Energy-Saving Design and Control of a Methyl Methacrylate

Article pubs.acs.org/IECR

Energy-Saving Design and Control of a Methyl Methacrylate Separation Process Wei-Lun Chang and I-Lung Chien* Department of Chemical Engineering, National Taiwan University, Taipei 10617, Taiwan ABSTRACT: An alternative design for the separation of a mixture including methyl methacrylate, methanol, and water is investigated as compared to a previous design using the same process units. In this alternative design, top and bottom products of the distillation column are designed to be at unstable and stable nodes, respectively. Thus, significant savings in the steam cost (16.3%) can be realized as opposed to the previous design of placing the bottom product at a saddle point. Another benefit is that the loss of methyl methacrylate product through the water outlet stream is also less than that of the previous design. This represents another 9.6% savings of the operating cost for this alternative design. Furthermore, an overall control structure for this alternative design is also devised based on a novel way of using open-loop and closed-loop sensitivity tests. By the control of a temperature difference at two trays of the distillation column and another single-tray temperature at the stripper, both MMA and water products can be maintained at high purities despite large variations in feed flow rate and feed composition changes. feed, such as pyridine in water or acetic acid in water.13,14 The hybrid process could be viewed as a modification of HAD process with an extractor, which was also shown to be economically favorable. In the above HAD examples, decanters were all designed at the top of the column so that the newly formed lightest azeotrope could naturally separate into two liquid phases. However, for the studied system of separating methyl methacrylate (MMA), methanol (MeOH), and water (H2O), which is needed in the production of MMA processes, a heterogeneous MMA/H2O azeotrope at a saddle point is presented. Wu et al.15 devised a two-column design with a bottom decanter. The reproduced flowsheet and the material balance lines are shown in panels a and b of Figure 1, respectively. The bottom composition of the distillation column was designed to be near MMA/H2O azeotrope so that a decanter can be designed to obtain aqueous and organic streams. The aqueous stream is pure enough to draw out of the system while another small stripper was needed to purify the organic stream to MMA product specifications. This design flowsheet was further simplified by combining two columns into one column with a middle decanter. The drawback of the above design is that the bottom composition of the distillation column needs to be placed near a saddle point (MMA/H2O azeotrope) of this ternary system.

1. INTRODUCTION Heterogeneous azeotropic distillation (HAD) is a separation method commonly used in industry to separate azeotropic mixtures or close-boiling mixtures. By adding another light component, a binary (or ternary) minimum boiling heterogeneous azeotrope can be formed. The top overhead of the HAD column is usually designed to be near the heterogeneous azeotrope, while the bottom is designed as high-purity product. The overhead separates into two liquid phases in a decanter: an organic phase sent back to the HAD column, and an aqueous phase drawn out of the process or fed to a recovery column depending on the desired purity. For example, ethanol dehydration1,2 by HAD utilized another recovery column to meet the specification on the aqueous outlet. On the other hand, the ethylenediamine dehydration process3 was implemented without any recovery column because of an already high purity in the aqueous outlet from the decanter. In the acetic acid dehydration process,4,5 the specification on the aqueous outlet could be met by sending portions of the aqueous pahse back to the HAD column rather than using another recovery column. For the purpose of process intensification, the dividing-wall column designs or thermally coupled techniques were commonly incorporated into the HAD process with a recovery column,6−10 showing the economic potential. Another savings on the utility usage could be realized by vapor recompression or heat pump.11,12 In addition, the recovery columns were further combined with the preconcentrator for diluted feed.7−10 Some authors even found good extraction solvents and proposed hybrid extraction−distillation processes to separate the diluted © XXXX American Chemical Society

Received: January 27, 2016 Revised: February 27, 2016 Accepted: February 29, 2016

A

DOI: 10.1021/acs.iecr.6b00391 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

Figure 1. (a) Optimal design flowsheet of original design. (b) Material balance lines of original design.

