Chem. Engng, 62, 108-116. Rangaiah, G. P. and Kariwala, V. (2012). Plantwide control, chapter 9, Wiley. Seki, H., Ogawa, M., Itoh, T., Ootakara, S., Murata, H., Hashimoto, Y., Kano, M. (2010). Plantwide control system design of the benchmark vinyl acetate monomer production plant, Computers Chem. Engng, 34, pp.12821295. Tu, T., Matthew, E., and Chistofides, P. D. (2013). Model predictive control of a nonlinear large-scale process network used in the production of vinyl acetate. Ind. Eng. Chem. Res., 52, 12463-12481. Yumoto, T., Ootakara, S., Seki, H., Hashimoto, Y., Murata, H., Kano, M., Yamashita, Y. (2010). Rigorous Dynamic Simulator for Control Study of the Large-scale Benchmark Chemical Plant, Dynamics and Control of Process Systems (DYCOPS 2010), pp.49-54.
4.3 Production load-up/load-down operation The users can execute load-up/load-down operation from the steady-state. Production load of the VAM process can be increased by raising the O2 feed concentration and the reactor temperature. Load-down operation is totally opposite. After changing operation condition of the reactor, the distillation column condition should be adjusted in conformity to reactor load. This operation scenario will be a good example to investigate the plant-wide control loop configuration. 4.4 Abnormal situation operation The users can start a simulation of the VAM plant model from the steady-state initial condition and experience abnormal situation operation by triggering disturbances and malfunctions at anytime. The users can study dynamic behavior caused by such disturbances and malfunctions, and also they can configure control logic and fault detection logic on the custom calculation blocks which are supported on the VAM plant model. 5. SUPPORTING ENVIRONMENT FOR THE USERS The following elements are available on the Omega Simulation Co., web site for free. -
The VAM plant model with a free limited license of Visual Modeler. - Time-series data sets for data analysis study including fault detection and diagnosis. - Explanation documents to use the VAM plant model, which include process description, operation procedure, system manual, and so on. We are also developing interface which links the VAM plant model and MATLAB®. This interface will be helpful to test control logics developed on MATLAB® as if it is applied to the real plant. 6. CONCLUSIONS A rigorous plant model of vinyl acetate monomer (VAM) process was developed. The users can get feelings of the real plant operation through this model. It can also be used as a test environment for various studies in the process control field, such as plant-wide control, process monitoring, optimal operation, and so on. The developed VAM plant model with a free limited license of Visual Modeler is already available through the web site of Omega Simulation Co., Ltd.: http://www.omegasim.co.jp/contents_e/product/vm/cnt4/ (Official site : http://www.omegasim.co.jp/index_e.htm) 538