CHEMICAL ENGINEERING TRANSACTIONS VOL. 57, 2017 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Sauro Pierucci, Jiří Jaromír Klemeš, Laura Piazza, Serafim Bakalis Copyright © 2017, AIDIC Servizi S.r.l. ISBN 978-88-95608- 48-8; ISSN 2283-9216 Significant Profitability Improvement for Methyl Decanoate Production using Different Types of Batch Distillation Columns Dhia Y. Aqara,b, Nejat Rahmaniana, Iqbal M. Mujtaba*a a Chemical Engineering Division, School of Engineering, University of Bradford, West Yorkshire BD7 1DP, UK b South Refineries Company, Basra, Iraq I.M.Mujtaba@bradford.ac.uk Esterification of fatty acid (Decanoic Acid) is a common practice in the chemical industries, which can be widely used in several applications such as lubricants agent, plasticizers in polymer industries, pharmaceutical and food industries, solvents for inks and paint removal, and detergents. Methyl decanoate (MeDC) as fatty ester is an important chemical component. Reactive distillation columns have been successfully employed in many reaction systems where reaction and separation takes place simultaneously. In this paper, the performance of esterification of decanoic acid to produce MeDC is considered using the newly developed integrated conventional batch (i-CBD) and semi-batch distillation processes (SBD). The performances of these column configurations are evaluated in terms of profitability via minimisation of production time for a given separation task. Additional operational constraints into SBD column are posed in the optimisation problem to avoid overflowing of still pot at any time of the process due to continuous feeding of methanol into the bottom of column as the reboiler is initially charged to its full capacity. The optimization results clearly demonstrate that the SBD system is more attractive operation in terms of batch time and energy consumption savings, maximum conversion level and higher annual profit as compared to the i-CBD column. 1. Introduction The investments in fatty acid alkyl esters (biodiesel) as biodegradable fuels produced from renewable resources are increasingly receiving attention in recent years due to the global warming problems and the economic issues. A large number of scholars (Karacan and Karacan, 2015) have previously studied the esterification of fatty acids (such as dodecanoic, and oleic acids) with methanol to form fatty methyl ester with only few (Machado et al., 2011) looking at methyl decanoate synthesis by the esterification of decanoic acid with methanol which is the main focus of this study. Batch operation is becoming more important in numerous chemical plants as the trend to specialty, low-scale production of high-added value chemicals continues. The integration of reaction and separation into one operation unit (reactive distillation) can reduce the capital investment, operational costs, and environment emissions as well as the energy consumption, improved productivity and selectivity. A number of studies (Edreder et al., 2009) discussed optimal design and operation polices of batch distillation column under fixed product demand scenario and strict purity considerations. More recently, Aqar et al. (2016a) presented new integrated batch reactive distillation columns for optimal synthesis of methyl lactate. For a defined separation task, they compared the performances of each of these column (i- CBD and i-SBD) configurations in terms of profitability. Their results visibly indicated that the i-SBD system provides much better performance than the i-CBD for methyl lactate synthesis. In this work, the performances of i-CBD and SBD processes for MeDC synthesis in terms of maximum profit function via minimizing processing batch time are determined by formulating and solving an optimization problem incorporating a detail dynamic model for the process. The dynamic optimization case is transformed