CHEMICAL ENGINEERING TRANSACTIONS VOL. 63, 2018 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Jeng Shiun Lim, Wai Shin Ho, JiΕ™Γ­ J. KlemeΕ‘ Copyright Β© 2018, AIDIC Servizi S.r.l. ISBN 978-88-95608-61-7; ISSN 2283-9216 Production Decision Support System for Multi-product with Multiple Different Size Processors Lee Pheng Cheea,b, Sharifah Rafidah Wan Alwia,b,*, Jeng Shiun Lima,b aProcess Systems Engineering Centre (PROSPECT), Research Institute of Sustainable Environment, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Johor, Malaysia bFaculty of Chemical & Energy Engineering, Universiti Teknologi Malaysia, 81310 UTM Johor Bahru, Johor, Malaysia syarifah@utm.my Aggregate planning is an operational activity with the objective of providing upfront information on quantity of material to be procured and resources to be secured. At a point of time, it might also influence both demand and supply. This is where the sales division will work closely with operation on aggregate planning to deliver maximum profit. Aggregate planning does not only serve as a master plan for the production planner, it is also closely linked to organisational decision-making. Realising its importance, researchers have worked on this subject consistently since 1950s but due to complexity and practicality issue, industry did not manage somehow to adopt the research work. In 2016, the concept of Production Decision Support System (PDSS) was introduced following the Pinch Analysis extended into supply chain area. In this work, the PDSS is applied to a batch industry case which involve multi-products with multiple different size processors. From the assessment, the PDSS has not only demonstrated its practicality but also helped the plant to realise their potential capacity. This has assisted the plant management to realign the strategy and avoided the original intention of expensive expansion. 1. Introduction Aggregate planning is an operational activity. It is normally performed as part of a production process for an advanced period of two to 18 months with the objective of providing upfront information on quantity of material to be procured and resources to be secured (Chakrabortty et al., 2015). According to Chinguwa et al. (2013), production planning starts with a business plan where capacity investment is matched against projected market requirement. This is where the sales division will come into play and work closely with operation on aggregate planning to deliver maximum profit. This will then result aggregate planning to influence both demand and supply. Aggregate planning does not only serve as a master plan for the production planner, it is also closely linked to organisational decision-making, especially during budgeting development and very often the budget revision can only be warranted with the agreement gained from aggregate planning. The importance of aggregate planning has attracted the interest of many researches ever since the 1950s, starting with Linear Decision Rule (Holt et al., 1955). All those research works including some other mathematical and heuristic based techniques have been summarised in the review paper by Martinez in year 2014 (Martinez-Costa et al., 2014). Despite all this, industry still somehow could not manage to appreciate completely on those good research works (Ramezanian et al., 2015). Pinch Analysis from Process Integration Approach was applied into aggregate planning by Singhvi and Shenoy (2002). In the year 2016, Production Decision Support System (PDSS) was introduced by Chee et al. (2016) for dealing with multiple production contributory factors such as cycle time, batch size, plant availability, and number of stream and product mixes. In Chee et al. (2017), the method was extended for multiple product types and product grades with multiple reactors. The objective of this paper is to implement the Production Decision Support System concept to a batch industry case which involves multi-products with multiple different size processors. DOI: 10.3303/CET1863086 