Also, the composition profile in the distillation column goes along the distillation boundary formed by two azeotropes (will be shown later). Thus, a larger reboiler duty was required at the distillation column. Some contributions16−20 mentioned a mix with water in a decanter or extractor followed by distillation methods for this separation purpose, but they all lacked details about the process and also the dynamic control of the process. In this work, this concept will be implemented by simulations and compared to the original design proposed by Wu et al.15 Moreover, a feasible control structure will be developed and tested for the rejection of changes in feed composition or throughput.

Table 1. UNIQUAC Model Parameters of the Studied Systemsa

a

comp. i

MeOH

MeOH

comp. j

water

MMA

water

aij aji bij (K) bji (K)

0 0 165.2623 −254.7308

0 0 44.6284 −411.6194

0 0 −474.3300 −194.0000

Aspen Plus UNIQUAC: ln γi = ln

2. ENERGY-SAVING DESIGN 2.1. Proposed Design Flowsheet. To correctly predict the thermodynamic behaviors of this ternary system, the UNIQUAC model was selected as in previous work done by Wu et al.15 The UNIQUAC model and the binary parameters are shown in Table 1. The proposed conceptual design flowsheet using the same process units by avoiding the saddle point is shown in Figure 2a with the material balance lines shown in Figure 2b. The fresh feed, containing 69.79 mol % MeOH, 12.5 mol % water, and 17.71 mol % MMA, is first mixed with some water and condensed stripper overhead so that the composition of this mixture is pulled within the liquid− liquid equilibrium (LLE) envelope. After liquid−liquid separation in the decanter at 50 °C, the aqueous stream can further be separated in a distillation column with top product of

MMA

−

Φi xi

θj′τij Φi θ z + qi ln i − qi′ln ti′ − qi′ ∑ + li + qi′ xi t ′j 2 Φi j

∑ xjlj j

where θi =

qixi

; qT =

∑ qkxk ; θi′ =

qi′xi

; qT′ =

∑ qk′xk ; Φi =

qT′ k z = ∑ rkxk ; li = (ri − qi) + 1 − ri ; ti′ = ∑ θk′τki ; τij 2 k k qT

k

rx i i ; rT rT

⎛ bij ⎞ = exp⎜aij + ⎟; and z = 10 T⎠ ⎝

mostly MeOH (MeOH/MMA azeotrope at the lowest temperature in that distillation region) and the bottom product B


Article

Industrial & Engineering Chemistry Research Table 2. Basis of Economics and Equipment Sizing

column diameter (D): Aspen tray sizing column length (L): NT trays with 2 ft spacing plus 20% extra length column and other vessel (D and L are in meters) capital cost = 17 640(D)1.066(L)0.802 condensers (area in m2) heat-transfer coefficient = 0.852 kW/K·m2 differential temperature = reflux-drum temperature −315 K capital cost = 7296(area)0.65 reboilers (area in m2) heat-transfer coefficient = 0.568 kW/K-m2 differential temperature = steam temperature − base temperature (ΔT > 20 K) capital cost = 7296(area)0.65 coolers (area in m2) heat-transfer coefficient = 0.852 kW/K-m2 differential temperature = LMTD of (inlet or outlet temperature −315 K) capital cost = 7296(area)0.65 energy cost HP steam = $9.88/GJ (41 barg, 254 °C) MP steam = $8.22/GJ (10 barg, 184 °C) LP steam = $7.78/GJ (5 barg, 160 °C) cooling water = $0.354/GJ electricity = $16.9/GJ MMA price = $2/kg (approximated from ICIS indicative chemical price) TAC = (capital cost/payback period) + energy cost payback period = 3 yr

Figure 2. (a) Conceptual design flowsheet of alternative design. (b) Material balance lines of alternative design.