into a nonlinear programming problem (NLP) and is solved utilizing Control Vector Parameterization (CVP) and Successive Quadratic Programming (SQP) based optimization approach available within gPROMS software. The product DOI: 10.3303/CET1757180 Please cite this article as: Aqar D., Rahmanian N., Mujtaba I.M., 2017, Significant profitability improvement for methyl decanoate production using different types of batch distillation columns, Chemical Engineering Transactions, 57, 1075-1080 DOI: 10.3303/CET1757180 1075 amount and its purity are used as constraints and the reflux ratios, and methanol recycle rate (only for the i- CBD) and methanol feed rate (for the SBD system) are optimized. 2. Column configurations and process models With reference to two column configurations (i-CBD and SBD) shown in Figure 1, the detailed dynamic model (Aqar et al., 2016) includes mass and energy balance equations, column holdup, rigorous thermodynamic properties, and chemical reaction on the trays, in the reboiler and in the condenser is used here. The process model assumes no vapour holdup, constant molar holdup on the trays and in the condenser, no heat loss, perfect mixing on all trays, fast energy dynamics, constant column pressure and total condensation with no sub-cooling. Note, further details can be found in Mujtaba (2004). Reflux Condenser Reboiler Accumulator Tank Reflux Drum (i-CBD) (a) MeOH Recycled Column Stages Reflux Column Stages Accumulator Tank Reflux Drum (SBD) (b) Condenser MeOH Charge Reboiler Figure 1: Schematic diagram of two column configurations for producing methyl decanoate: (a) the integrated conventional (i-CBD), and (b) semi-batch distillation (SBD) 3. Optimization problem formulation In this study, the optimum operations of i-CBD and SBD columns are evaluated in terms of maximum yearly revenue for a given product amount and desired purity of MeDC. 3.1 Maximum Profit Problem The optimization problem can be described as follows: Given: The i-CBD/SBD column configurations, the feed composition, the condenser vapour load, The product purity and the amount of bottom product (methyl decanoate) Optimize: The reflux ratio (Ri-CBD), and the MeOH recycle rate (SMeOH) profiles (for i-CBD) Or, the reflux ratio (RSBD), and the MeOH feed rate (FMeOH) profiles (for SCBD) So as to maximise: The profit Subject to: Process constraints (reboiler overflowing, etc.), Model equations (Equality and inequality constraints) In mathematical terms, the optimization problem (OP1) can be represented as follow: 1076 OP1 Max P R i-CBD(t), SMeOH(t) (For i-CBD Column) Or (1) RSBD(t), FMeOH(t) (For SBD Column) Subject to : B = B * (Inequality Constraints) (2) xMeDC * - ɛ ≤ xMeDC ≤ xMeDC * + ɛ (Inequality Constraints) (3) The profit function equations and constants for both i-CBD and SBD systems can be shown as: P i-CBD = (CMeDC B - CRB0 - OC) × NB - ACC (4) PSBD = (CMeDC B - CRB0 - OC - CMeOH Charge) × NB - ACC (5) OC (The operating cost, $/Batch) = ( K3 VC AP ) × (tf + ts ) (6) NB (Number of Batches, Batch/ yr) = (PH / yr) (tf + ts ) (7) ACC ( Annual Capital Cost, $/ yr) = K1 (VC ) 0.5 (N) 0.8 + K2 (VC ) 0.65 (8) TYP ( Kmol/ yr) = NB × B (9) Where, K1 =1500; K2 = 9500; K3 = 180; the operating cost constant (AP) = 8000; setup time (ts) = 0.5 hr; PH = 8000 hr/yr (Miladi and Mujtaba, 2004). Since the number of plates (N) and the vapor load to condenser (VC) kept constant, the annual capital cost (ACC) is also fixed. For a given separation task (given product amount and purity of product per batch), the minimization of batch time will increase the number of batches (NB) and thus will increase the profitability. Therefore, the maximum profit problem can be converted to minimum batch time problem as shown below. 