DOI: 10.3303/CET2188146 Paper Received: 23 May 2021; Revised: 21 August 2021; Accepted: 13 October 2021 Please cite this article as: Tan Y.W., Andiappan V., Ng L.Y., Ng D.K., 2021, Mathematical Optimisation Approach for Improvement of Palm Oil Traceability, Chemical Engineering Transactions, 88, 877-882 DOI:10.3303/CET2188146 CHEMICAL ENGINEERING TRANSACTIONS VOL. 88, 2021 A publication of The Italian Association of Chemical Engineering Online at www.cetjournal.it Guest Editors: Petar S. Varbanov, Yee Van Fan, Jiří J. Klemeš Copyright © 2021, AIDIC Servizi S.r.l. ISBN 978-88-95608-86-0; ISSN 2283-9216 Mathematical Optimisation Approach for Improvement of Palm Oil Traceability Yi Wei Tan, Viknesh Andiappan, Lik Yin Ng, Denny K.S.Ng* School of Engineering and Physical Sciences, Heriot-Watt University Malaysia, 62200, Putrajaya, Wilayah Persekutuan Putrajaya, Malaysia Denny.Ng@hw.ac.uk Oil palm is recognised as the most productive oil crop with highest oil yield per hectare of agriculture land. The palm oil products generated from oil palm are widely used in our daily life, contributing from food processing, personal care and hygiene sectors as well as the biofuel industry. However, the reputation of oil palm has been declining as it was accused for causing environmental and social problems, such as deforestation, workers exploitation. In order to achieve sustainable production and consumption of palm oil products, traceability of the products is very important. With the complete traceability, the origin location and quantity of palm oil products distributed along the supply chain can be identified and tracked in both the forward and backward directions. In this work, a systematic approach is developed to evaluate the traceability of palm oil products from oil palm plantations to palm oil mills. A scoring system with qualitative measure is developed to quantify the traceability of the feedstock (fresh fruit bunches). A mixed integer non-linear programming model is then developed to maximise the economic performance while tracking the traceability of the fresh fruit bunches in a supply chain. A case study in Perak, Malaysia is solved to illustrate the proposed approach. Two scenarios are solved to determine the maximum traceability and economic performance. In the first scenario, the maximum economic performance of the supply chain is determined as RM 16.8 x 106 per annum with traceability of 0.688. In scenario 2, when optimisation objective is set as maximise traceability, the maximum traceability of 0.753 with economic performance of RM 12 x 106 per annum is obtained. 1. Introduction Crude Palm Oil (CPO) is an edible vegetable oil extracted from the mesocarp of the oil palm fruitlet. In 2020, palm oil is the largest production of vegetable oil globally, accounting for approximately 31 % of the global oil and fats production (Yusof, 2021). The global palm oil market is predicted to reach USD 78 x 109, expanding at a compound annual growth rate (CAGR) of 3.1 % by 2027 (Grand View Research, 2020). After fresh fruit bunches (FFB) is harvested from the plantation, FFB are sent to the palm oil mills and kernel crushing plant for extraction of crude palm oil (CPO) and crude palm kernel oil (CPKO) (Foong et al., 2019a). A number of biomasses which include kernel cake, empty fruit bunch and palm oil mill effluent are generated during the extraction process (Foong et al., 2019b). The palm-based biomasses can be used as feedstock for combined heat and power (CHP) plant after pre-treatment processes (Lam et al., 2010). CPO and CPKO will then be sent to the refinery to produce refined bleached and deodorized (RBD) palm oil and RBD palm kernel oil (Gibon et al., 2007). As reported by Malaysian