Please cite this article as: Lee Pheng Chee, Sharifah Rafidah Wan Alwi, Jeng Shiun Lim, 2018, Production decision support system for multi- product with multiple different size processors, Chemical Engineering Transactions, 63, 511-516 DOI:10.3303/CET1863086 511 2. Case Study This plant has twelve reactor streams. Not all the streams share the same size. Four of the twelve reactors have half the reactor size of the rest of the streams. Out of these four smaller-sized reactors, two of the reactors have been taken out and are now being revamped to prepare for different product production. In short, there are ten available reactors: two small and eight big reactors. This emulsion plant operates 24 h / 7 d a week, which is the same as the other plant. This plant has six grades but only produces one type of product, which is Type I. The sales forecast is tabulated as per Table 1 and the respective processing time as well as the batch sizes have been compiled and consolidated in Table 2. Table 1: Annual sales forecast in t for the emulsion plant Product, j Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec 61A 1,760 1,545 1,521 1,039 1,031 1,420 1,116 1,122 1,173 1,134 1,134 1,134 62B 2,492 2,344 3,075 2,549 2,922 3,330 3,741 3,301 3,301 3,278 3,255 3,255 67C 1,395 1,390 1,250 888 1,328 1,503 1,392 1,505 1,505 1,447 1,493 1,447 60D 69 92 - 69 - - - - - - - - 69E 1,816 1,778 1,761 1,855 2,115 2,152 2,000 2,156 2,098 2,098 2,098 2,098 68D 93 59 177 106 23 118 61 106 83 83 106 83 Total 7,624 7,208 7,783 6,507 7,419 8,524 8,310 8,190 8,160 8,040 8,086 8,017 Table 2: Processing data of the product grade for the emulsion plant Grade, i Reactor conditioning, RC (h) Recipe Time, RT (h) Reactor Transfer time, TT (h) Adjustment, AT (h) Stripping time, ST (h) Sieving time, FT (h) 61A (S) 1.0 10.0 0.4 0.9 6.0 1.3 62B (S) 0.5 12.0 0.4 1.3 9.5 1.5 67C (S) 0.5 10.0 0.4 1.5 7.0 1.4 60D (S) 0.6 10.0 0.4 0.8 8.0 1.4 68D (S) 0.9 10.5 0.4 1.0 8.0 1.3 61A (L) 1.3 11.0 0.5 1.0 8.0 1.3 69E (L) 0.8 12.5 1.0 3.0 9.0 1.3 62B (L) 0.8 12.5 1.0 3.0 9.0 1.3 67C (L) 0.5 10.0 0.4 1.5 7.0 1.4 There are several product grades that can be produced with either the small or big reactors. The same grade produced from the different-sized reactors will be treated as two different grades during plant data collection. In the sales forecast, this parameter is treated the same. An overview of case study is illustrated in Figure 1. 3. Data translation and problem formulation The step by step procedure to calculate the effective plant capacity based on sales forecast are described next. 3.1 Determine of minimum Cycle Time, Ctmin Following the definition from Biegler et al. (1997), both reactor and stripper cycle time for this case study can be rewritten as Eq(1) and Eq(2). 𝐢𝑇𝑗 𝑅 = 𝑅𝐢𝑗 + 𝑅𝑇𝑗 + 𝑇𝑇𝑗 (1) 𝐢𝑇𝑗 𝑆 = 𝑇𝑇𝑗 + 𝐴𝑇𝑗 + 𝑆𝑇𝑗 + 𝐹𝑇𝑗 (2) 512 Figure 1: Product arrangement overview of the case study The minimum cycle time, Ctmin can be determined by using Eq(3), and the values are presented in Table 3. The processing time of the same grade products produced by the different reactor sizes are different since the reaction time is not the same. Ctmin = max (CTj R, CTj S) (3) Table 3: Simplified cycle time using Eq(1) to Eq(3) Grade, j Change-over time, CO (d) Reactor Cycle time, CTRj (h) Stripper cycle time, CTSj (h) Minimum Cycle time, Ctmin 61A (S) 0 11.417 8.583 11.417 62B (S) 0 12.917 12.750 12.917 67C (S) 0 10.917 10.333 10.917 60D (S) 0 11.000 10.667 11.000 68D (S) 0 11.833 10.667 11.833 61A (L) 0 12.833 10.750 12.833 69E (L) 0 14.250 14.333 14.333 62B (L) 0 14.250 14.333 14.333 67C (L) 0 10.917 10.333 10.917 Note: β€œS” indicate a small size reactor, β€œL” indicates big size reactor 3.2 Determine of plant availability, AD In this plant, the number of days the plant is available is calculated according to the reactor size group, as per Eq(4). Two of the small reactors have been taken out for revamping (out of service for 2 y). The availability of the first two small reactors are indicated as zero available days. Based