1.Introduction Industrial Symbiosis (IS) can be seen as a way to foster the cooperation required in a circular economy perspective to go beyond the linearity in the current production system. In particular IS implies that cooperation to interchange materials, energy, water and by-products should produce major benefits than the sum of individual achievable benefits from isolated entities [1]. Despite great potentialities for symbiosis, however, it still meet different barriers in its implementation. IS parks are generally composed by private industries and the economic driver appears to be the main driver for the network setting. At the same time, flows circularization can lead to worsen the performances of indicators with lower priority, like environmental or social ones [2], in case the flow allocation follows mere economic indicators. Moreover, due to local constraints and policies it appears difficult to translate symbiosis policies into environmental effects according to a life-cycle perspective: this understanding requires the modelling of a real case study and the further optimization of key sustainability indicators. 2.Materials and methods 2.1.Symby-net approach The research project Symbioptima is facing the problem of optimization of symbiosis activities in order to identify clustering opportunities for symbiotic collaboration within an industry network [3]. A series of specific methodologies for symbiotic assessment have been implemented within the software Symby-Net. Such tool focuses specifically on symbiotic flows that can be potentially or are actually circularized within the company network on a physical basis. Each symbiotic activity that is related to specific symbiotic flows contributes to the overall impact through a cradle-to-gate additional impact. Information is integrated as Independent Information Modules (IIMs) in order to calculate the LCIA of the IS for a specific configuration of industrial symbiosis [4]. The IS is intended as a flexible network in which quantities of exchanged symbiotic flows can vary and in which new players can be added to the network. 2.2.Optimization The problem of optimization for industrial symbiosis rises from different constraints in LCA modelling for IS. From modelling perspective, the IS can be complex and variance assessment should include actual limits for flow substitution and overall possible configuration of the IS. From a methodological perspective, midpoint environmental impact categories concurr competitively to determine criteria for optimization, furthermore these impact categories concurr competitively with economic and social impact categories. According to literature, the impact category normalization and weighting steps suffer from ex-ante policy methods that can interfere with single player policy objectives. Application of optimization to environmental effect of IS networks implies a solution for a multi-objective problem. The general objective is the impact minimization of a given set of impact categories related to sustainability of symbiosis activities within the IS. The user has the possibility to select which impact categories should be optimized. Since impact categories may have different units of measurement, one objective function is optimized, while the others are used as constraints following the epsilon-constraints method [5]. To solve the problem, a heuristic approach is developed according to two goals: the efficient identification of the epsilon thresholds used in the epsilon-constraints method and the smart modification of these epsilons in order to quickly obtain optimal solutions. The output of the optimization is the identification of a set of sub-optimal symbiotic scenarios: e.g. sub-optimal allocation of flows among IS companies. In general, the dimension of the problem can be as complex as the number of the interchangeable flows. For this kind of problem it is seldom possible to find a feasible solution that minimizes all the objective functions simoultaneously. Hence, the concept of Pareto dominance is introduced. Using this concept, a solution is said to be optimal if there is no other feasible solution that dominates it. 3.Results and discussion The formulated problem is still a max flow one to which more constraints have been added. These additional constraints are objective functions, one for each optimized indicators, whose values should be equal or lower of a certain threshold epsilon. This gives the name to the scientific method of the epsilon-constraints. In order to identify a set of Pareto dominating solutions, a heuristic method is implemented. The output of this algorithm is twofold: from one side it returns the Pareto dominating solutions and on the other side exploits unfeasibility. This aspect is very relevant from this point of view since can inform decision makers about the unfeasible frontier, namely the set of epsilon-values for which no feasible solution can be found. Indeed, if a threshold is too strict, it is possible that no solutions could be found while if the threshold is too loose, misleading conclusions could be obtained. As a matter of fact, the latter case would mean to accept that the optimal solution of the problem is the one provided by the algorithm while an indicator could have been further improved. Since the tool is thought for decisional support, accepting this solution could lead to avoidable impacts to the environment. For this reason, the proposed algorithm tries to dynamically identify the feasible region, in order to recognize the sensitivity of the threshold values and how these impacts on the selected objective function value. This produces a huge number of sub-problems to be solved: hence, it is necessary a smart way to generate the threshold values. 4.Conclusions The epsilon-constraints method is useful when working with the multi-objective optimization typical of the LCA field since it can deal with several units of measurement at the same time. This opens the possibility to optimize midpoint impact indicators, without the need of the normalization and weighting steps that could increase uncertainty. To obtain valuable data to support decisions in the IS, the threshold management is crucial. 5.References [1] Chertow M. 2000. Industrial symbiosis: Literature and Taxonomy. Annu. Rev. Energy Environ. 25:313-37. [2] Pakarinen S, Mattila T, Melanen M, Nissinen A, Sokka L. 2010. Sustainability and industrial symbiosis-The evolution of a Finnish forest industry complex. Resour. Conserv. Recycl. 54:1393-1404. [3] http://symbioptima.eu/index.php/project/description. Accessed 29/11/2017. [4] Buxmann K, Kistler P, Rebitzer G. 2009. Independent information modules - a powerful approach for life cycle management. Int. J. Life Cycle Assess. 14:92-100. [5] Miettinen K. 1998.Nonlinear Multiobjective Optimization. Springer. ISBN 978-0-7923-8278-2.
