PRES22_0231.docx DOI: 10.3303/CET2294209 Paper Received: 26 May 2022; Revised: 05 June 2022; Accepted: 13 June 2022 Please cite this article as: Chin H.H., Varbanov P.S., Klemeš J.J., Tan R.R., Aviso K.B., 2022, Plastic Waste Circularity with Data-Driven Approach Considering Polymer Heterogeneity , Chemical Engineering Transactions, 94, 1255-1260 DOI:10.3303/CET2294209 CHEMICAL ENGINEERING TRANSACTIONS VOL. 94, 2022 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, Sandro Nižetić Copyright © 2022, AIDIC Servizi S.r.l. ISBN 978-88-95608-93-8; ISSN 2283-9216 Plastic Waste Circularity with Data-Driven Approach Considering Polymer Heterogeneity Hon Huin China,*, Petar Sabev Varbanova, Jiří Jaromír Klemeša, Raymond R. Tanb, Kathleen B. Avisob aSustainable Process Integration Laboratory – SPIL, NETME Centre, Faculty of Mechanical Engineering, Brno University of Technology - VUT Brno, Technická 2896/2, 616 69 Brno, Czech Republic bDepartment of Chemical Engineering, Gokongwei College of Engineering, De La Salle University, 2401 Taft Avenue, 0922 Manila, Philippines chin@fme.vutbr.cz Plastic debris has been a consistent issue in the global environment, and plastic waste recycling is the most promising option to avoid further accumulation. The waste quality serves as a crucial restriction for recycling planning and requires proper definition. This work aims to extend the previously developed data-driven approach to quantifying the recyclability clusters of plastic waste to consider polymer heterogeneity in the evaluation. Heterogeneity estimation is conducted by identifying the compatibility between polymers based on their surface tension. The applicability of the polymers mix (targeted and non-targeted polymers) can then be decided based on the Q-value approach. The method identifies the quality class of the plastic mixture based solely on the compatibility of the polymer, where the quality trend varies for each identified compatibility class. This gives insights into the suitability of mixing different polymers type prior to recycling. The recycling potential/circularity of the plastic waste can then be identified based on the Plastic Pinch Analysis, which outputs the ideal maximum external plastic demands with a certain threshold grade of the plastic (Pinch Quality). A case study is shown using three types of polymers: Polyethylene Terephthalate (PET), Polyethylene (PE) and Polypropylene (PP), to showcase the polymers heterogeneity evaluation. The results show that around 32.4 % of disposed PP waste could potentially be mixed with PE to have a compatible mixture. However, it is also crucial to check the properties of the mixed polymers to fulfil the demands of site requirements prior to recycling. 1. Introduction Plastic waste accumulation has been a worldwide issue for decades and requires considerable attention to solve the problem systematically. The main priority is to design a systematic plastic reusing and recycling framework to reduce further the environmental impacts of plastic waste, especially when this issue became more alarming during the COVID-19 pandemic due to high levels of personal protective equipment use, overpackaging of delivered goods, and mass vaccination (Klemeš et al., 2021). Previous plastic recycling studies can be found. Huysman et al. (2017) proposed that the compatibility between polymers plays the main role in deciding the recycling options for plastic waste. The waste substitution ratio is also mainly decided by the compatibility between polymers. Gradus et al. (2017) found that the cost of plastic treatment in Dutch recycling facilities can be reduced using post-separation of waste rather than source separation. Eriksen et al. (2019) derived several categories for plastic waste mainly based on their application types, including food application, medicine or toys. Their assessment is mainly dependent on two main properties: polymeric composition and other general residues of the polymers. Faraca and Astrup (2019) analysed the waste in Denmark facilities and categorised the waste in terms of applicability, impurity, lifetime, and polymeric compositions. Brouswer et al. (2020) proposed a qualitative categorisation of plastic package waste depending on the physical and chemical properties, also mainly depends on the application categories for the plastic packages. Min et al. (2020) have collected the critical factors for ocean plastic degradation from various literature sources and produced a decision support model to evaluate the critical factors in identifying 1255 plastic degradation. Varbanov et al. (2021) have recently proposed a data-driven prototype for quality-oriented plastic waste recycling, and extension by Chin et al. (2022) to consider various properties. The approach gathers different complex properties of plastic waste and identifies the clusters based on the available data set, where each cluster represents quality grades. This can be subjective to different opinions as there are various properties that can be considered the critical