on the historical data, this plant requires five unplanned shut down days per month and 14 d of planned shut down for maintenance. The calculated plant availability for the respective reactor group is summarised in Table 4. ADt,r = βˆ‘ (FDt,m βˆ’m SDt,m) Nm , βˆ€r, βˆ€t (4) 513 where SDt,m is the total number of planned shutdown days of identical production line m for the month t, Nm is the total number of identical production lines, m, FDt,m is the total number of calendar days in the month t. Table 4: The available days of all production streams 3.3 Determine the reactor stream number, SR The reactor stream is only meant for one type of product. As such, each reactor shares one full production reactor stream, which is the same reactor stream. 3.4 Determine product mix From the annual sales forecast, the product mix ratio is calculated as per Eq(5) and then tabulated in Table 5. SRj,t = Sj,t SFt , βˆ€t (5) where Sj,t is the total sales of the product j for the month t, and SFt is the total sales forecast for the month t. Table 5: Product mix ratio on respective grades as calculated based on sales forecast Product, j Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec 61A 0.231 0.214 0.195 0.160 0.139 0.167 0.134 0.137 0.144 0.141 0.140 0.141 62B 0.327 0.325 0.395 0.392 0.394 0.391 0.450 0.403 0.405 0.408 0.403 0.406 67C 0.183 0.193 0.161 0.136 0.179 0.176 0.168 0.184 0.184 0.180 0.185 0.181 60D 0.009 0.013 0.000 0.011 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 69E 0.238 0.247 0.226 0.285 0.285 0.252 0.241 0.263 0.257 0.261 0.259 0.262 68D 0.012 0.008 0.023 0.016 0.003 0.014 0.007 0.013 0.010 0.010 0.013 0.010 Total 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Month, t Total days for the month, FDt (d) Shut down, SD Available Day for the respective stream, AD (d) Total Available Day, ADt,r (d) Unplanned shutdown /reactor (d) Planned shutdow n/ reactor (d) R1S R2S R3S R4S R5L R6L R7L R8L R9L R10L R11L R12L S L Jan-15 31 5 0 0 0 26 26 26 26 26 26 26 26 26 26 26 26.0 Feb-15 28 5 0 0 0 23 23 23 23 23 23 23 23 23 23 23 23.0 Mar-15 31 5 14 0 0 12 26 26 26 26 26 26 26 26 26 19 26.0 Apr-15 30 5 14 0 0 25 11 25 25 25 25 25 25 25 25 18 25.0 May-15 31 5 14 0 0 26 26 12 26 26 26 26 26 26 26 26 24.3 Jun-15 30 5 14 0 0 25 25 25 11 25 25 25 25 25 25 25 23.3 Jul-15 31 5 14 0 0 26 26 26 26 12 26 26 26 26 26 26 24.3 Aug-15 31 5 14 0 0 26 26 26 26 26 12 26 26 26 26 26 24.3 Sep-15 30 5 14 0 0 25 25 25 25 25 25 11 25 25 25 25 23.3 Oct-15 31 5 14 0 0 26 26 26 26 26 26 26 12 26 26 26 24.3 Nov-15 30 5 14 0 0 25 25 25 25 25 25 25 25 11 25 25 23.3 Dec-15 31 5 14 0 0 26 26 26 26 26 26 26 26 26 12 26 24.3 Total 365 60 140 0 0 291 291 291 291 291 291 291 291 291 291 291 291 514 3.5 Calculate the Effective Plant Capacity (EPC) Given the plant availability (ADt) is equal to Total Required Production day (TDt) for this case study; the Effective Plant Capacity EPCt adopted the calculation for Production Volume PAt where product mix, minimum cycle time, and the batch size are taken into consideration (Chee et al., 2016), as shown in Eq(6). The Production Volume PAt is calculated by using Eq(7). The calculated EPC is summarised in Table 6. EPCt,r = βˆ‘ PAt,m x ADt,mm TDt,m , βˆ€r, βˆ€t (6) PAt = βˆ‘ (NB m j,t,m Γ— BSj,t,m), βˆ€j, βˆ€t (7) Table 6: Summary of calculated Effective Plant Capacity based on sales forecast Month, t Total days for the month, FDt Total Available Day, ADt (d) Time required, TDt (d) Number of batches, NBt Sales Forecast , SFt Production Vol, PAt Accumulated Inventory of the month, AIt S L S L S L Total Jan-16 31 26.0 26.0 26.0 26.0 425 7,624 1,338 9,617 10,955 3,331 Feb-16 28 23.0 23.0 23.0 23.0 378 7,208 1,188 8,492 9,680 5,803 Mar-16 31 19.0 26.0 19.0 26.0 408 7,783 966 9,543 10,509 8,529 Apr-16 30 18.0 25.0 18.0 25.0 387 6,507 925 9,083 10,007 12,029 May-16 31 26.0 24.3 26.0 24.3 395 7,419 1,339 8,758 10,097 14,707 Jun-16 30 25.0 23.3 25.0 23.3 381 8,524 1,281 8,457 9,739 15,922 Jul-16 31 26.0 24.3 26.0 24.3 391 8,310 1,336 8,739 10,075 17,687 Aug-16 31 26.0 