Optimization method to deal with LCIA trade-offs: the case of industrial symbiosis
Simone Cornago;Carlo Brondi;
2018
Abstract
1.Introduction Industrial Symbiosis (IS) can be seen as a way to foster the cooperation required in a circular economy perspective to go beyond the linearity in the current production system. In particular IS implies that cooperation to interchange materials, energy, water and by-products should produce major benefits than the sum of individual achievable benefits from isolated entities [1]. Despite great potentialities for symbiosis, however, it still meet different barriers in its implementation. IS parks are generally composed by private industries and the economic driver appears to be the main driver for the network setting. At the same time, flows circularization can lead to worsen the performances of indicators with lower priority, like environmental or social ones [2], in case the flow allocation follows mere economic indicators. Moreover, due to local constraints and policies it appears difficult to translate symbiosis policies into environmental effects according to a life-cycle perspective: this understanding requires the modelling of a real case study and the further optimization of key sustainability indicators. 2.Materials and methods 2.1.Symby-net approach The research project Symbioptima is facing the problem of optimization of symbiosis activities in order to identify clustering opportunities for symbiotic collaboration within an industry network [3]. A series of specific methodologies for symbiotic assessment have been implemented within the software Symby-Net. Such tool focuses specifically on symbiotic flows that can be potentially or are actually circularized within the company network on a physical basis. Each symbiotic activity that is related to specific symbiotic flows contributes to the overall impact through a cradle-to-gate additional impact. Information is integrated as Independent Information Modules (IIMs) in order to calculate the LCIA of the IS for a specific configuration of industrial symbiosis [4]. The IS is intended as a flexible network in which quantities of exchanged symbiotic flows can vary and in which new players can be added to the network. 2.2.Optimization The problem of optimization for industrial symbiosis rises from different constraints in LCA modelling for IS. From modelling perspective, the IS can be complex and variance assessment should include actual limits for flow substitution and overall possible configuration of the IS. From a methodological perspective, midpoint environmental impact categories concurr competitively to determine criteria for optimization, furthermore these impact categories concurr competitively with economic and social impact categories. According to literature, the impact category normalization and weighting steps suffer from ex-ante policy methods that can interfere with single player policy objectives. Application of optimization to environmental effect of IS networks implies a solution for a multi-objective problem. The general objective is the impact minimization of a given set of impact categories related to sustainability of symbiosis activities within the IS. The user has the possibility to select which impact categories should be optimized. Since impact categories may have different units of measurement, one objective function is optimized, while the others are used as constraints following the epsilon-constraints method [5]. To solve the problem, a heuristic approach is developed according to two goals: the efficient identification of the epsilon thresholds used in the epsilon-constraints method and the smart modification of these epsilons in order to quickly obtain optimal solutions. The output of the optimization is the identification of a set of sub-optimal symbiotic scenarios: e.g. sub-optimal allocation of flows among IS companies. In general, the dimension of the problem can be as complex as the number of the interchangeable flows. For this kind of problem it is seldom possible to find a feasible solution that minimizes all the objective functions simoultaneously. Hence, the concept of Pareto dominance is introduced. Using this concept, a solution is said to be optimal if there is no other feasible solution that dominates it. 3.Results and discussion The formulated problem is still a max flow one to which more constraints have been added. These additional constraints are objective functions, one for each optimized indicators, whose values should be equal or lower of a certain threshold epsilon. This gives the name to the scientific method of the epsilon-constraints. In order to identify a set of Pareto dominating solutions, a heuristic method is implemented. The output of this algorithm is twofold: from one side it returns the Pareto dominating solutions and on the other side exploits unfeasibility. This aspect is very relevant from this point of view since can inform decision makers about the unfeasible frontier, namely the set of epsilon-values for which no feasible solution can be found. Indeed, if a threshold is too strict, it is possible that no solutions could be found while if the threshold is too loose, misleading conclusions could be obtained. As a matter of fact, the latter case would mean to accept that the optimal solution of the problem is the one provided by the algorithm while an indicator could have been further improved. Since the tool is thought for decisional support, accepting this solution could lead to avoidable impacts to the environment. For this reason, the proposed algorithm tries to dynamically identify the feasible region, in order to recognize the sensitivity of the threshold values and how these impacts on the selected objective function value. This produces a huge number of sub-problems to be solved: hence, it is necessary a smart way to generate the threshold values. 4.Conclusions The epsilon-constraints method is useful when working with the multi-objective optimization typical of the LCA field since it can deal with several units of measurement at the same time. This opens the possibility to optimize midpoint impact indicators, without the need of the normalization and weighting steps that could increase uncertainty. To obtain valuable data to support decisions in the IS, the threshold management is crucial. 5.References [1] Chertow M. 2000. Industrial symbiosis: Literature and Taxonomy. Annu. Rev. Energy Environ. 25:313-37. [2] Pakarinen S, Mattila T, Melanen M, Nissinen A, Sokka L. 2010. Sustainability and industrial symbiosis-The evolution of a Finnish forest industry complex. Resour. Conserv. Recycl. 54:1393-1404. [3] http://symbioptima.eu/index.php/project/description. Accessed 29/11/2017. [4] Buxmann K, Kistler P, Rebitzer G. 2009. Independent information modules - a powerful approach for life cycle management. Int. J. Life Cycle Assess. 14:92-100. [5] Miettinen K. 1998.Nonlinear Multiobjective Optimization. Springer. ISBN 978-0-7923-8278-2.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.