3.2 Minimum Batch Time Problem The optimization problem can be described as follows: Given: The i-CBD/SBD column configurations, the feed composition, the condenser vapour load, The product purity and the amount of bottom product (methyl decanoate) Optimize: The reflux ratio (Ri-CBD), and the MeOH recycle rate (SMeOH) profiles (for i-CBD) Or, the reflux ratio (RSBD), and the MeOH feed rate (FMeOH) profiles (for SCBD) So as to minimise: The operating batch time Subject to: Process constraints (reboiler overflowing, etc.), Model equations (Equality and inequality constraints) Mathematically, the optimization problem (OP2) can be stated as follow: OP2 Min tf R i-CBD(t), SMeOH(t) (For i-CBD Column) Or (10) RSBD(t), FMeOH(t) (For SBD Column) Subject to : B = B * (Inequality Constraints) (11) xMeDC * - ɛ ≤ xMeDC ≤ xMeDC * + ɛ (Inequality Constraints) (12) Where B, xMeDC are the amount of bottom product (2.4 kmol for both columns), and purity of methyl decanoate at the final batch time tf (denotes that the B * , xMeDC * are specified). RSBD (t) and Ri-CBD (t) are the optimal reflux ratio profiles for both columns (SBD and i-CBD) which are optimized and  is small positive numbering the order of 10-3. Note, all prices of reactant were taken from (Alibaba Trade, 2016) and the prices of methyl decanoate at other purities are estimated based on the exponential trend method used in (Aqar et al., 2016b). 1077 The costs of chemical reaction (DeC and MeOH) and product (MeDC) at various product purities values are tabulated in Table 1. Note also, the calculations of operational constraints policy in terms of overloading of reboiler (B, xMeDC) for the SBD column are similar to those used in (Aqar et al., 2016b). Table 1: The prices of feed and product Pure MeOH Reactant Price 12.84 MeDC Price at 92.5% Purity 250 MeOH Charge Price at 95% Purity 12.20 MeDC Price at 95.0% Purity 280 Pure DeC Reactant Price 87.00 MeDC Price at 96.5% Purity 335 MeDC Price at 90.0% Purity 240.0 MeDC Price at 97.5% Purity 380 4. The synthesis of methyl decanoate system 4.1 Problem description The case study is implemented in a 10-plates batch column (including a total condenser and a reboiler) with the condenser vapour load, VC = 2.5 kmol/hr. The column stages are numbered from the top down. Four percent of the initial feed is the total column holdup. Fifty percent of this total holdup is taken as the condenser holdup and the rest is taken as the plate holdup (equally divided). This strategy of column holdups has been applied for both columns (i-CBD and SBD). Note, the same strategy is used for the catalyst loading distribution. The total initial amount of feed is 5 kmol with the feed composition < DeC, MeOH, MeDC, H2O > is : <0.5, 0.5, 0.0, 0.0>, respectively. 4.2 Chemical reaction and kinetics The esterification reaction of decanoic acid (DeC) with methanol (MeOH) forms methyl decanoate (MeDC) and water (H2O) together with the boiling points of the components can be expressed as follows: DeC (1) + MeOH (2) <=> MeDC (3) + H2O (4) (13) B.P (K) 543.15 337.15 505.15 373.15 The modified Langmuir-Hinshelwood-Hougen-Watson (LHHW) activity (ai = i xi) based kinetic model is used (taken from Steinigeweg and Gmehling, 2003) and can be written as: - r = MCat (3.1819 × 10 6 exp ( 72230 RT ) × ( a1a2 (2.766 a 4 ) 2 ) - 3.5505 × 10 5 exp ( 71900 RT ) × ( a3 2.766 a4 )) (14) 4.3 Vapour-liquid equilibrium (VLE) K-values (VLE constants) are calculated from (Eq. 15) where i is calculated from the NRTL method, the saturated vapor pressure (Psat) for each pure substances is estimated by using Antoine’s equation. The NRTL binary interaction parameters were taken from Aspen HYSYS package and Antoine coefficients were taken from Steinigeweg and Gmehling (2003). The vapor enthalpies are computed using empirical equations from Aspen HYSYS (2016) and the liquid enthalpies were computed by subtracting enthalpy of vaporization from the vapor enthalpies. K = ( γ i Pi sat P ) (15) 5. Results and discussions i-CBD process: Table 2 summarizes