Palm Oil Council (MPOC), palm oil is the oil crop with highest yield for a given size of land (MPOC, 2016). However, the reputation of palm oil industry has been declining as it was labelled as the cause of environmental and social problems, such as deforestation and workers exploitation. In order to ensure the sustainable production and consumption of palm oil products, traceability within the entire value chain is important. According to ISO 9000:2005, the term traceability means the ability to trace the history, application or location of that which is under consideration (ISO, 2009). With the implementation of traceability, the origin location and quantity of palm oil products distributed along the supply chain can be identified in both the forward and backward directions. However, to date, limited literature focus on quantifying and qualifying traceability in a supply chain (Lo et al., 2020). Based on the previous studies, traceability is also yet to be defined qualitatively which can optimised via 877 optimisation model. The purpose of this work is to propose a systematic method based on qualitative and quantitative measures to analyse the impact of traceability on the economic performance of a supply chain. A mathematical optimisation model is developed via a commercial optimisation software (Lingo 18.0 with global solver) to demonstrate the proposed approach. The traceability analysis, mathematical formulations, case study, conclusion and future work are presented in the following sections. 2. Traceability Analysis In this work, a new traceability analysis is developed. Three scoring systems named Operation Confidence Level (OCL), Interaction Confidence Level (ICL) and Traceability Confidence Level (TCL) are introduced to measure the traceability of a material in a supply chain qualitatively and quantitatively. In this work, the traceability of FFB from the plantations to the palm oil mills is presented. OCL, ICL and TCL of such supply chain is showed in Tables 1 – 3. In this work, Operation Confidence Level (OCL) is defined as operability of an entity, it can be determined based on the track record, certification status of an entity and reliability of the feedstock. Certified entity is refers to as business entity which complies under international or national schemes, such as Roundtable Sustainable Palm Oil (RSPO), Malaysia Sustainable Palm Oil (MSPO). The audit record refers to previous records on transection of FFB. The detailed description for OCL is showed in Table 1. Table 1: OCL 9-point scales Score Description 1 Uncertified entity with no audit record 2 Uncertified entity with non-proven audit record (record < 3 year) 3 Uncertified entity with proven audit record (record > 3 years) but unreliable FFB stocks (deviation of average yearly production > 30 %) 4 Uncertified entity with proven audit record (record > 3 years) but unreliable FFB stocks (deviation of average yearly production <30 %) 5 Uncertified entity with proven audit record (record > 3 years) and reliable FFB stocks (deviation of average yearly production <20 %) 6 Certified entity with proven audit record (record > 3 years) but unreliable FFB stocks (deviation of average yearly production >30 %) 7 Certified entity with proven audit record (record > 3 years) but unreliable FFB stocks (deviation of average yearly production <30 %) 8 Certified entity with proven audit record (record > 3 years) and reliable FFB stocks (deviation of average yearly production <20 %) 9 Certified entity with proven audit record (record > 3 years) and high reliability of FFB stocks (deviation of average yearly production <10 %) Interaction Confidence Level (ICL) defines the interaction between two interconnecting