Control of a Train of Distillation Columns for the Separation of Natural

Subscriber access provided by RMIT University Library

Article

Control of a Train of Distillation Columns for the Separation of NGL William L. Luyben Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/ie400869v • Publication Date (Web): 05 Jul 2013 Downloaded from http://pubs.acs.org on July 9, 2013

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Paper submitted to Ind. Eng. Chem. Research

Control of a Train of Distillation Columns for the Separation of NGL

William L. Luyben Department of Chemical Engineering Lehigh University Bethlehem, PA 18015 USA

March 22, 2013 Revised May 23, 2013 Revised June 5, 2013 Revised June 25, 2013

[email protected]; 610-758-4256; FAX 610-758-5057 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 36

Abstract The recovery of valuable heavier hydrocarbons from natural gas is accomplished in a series of five distillation columns. The first and second columns are cryogenic highpressure columns in which the methane and ethane are sequentially taken overhead. The third column produces a propane distillate product. The fourth column removes isobutane and normal butane overhead for separation in the fifth column. The bottoms product from the fourth column is a mixture of hydrocarbons that are heavier than the C4 components. This paper studies the dynamic control of this train of distillation columns. Effective control structures are developed and tested on all columns using conventional PI controllers. The product-quality loops in three of the column use tray temperature control, but the first and last column require composition controllers to effectively handle disturbances in feed composition. The difficult separation in the large, high reflux ratio de-isobutanizer is well controlled by an unusual control structure that uses reboiler duty to control reflux-drum level.

1. Introduction Natural-gas processing is becoming more important in chemical engineering because of the increased supply and decreased cost of natural gas that have resulted from improved production methods. New chemical plants are being considered at locations near the Marcellus and Utica natural gas supply areas. The composition of the natural gas varies from source to source, but it mostly methane (85 mol% C1) with small amounts of ethane (C2), propane (C3), isobutane and normal butane (iC4 and nC4) and heavier hydrocarbons down to C7’s. The C2 and heavier components are called natural gas liquid (NGL). They are more valuable than methane, so their recovery is important. It is achieved using a series of distillation columns. The methane is separated out first in a high-pressure (25 atm) cryogenic distillation column using expansion to generate the low temperatures (180 K). A complex configuration of conventional and intermediate reboilers is used to provide vapor boilup and cool the feed. Two process-process heat exchangers, an expander, two flash drums and two throttling valve are used to produce four different feed streams to the column. This column is called a demethanizer. It is designed to achieve two objectives: (1) recover a specified desired percent of the ethane in the feed and (2) keep a low concentration of methane in its bottoms product. It should noted that the demethanizer configuration used in this paper is a new design proposed by Nawaz et al1 that has economic advantages over some existing flowsheets (lower energy consumption). The second column is called a de-ethanizer, and its function is to recover the ethane. It operates at 21 atm and 264 K, using refrigeration in the condenser. It is designed to achieve specified concentrations of propane impurity in the distillate and ethane impurity in the bottoms.

ACS Paragon Plus Environment

2

Page 3 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The third column is a depropanizer whose function is to produce a propane distillate product. It operates at 17 atm with a reflux-drum temperature of 322 K so cooling water can be used in the condenser. It is designed to achieve specified concentrations of isobutane impurity in the distillate and propane impurity in the bottoms. The fourth column is a debutanizer whose function is to take the isobutane and normal butane overhead for further separation in the downstream column. This iC4/nC4 separation is the most difficult (lowest relative volatility) of all the distillations, so it is economical to perform it after all other lighter and heavier components have been removed. The column operates at 7.1 atm with a reflux-drum temperature of 322 K so cooling water can be used in the condenser. It is designed to achieve specified concentrations of isopentane impurity in the distillate and normal butane impurity in the bottoms. The final column is a de-isobuanizer whose function is to produce an isobutane distillate product and a normal butane bottoms. The column operates at 6.6 atm with a reflux-drum temperature of 322 K so cooling water can be used in the condenser. It is designed to achieve specified concentrations of normal butane impurity in the distillate and isobutane impurity in the bottoms. Since distillation is the premier separation method in the chemical and petroleum industries, the subject of distillation control has been extensively studies for many decades. Hundreds of papers and several books have explored many types of columns and many types of control structures. Most of these studies have looked at individual columns in isolation, but multi-column systems have been examined in cases where recycles streams are present that connect the columns. Important examples are in azeotropic distillation system (pressure-swing, extractive and heterogeneous azeotropic units). Multiple columns in a plantwide environment have been explored in a variety of case studies of multi-unit processes with reaction and separations sections. Because of the importance and widespread application of the natural gas NGL separation system, we would expect the control of this system to be extensively studied. Many papers have explored the economic sequencing of the separation from the standpoint of steady-state design. However, the author has been unable to find any studies in the open literature that develop effective control structures for the train of distillation columns. That is the purpose of this paper. The only study (Zu et al2) of this particular sequence of distillation columns found in the literature deals with emergency flaring issues, not normal control structures. Some dynamic control studies of cryogenic distillation systems have been reported. Hanakuma et al3 reported the use of a self-tuning fuzzy control system to control bottoms temperature by manipulate reboiler duty in a demethanizer in an ethylene plant. Dynamic modeling of a somewhat similar cryogenic liquefied natural gas process was reported by Michelsen et al4. The openloop dynamic behavior was explored but no closedloop control studies were reported. Panahi and Skogestad5 discuss control of a natural gas to liquid plant. A control structure for a cryogenic ethylene splitter is described by Song et6. Many studies of the control of cryogenic heat-integrated liquid air columns have appeared7,8. Many papers dealing with individual de-ethanizers, depropanizers and deisobutanizers have appeared9,10,11. These separations are widely used in application examples for various techniques for control, identification and optimization.


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 36

2. Process Studied The numerical example used in this paper uses the natural gas composition reported by Lee et al13 and the demethanizer configuration developed by Nawaz et al1. Figures 1 to 3 give the flowsheets of the five distillation columns in the system with their steady-state design conditions, energy duties and equipment sizes. In the following sections, each of the columns is described in detail. Aspen simulations are used in this study with Peng-Robinson physical properties. Dynamic control studies used the Gear numerical integration algorithm.