of pure water (at the highest temperature in the same distillation region). The MMA product can be obtained from further purifying the organic stream in a small stripper. Note that in this alternative design, top and bottom products of the distillation column are designed to be at the unstable and stable nodes, respectively. For an easy comparison of these two designs, the total numbers of stages of the distillation column and the stripper are assumed to be the same as in the previous design. The remaining design variables considered in this study are only two, including the recycled water flow rate (REC) and the feed stage of the distillation column (NF1). The three product purity specifications are set the same as in the previous design, which are 92.9946 mol % MeOH at distillate of the distillation column, 99.5 mol % H2O at the bottom of the distillation column, and 99.9 mol % MMA at the bottom of the small stripper. A sequential iterative optimization procedure is performed with REC as the outer iteration loop to minimize the total annual cost (TAC) of this process and NF1 as the inner loop to minimize total reboiler duty. The TAC includes total operating cost and annualized total capital cost, where a three year payback period is assumed. Operating cost includes the cost of steam and cooling water, while the capital cost includes the cost of columns and heat exchangers (reboilers and condensers). Table 2 summarizes the formulas for the TAC calculation.21 Figure 3 shows the results of the optimization procedures. Note that the recycled water flow rate is represented by the ratio of recycled water flow rate to the water product flow rate (REC/ H 2 O) as a representative design variable during the

Figure 3. Summary of TAC plots for this alternative design. C


Article


Figure 4. Optimal design flowsheet of alternative design.

Table 3. Itemized TAC Terms of Various Systems

C-1 column (1000 USD) C-1 condenser (1000 USD) C-1 reboiler (1000 USD) C-2 column (1000 USD) C-2 reboiler (1000 USD) cooler (1000 USD) capital cost (1000 USD) LP steam cost (1000 USD/y) cooling water cost (1000 USD/y) operating cost (1000 USD/y) MMA lost from water outlet (1000 USD/y) TAC (1000 USD/y)

original design

alternative design (REC/H2O = 23)

267.59 164.48 104.96 8.75 11.29 18.24 575.31 567.80 24.37 592.17

209.79 128.69 109.01 12.66 17.64 59.38 537.18 475.63 20.02 495.65 (−16.3%) −57.10 (−9.6%)

783.94

617.61 (−21.2%)

optimization. The optimal value of this ratio is 23 with the feed on the 28th stage of the distillation column. The optimal process flowsheet is shown in Figure 4. Also, the itemized TAC of these two designs is shown in Table 3. Despite large amount of water recirculated in this system, a significant saving in the steam cost (16.3%) is realized by using this design. Another benefit is that the MMA product loss through the water product stream is also less than previous design. This represents another 9.6% savings of the operating cost. 2.2. Composition Profiles in the Distillation Column. Figure 5a,b shows the liquid-phase composition profiles in the distillation column for both designs. Two isovolatility curves divide the ternary map into three regions where different volatility rankings to these three components occur. For the original design, not only is the bottom composition designed near a saddle point, but also the composition profile goes along the distillation boundary formed by two azeotropes. This profile is located in a region where methanol is the most volatile component followed by MMA and then water. Results show that this profile has so close volatility that it requires a large energy duty for separation. However, for the alternative design, most of the profile is located in another region where MMA is the most volatile component followed by methanol and then water. Results show that it requires less energy duty because most of the stages are separating just methanol and water, a binary pair without

Figure 5. (a) Liquid-phase composition profile of C-1 column in original design. (b) Liquid-phase composition profile of C-1 column in alternative design.

azeotrope. More accurately, as long as the MMA content in the feed to C-1 column all goes out from the top distillate with D


Article

Industrial & Engineering Chemistry Research Table 4. Results in Process Variables from Closed-Loop Sensitivity Testsa deviation base case (kmol/h) (kW) REC R1 QR1 QR2 QC1 V1 D1 B1 H2O AQ OR V2 B2 MIX QC

286.11 87.85 1991.76 131.00 −1592.41 162.82 74.97 298.55 12.44 373.52 21.38 8.79 12.59 394.89 −371.24

RR R1/AQ QR1/B1 QR1/AQ QR2/OR QR2/B2

1.172 0.235 6.672 5.332 6.092 10.405

MeOH + 10% Process Variables 12.34% 10.04% 10.44% −36.89% 10.04% 10.04% 10.04% 10.84% −23.54% 10.68% −36.66% −36.86% −36.52% 8.12% 3.31% Ratios 0.00% −0.58% −0.36% −0.22% −0.36% −0.58%

MeOH − 10%

feed + 20%

feed − 20%

−11.87% −10.03% −10.30% 37.21% −10.04% −10.03% −10.03% −10.40% 23.45% −10.32% 36.71% 36.91% 36.57% −7.78% −2.78%