factor. A critical gap remains in the lack of a well-defined plastic waste recyclability index considering the heterogeneity of polymers. Huysman et al. (2017) demonstrated a possible method for evaluating the compatibility between polymers, but it was not indicative enough for the quality of plastic waste as the properties of the polymers are still the major restrictions. Plastic waste trading is still restricted by the current recycling capacity due to the lack of a plastic chemical information system and compatibility assessment. This work aims to extend the work from Chin et al. (2022) in quantifying the qualities of the plastic waste and coupled with the compatibility estimation from Huysman et al. (2017) to derive a proper plastic waste grading system. 2. Data-driven Plastic Pinch Analysis The use of a data-driven framework to quantify the quality clusters for plastic waste streams is proposed. The plastic waste samples are first distinguished into various polymer types. The property data samples for each polymer type are then fed into the Machine Learning model to learn their quality patterns. The Pinch Analysis method followed up by Machine Learning allows users to gain crucial intuitions on the resource quality bottlenecks of the current production or recycling system. The quality grades from the Machine Learning model can be used as the quality indicator in the Pinch Analysis framework. Pinch Analysis is an efficient framework to target minimum external resource requirements, and the solutions can be presented in a graphical manner – see Figure 1. More information on the Machine Learning approach can be found in Chin et al. (2022). Figure 1: Plastic Pinch Analysis framework in evaluating minimum virgin plastic and disposed waste 3. Polymer heterogeneity Plastic Pinch Analysis could potentially identify the minimum virgin plastic flow, but it ignores the heterogeneity of the polymers mix. In reality, it is difficult to achieve full separation of the plastic waste streams into individual polymer types. Different polymer types can have different chemical natures. Mixing them could cause an immiscible mixture which can jeopardise the properties of the waste, especially mechanical properties (Lei et al., 2009). Huysman et al. (2017) proposed the waste quality prediction approach mainly based on the compatibility between polymers in a mix. The assumptions of the approach are the mixture is binary (two polymer types), the effect of thermo-mechanical degradation is not accounted for, and no additional agent is added to the mixture. The compatibility is determined by the interfacial tension between polymers in a molten state. Other approaches are available such as Flory-Huggins parameters, Hildebrand solubility parameters or mixing enthalpy function. The interfacial tension method is more straightforward and can identify the miscibility between two fluids easily. Other approaches can be used, and results comparison can be made. The interfacial tension can be regarded as the surface energy (γij) for the polymers ‘i’ and ‘j’, which are contributed by the intermolecular forces, dispersion forces and polar forces (Fowkes, 1964). Wu (1971) has proposed a geometric average mixing rule for surface tension estimation for binary polymers to account for the dispersion (γid) and polar components (γip) of the polymers, where ‘i’ represents the polymer types – see Eq(1), where γ12, γ1, γ2 represent the respective surface tension for mixture, polymer 1 and polymer 2. The estimation has an accuracy of close to 96 % as the polar interaction is accounted. This mixing rule is more commonly used for predicting the surface energy of polymers as compared to other available methods. 1256 𝛾𝛾12 = 𝛾𝛾1 + 𝛾𝛾2 − 4 � 𝛾𝛾1𝑑𝑑𝛾𝛾2𝑑𝑑 𝛾𝛾1𝑑𝑑 + 𝛾𝛾2𝑑𝑑 � − 4 � 𝛾𝛾1𝑝𝑝𝛾𝛾2𝑝𝑝 𝛾𝛾1𝑝𝑝 + 𝛾𝛾2𝑝𝑝 � (1) Huysman et al. (2017) conducted the calculation of the binary surface tension between the commonly found polymers, and their results are presented in Table 1. They have derived various compatibility classes based on the range of interfacial tension γ12 for binary mixture, i.e., perfectly compatible, reasonably compatible, limited compatible and incompatible. In addition to the surface tension estimation, they have proposed a quality value called the Q-values for each compatibility class. The Q-value represents the quality of the plastic waste based solely on the polymer compatibility. The range of surface tension for each compatibility class is presented in Table 2. For example, the perfectly compatible mixture is the LPE/PE since their surface tension falls on the ‘perfectly compatible’ class. PP/PE pair is still reasonable compatible, but PP pairs with PS have limited compatibility. The completely not compatible mixture is PE, PP and LPE with PA. Table 1: Surface tension for common polymers (γ12), in mN/m, data obtained from Huysman et al. (2017) Low-density polyethylene (LPE) Polyethylene (PE) Polypropylene (PP) Polystyrene (PS) Polyvinyl chloride (PVC) Polyamide (PA) Polyethylene Terephthalate (PET) LPE - 0.002 0.477 