2.1 Demethanizer The design of the demethanizer is based on the following specifications and assumptions: 1. The recovery of C2 in the column bottoms NGL stream is set at 80 %. This achieved by varying the exit temperature from the externally cooled heat exchanger. Lower temperatures increase C2 recovery but increase refrigeration duty. 2. The composition of the bottoms stream is specified to produce a ratio of C1 to the sum of C1 and C2 of 0.02. Thus the bottoms composition on a C3-and-heavierfree basis is 2 mol% C1. 3. The column has 31 stages with four feed streams fed on different stages where stream and stage C1 compositions are similar. 4. The two side reboilers have liquid pumparound flowrates of 2000 kmol/h and are located at Stages 24 and 28. 5. The pressure at the top of the column is 25 atm and stage pressure drop is 0.01 atm per tray. 6. Minimum approach temperature differentials at either the hot or the cold end of the heat exchangers are 2 K. Plate heat exchangers are used because of the large area requirements. 7. The assumed overall hear-transfer coefficients for sizing gas-gas systems and gasliquid systems are 0.17 kW K-1 m-2 and 0.28 kW K-1 m-2, respectively. The natural-gas feed flowrate is 5269 kmol/h with a composition of 86.4 mol% C1 and 47.79 mol% C2. There is a small amount of nitrogen (1.54 mol%). The remaining components run from C3 (2.87 mol%) down to a small amount of nC7. See Figure 1 for a detailed list of the feed composition. About 10% of the feed gas leaves the bottom of the demethanizer as valuable NGL. The supply pressure is 59 atm, and a small feed compressor boosts the pressure to 61 atm. The gas feed stream at 313 K is hotter than the temperatures in the stripping section of the column, so the use of intermediate (side) reboilers is attractive. Side reboilers have two advantages. They reduce energy consumption in the steam-heated


4

Page 5 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


reboiler at the base of the column, and they cool the feed so that less refrigeration is required. Two side reboilers are used. The first takes a large pumparound liquid stream (2000 kmol/h) from Stage 28 at 303.5 K and heats it to 306.4 K before returning it back to the column. The heat transferred in this heat exchanger is 538.3 kW, requiring 505 m2 of area assuming an overall heat-transfer coefficient of 0.28 kW K-1 m-2 in this gas-liquid system and a minimum differential approach temperature of 2 K. The feed gas is cooled from 313 K to 305.4 K and flows to the second side reboiler located higher in the column where the tray temperature is lower. The second side reboiler takes a second large pumparound liquid stream (2000 kmol/h) from Stage 24 at 287.9 K and heats it to 297.0 K. The heat transferred in this heat exchanger is 1228 kW, requiring 995 m2 of area. The feed gas is cooled from 305.4 K to 289.8 K. As shown in Figure 1, a small amount of the ambient-temperature feed gas (57 kmol/h) is bypassed around the two side reboilers. As discussed later in this paper, the side reboilers produce too much vapor boilup under some conditions (varying feed composition), and the duty of the steam heated reboiler would go to zero unless some bypassing is used. A control structure is developed that uses split-ranged valves to manipulate the two variables (reboiler steam and bypassing). The mixed main and bypass streams flow to the next heat exchanger in which the cold methane product stream is used to further cool the feed. The cold stream to this feed precooler is at 245.3 K with a flowrate of 4762 kmol/h. This is the gas product with a composition of 1.70 mol% nitrogen, 97.14 mol% C1, 1.11 mol%C2 and a trace of C3. This stream is heated to 288.0 K by the feed stream, which is cooled from 290.0 K to 266.3 K. The heat-transfer rate is 2156 kW, requiring 1565 m2 for an overall heat-transfer coefficient of 0.17 kW K-1 m-2 in this gas-gas system. The feed then goes to a cooler that uses an external refrigerant to further cool the feed to 252.0 K, requiring 1500 kW of refrigeration duty. At 250.0 K and 60 atm, some liquid is formed. The feed stream goes to the first flash drum. The liquid phase at 294.8 kmol/h is flashed through Valve V1 to near the operating pressure of the column (25 atm) and fed onto Stage 18 of the 31-stage column. The temperature of the stream after the flash is 237.2 K with a vapor fraction of 0.2855. The gas phase from the first flash drum is split into two streams. One is fed to an expander (3237 kmol/h) that produces 877.1 kW of power, which is used to help drive the main product compressor. The temperature of the expander exit is 212.9 K with a vapor fraction of 0.952. The stream leaving the expander goes to a second flash drum. The vapor from this drum (3079 kmol/h) is fed on Stage 5. The liquid (154.7 kmol/h) is fed on Stage 10. The other portion of the gas stream from the first flash drum (1740 kmol/h) goes to a heat exchanger in which it is cooled and condensed at 181.8 K by the cold overhead vapor stream from the top of the column at 179.8 K. The heat exchanger transfers 3679 kW with an area of 5876 m2 assuming an overall heat-transfer coefficient of 0.17 kW K-1 m-2 in this gas-gas system. The feed stream leaving this heat exchanger is all liquid. It flashes through Valve V2 from 60 atm to 25 atm, which drops the temperature to 174.3 K and produces a vapor fraction of 0.1015.


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 36

At the base of the column a steam-heated reboiler has a heat duty of 333.4 kW. The base temperature is 308.2 K, so a low-temperature heat source can be used. Notice that the heat added in the steam-heated reboiler is only 16% of the total heat added to the column in the three reboilers. The effect of this on dynamics and control are discussed in a later section of this paper. Column diameter is 1.50 m. An important design optimization variable in the demethanizer column is how the vapor stream leaving the first flash tank is split. Some is sent to the expander and some is sent to the heat exchanger to be cooled by the overhead vapor. In the flowsheet given in Figure 1, the stream fed to the expander is 65% of the gas stream. The optimum split is determined by looking at the effects of the split on the recovered expander power, the reboiler duty, the refrigerated cooler duty and the NGL production rate. The C2 recovery is kept constant by adjusting the refrigerated cooler temperature, and the composition of the NGL is kept constant at 2 mol% C1 on a C3-and-heavier-free basis. The minimum in the net energy (cooler + reboiler – expander) occurs at 65%, so this is selected as the operating point. The demethanizer is clearly the most complex of the columns. In the sections below, the designs of each of the four subsequent columns are discussed.

2.2 De-Ethanizer The column on the left in Figure 2 gives the flowsheet for the de-ethanizer column. It operates at 21 atm so as to be well under the critical pressure of ethane (48 atm) and avoid liquid/vapor hydraulic problems. At this pressure, the reflux-drum temperature is 264 K, so refrigeration is required (1.734 MW). The column base temperature is 363 K, so a low-temperature heat source is required (2.108 MW). The C2/C3 separation is not difficult, and a column with 31 stages is used with feed introduced onto Stage 15, the feed-tray location that minimizes the refrigerated condenser duty. The design specifications are 1 mol% C3 in the distillate and 0.34 mol% C2 in the bottoms. The required reflux ratio is 1.258. The purity of the ethane product D2 is 97.06 mol%. Column diameter is 1.23 m.