entities where transactions of FFB happen. ICL can be determined based on the nature of OCL of each involved entity. The ICL score can be calculated via Eq(1) and Table 2 shows the description of ICL based on a 9-point scales. Note that the ICL score will be rounded down to the nearest value for a more conservative approach. In order to achieve ICL of 9, both involved entities are required to be certified with proven audit record for more than three years and high reliability of FFB stocks. ICL = OCL of entity 1 × OCL of entity 2 Highest OCL out of the two entities involved (1) Traceability Confidence Level (TCL) defines the range of achievable traceability of a supply chain. TCL of each entity can be calculated based on OCL and ICL scores which will be discussed in detailed in the next section. The overall TCL of a supply chain, TCLOverall can be determined by taking the average TCL of every entity involved in the supply chain. Table 3 shows the description for TCL point-scales. 878 Table 2: ICL 9-point scales Score Description 1 Both entities with OCL = 1 At least one of the entities with OCL = 2 At least one of the entities with OCL = 3 At least one of the entities with OCL = 4 At least one of the entities with OCL = 5 At least one of the entities with OCL = 6 At least one of the entities with OCL = 7 At least one of the entities with OCL = 8 Both entities with OCL = 9 2 3 4 5 6 7 8 9 Table 3: TCL point-scales Score Description 0.10 – 0.39 Minimum confidence level in traceability of palm products Undesirable confidence level in traceability of palm products Average confidence level in traceability of palm products Desirable confidence level in traceability of palm products Maximum confidence level in traceability of palm products 0.40 – 0.59 0.60 – 0.79 0.80 – 0.89 0.90 – 1.00 3. Mathematical Formulation The mathematical model is formulated based on mass balance of FFB between different entity from one level to another (plantation p –> collection hub ch –> palm oil mill om) as shown in the superstructure in Figure 1. As shown FFB from plantation p both certified and non-certified are first transferred to collection hub ch before sending to palm oil mill om. Each palm oil mill is given with the minimum and maximum operating capacity. The transportation cost for transferring FFB to palm oil mill along the supply chain can then be calculated. Plantations Collection hubs Palm Oil Mills Figure 1: Superstructure of the model Following the proposed method, OCL of each entity is first defined based on the guideline in Table 1. Next, the ICL of each entity can be calculated using Eq(1), and the TCL of plantation p, collection hub ch and palm oil mill om can be determined via Eqs(2 – 4). 𝑇𝐶𝐿𝑝 = 𝑏𝑝 𝑛𝑝 ∑ [𝑏𝑝,𝑐ℎ ( 𝐼𝐶𝐿𝑝,𝑐ℎ 9 ) ( 𝑂𝐶𝐿𝑐ℎ 9 )] 𝑐ℎ ∀𝑝 (2) 𝑇𝐶𝐿𝑐ℎ = 𝑏𝑐ℎ 𝑛𝑐ℎ ∑𝑝,𝑜𝑚 {[𝑏𝑐ℎ,𝑝 ( 𝐼𝐶𝐿𝑐ℎ,𝑝 9 ) ( 𝑂𝐶𝐿𝑝 9 )] + [𝑏𝑐ℎ,𝑜𝑚 ( 𝐼𝐶𝐿𝑐ℎ,𝑜𝑚 9 ) ( 𝑂𝐶𝐿𝑜𝑚 9 )]} ∀𝑐ℎ (3) 𝑇𝐶𝐿𝑜𝑚 = 𝑏𝑜𝑚 𝑛𝑜𝑚 ∑𝑐ℎ [𝑏𝑜𝑚,𝑐ℎ ( 𝐼𝐶𝐿𝑜𝑚,𝑐ℎ 9 ) ( 𝑂𝐶𝐿𝑐ℎ 9 )] ∀𝑜𝑚 (4) p = 1 p = 2 p = n ch = 1 ch = 2 om = 1 om = 2 om = 3 879 where, 𝑇𝐶𝐿𝑝, 𝑇𝐶𝐿𝑐ℎ and 𝑇𝐶𝐿𝑜𝑚 are the traceability for certified and non-certified plantations, collection hubs and palm oil mills. 𝐼𝐶𝐿𝑝,𝑐ℎ , 𝐼𝐶𝐿𝑐ℎ,𝑝 , 𝐼𝐶𝐿𝑐ℎ,𝑜𝑚 and 𝐼𝐶𝐿𝑜𝑚,𝑐ℎ are the ICL for that involving entity and the other entity. 𝑂𝐶𝐿𝑐ℎ , 𝑂𝐶𝐿𝑝 and 𝑂𝐶𝐿𝑜𝑚 are the pre-determined OCL score for each entity. Note that 𝐼𝐶𝐿 = 0 if there is no interaction between the entities while 𝐼𝐶𝐿 = 9 (assumed to be highest possible score) if it is a self-interacting relationship. 𝑏𝑝 , 𝑏𝑐ℎ and 𝑏𝑜𝑚 are binary value and such binary will be equal to 1 for the involved entity. 𝑛𝑝 = ∑𝑐ℎ 𝑏𝑝,𝑐ℎ ∀𝑝 (5) 𝑛𝑐ℎ = ∑𝑝,𝑜𝑚 (𝑏𝑐ℎ,𝑝 + 𝑏𝑐ℎ,𝑜𝑚 ) ∀𝑐ℎ (6) 𝑛𝑜𝑚 = ∑𝑐ℎ 𝑏𝑜𝑚,𝑐ℎ ∀𝑜𝑚 (7) where, 𝑛𝑝 , 𝑛𝑐ℎ and 𝑛𝑜𝑚 are the total numbers of interaction between the involving entity and other entity. 