21.70% 20.01% 20.88% 19.60% 20.02% 20.01% 20.01% 21.62% 19.70% 21.30% 19.76% 19.09% 20.22% 21.21% 24.32%

−20.61% −20.00% −20.44% −19.89% −20.00% −20.00% −20.00% −20.58% −19.93% −20.46% −19.91% −19.67% −20.08% −20.43% −22.18%

0.00% 0.32% 0.11% 0.03% 0.37% 0.47%

0.00% −1.06% −0.61% −0.34% −0.13% −0.52%

0.00% 0.58% 0.18% 0.03% 0.03% 0.24%

a [REC], recycled water flow rate; [R1], reflux rate of C-1; [QR1], reboiler duty of C-1; [QR2], reboiler duty of C-2; [QC1], condenser duty of C-1; [V1], overhead vapor flow rate of C-1; [D1], top distillate flow rate of C-1; [B1], bottom flow rate of C-1; [H2O], water product flow rate; [AQ], aqueous phase flow rate; [OR], organic phase flow rate; [V2], overhead vapor flow rate of C-2; [B2], bottom flow rate of C-2; [MIX], mix stream flow rate of fresh feed, REC, and V2; [QC], cooler duty.

with 50% total liquid level is used because of a smaller density difference in two liquid phases. Then this steady-state simulation is exported to the dynamic simulations of Aspen Plus Dynamics. Pressure-driven simulations in Aspen Plus Dynamics is selected with the top stage pressures of every column set at atmospheric pressure. Pressure drops inside the columns will be automatically calculated by Aspen Plus Dynamics. 3.1. Inventory Control Loops. The intuitive regulatory control and inventory control loops are designed as follows. Fresh feed flow rate is flow-controlled with a PI controller of Kc = 0.5 and τI = 0.3 min. Both column base levels are controlled by manipulating their bottom flow rates separately. Reflux drum level is controlled by manipulating distillate flow rate. As for the decanter, the aqueous phase level is controlled by manipulating aqueous outlet flow, while the organic phase level is controlled by manipulating organic outlet flow. P-only controllers are used in level control loops. To speed up closedloop control behavior in the recycle loop, the settings Kc = 10 for the organic and aqueous phase level loops to manipulate organic or aqueous flow and Kc = 5 for the base level loop of distillation column to manipulate bottom flow rate are used. Otherwise, Kc = 2 is used in the other level control loops. The pressure of the distillation column is controlled by manipulating the condenser duty, while the pressure of the stripper is controlled by manipulating the overhead vapor flow. In addition, there are two temperature controllers controlling temperature of the condensed overhead flow and the stream entering the decanter at 50 °C by manipulating the cooler duty. These pressure and temperature PI controller parameters are

methanol, the profile will be in that region where MMA is the lightest component. Otherwise, if there is so much MMA fed to the column that some of MMA should go out from the bottom, MMA must be less volatile than methanol, which can be achieved only with a high-energy-required composition profile similar to that in the original design. The maximum MMA content allowed for the profile as in Figure 5b depends on the product specifications and material balance. This is why the TAC with REC/H2O = 21 in Figure 3 exceeds that of the original design. The composition profile changes to the highenergy-required one because the amount of recycled water is not enough to separate out most of the MMA in the decanter.

3. OVERALL CONTROL STRATEGY DEVELOPMENT In this section, a proper control strategy for this optimal design in Figure 4 will be devised based on open-loop and closed-loop sensitivity tests with Aspen Plus steady-state simulations. No online composition measurement will be used in the quality control loops for wider industrial applications. This control strategy is expected to hold the product purities at specifications despite any feed composition change or throughput change. The tray sizing option in Aspen Plus is utilized to calculate the column diameters with the assumed tray spacing of 0.6096 m and weir heights of 0.0508 m. The resulting column diameters are 0.890 m for the distillation column and 0.327 m for the stripper. Other equipment sizing follows the recommendations by Luyben and Chien.22 Volumes of the column bases and the reflux drum are sized to provide 10 min holdup with 50% liquid level. As for the decanter, it is sized bigger to allow the separation of two liquid phases. The holdup time of 40 min E