the optimal operation strategy, including methanol recycle rates, reflux ratios, total amount of recycled methanol, the operating time, heat usage rate, the conversion rate of DeC, the number of batches, and annual production rate, as well as the maximum achievable profit for the i-CBD column. As can be noted form Table 2 that increasing the quality of MeDC product increases all reflux ratios, and the processing-batch time together with total energy consumption, and the total amount of methanol recycled to column. Clearly, increasing the operating time can lead to increase the conversion ratio of DeC into MeDC. It can be noticed also from Table 2 that as the product composition and production time increase, the number of batches produced over the year and total yearly product progressively decrease (see Eqs. 7 and 9). As the bottom product purity constraint increases form (0.90 to 0.975 mole fractions) together with price of MeDC product, the annual profit increases progressively. However, note, for 0.975 of product purity, there is a sharp decrease in the product revenue due to huge increase in the production batch time and reflux ratio (although the recycle rate of methanol reduces). This makes i-CBD uncompetitive process (compared to others) at higher MeDC quality and hence the suggested SBD mode. SBD mode: For the five product purities considered, the summary of optimization results including optimal methanol feed rates and reflux ratios profiles, maximum reflux ratios, total amount of charged methanol, the 1078 operating batch time, total thermal heat rate, the conversion rate of DeC, the number of batches, and annual production rate, and the net profit for the SBD column are presented in Table 3. It is clear form Table 3 that the use of SBD mode caused big savings in the operating batch time and the total energy demand, and higher improvement in the DeC conversion and the process revenue as compared to the i-CBD operation (Table 2). For instance, at 0.975 mole fraction purity the reductions in the batch processing time, and the thermal energy requirement are almost 81.16%, and 81.57%, and the improvement in conversion rate is about 0.79%, as compared to i-CBD column. It is seen form Table 3 that the total annual product demand upgraded is about 79.30% at product purity of 97.50% compared to that obtained by the i-CBD operation. In addition, for the 0.975 of MeDC composition case, comparison of the maximum yearly profit for the SBD column with those obtained using i-CBD reveals 80.06% more profit due to low production time and thermal heat demanded to achieve the desired purity requirements. Clearly, the SBD is found to outperform the i-CBD operation in many respects. Figure 2 compares the processing batch time for all MeDC mole fractions for both column configurations. For i-CBD column up to 0.965 purity of MeDC, the increasing product value dominates over the batch time. Beyond 0.965 purity, the batch time dominates over the product value thereby reducing the profitability. In such circumstances, probably higher number of column trays are necessary. Note, the values of total annualized capital cost and total operating cost remained the same for all MeDC purity (both columns) which are 32198 $/yr and 450 $/yr, respectively. It can be observed from Table 3 for each MeDC purity specification that the values of maximal reflux ratios (RMax) are found to be larger than normal reflux ratios ensuring no overflowing of still pot. Figure 2: The final operating time profile for both i-CBD and SBD systems Table 2: Summary of optimization results for i-CBD column Purity of MeDC 0.90 0.925 0.950 0.965 0.975 Recycle Rate, SMeOH, kmol/hr 1.26 1.28 1.11 0.91 0.86 Reflux Ratio, Ri-CBD 0.107 0.180 0.355 0.526 0.612 Total MeOH Amount, St, kmol 3.13 3.99 5.25 7.94 18.40 Batch Time, tf, hr 2.48 3.12 4.75 8.72 21.40 Heat Usage, QTot, mkJ 0.233 