𝑏𝑝,𝑐ℎ, 𝑏𝑐ℎ,𝑝, 𝑏𝑐ℎ,𝑜𝑚 and 𝑏𝑜𝑚,𝑐ℎ denote the binary variables on the existence of interaction between the involving entity and the other entity. In another words, the existence of transaction of FFB between the entities. Next, the overall traceability of the supply chain can then be determined in Eq(8) by taking the average 𝑇𝐶𝐿 of every entity involved as below. The value computed by the optimisation in Eq(8) will be evaluated against the scale in Table 3 to determine the category of the traceability for the supply chain. 𝑇𝐶𝐿𝑂𝑣𝑒𝑟𝑎𝑙𝑙 = 1 𝑁 (∑𝑝𝑇𝐶𝐿𝑝 + ∑𝑐ℎ 𝑇𝐶𝐿𝑐ℎ + ∑𝑜𝑚 𝑇𝐶𝐿𝑜𝑚) (8) 𝑁 = ∑𝑝𝑏𝑝 + ∑𝑐ℎ 𝑏𝑐ℎ + ∑𝑜𝑚 𝑏𝑜𝑚 (9) where 𝑁 refers to the total number of entities involved in the supply chain. The economic performance (𝐸𝑃) of the supply chain can then be determined via Eqs(10 – 13). 𝐺𝑝 = 𝐴1 ∑𝑐𝑝 𝐹𝑐𝑝,𝑐ℎ + 𝐴2 ∑𝑛𝑐𝑝𝐹𝑛𝑐𝑝,𝑐ℎ ∀𝑝 (10) 𝐺𝑐ℎ = 𝐴3 ∑𝑐ℎ 𝐹𝑐ℎ ∀𝑐ℎ (11) 𝐺𝑜𝑚 = 𝐴4 ∑𝑜𝑚(𝐹𝑜𝑚 𝑋𝑜𝑚) ∀𝑜𝑚 (12) 𝐸𝑃 = 𝐺𝑃 + 𝐺𝑐ℎ + 𝐺𝑜𝑚 − 𝐶𝑇𝑡 (13) where, 𝐴1 and 𝐴2 refer to the profit of selling certified FFB and non-certified FFB. 𝐹𝑐𝑝,𝑐ℎ and 𝐹𝑛𝑐𝑝,𝑐ℎ refer to the amount of FFB distributed from the certified plantations and non-certified plantations to collection hubs ch. 𝐹𝑐ℎ refers to the total amount of received FFB in collection hubs, 𝑐ℎ. 𝐴3 refers to the profit of selling FFB to palm oil mills. 𝐹𝑜𝑚 refers to the total amount of received FFB in palm oil mills. 𝐴4 refers to the profit of selling CPO. 𝑋𝑜𝑚 refers to the conversion of FFB to CPO. 𝐺𝑝, 𝐺𝑐ℎ and 𝐺𝑜𝑚 refers to the gross profit for both certified and non- certified plantations, collection hub and palm oil mill. 𝐶𝑇𝑡 refers to the total cost of transportation. The optimisation objectives of the proposed model is to maximise economic performance Eq(14) of the supply chain while understand the traceability of FFB and palm oil products. Maximise 𝐸𝑃 (14) 4. Case Study An oil palm supply chain case study in Perak is solved to illustrate the proposed approach. In this case study, six certified plantations, five non-certified plantations, two collection hubs and three palm oil mills are considered. Figure 2 shows the location while Table 4 shows the OCL of each participating entities in Perak. Table 5 shows the estimated production of FFB in every participating plantation. Two scenarios are considered in the case study. Scenario 1 is purely maximise EP; while scenario 2 considers maximise TCLoverall. It is to investigate the maximum achievable economic performance and traceability of the supply chain as well as evaluating the relationship between the two parameters. 880 Table 4: OCL score for each participating entity Entities OCL Entities OCL Entities OCL Entities Entities CP1 9 CP5 9 NCP3 4 CHb 8 CP2 8 CP6 8 NCP4 4 OM1 9 CP3 7 NCP1 5 NCP5 3 OM2 8 CP4 7 NCP2 3 CHa 9 OM3 9 Table 5: Production of FFB in every participating plantation Plantations FFB Production (tonnes/day) CP1 121 85 180 22 222 480 75 33 37 CP2 CP3 CP4 CP5 CP6 NCP1 NCP2 NCP3 NCP4 87 NCP5 40 Figure 2: Location of participating entities In scenario 1, the optimisation objective is set as maximise economics performance without the consideration of TCL. Solving Eq(14) subjected to Eq(1) – Eq(13) in Lingo 18.0 with global solver, the economic performance of the supply chain is determined as RM 562,286 per batch assuming the FFB are harvested in every 10 days after accounting for the cost of transportation. With this, the economics performance of the supply chain is estimated to be approximately RM16.8 x 106 per annum. The TCLOverall of the supply chain is determined as 0.688. Referring to Table 3, TCLOverall of this supply chain falls within the average traceability confidence level. Comparing the FFB production in Table 5 and allocation of FFB in Figure 3(a), it is noted that almost all