2.3 Depropanizer The column on the right in Figure 2 gives the flowsheet for the depropanizer column. It operates at 17 atm so that a reflux-drum temperature of 322 K is achieved, which permits the use of cooling water in the condenser. The C3/C4 separation is more difficult, so a column with 51 stages is used. The feed tray location that minimizes reboiler duty is Stage 24. The design specifications are 0.6 mol% C4 (sum of iC4 and nC4) in the distillate and 0.1 mol% C3 in the bottoms. The required reflux ratio is 2.114 and the reboiler duty is 1.656 MW of low-pressure steam. Column diameter is 2.114 m.

2.4 Debutanizer The column on the left in Figure 3 gives the flowsheet for the debutanizer column whose function is to take the iC4 and nC4 components overhead for subsequent separation in the downstream column. Because the iC4/nC4 separation is quite difficult (1.3 relative volatility at 322 K), conventional distillation wisdom recommends that this


6

Page 7 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


separation be performed when the mixture is essentially binary so as to minimize energy requirements. The separation in the debutanizer is between the nC4 and the iC5 and is fairly easy with a relative volatility of 2.2 at 322 K. A column with 31 stages is used, operating at 7.1 atm to give a reflux-drum temperature of 322 K. The feed tray location that minimizes reboiler duty is Stage 16. The design specifications are 0.2 mol% iC5 in the distillate and 0.2 mol% nC4 in the bottoms. The required reflux ratio is 2.20 and the reboiler duty is 1.02 MW of lowpressure steam. Column diameter is 0.845 m.

2.5 De-Isobutanizer The column on the right in Figure 3 gives the flowsheet for the de-isobutanizer column whose function is to make the difficult separation between iC4 and nC4. A column with 81 stages is used, operating at 6.6 atm to give a reflux-drum temperature of 322 K. The feed tray location that minimizes reboiler duty is Stage 39. The design specifications are 2 mol% nC4 in the distillate and 2 mol% iC4 in the bottoms. A high reflux ratio of 7.94 is required. Reboiler duty is 1.62 MW of lowpressure steam. Column diameter is 0.990 m.

3. Control Structure Development The development of control structures for each of the five distillation columns is discussed in the following sections. All liquid holdups are sized to provide 5 minute of holdup when half full. All level controllers are proportional with KC = 2 except in the deisobutanizer, as discussed later.

3.1 Demethanizer Control Structure A recent paper14 studied the demethanizer control issues, and the proposed control scheme is shown in Figure 4. The specific loops are described below. 1. Feed flowrate is controlled by manipulating the power to the feed compressor. 2. Column pressure is controlled by manipulating power to the gas product compressor, which discharges into a 60 atm pipeline. 3. The temperature in the first flash drum is controlled by manipulating the refrigerant duty in the cooler. 4. The pressure in the first flash drum is controlled by the Valve V2 in the reflux line to the column. 5. The level in the first flash drum is controlled by manipulating valve in the liquid line from the drum. 6. The flowrate through the expander is flow controlled by manipulating speed and is ratioed to the reflux flowrate. The setpoint of the flow controller comes from a multiplier whose two inputs are the reflux flowrate and a constant 1.858 (the steady-state design ratio to keep the vapor split at 65% through the expander). 7. The level in the second flash drum is controlled by manipulating the valve in the liquid line leaving the drum.


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 36

8. Liquid level in the bottom of the column is controlled by manipulating the bottoms NGL product. 9. The valves in the side-reboiler pumparound streams are held in fixed positions. 10. The methane impurity composition of the bottoms is controlled by manipulating the reboiler duty and the bypass flowrate, using a split-range valve configuration. When the duty of the steam-heated reboiler is reduced to a low level, the flowrate of the feed gas bypassed around the side reboilers is increased. The first flash-drum temperature controller has 1-minute deadtimes and is tuned by running relay-feedback tests and using Tyreus-Luyben settings. Its tuning constants are KC = 3.51 and τI = 9.2 min with a 100 K temperature transmitter span and a maximum heat removal of 715,500 cal/s. A 3-minute deadtime is used in the composition controller (CC1), which is similarly tuned giving KC = 0.785 and τI = 41 min with a 5 mol% C1 composition transmitter span. The use of tray temperature controllers instead of a bottoms composition controller was explored in the previous paper but found to be ineffective for maintaining bottoms composition, particularly for disturbances in feed composition.

3.2 De-Ethanizer Control Structure The column is separating C2 from C3. The possibility of using a simple singleend tray-temperature control structure is explored by first looking at the temperature and composition profiles. The C2 and C3 compositions are changing fairly rapidly around Stage 23, where the temperature profile is also changing. So Stage 23 is selected to test. If a single-end control structure is to be effective, the remaining control degree of freedom must be fixed in an appropriate manner. The use of feed composition sensitivity analysis provides guidance in this determination. Using steady-state Aspen Plus simulation, the C2 and C3 compositions of the feed to the column are changed from their design values while maintaining the specified key-component impurity levels in the distillate and bottoms by using two Aspen Design spec/vary functions. The upper part of Table 1 gives the results of these calculations for the de-ethanizer column. The percent variation of the required reflux flowrate and reflux ratio over the range of feed compositions is shown to be 8.1 % and 29.0 %, respectively. This indicates that a singleend control structure with a fixed reflux-to-feed ratio should be better at handling feed composition changes than a fixed reflux ratio structure. The control structure for the de-ethanizer is shown on the left side of in Figure 5 and has the following loops: 1. Feed comes in on level control from the upstream column. 2. Pressure is controlled by manipulating refrigeration duty in the condenser. 3. Reflux is ratioed to feed. 4. Reflux-drum level is controlled by manipulating the flowrate of the distillate product. 5. Base level is controlled by manipulating the flowrate of the bottoms. 6. Stage 23 temperature is controlled by reboiler duty. The controller has a 1-minute deadtime and is tuned by a running relay-feedback test and using Tyreus-Luyben settings. Tuning constants are KC = 0.59 and τI = 16 min with a 100 K temperature transmitter span and a maximum heat removal of 1,010,000 cal/s.