Article

Industrial & Engineering Chemistry Research set at default Kc = 20 and τI = 10 min for the purpose of tight control. After the above regulatory control and inventory control loops are established, there are four remaining degrees of freedom in this system, including reboiler duty and reflux rate of the distillation column, reboiler duty of the stripper, and the recycled water flow rate, which are related to quality control. Thus, in sections 3.2 and 3.3, it will be illustrated how the additional four control loops are determined for manipulating these four remaining degrees of freedom. 3.2. Closed-Loop Sensitivity Test. A control strategy is desirable if it is able to hold the product purities at

specifications despite any disturbance. To save time, closedloop sensitivity test is a test that can help us understand what kind of combinations of quality control loops will lead to better performance in closed-loop control because this test can be easily done with Aspen Plus steady-state simulation. This test assumes that the overall process will finally achieve perfect control after the anticipated disturbances are introduced to the system. Thus, what we should do is to change the input data, such as feed flow rate or feed composition, as disturbance and then use the Design Spec option in Aspen Plus to vary the degrees of freedom to hold the product purities at specification. Finally, this process will go to a new steady-state equal to perfect control if the Design Specs options all work successfully. The last step is to identify any process variables or ratios of process variables that remain almost unchanged before and after the introduced disturbance, which are suitable to be selected as controlled variables. For this process shown in Figure 4, as mentioned in the section 3.1, there are four remaining degrees of freedom. However, the number of product purities of interest are just three, which means there is still one degree of freedom untreated if we hold all of the product purities. Thus, a quality control loop for one of the degrees of freedom should be guessed again and again until all Design Spec options work successfully. The process is tested under two types of disturbances: one is ±20% throughput changes (feed ± 20%), and the other is ±10% methanol changes in the fresh feed composition (MeOH ± 10%). In the feed composition change, MMA and water composition increase or decrease simultaneously and proportionally to each other. The successful guessed control loop is to fix the reflux ratio in the distillation column; then the remaining three degrees of freedom are varied to hold the purities. Parts of the results of the closed-loop sensitivity test are listed in Table 4, including process variables and some important ratios. Base case represents the results before any disturbance, while the deviation means the results of perfect control compared to the base case. The other results of the closed-loop sensitivity test, tray temperatures in the columns, will be used in section 3.3. Table 4 shows that every degree of freedom should change for the sake of perfect control, but some of the ratios can be fixed. These ratios which change least are the reflux ratio in the distillation column (RR), ratio of reboiler duty in C-1 to the aqueous phase flow rate (QR1/AQ), and ratio of reboiler duty in the stripper to the organic phase flow rate (QR2/OR). Thus, fixing these three ratios is feasible for perfect control. The remaining degree of freedom is the recycled water flow rate, which must be manipulated with the aid of a tray temperature controller.

Figure 6. (a) Closed-loop sensitivity test on temperature profile in C-2 stripper. (b) Open-loop sensitivity test with reboiler duty (QR2) on temperature profile in C-2 stripper.

Table 5. Results in Tray Temperatures from Closed-Loop Sensitivity Tests deviation base case (kmol/h) (kW) QR2/OR T2 T36−T33 T33−T28

6.092 81.061 11.126 9.573

MeOH + 10% Ratio and Temperatures −0.36% 0.00% −0.15% 1.18%

MeOH − 10%

feed + 20%

feed − 20%

0.37% 0.06% −0.05% −1.12%

−0.13% 0.13% −0.18% 2.38%

0.03% −0.08% −0.13% −2.12%

[T2], temperature on 2nd stage in C-2; [T36−T33], temperature difference between 36th and 33rd stages in C-1; [T33−T28], temperature difference between 33rd and 28th stages in C-1. F