available FFB from the certified and non-certified plantations are processed in palm oil mills. Note that, the shortest travel distance for the distribution of FFB is chosen in order to reduce the transportation cost. In scenario 2, the optimisation objective is set as maximise TCLoverall Eq(8). Based on the optimised results, TCLoverall score of 0.753 and EP of RM 12 x 106 per annum are obtained. It is noted that the TCLoverall has increased from 0.688 to 0.753 as compare with scenario 1. According to Table 3, TCL of the supply chain is also within the average traceability confidence level. Figure 3 shows the distribution and allocation of FFB across the supply chain for both scenarios. From Figure 3(b), only the plantations and collection hub with the highest OCL score are selected. The existence of interaction only happens between entities that give the highest ICL score in order to achieve maximum TCLOverall while fulfilling the required FFB of each palm oil mills. 881 (a) (b) Figure 3: Distribution and allocation of FFB for case study (a) scenario 1 (b) scenario 2 5. Conclusion and future works A mathematical model is developed to maximise economic performance while analyse traceability of palm oil supply chain. An effective quantitative scale and qualitative measure is presented to define traceability clearly. A case study with two scenarios is solved to illustrate the proposed approach. Based on the optimised result, the maximum economic performance of the investigated supply chain is reported as RM 16.8 x 106 per annum while the maximum traceability of the supply chain is reported as 0.753. The model can be used as an effective tool to quantify the traceability of a supply chain. For future work, multiple objective optimisation can be applied to trade-off between economics performance as well as traceability of a supply chain. The mathematical model can be further extended to consider downstream activities of the palm oil supply chain such as the refineries, oleochemicals as well as incorporates the carbon footprint across the supply chain. Works can also be done to implement this model in the other relating industries which account for traceability. References Yusof A., 2021, Palm Oil to still be a Powerhouse of Oils and Fats Market: MPOB, New Straits Times accessed 07.12.2020. Foong S.Z.Y., Goh C.K.M., Supramaniam C.V., Ng D.K.S., 2019a, Input-Output Optimisation Model for Sustainable Oil Palm Plantation Development, Sustainable Production and Consumption, 17, 31-46. Foong S.Z.Y., Andiappan V., Tan R.R., Foo D.C.Y., Ng D. K. S., 2019b, Hybrid Approach for Optimisation and Analysis of Palm Oil Mill, Processes, 7(2), 100. Gibon V., De Greyt W., Kellens M., 2007, Palm Oil Refining, European Journal of Lipid Science and Technology, 109(4), 315–335.. ISO, 2009, ISO 9000 Quality management- Fundamentals and Vocabulary’, 1–56. International Organization for Standardization accessed 25.11.2020. Lam H. L., Foo d. C. Y., Mustafar K., Klemeš J., 2010, Synthesis of Regional Energy Supply Chain Based on Palm Oil Biomass, Chemical Engineering Transactions, 21, 589–594, DOI: 10.3303/CET1021099. Lo S.L.Y., How B.S., Leong W.D., Teng S.Y., Rhamdhani M.A., Sunarso J., 2020, Techno-economic Analysis for Biomass Supply Chain: A state-of-the-art review. Renewable and Sustainable Energy Reviews, 135, 110164. Malaysian Palm Oil Council (MPOC), 2016, Palm Oil Fact , Malaysian Palm Oil Council accessed 23.05.2021. Grand View Research, Inc., 2020, Market Analysis Report: Palm Oil Market Size, Share & Trends Analysis Report By Origin (Organic, Conventional), By Product (Crude, RBD, Palm Kernel Oil, Fractionated), By End Use, By Region, And Segment Forecasts, 2020 - 2027 accessed 10.05.2021. 882