8

Page 9 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


A variety of disturbances are introduced into the process to test the effectiveness of this single-end tray-temperature control structure, and results are presented in a later section of this paper.

3.3 Depropanizer Control Structure The key components in this column are C3 and iC4. The possibility of using a simple single-end tray-temperature control structure is explored by first looking at the temperature and composition profiles. The C3 and iC4 compositions are changing fairly rapidly around Stage 33, where the temperature profile is also changing. So Stage 33 is selected to test. Feed composition sensitivity analysis is used to see if a single-end control structure could be effective. Using steady-state Aspen Plus simulation, the C3 and iC4 compositions of the feed to the column are changed from their design values while maintaining the specified key-component impurity levels in the distillate and bottoms by using two Aspen Design spec/vary functions. The second part of Table 1 gives the results of these calculations for the depropanizer column. The percent variation of the required reflux flowrate and reflux ratio over the range of feed compositions is shown to be 8.9 % and 20.4 %, respectively. This indicates that a single-end control structure with a fixed reflux-to-feed ratio should be better at handling feed composition changes that a fixed reflux ratio structure. The control structure for the Depropanizer is shown on the right side of Figure 5 and has the following loops: 1. Feed comes in on level control from the upstream column. 2. Pressure is controlled by manipulating condenser duty. This is shown as manipulating cooling water but other alternative setups are widely used such as flooded condensers, hot-vapor bypass and vent-bleed systems. 3. Reflux is ratioed to feed. 4. Reflux-drum level is controlled by manipulating the flowrate of the distillate product. 5. Base level is controlled by manipulating the flowrate of the bottoms. 6. Stage 33 temperature is controlled by reboiler duty. The controller has a 1-minute deadtime and is tuned by a running relay-feedback test and using Tyreus-Luyben settings. Tuning constants are KC = 0.89 and τI = 13 min with a 100 K temperature transmitter span and a maximum heat removal of 798,500 cal/s. Evaluation of the effectiveness of this single-end tray-temperature control structure is presented in a later section of this paper.

3.4 Debutanizer Control Structure The key components in this column are nC4 and iC5. The possibility of using a simple single-end tray-temperature control structure is explored by first looking at the temperature and composition profiles. The nC4 and iC5 compositions are changing fairly rapidly around Stage 19, where the temperature profile is also changing. So Stage 19 is selected to test.


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 36

Feed composition sensitivity analysis is used to see if a single-end control structure could be effective. Using steady-state Aspen Plus simulation, the nC4 and iC5 compositions of the feed to the column are changed from their design values while maintaining the specified key-component impurity levels in the distillate and bottoms by using two Aspen Design spec/vary functions. The third part of Table 1 gives the results of these calculations for the debutanizer column. The percent variation of the required reflux flowrate and reflux ratio over the range of feed compositions is shown to be 11.5 % and 29.5 %, respectively. This indicates that a single-end control structure with a fixed refluxto-feed ratio should be better at handling feed composition changes that a fixed reflux ratio structure. The control structure for the debutanizer is shown on the left side of Figure 6 and has the following loops: 1. Feed comes in on level control from the upstream column. 2. Pressure is controlled by manipulating condenser duty. 3. Reflux is ratioed to feed. 4. Reflux-drum level is controlled by manipulating the flowrate of the distillate product. 5. Base level is controlled by manipulating the flowrate of the bottoms. 6. Stage 19 temperature is controlled by reboiler duty. The controller has a 1-minute deadtime and is tuned by running a relay-feedback test and using Tyreus-Luyben settings. Tuning constants are KC = 0.77 and τI = 16 min with a 100 K temperature transmitter span and a maximum heat removal of 488,000 cal/s. Evaluation of the effectiveness of this single-end tray-temperature control structure is presented in a later section of this paper.

3.5 De-Isobutanizer Control Structure This column represents the most difficult separation and the most challenging control problem. The reflux ratio in this superfractionator is large, so conventional distillation control wisdom recommends controlling reflux-drum level using reflux. The temperature profile is flat because the changes in compositions from tray to tray are very small. So the use of temperatures to infer compositions is not going to be effective. The key components in this column are iC4 and nC4. Feed composition sensitivity analysis is used to see if a single-end composition control structure could be effective or if a dual-composition control scheme is required. Using steady-state Aspen Plus simulation, the iC4 and nC4 compositions of the feed to the column are changed from their design values while maintaining the specified key-component impurity levels in the distillate and bottoms by using two Aspen Design spec/vary functions. The bottom part of Table 1 gives the results of these calculations for the de-isobutanizer column. The percent variation of the required reflux flowrate and reflux ratio over the range of feed compositions is shown to be 0.8 % and 19 %, respectively. This indicates that a singleend composition control structure with a fixed reflux-to-feed ratio should be significantly better at handling feed composition changes that a fixed reflux ratio structure. It is important to note that there are two competing rules from distillation control wisdom. The first suggests that the reflux-drum level should be controlled by reflux because of the very high reflux ratio. The second suggests that the reflux should be


10

Page 11 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ratioed to the feed. It appears that we cannot follow both of these rules. However, there is a control structure that permits both. The control structure for the debutanizer is shown on the right side of Figure 6 and has the following loops: 1. Feed comes in on level control from the reflux drum of the upstream column. 2. Pressure is controlled by manipulating condenser duty. 3. Reflux is ratioed to feed. 4. Reflux-drum level is controlled by manipulating reboiler duty using a proportional controller with a gain of 10. 5. Base level is controlled by manipulating the flowrate of the bottoms. 6. Distillate nC4 impurity is controlled by manipulating the control valve in the distillate line. The controller has a 3-minute deadtime and is tuned by running a relay-feedback test and using Tyreus-Luyben settings. Tuning constants are KC = 1.6 and τI = 50 min with a 5 mol% composition transmitter span. This control structure is unusual because of the use of reboiler heat input to control reflux-drum level. Since the reflux ratio is large, the reflux-drum level would normally be controlled by manipulating the reflux flowrate. However, the reflux-to-feed control structure is indicated by the feed composition sensitivity analysis to require very small changes in this ratio, so it is logical to fix this ratio. However, some other effective variable must be selected to control the liquid level in the reflux drum. Using distillate would be ineffective because its flowrate is much smaller than the reflux. Using reboiler duty works well for this job because the changes in vapor flowrates in the column are quite rapid. The distillate composition is controlled by the flowrate of distillate. A gain of 10 is used in the reflux-drum level controller to give faster response of vapor boilup to level changes since the changes in distillate flowrate need to be quickly detected by the level controller, which then changes the vapor rate up the column and affects distillate composition. Notice that the pressure controller manipulating condenser duty is “nested” inside the composition control loop. The pressure controller must be on automatic for the composition controller to work. Evaluation of the effectiveness of this single-end composition control structure is presented in the next section of this paper.