Article

Industrial & Engineering Chemistry Research 3.3. Selection of Temperature Control Stage(s). Openloop and closed-loop sensitivity tests are used to determine the tray temperature control point for both columns. The openloop sensitivity test is performed in which ±0.1% changes in one of the degrees of freedom is given when others are fixed. An appropriate temperature control point should have large

sensitivity (large temperature deviation in open-loop sensitivity test) to the manipulated variables. On the other hand, as mentioned before, an appropriate temperature control point should also be almost unchanged (small deviation in closed-loop sensitivity test) for the sake of perfect control. Thus, the control point is determined based on both open-loop and closed-loop sensitivity tests. For the C-2 stripper, the results of tray temperature deviation in the closed-loop sensitivity test are shown in Figure 6a, and the results in the open-loop sensitivity test are shown in Figure 6b. Although temperature on the first stage deviates the least under perfect control, it has less sensitivity to the reboiler duty. The best temperature control point is the second stage considering both requirements discussed above. However, in section 3.2, we know that the reboiler duty can also be manipulated with fixed ratio to the organic phase flow rate. In Table 5, deviations of this ratio and the temperature on second stage (T2) in closedloop sensitivity test are compared. This temperature is better as a controlled variable because of less deviation, while this ratio can still be applied as an inner-loop ratio control for better dynamic performance. As for the C-1 distillation column, results of both tests are shown in Figure 7a,b. The selection is not obvious because stages with larger open-loop sensitivity all have much temperature deviation under perfect control. Large deviation in product purities is further confirmed after the introduction of any disturbance by using either single-end or dual-end temperature control. However, temperatures on stages 28−36 have similar temperature deviations under perfect control. This means that a smaller deviation in temperature difference between any two of these stages can be achieved. If the temperature difference between any two stages is selected as a controlled variable, it should also have enough open-loop sensitivity to the manipulated variables, which is the subtraction of two sensitivities for these two stages. Thus, one of the stage temperatures should have larger sensitivity, while the other one should have smaller sensitivity. Figure 7b shows that temperature difference between stages 28 and 33 (T33 − T28) has the largest open-loop sensitivity. However, the final selected temperature difference is that between stages 36 and 33 (T36 − T33) for better closed-loop performance as listed in Table 5, at the cost of just a little decrease in the open-loop sensitivity, shown in Figure 7b. Here, controlling a temperature difference

Figure 7. (a) Closed-loop sensitivity test on temperature profile in C-1 column. (b) Open-loop sensitivity test with recycled water flow rate (REC) on temperature profile in C-1 column.

Figure 8. Overall proposed control structure for this alternative design. G


Article


Figure 9. Closed-loop dynamic responses to ±10% methanol feed composition changes.

fixed ratio of reboiler duty in C-1 column to aqueous phase flow rate (QR1/AQ), controlling temperature on second stage (T2) in C-2 stripper by manipulating the ratio of its reboiler duty to organic phase flow rate (QR2/OR), and controlling the temperature difference between the 36th

can also be viewed as a single-tray temperature control with a floating set point from another tray. 3.4. Closed-Loop Dynamic Control Behaviors. In summary of sections 3.2 and 3.3, the four quality control loops are determined: fixed reflux ratio in C-1 column (RR), H


Article


Figure 10. Closed-loop dynamic responses to ±20% throughput changes.

and 33rd stages (T36-T33) in C-1 column by manipulating the recycled water flow rate. PI controllers are used in these two temperature control loops with 1 min deadtime. They are iteratively tuned by relay feedback test provided in Aspen Plus Dynamics with Tyreus−Luyben tuning rules.23

For the single-tray temperature controller, the resulting parameters are Kc = 0.931 and τI = 10.56 min with reverse action. For the controller controlling temperature difference, the resulting parameters are Kc = 0.295 and τI = 65.34 min with direct action. I


Article


between the 36th and 33rd stages in the C-1 column, and the temperature at the second stage in the C-2 stripper. Two innerloop ratio controls are added to enhance the dynamic performance. Results show that with the aid of this control strategy, the process quickly reaches another steady state within 6 h and all the product purities are still maintained near their specifications despite any disturbance, such as feed composition changes or throughput changes.