6.121 2.192 14.15 9 PE 0.002 - 0.413 6.109 2.236 14.12 9.001 PP 0.477 0.413 - 6.4 3.27 14.09 9.46 PS 6.121 6.109 6.4 - 2.413 3.16 0.574 PVC 2.192 2.236 3.27 2.413 - 9.68 4.657 PA 14.15 14.116 14.092 3.165 9.681 - 0.012 PET 9 9.001 9.46 0.574 4.657 0.012 - Table 2: Compatibility class for different ranges of binary surface tension. Data obtained from Huysman et al. (2017) Compatibility Min γ12 Max γ12 Perfectly compatible 0 0.1 Reasonably compatible 0.1 1 Limited compatible 1 10 Not compatible 10 - In this work, the Q-value approach is utilised to analyse the potential sources and sinks in the plastic recycling system. Based on the Plastic Pinch Analysis, the disposed of plastic waste from different polymers could be mixed to achieve a potential recyclable plastic for other processes. This study proposes and demonstrates the evaluation of polymers heterogeneity based on the results from previous Plastic Pinch Analysis. Figure 2 shows the potential integration of both methods, where the disposed waste identified from Pinch Analysis can be mixed to produce a new stream of sources that can be used for other processes. Figure 2: Potential Integration of Plastic Pinch Analysis and Q-value approach 1257 The Q-value is dependent on the number of contaminated polymers. Depending on a certain range of the Q- value, the recycling options for the plastic waste can be decided, and they have suggested four classes for the plastic waste per compatibility class. Figure 3 shows the graphical correlations of Q-value between polymers contamination for each compatibility class. Closed-loop represents the waste that can fully substitute virgin plastic to manufacture the same product. Waste in a semi-closed loop can partially substitute virgin plastic for the same product. Waste in an open-loop cannot be reused for the same product but can be used for other products, while incineration represents the waste is not reusable/recyclable. It is worth noting that Q-value drops at a faster rate when the compatibility is poorer. Note that the Q-value is based on the results from previous literature and should be validated with an empirical approach in the future. However, this preliminary model could shed light on the heterogeneity of polymers. Note that the perfectly compatible mixture is assumed to be of high quality than can be recycled close-loop – see definition of the recycling options in Huysman et al. (2017). (a) Reasonable compatible (b) Limited compatible (c) Not compatible Figure 3: Q-values vs contaminating polymers for each compatibility class, adapted from Huysman et al. (2017). 1258 4. Case study demonstration A simple demonstration of the Q-values approach in evaluating the compatibility between waste sources is conducted based on the results of the Pinch Analysis from Chin et al. (2022). The authors have identified the minimum external plastics required for a case study considering three polymer types: PE, PET and PP, where each polymer type has disposed of plastics waste that is not usable for the demands. The disposed waste can be mixed, and their compatibility can be evaluated. Not only the disposed waste, the sources that are recycled can also be mixed to enhance the polymer grades if possible, and in turn reduce further the virgin plastic demands``. This work considers different scenarios of a plastic mix between polymer types. Table 3 below shows the extracted plastic source data used in this work. The data from Table 3 indicates the results from Pinch Analysis, where about 500 t of PET is disposed of, and 100 t of PP is disposed of, while PE is fully reused. The disposed waste can be mixed to check for the compatibility class. However, since PET and PP have limited compatibility due to their binary surface tension being 9.001 – see Tables 2 and 3. Mixing them should not produce a mixture that can be fully recycled. Figure 4 shows the Q-values of the PET/PP mixture when the mixture is mixed with 50 % of SR3-PET and 50 % of SR3-PP. It can be clearly seen that the mixture falls into the ‘incineration’ class, which seems it is unwise to mix them. A similar case is observed with mixing SR1-PE even with grade 1. For a good quality PET/PE/PP mixture, one of PET or PE/PP should be as low as 10 % to be usable. Table 3: Sources data with their polymeric compositions. Data retrieved from Chin et al. (2022). Binary mixture surface tension between two major polymers is considered only Targeted polymer types Grades Compositions (wt %) Q-value Sources PET PP PE SR1- PET 2 0.661 0.008 0.330 0.2-0.7 (Open loop) SR2 PET 2 0.930 0.001 0.070 0.7-0.9 (Semi-closed) SR3 PET (Disposed 500 t) 3 0.968 0.000 0.031 0.7-0.9 (Semi-closed) SR1 PP 1 0.001 0.845 0.154 0.7-0.9 (Semi closed) SR2 PP 1 0.001 0.925 0.074 0.9-1 (Closed) SR3 PP (Disposed 100 t) 4 0.000 0.984 0.016 0.9-1 (Closed) SR4 PP 1 0.001 0.914 0.085 0.9-1 (Closed) SR1 PE 2 0.189 0.001 0.811 0.7 – 0.9 (Semi closed) SR2 PE 1 0.046 0.000 0.954 0.9 – 1 (Closed) SR3 PE 4 0.047 0.001 0.952 0.9 – 1 (Closed) SR4 PE 2 0.437 0.087 0.476 0.2 – 0.7 (Open) As for