24.3 26.0 24.3 395 8,190 1,333 8,777 10,110 19,606 Sep-16 30 25.0 23.3 25.0 23.3 379 8,160 1,283 8,423 9,706 21,152 Oct-16 31 26.0 24.3 26.0 24.3 395 8,040 1,328 8,780 10,108 23,220 Nov-16 30 25.0 23.3 25.0 23.3 378 8,086 1,282 8,422 9,703 24,837 Dec-16 31 26.0 24.3 26.0 24.3 393 8,017 1,334 8,776 10,110 26,930 Total 365 291 291 291 291.0 4,623 93,869 14,365 104,505 120,799 4. Results and Discussion Both demand and supply are translated into a Composite Curves and Grand Composite Curve, as shown in Figure 2 and 3. Figure 2: Composite Curve β€œas it is” 515 Based on the Effective Plant Capacity calculation as well as the Composite Curves, this batch plant should be able to cater for a sales volume of 120,800 t/y and stock-out scenarios should not exist. This is calculated based on the current product mix and at 80 % production stream availability. This result suggests that the local site management team look into their production stream availability in terms of reliability; the actual production stream availability is below 65 %. According to world-class production overall equipment effectiveness, the plant should achieve at least 85 % availability. From this, it can be concluded that much improvement can be done via an in-depth study of underlying causes to unlock any unutilised capacity before proceeding to future plant expansions. Figure 3: Grand Composite Curve β€œas it is” 5. Conclusions Through this study, the Production Decision Support System has been applied for aggregate planning for multi- product with multiple different size processor. It demonstrated a much simpler way of tactical planning in batch processes. It provides a fast and true holistic overview of plant capability and helps plant managers to arrive at an effective decision in a timely manner. No specialised knowledge is needed and the Microsoft Excel tool, the only prerequisite for this system, is also widely available. Because of its flexibility, this system is also being used as a tool to set team site performance targets. Acknowledgments The authors would like to thank Universiti Teknologi Malaysia (UTM) through the UTM Flagship Grant Scheme (UTMF), Vote number: R.J130000.7709.4J233 and Fundamental Research Grant Scheme from Malaysia Ministry of Education, Vote number: R.J130000.7809.4F918 for providing the financial support of this work. Reference Biegler L.T., Grossmann I.E., Westerberg A.W., 1997, Systematic Methods of Chemical Process Design, Prentice Hall, New Jersey, USA. Chakrabortty R.K., Hasin M.A.A., Sarker R.A., Essam D.I., 2015, A possibilistic environment based particle swarm optimization for aggregate production planning, Computers & Industrial Engineering, 88, 366-377. Chee L.P., Wan Alwi S. R., Lim J.S., 2016, Production decision support system for batch processes, Journal of Advanced Research in Applied Sciences and Engineering Technology, 5 (1), 7-21. Chee L.P., Wan Alwi S.R., Lim J.S., Siwayanan P., 2017, Production decision support system: Real-life emulsion plant case study, Chemical Engineering Transactions, 56, 1267-1272. Chinguwa S., Madanhire I., Musoma T., 2013, A decision framework based on aggregate production planning strategies in a multi-product factory: A furniture industry case study, International Journal of Science and Research, 2 (2), 370-383. Holt C.C., Modigliani F., Simon H.A., 1955, A linear decision rule for production and employment scheduling, Management Science, 2 (1), 1-30. MartΓ­nez-Costa C., Mas-Machuca M., Benedito E., Corominas A., 2014, A review of mathematical programming models for strategic capacity planning in manufacturing, International Journal of Production Economics, 153, 66-85. Ramezanian R., Rahmani D., Barzinpour F., 2012, An aggregate production planning model for two phase production systems: Solving with genetic algorithm and tabu search, Expert Systems with Applications, 39 (1), 1256-1263. Singhvi A., Shenoy U.V., 2002, Aggregate planning in supply chains by pinch analysis, Chemical Engineering Research and Design, 80 (6), 597-605. 516