The overall proposed control structure is shown in Figure 8. Note that two inner-loop ratio controls are added for faster dynamic response: ratio of recycled water flow rate to fresh feed flow rate and ratio of reboiler duty in the stripper to organic phase flow rate. Thus, both temperature controllers adjust these two ratios separately, rather than directly manipulating recycled water flow rate or the reboiler duty. Two types of disturbances will be used to test the proposed control strategy. The tough disturbance of unmeasurable fresh feed composition changes is considered first. Large variations of ±10% methanol changes (MeOH ± 10%) in the fresh feed composition are tested, and MMA and water in feed composition increase or decrease simultaneously and proportionally to each other. For example, in the +10% MeOH case, the feed composition changes from 69.79 mol % MeOH, 12.50 mol % H2O, and 17.71 mol % MMA to 76.77 mol % MeOH, 9.61 mol % H2O, and 13.62 mol % MMA. Figure 9 shows the closed-loop dynamic responses to feed composition changes. It is observed that the temperature in the C-2 stripper and the temperature difference in the C-1 column are quickly controlled back to their set point values within 6 h. Three product flow rates increase or decrease according to the unmeasurable feed composition changes. Most importantly, all three product purities are maintained near their specifications despite these unmeasurable disturbances. The other tested disturbance is the ±20% throughput changes (feed ± 20%). Although throughput changes can be considered as known disturbance, any set point in this control strategy has no need to adjust with the throughput changes. Figure 10 displays the dynamic responses of this disturbance. All three product purities are also maintained near their specifications despite large throughput changes. When the closed-loop performances of this proposed design are comared with that of the original design in Wu et al.,15 the ranges of feed composition and feed rate changes for the closed-loop tests in this paper are almost doubled. In spite of these larger disturbance changes, the product purities are still kept very close to their original values even during transient conditions.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +886-3-3366-3063. Fax: +886-2-2362-3040. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research funding from the Ministry of Science and Technology of R.O.C. under Grant MOST 104-2218-E-002006 is greatly appreciated.

■

REFERENCES

(1) Luyben, W. L. Economic Optimum Design of the Heterogeneous Azeotropic Dehydration of Ethanol. Ind. Eng. Chem. Res. 2012, 51, 16427−16432. (2) Pla-Franco, J.; Lladosa, E.; Loras, S.; Montón , J. B. Thermodynamic Analysis and Process Simulation of Ethanol Dehydration via Heterogeneous Azeotropic Distillation. Ind. Eng. Chem. Res. 2014, 53, 6084−6093. (3) Yu, H.; Ye, Q.; Xu, H.; Dai, X.; Suo, X.; Li, R. Comparison of alternative distillation processes for the maximum-boiling ethylenediamine dehydration system. Chem. Eng. Process. 2015, 97, 84−105. (4) Chien, I. L.; Zeng, K. L.; Chao, H. Y.; Liu, J. H. Design and control of acetic acid dehydration system via heterogeneous azeotropic distillation. Chem. Eng. Sci. 2004, 59, 4547−4567. (5) Wang, S. J.; Huang, K. Design and control of acetic acid dehydration system via heterogeneous azeotropic distillation using pxylene as an entrainer. Chem. Eng. Process. 2012, 60, 65−76. (6) Sun, L. Y.; Chang, X. W.; Zhang, Y. M.; Li, J.; Li, Q. S. Reducing Energy Consumption and CO2 Emissions in Thermally Coupled Azeotropic Distillation. Chem. Eng. Technol. 2010, 33, 395−404. (7) Wu, Y. C.; Huang, H. P.; Chien, I. L. Investigation of the EnergySaving Design of an Industrial 1,4-Dioxane Dehydration Process with Light Feed Impurity. Ind. Eng. Chem. Res. 2014, 53, 15667−15685. (8) Wu, Y. C.; Lee, H. Y.; Huang, H. P.; Chien, I. L. Energy-Saving Dividing-Wall Column Design and Control for Heterogeneous Azeotropic Distillation Systems. Ind. Eng. Chem. Res. 2014, 53, 1537−1552. (9) Wang, S. J.; Chen, W. Y.; Chang, W. T.; Hu, C. C.; Cheng, S. H. Optimal design of mixed acid esterification and isopropanol dehydration systems via incorporation of dividing-wall columns. Chem. Eng. Process. 2014, 85, 108−124. (10) Yu, H.; Ye, Q.; Xu, H.; Zhang, H.; Dai, X. Design and Control of Dividing-Wall Column for tert-Butanol Dehydration System via Heterogeneous Azeotropic Distillation. Ind. Eng. Chem. Res. 2015, 54, 3384−3397. (11) Liu, Y.; Zhai, J.; Li, L.; Sun, L.; Zhai, C. Heat pump assisted reactive and azeotropic distillations in dividing wall columns. Chem. Eng. Process. 2015, 95, 289−301. (12) Shi, L.; Huang, K.; Wang, S. J.; Yu, J.; Yuan, Y.; Chen, H.; Wong, S. H. Application of Vapor Recompression to Heterogeneous Azeotropic Dividing-Wall Distillation Columns. Ind. Eng. Chem. Res. 2015, 54, 11592−11609. (13) Kürüm, S.; Fonyo, Z.; Kut, Ö . M. Design Strategy for Acetic Acid Recovery. Chem. Eng. Commun. 1995, 136, 161−176.