the option of mixing disposed PP with PE waste (SR1-PE), it can be seen that the mixture (with a 50:50 mix ratio of two streams) is still considered to be open-loop recycling waste even though they are reasonably compatible. In order to achieve a quality of waste to be at least usable for semi-closed loop recycling, the PP waste streams should be reduced. Based on Figure 5, the amount of SR3-PP should be reduced to 32.4 %. Mixing different polymers could cause compatibility issues, but mixing polymers with the same types should not be a major issue. However, the mixture compatibility is still to be checked since they are contaminated with other polymers. Mixing SR3-PP with the high-quality SR2-PP seems to produce a compatible and high-quality waste, even with a 50:50 mix ratio between two streams- see Figure 5. In this case, the disposed waste can be reduced. However, this analysis only shows the compatibility evaluation, and the mixture properties are still to be experimentally validated, and the final properties of the plastic mixture should be checked prior to recycling. Figure 4: Q-values of mixing PET waste streams with PP or PE streams 1259 Figure 5: Q-values of mixing PP waste streams with PP or PE streams 5. Conclusions This work has developed a simple evaluation framework to evaluate the compatibility of the polymers between two waste streams. The method is based on the Q-values as a function of contaminated polymers. To facilitate a systematic recycling planning, the data-driven Plastic Pinch Analysis can be integrated with the Q-values approaches to identify the quality predictors of the plastic waste streams, which in turn ideally minimise the virgin plastic demands for any recycling system. From the case study, it is shown that mixing PE with PP is plausible, but the mass share of the PP waste streams should be reduced to about 32.4 %. Mixing two limited compatible waste is possible but only with a minute amount. However, this framework is preliminary and requires more advanced improvements and experimental validations. Mixtures with multi-polymers with proper surface tension mixing rules should be investigated in the future. The Q-value approach mainly accounts for polymers compatibility. The final properties of the plastic waste should be evaluated prior to the recycling decision. Acknowledgements Gratefully acknowledged Support from the Grant Agency of the Czech Republic under Project No. 21-45726L in a bilateral collaborative project with the Ministry of Education, Science and Sport of the Republic of Slovenia (project No. 5442-1/2018/106), the Slovenian Research Agency (research core funding No. P2-0412 and P2- 0421 and project No. J7-3149). References Brouswer M., Velzen E.U.T.V., Ragaer, K., Klooster R.T, 2020. Technical Limits in Circularity for Plastic Packages. Sustainability, 12, 10021. Chin H.H., Varbanov P.S., You F., Sher F., Klemeš J.J., 2022. Plastic Circular Economy Framework with Hybrid of Machine Learning and Pinch Analysis. Resources, Conservation and Recycling, 184, 106387. Eriksen M.K., Christiansen J.D., Daugaard A.E., Astrup, T.F, 2019. Closing the loop for PET, PE and PP waste from households: Influence of material properties and product design for plastic recycling. Waste Management, 96, 75 –85. Faraca G., Astrup T., 2019. Plastic waste from recycling centres: Characterisation and evaluation of plastic recyclability. Waste Management, 95, 388-398 Fowkes F.M., 1964. Attractive forces at interfaces. Industrial & Chemistry Engineering Research, 56, 40–52. Gradus R.H.J.M., Nillesen P.H.L., Dijkgraaf E, van Koppen E.J., 2017. A cost-effectiveness analysis for incineration or recycling of Dutch household plastic waste. Ecological Economics, 135, 22-28. Huysman S., Schaepmeester J. D., Ragaert K., Dewulf J., Meester S.D., 2017. Performance indicators for a circular economy: A case study on post-industrial plastic waste. Resources, Conservation and Recycling, 120, 46-54. Klemeš J.J., Jiang P., Fan Y.V., Bokhari A., Wang X-C, 2021. COVID-19 pandemics Stage II – Energy and environmental impacts of vaccination. Renewable and Sustainable Energy Reviews, 150, 111400. Min K., Cuffi J.D., Mathers R.T., 2020. Ranking environmental degradation trends of plastic marine debris based on physical properties and molecular structure. Nature Communications, 11, 727. Wu S., 1971. Calculation of interfacial tension in polymer systems. Journal of Polymer Science Part C: Polymer Symposia, 34(1), 19–30. Varbanov P.S., Chin H.H., Jia X., Klemeš J.J., 2022. Environmental Impacts of Plastic Materials Flow Minimisation Using Data-Driven Pinch Method. Chemical Engineering Transactions, 88, 967-972. 1260 https://www.sciencedirect.com/science/article/pii/S0956053X1930426X https://www.sciencedirect.com/science/article/pii/S0956053X1930426X https://www.sciencedirect.com/science/article/pii/S0921800916306462 https://www.sciencedirect.com/science/article/pii/S0921800916306462 PRES22_0475.pdf Plastic Waste Circularity with Data-Driven Approach Considering Polymer Heterogeneity