4. Dynamic Performance The effectiveness of the control structures for the five columns is tested by making large changes in throughput and in feed composition..

4.1 Feed Flowrate Disturbances Figure 7 gives dynamic responses for 20 % disturbances in the setpoint of the feed gas flow controller at time equal 0.5 hours. The solid lines are for 20% increases, and the dashed lines are for 20% decreases. Figure 7A shows how all the product compositions leaving the process respond to these large disturbances. The upper left graph shows how the C1 composition yD1(C1) of the product gas stream going overhead in the demethanizer


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 36

responds. This composition is not controlled, but it changes very little from the normal 96.76 mol% C1. The upper right graph shows that the composition of the ethane product xD2(C2) also stays very close to is normal 97.06 mol% C2. Remember that this composition is not directly controlled, but Stage 23 temperature is controlled and a reflux-to-feed ratio is maintained. This structure works well for throughput disturbances. The middle left graph in Figure 7A shows that the composition of the propane product xD3(C3) is also held quite close to the design value of 98.75 mol%C3 when using a Stage-33 temperature controller and a reflux-to-feed ratio control structure. The middle right graphs shows that the losses of nC4 in the bottoms from the debutanizer xB4(nC4) remain small. Note the composition is given in mol% for this impurity. The bottoms two graphs in Figure 7A show the compositions of the two product streams leaving the de-isobutanizer. The distillate purity xD5(iC4) is not directly controlled at 97.6 mol% iC4, but the impurity of nC4 is directly controlled by manipulate distillate flowrate. Reflux-drum level is controlled by manipulating reboiler duty and a reflux-tofeed ratio is maintained. Distillate xD5(iC4) purity stays very close to it specification. The bottoms purity xB5(nC4) is not directly controlled, but it is maintained fairly close to the normal 97.59 mol% nC4. Figures 7B and 7C show how the three temperature controllers in the de-ethanizer, depropanizer and debutanizer columns respond by manipulating reboiler duties. Also shown are the two composition loops. One in the demethanizer controls bottoms C1 impurity xB1(C1) by manipulating both reboiler duty and bypass flow (shown in the bottom left graph in Figure 7C).. The second composition controller is in the de-isobutanizer, which maintains nC4 impurity in the distillate xD4(nC4) by manipulating distillate flowrate These results demonstrate that the proposed control structure handles large throughput disturbances quite well.

4.2 Feed Composition C2 Disturbances Figure 8 gives results when the composition of the feed gas is changed in terms of C2. The solid lines are when C2 feed composition is increased from the design 6.47 to 8.47 mol% C2. The composition of the C1 in the feed is corresponding reduced. The dashed lines are when C2 feed composition is decreased from the design 6.47 to 4.47 mol% C2. The composition of the C1 in the feed is corresponding increased. Note that these are large disturbance on a relative basis (31 %). Figure 8A shows that there is a decrease in the C1 composition of the gas product from the demethanizer when there is more C2 and less C1 in the feed. The purity of the ethane product and the purity of the propane product shift up or down slightly when there is more or less C2 in the feed. The other product compositions are essentially unaffected. Higher C2 feed concentration results in a lower heat input to the demethanizer reboiler (Figure 8B) and more bypassing (Figure 8C). Reboiler duty in the ethanizer increases.

4.3 Feed Composition C3 Disturbances Figure 9 gives results when the composition of the feed gas is changed in terms of C3. The solid lines are when the C3 feed composition is increased from 2.87 to 3.37 mol% C3 with a corresponding reduction in the C2 feed composition from 6.47 to 5.97


12

Page 13 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


mol% C2. The dashed lines are when the C3 feed composition is decreased from 2.87 to 2.37 mol% C3 with a corresponding increase in the C2 feed composition from 6.47 to 6.97 mol% C2. Note that these are large disturbance on a relative basis (17 %). Figure 9A shows that there are quite small shifts in the product composition. Figure 9B shows that more energy is required in the depropanizer reboiler and less in the ethanizer reboiler when C3 feed composition increases. Reboiler duty in the demethanizer is higher and bypassing is lower (Figure 9C).

4.4 Feed Composition C4 Disturbances Figure 10 gives results when the composition of the feed gas is changed in terms of iC4 and nC4. The solid lines are when the nC4 feed composition is increased from 0.82 to 1.02 mol% nC4 with a corresponding reduction in the iC4 feed composition from 0.72 to 0.52 mol% iC4. The dashed lines are when the nC4 feed composition is decreased from 0.82 to 0.62 mol% nC4 with a corresponding increase in the iC4 feed composition from 0.72 to 0.92 mol% C2. Note that these are large 27% disturbances on a relative basis. Figure 10A shows that the purities of all product streams are maintained quite close to their specifications. Heat duties in all columns except the de-isobutanizer are little affected (Figure 10B). In the de-isobutanizer, reboiler duty and distillate flowrate decrease when there is more nC4 and less iC4 in the feed. All of these disturbances are effectively handled by the control structures used on the five columns. Stable regulatory plantwide control is achieved with product compositions maintained close to their specifications.

4.5 Alternative DIB Control Structure Since the control structure on the DIB is somewhat unusual (reflux-drum level control using reboiler duty), it might be useful to compare its performance with a conventional control structure [L,B] recommended by Horowitz et al15. The [L,B] structure is a dual-composition control structure in which distillate impurity xD5(nC4) is controlled by manipulating reflux flowrate and bottoms impurity xB5(iC4) is controlled by manipulating reboiler duty. Reflux-drum level is controlled by manipulating distillate flowrate, and base level is controlled by manipulating reboiler duty. Note that the reflux is ratioed to the feed with the distillate composition controller changing the ratio. The two composition loops contained 3-minute deadtimes and were tuned sequentially using relay-feedback tests and Tyreus-Luyben tuning rules. The resulting integral times are very large (118 minutes in the xD5 loop and 100 minutes in the xB5 loop). Remember that only one composition controller is used in the proposed structure. Figure 11 gives a comparison of the two alternative control structures for a 20% increase in feed flowrate to the demethanizer. The LB results show larger peak transients in the purities of the two product streams from the DIB because of the large integral times in the two composition controllers and the interaction between the two composition loops.