4. CONCLUSIONS In this work, an alternative design of a methyl methacrylate separation process is rigorously studied and compared to the original design. Both designs utilize a distillation column, a stripper, and a decanter. In the alternative design, the fresh feed is first mixed with some water and condensed stripper overhead so that the composition of this mixture is pulled within the LLE envelope. After liquid−liquid separation in a decanter, the distillation column is used to separate the aqueous stream, while the organic stream is purified in a stripper. It shows that significant saving in the steam cost (16.3%) is realized. Another benefit is that the MMA product loss through the water product stream is also less than that of the original design. This represents another 9.6% savings of the operating cost. Also, the composition profiles in the distillation column are compared. It shows that these two profiles are in different regions with different volatility rankings, resulting in different relative volatility as well as energy usage for separation. An overall control structure is also devised based on a novel way to use closed-loop and open-loop sensitivity tests. The four controlled variables in quality control loops include the reflux ratio in the C-1 column, the ratio of reboiler duty in the C-1 column to aqueous phase flow rate, the temperature difference J


Article

Industrial & Engineering Chemistry Research (14) Chen, Y. C.; Li, K. L.; Chen, C. L.; Chien, I. L. Design and Control of a Hybrid Extraction−Distillation System for the Separation of Pyridine and Water. Ind. Eng. Chem. Res. 2015, 54, 7715−7727. (15) Wu, Y. C.; Hsu, C. S.; Huang, H. P.; Chien, I. L. Design and Control of a Methyl Methacrylate Separation Process with a Middle Decanter. Ind. Eng. Chem. Res. 2011, 50, 4595−4607. (16) Croom, C. J. W. Production of esters of methacrylic acid. U.S. Patent 2,056,771, 1936. (17) Pai, V. K.; Hyman, D.; Witheford, J. M. Process for the production of methyl methacrylate. U.S. Patent 3,821,286, 1974. (18) Higuchi, H.; Kida, K. Process for producing methyl methacrylate. U.S. Patent 5,087,736, 1992. (19) Nagai, K. New developments in the production of methyl methacrylate. Appl. Catal., A 2001, 221, 367−377. (20) Sugiyama, H.; Fischer, U.; Hungerbühler, K.; Hirao, M. Decision Framework for Chemical Process Design Including Different Stages of Environmental, Health, and Safety Assessment. AIChE J. 2008, 54, 1037−1053. (21) Luyben, W. L. Comparison of extractive distillation and pressure-swing distillation for acetone/chloroform separation. Comput. Chem. Eng. 2013, 50, 1−7. (22) Luyben, W. L.; Chien, I. L. Design and Control of Distillation Systems for Separating Azeotropes; Wiley: Hoboken, NJ, 2010. (23) Tyreus, B. D.; Luyben, W. L. Tuning PI Controllers for Integrator/Dead Time Processes. Ind. Eng. Chem. Res. 1992, 31, 2625−2628.

K


Energy-Saving Design and Control of a Methyl Methacrylate

Recommend Documents