0.293 0.446 0.822 1.992 Conversion of DeC, % 93.00 95.14 96.86 98.00 98.72 Number of Batches, NB, batch/yr 2688 2211 1525 867 366 Total Yearly Product, kmol/yr 6450 5306 3659 2081 878 Annual Profit, $/yr 173675 190147 230784 231672 118332 0 5 10 15 20 25 30 0,875 0,9 0,925 0,95 0,975 1 O pe ra tin g B at ch T im e (h r) MeDC Purity (Molefraction) i-CBD Operation SBD Operation 1079 Table 3: Summary of optimization results for SBD column Purity of MeDC 0.90 0.925 0.950 0.965 0.975 Feed Rate, FMeOH, kmol/hr 0.78 0.70 1.35 1.32 1.19 Reflux Ratio, RSBD 0.00 0.118 0.069 0.185 0.286 Maximum Reflux Ratio, RMax 0.689 0.720 0.462 0.472 0.524 Total Fed Amount, Ft, kmol 1.09 1.12 3.29 4.42 4.79 Batch Time, tf, hr 1.39 1.60 2.44 3.35 4.03 Heat Usage, QTot, mkJ 0.130 0.143 0.225 0.305 0.367 Conversion of DeC, % 93.00 94.51 98.50 99.32 99.51 Number of Batches, NB, batch/yr 4223 3818 2716 2079 1767 Total Yearly Product, kmol/yr 10136 9162 6520 4990 4241 Annual Profit, $/yr 235626 300044 327672 488928 593531 6. Conclusions In this study, the synthesis of methyl decanoate via esterification of decanoic acid is investigated in different types of batch reactive distillation processes. The performances of i-CBD and SBD systems in terms of profitability are determined using model-based methods where a rigours process model is embedded with the dynamic optimization problem in gPROMS modelling software. Piecewise constants reflux ratios and methanol recycle rate profiles (only for i-CBD), and methanol feed rate (for SBD column) are considered. The results indicate that the employment of SBD operation is more promising option and quite interesting compared to the i-CBD system in terms of minimum batch time and thus energy consumption, highest conversion of DeC, and maximum achievable profit improvement. As an example, the operating time and thus energy usage rate are saved by an average 81.16%, and 81.57%, and the highest achievable conversion and maximum revenue are improved by almost 0.79% and 80.06% respectively at product purity of 0.975 mole fraction compared to that obtained by the i-CBD mode. Reference Aqar D.Y., Rahmanian N., Mujtaba I.M., 2016a, Integrated batch reactive distillation column configurations for optimal synthesis of methyl lactate, Chemical Engineering and Processing: Process Intensification, 108, pp.197-211. Aqar D.Y., Rahmanian N., Mujtaba I.M., 2016b, Methyl lactate synthesis using batch reactive distillation: Operational challenges and strategy for enhanced performance, Separation and Purification Technology, 158, pp.193-203. Alibaba Trade, 2016, Available at: http://www.alibaba.com/trade accessed 18.11.2016 Edreder E., Mujtaba I.M., Emtir M., 2009, Profitability analysis for batch reactive distillation process based on fixed product demand, Chemical Engineering Transactions, 18, pp.701-706. gPROMS, 2016, Advanced User Guide, Process Systems Enterprise Ltd., London. Hysys Reference Manual, 2016, Hyprotech Ltd., Cambridge. Karacan S., Karacan F., 2015, Steady-state optimization for biodiesel production in a reactive distillation column, Clean Technologies and Environmental Policy, 17(5), pp.1207-1215. Machado G.D., Aranda D.A., Castier M., Cabral V.F., Cardozo-Filho L., 2011, Computer simulation of fatty acid esterification in reactive distillation columns, Industrial & Engineering Chemistry Research, 50(17), pp.10176-10184. Miladi M.M., Mujtaba I.M., 2004, Optimisation of design and operation policies of binary batch distillation with fixed product demand, Computers & chemical engineering, 28(11), pp.2377-2390. Mujtaba, I.M., 2004. Batch Distillation: Design and Operation, Series on Chemical Engineering, Vol.3, Imperial Collage Press. Steinigeweg S., Gmehling J., 2003, Esterification of a fatty acid by reactive distillation, Industrial & Engineering Chemistry Research, 42(15), pp.3612-3619. 1080 http://www.alibaba.com/trade