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 36

5. Conclusion A control structure for the 5-column distillation train in a natural gas NGL refining process has been developed and tested. Effective product-quality control is achieved for the six product streams. Conventional PI controllers provide stable regulatory dynamic responses. Temperature controllers in three of the columns provide effective inferential composition control. Direct composition measurement and control are required in the demethanizer and the de-isobutanizer. The difficult separation in the large, high reflux ratio de-isobutanizer is well controlled by an unusual control structure that used reboiler duty to control reflux-drum level. Some general comments about designing control structure for trains of distillation columns may be useful. In this process, all the columns operate in series with the flows moving downstream from column to column. Therefore each column can be studied in isolation in terms of developing an appropriate control structure and finding stable tuning parameters. Life gets much more difficult when there are recycle streams from downstream column back upstream. A good example is the use of a heavy naphtha absorber to recover propane and heavier components from atmospheric crude unit distillate product while rejecting ethane. In this process, a portion of the heavy naphtha bottoms from the second column is recycled back to the first column. The control structure must be developed considering both columns.

References 1. Nawaz, M., Jobson, M., Finn, A. “Synthesis and optimization of demethanizer flowsheets for NGL recovery” Spring 2012 AIChE meeting, Houston, TX. 2. Zu, Q., Yang, X., Liu, C., Li, K., Lou, H., Gossage, J. L. “Chemical plant flare minimization via plantwide dynamic simulation Eng Chem Res. 48, 2009, 35053512. 3. Hanakuma, Y., Irizuki, Y.,Adachi, M., Nakanishi, E. “Design of a self-tuning fuzzy control system and application to a distillation column” International Chem. Eng. 34, 1994, 91-96. 4. Michelsen, F. A., Halvorsen, I. J., Lund, B. F., Wahl, P. E. “Modeling and simulation for control of the TEALARC liquified natural gas process” Ind. Eng. Chem. Res. 49, 2010, 7389-7397. 5. Panahi, M., Skogestad, S. “Selection of controlled variables in a natural gas to liquid process” Ind. Eng. Chem. Res. 51, 2012, 10179-10190. 6. Song, G., Qiu, T., Zhao, J. “A flare minimization model for an ethylene splitter system’s shutdown” submitted to Energy, January 2013. 7. Blan, S., Khowihij, S., Henson, M. A., Belanger, P., Megan, L. “Compartmental modeling of high purity air separation columns” Comp. Chem. Eng. 29, 2005, 2096-2109. 8. Vinson, D. R. “Air separation control technology” Comp. Chem. Eng. 30, 2006, 1436-1446.


14

Page 15 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


9. Huang, H., Riggs, J. P. “Including levels in MPC to improve distillation control” Ind. Eng Chem Res. 41, 2002, 4048-4053. 10. Wang, K., Sltao, Z., Zhang, Z., Chen, Z., Fang, X., Zhou, Z., Chen, X., Qian, J. “Convergence depth control for process system optimization” Eng Chem Res. 46, 2007, 7729-7738. 11. Luyben, W. L. “Design and control of an autorefrigerated alkylation process” Ind. Eng Chem Res. 48, 2009, 11081-11093. 12. Kaniko, H., Funatso, K. “Discussion on time difference models and intervals of time difference for application of soft sensors” Eng Chem Res. 52, 2013, 13221324.. 13. Lee, S., Long, N. V. D., Lee, M. “Design and optimization of natural gas liquefaction and recovery processes for offshore floating liquefied natural gas plants” Ind. Eng Chem Res. 51, 2012, 10021-10030. 14. Luyben, W. L. “NGL demethanizer control” Ind. Eng Chem Res., submitted Feb. 8, 2013. 15. Hurowitz, S. E., Anderson, J., Duval, M., Riggs, J.B. “Distillation control configuration selection” J. Proc. Con, 13, 2003, 357-362.


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 36

Table 1 – Feed Composition Sensitivity Analysis De-Ethanizer Feed Ethane (mole fraction) 0.5278 Design

Propane (mole fraction) 0.2111

Reflux (kmol/h)

Reflux Ratio

341.7

1.114

0.4778

0.2611

357.7

1.258

0.4278

0.3111

370.8

1.487

8.12

29.0

% Change over range

Depropanizer

Design

Feed Propane (mole fraction) 0.4934

Feed iC4 (mole fraction) 0.1546

329.3

2.210

0.5234

0.1246

315.6

2.114

0.5534

0.0946

301.1

1.803

8.90

20.4

% Change over range

Debutanizer Feed nC4 (mole fraction) 0.2526


182.0

2.538

0.3027

0.1517

173.6

2.220

0.3526

0.1017

162.0

1.888

11.5

29.5

% Change over range

De-Isobutanizer

% Change over range


Feed nC4 (mole fraction) 0.4954

286.6

7.240

0.4605

0.5354

288.0

7.940

0.4205

0.5754

289.0

8.751

0.80

19.0


16

Page 17 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure Captions Figure 1 – Demethanizer flowsheet Figure 2 – De-ethanizer and depropanizer flowsheets Figure 3 – Debutanizer and de-isobutanizer flowsheets Figure 4 - Demethanizer control structure Figure 5 – De-ethanizer and depropanizer control structures Figure 6 - Debutanizer and de-isobutanizer control structures Figure 7 – Feed flowrate disturbances Figure 8 – Feed composition C2 disturbances Figure 9 – Feed composition C3 disturbances Figure 10 – Feed composition C4 disturbances Figure 11 – Comparison of proposed and LB control structures: +20% feed flowrate


17

Industrial & Engineering Chemistry Research Fig. 1 – Demethanizer Flowsheet

Page 18 of 36

245.3 K

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

25 atm 179.8 K

3679 kW 5876 m2 0.17 kW K-1 m-2

1740 kmol/h 181.8 K

V2

1

3234 kmol/h 4974 kmol/h 2158 kW 1565 m2 0.17 kW K-1 m-2

25.1 atm 212.9 K 3079 kmol/h 877.1 kW

266.3 K

T1

5 10

60 atm 251.4 K

154.7 kmol/h

1500 kW

290.0 K

24.6 atm 288.0 K

Tcool=252 K

18

294.8 kmol/h V1 289.8 K

1228 kW 995 m2 0.28 kW K-1 m-2 3855 kW

Gas 4699 kmol/h 60 atm, 362 K 0.0173 N2 0.9676 C1 0.0146 C2 0.0005 C3

Fby = 57 kmol/h

Feed

538.3 kW 505 m2 0.28 kW K-1 m-2

ID = 1.50 m 2000 kmol/h

24

297.0 K

287.9 K 305.4 K 2000 kmol/h

28

306.4 K 61 atm 313 K

303.5 K

NGL 569.9 kmol/h

5269 kmol/h 25.1 atm, 308.2 K 59 atm, 310 K 134.1 kW 0.0096 C1; 0.4779 C2 0.0154 N2; 0.8640 C1 0.2613 C3; 0.0664 iC4 0.0647 C2; 0.0287 C3 0.0757 nC4; 0.0379 iC5 0.0072 iC4; 0.0082 nC4 0.0287 nC5; 0.0287 nC6 ACS Paragon Plus Environment 0.0041 iC5; 0.0031 nC5 0.0139 nC7 0.0031 nC6; 0.0015 nC7

30

333.4 kW

& Engineering Flowsheets Chemistry Research Figure 2 – De-Ethanizer and Industrial Depropanizer

Page 19 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

1.734 MW (Refrigeration)

1.611 MW

B1 569.9 kmol/h 0.0096 C1 0.4779 C2 0.2613 C3 0.0664 iC4 0.0757 nC4 0.0379 iC5 0.0287 nC5 0.0287 nC6 0.0139 nC7

RR=1.258 ID=1.23 m

21.3 atm 363.1 K

17 atm 322 K

21 atm 264 K

T2

T3 24

15

D2 279.6 kmol/h 30

D3 147.9 kmol/h

RR=2.114 ID=0.970 m

0.0196 C1 0.9706 C2 0.0097 C3

50

1.656 MW

B3 142.4 kmol/h

2.108 MW

B2 290.3 kmol/h

17.5 atm 398.4 K


0.0034 C2

0.0066 C2 0.9875 C3 0.0057 iC4 0.0002 nC4

0.0010 C3 0.2597 iC4 0.3027 nC4 0.1517 iC5 0.1147 nC5 0.1147 nC6 0.0555 nC7

Industrial & Engineering Chemistry Research Figure 3 – Debutanizer and De-Isobutanizer Flowsheets 1 2 3 4 5 6 B3 7 8 142.4 kmol/h 9 0.0010 C3 10 110.2597 iC4 120.3027 nC4 13 140.1517 iC5 150.1147 nC5 16 0.1147 nC6 17 180.0555 nC7 19 20 21 22 23 RR=2.22 24 25 ID=0.8448 m 26 27 28 29 30 7.40 atm 31 394.6 K 32 33 34 35 36 37 38 39 40 41 42

Page 20 of 36

1.622 MW 1.301 MW

6.6 atm 322 K 7.1 atm 322 K

D4 80.30 kmol/h

T4

0.0018 C3 0.4605 iC4 0.5354 nC4 0.0020 iC5 0.0002 nC5

16

T5

39

D5 37.00 kmol/h RR=7.94 ID=0.990 m

30

0.0040 C3 0.9760 iC4 0.0200 nC4

1.02 MW 80

B4 62.12 kmol/h 0.0001 iC4 0.0020 nC4 0.3451 iC5 0.2627 nC5 7.399 atm 0.2629 nC6 ACS Paragon Plus Environment 339.7 K 0.1272 nC7

1.620 MW

B5 43.28 kmol/h 0.0200 iC4 0.9759 nC4 0.0038 iC5 0.0003 nC5

Fig. 4 – Demethanizer Control Structure Industrial & Engineering Chemistry Research

Page 21 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

PC SP

x

FT

1

FC Refrig (233 K)

PC

V2

5 10

TC LC LC

V1

18

24 26 28

Gas

30 FC

NGL

Feed ACS Paragon Plus Environment

CC

LC

& Engineering Control Chemistry Research Figure 5 – De-Ethanizer and Industrial Depropanizer Structures 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

PC

PC

x

x

SP

SP

FC

FC

T2

LC LC

T3

LC FT

24

FT

15

Page 22 of 36

33 23 TC 50

30 LC

LC


TC

Page 23 of 36


Figure 6 – Debutanizer and De-Isobutanizer Control Structures 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 LC 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

PC

x

SP

PC

x

FC LC

T5

SP

FC

T4

LC

CC

16

FT

39

FT 19

TC

30 80

LC LC



Page 24 of 36

Fig. 7A – Feed Flowrate Disturbances; Product Compositions Feed Flowrate Disturbances 0.98

xD2(C2)

yD1(C1)

0.97

0.968 0.966 0

+ΔF 2

4

6

0.98 0

2

6

0.98

0.975 2

4

6

8

+ΔF 2

4

6

8

2

4

6

8

2

4

6

8

1 0.5 0

8

0.985

0.97 0

0.97

xB4(%nC4) 4

xD5(iC4)

xD3(C3)

+ΔF

0.985

0.975

0.965 0

8

0.99

xB5(nC4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

0.985 0.98

0.975 0.97 0

Time (h)

Time (h) ACS Paragon Plus Environment

Page 25 of 36


Fig. 7B – Feed Flowrate Disturbances; DM, DE and DP Loop Variables

1.4

QR1 (kW)

xB1(%C1)

Feed Flowrate Disturbances 1.2 1 0.8

0

2

4

6

QR2 (MW)

T23 (K)

360 355 0

2

4

6

QR3 (MW)

330

325 320 0

2

4

6

8

+ΔF

300 200 2

4

6

8

4

6

8

4

6

8

3

+ΔF 2.5 2 1.5

8

335

400

100 0

8

365

T33 (K)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

2

2

+ΔF

1.5 1

0

Time (h)

2



Page 26 of 36

Fig. 7C – Feed Flowrate Disturbances; DB and DIB Loop Variables Feed Flowrate Disturbances

360 2

4

6

2.5 2 1.5

0

2

4

6

50 0

0

+ΔF 2

4

6

8

+ΔF

1

0

2

4

6

8

4

6

8

4

6

8

50

+ΔF

40 30 20 0

8

100

1.5

0.5

8

D5 (kmol/h)

350 0

xD5(%nC4)

QR4 (MW)

370

QR5 (MW)

T19 (K)

380

Fby (kmol/h)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

2

2

+ΔF 1.5 1

0

Time (h)

2


Page 27 of 36


Fig. 8A – Feed C2 Composition Disturbances; Product Compositions Feed C2 Composition Disturbances

xD2(C2)

0.99

0.96

+ΔC2 0.94 0

2

4

6

xD3(C3)

+ΔC2 0.985

2

4

6

xD5(iC4)

0.9765 0.976 0.9755 0.975 0

2

4

6

8

2

4

6

8

2

4

6

8

2

4

6

8

0.4 0.3 0.2 0.1

8

+ΔC2

0.97 0.96 0

8

0.99

0.98 0

0.98

xB4(%nC4)

yD1(C1)

0.98

xB5(nC4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

0.978 0.977 0.976 0.975 0

Time (h)



Page 28 of 36

Fig. 8B – Feed C2 Composition Disturbances; DM, DE and DP Loop Variables

1.5

QR1 (kW)

xB1(%C1)

Feed C2 Composition Disturbances

1

+ΔC2 0.5

0

2

4

6

QR2 (MW)

T23 (K)

360 358 0

2

4

6

QR3 (MW)

330 325 320 0

2

4

6

8

300

+ΔC2

200 2

4

6

8

2.5

+ΔC2 2 1.5

8

335

400

100 0

8

362

T33 (K)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

2

4

6

8

0

2

4

6

8

1.8 1.6 1.4

Time (h)


Page 29 of 36


Fig. 8C – Feed C2 Composition Disturbances; DB and DIB Loop Variables Feed C2 Composition Disturbances

2

4

6

2.1 2 1.9

0

2

4

6

100

+ΔC2

50 0

0

2

4

6

8

1

0

2

4

6

8

2

4

6

8

2

4

6

8

40 38 36 34 0

8

150

1.1

0.9

8

D5 (kmol/h)

360 0

xD5(%nC4)

QR4 (MW)

365

QR5 (MW)

T19 (K)

370

Fby (kmol/h)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

1.7 1.6 1.5

0

Time (h)



Page 30 of 36

Fig. 9A – Feed C3 Composition Disturbances; Product Compositions Feed C3 Composition Disturbances 0.98

xD2(C2)

+ΔC3 0.968 0.966 0

2

4

6

8

xB4(%nC4)

yD1(C1)

0.97

0.985 0.98 0

+ΔC3 2

4

6

8

0.978

xD5(iC4)

xD3(C3)

0.99

xB5(nC4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0.977 0.976 0.975 0

+ΔC3 2

4

6

8

+ΔC3

0.96 0.94 0

2

4

6

8

0.24 0.22 0.2 0.18 0.16 0

2

4

6

8

6

8

0.98

+ΔC3 0.975 0.97 0

Time (h)

2

4


Page 31 of 36



1.5 1 0.5

0

2

4

6

2

4

6

8

328 326 0

2

4

6

8

2

4

6

8

4

6

8

4

6

8

2.2 2.15

+ΔC3

2.1 2.05 0

QR3 (MW)

330

+ΔC3

200 100 0

QR2 (MW)

360 358 0

300

8

362

T23 (K)

400

QR1 (kW)

xB1(%C1)

Feed C3 Composition Disturbances

T33 (K)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

2

2 1.8

+ΔC3

1.6 1.4

0

Time (h)

2



Page 32 of 36

Fig. 9C – Feed C3 Composition Disturbances; DB and DIB Loop Variables Feed C3 Composition Disturbances

363 2

4

6

8

2.1 2 1.9

0

2

4

6

0.95 0

100 50 0

+ΔC3 2

4

6

8

2

4

6

8

2

4

6

8

2

4

6

8

40 38 36 34 0

8

150

0

1

D5 (kmol/h)

362 0

xD5(%nC4)

QR4 (MW)

364

1.05

QR5 (MW)

T19 (K)

365

Fby (kmol/h)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

1.7 1.6 1.5

0

Time (h)


Page 33 of 36


Fig. 10A – Feed C4 Composition Disturbances; Product Compositions Feed nC4 Composition Disturbances

xD2(C2)

+ΔnC4

0.9676 0.9676 0.9675 0

2

4

6

8

+ΔnC4 0.98

2

4

6

0.97 0.968 0

8

1 0.99 0.98 0.97 0

+ΔnC4

xD5(iC4)

xD3(C3)

1

0.96 0

0.972

xB4(%nC4)

yD1(C1)

0.9677

xB5(nC4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

2

4

6

8

0.24 0.22 0.2 0.18 0.16 0

2

4

6

8

+ΔnC4

2

4

6

8

2

4

6

8

0.99 0.98 0.97 0

Time (h)

Time (h)



Page 34 of 36


0.966

QR1 (kW)

xB1(%C1)

Feed nC4 Composition Disturbances

0.964 0.962 0

+ΔnC4 2

4

6

QR2 (MW)

T23 (K)

360.05 360 0

2

4

6

QR3 (MW)

328 327.9 0

2

4

6

8

234

+ΔnC4

233

2

4

6

8

6

8

6

8

2.12

+ΔnC4 2.11 2.1

8

328.1

235

232 0

8

360.1

T33 (K)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

2

1.665

4

+ΔnC4

1.66 1.655 1.65 0

Time (h)

2

4

Time (h)


Page 35 of 36


Fig. 10C – Feed C4 Composition Disturbances; DB and DIB Loop Variables Feed nC4 Composition Disturbances

QR4 (MW)

364 363.5

2

4

6

2 1.5 0

2

4

6

150 100 50 0

1.02 1

0

2

4

6

8

0

2

4

6

8

50 40

+ΔnC4

30 20

8

+ΔnC4

1.04

8

2.5

1

1.06

D5 (kmol/h)

xD5(%nC4)

363 0

QR5 (MW)

T19 (K)

364.5

Fby (kmol/h)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

0

2

4

6

8

1.7 1.65

+ΔnC4

1.6 1.55 0

Time (h)

2

4

Time (h)


6

8


Page 36 of 36

Fig. 11 – Comparison of Proposed and LB Control Structures: + 20% Feed Flowrate

0.985

LB 0.98 0.975 0.97 0

2

4

380

R5 (kmol/h)

xB5(nC4)

20% Increase in Feed: LB structure versus Proposed 0.99

6

8

10

12

360 340 320 300 280 0

0.98

0.975

4

6

8

10

12

2

4

2

4

6

8

10

12

2 1.9 1.8 1.7

LB 0.97 0

2

2.1

QR5 (MW)

0.985

xD5(iC4)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

6

8

10

12

1.6

0

Time (h)


Control of a Train of Distillation Columns for the Separation of Natural

Recommend Documents