Collaboration Platform for Enabling Industrial ... - ScienceDirect.com

Available online at www.sciencedirect.com

ScienceDirect Procedia CIRP 61 (2017) 263 – 268

The 24th CIRP Conference on Life Cycle Engineering

Collaboration Platform for Enabling Industrial Symbiosis: Application of the By-product Exchange Network Model Benjamin Raabea, Jonathan Sze Choong Lowb*, Max Jurascheka, Christoph Herrmanna, Tobias Bestari Tjandrab, Yen Ting Ngb, Denis Kurlea, Felipe Cerdasa, Jannis Lueckengaa, Zhiquan Yeob, Yee Shee Tanb a

Sustainable Manufacturing & Life Cycle Engineering, Institute of Machine Tools and Production Technology (IWF), Technische Universität Braunschweig, Langer Kamp 19b, 38106 Braunschweig, Germany Singapore Institute of Manufacturing Technology, Agency for Science, Technology and Research, 2 Fusionopolis Way, #08-04, Innovis, Singapore 138634

b

* Corresponding author. Tel.: +65-6319-4430; fax: ++65-6250-3659. E-mail address: [email protected]

Abstract Industrial symbiosis is enabled when physical exchanges of by-products take place between companies within and across different industries. It can be an important strategy in realising the vision of “zero waste to landfill” especially for a land-scarce country like Singapore. The success of industrial symbiosis relies on companies working collectively and collaboratively with the common goal of mutual economic and environmental sustainability. However, one major barrier faced by companies is the lack of information and knowledge of what are possible and available as by-product exchanges. With this as the motivation, this paper introduces the system architecture of a collaboration platform for enabling industrial symbiosis. Within the system architecture, we focus on the by-product exchange network (BEN) model. Then using a case study of food waste in Singapore, we demonstrate the application of the model as a decision support tool for companies to evaluate the economic viability of industrial symbiosis. © 2017 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license © 2017 The Authors. Published by Elsevier B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). under responsibility of the scientifiof c committee of the 24th CIRP Conference LifeCIRP Cycle Engineering Peer-review Peer-review under responsibility the scientific committee of theon 24th Conference Keywords: industrial ecology; urban metabolism; eco-industrial park; Circular Economy; resource productivity

1. Introduction In less than 10 years from 2003 to 2012, global waste generation per capita has increased by more than 87% [1]. According to a 2013 Nature article by Hoornweg et al., the world is generating waste faster than any other environmental pollutants including greenhouse gases [2]. And at this rate, waste generation is expected to triple and exceed 11 million tonnes per day by 2100. Because of this waste situation, Singapore and governments around the world are seeking better strategies to deal with the escalating cost of waste management. The challenge with the waste situation is even more acute in land-scarce countries like Singapore. In 2015, 1.4 tonnes of waste were generated per capita. This amounted to almost 7.7 million tonnes of waste generated in the small island city-state of only 719.1km2 for that year alone [3]. Of the total amount,

on Life Cycle Engineering.

61% was sent for recycling and the rest incinerated. Although the recycling rate seems high, in order to keep up with Singapore’s growing population coupled with the increasing rate of consumption per capita, a new incineration plant will need to be built every five to seven years and a new landfill every 25 to 30 years. This is obviously unsustainable especially for a land-scarce country like Singapore. But, in the several million tonnes of waste generated potentially lies value and opportunities for companies to exploit. Besides the obvious recycling of metals and plastics, companies can potentially use by-products, which are traditionally regarded as wastes, as substitutes for raw materials. This can create what is known as industrial symbiosis whereby by-products and wastes are physically exchanged between different companies from within and across industries.

2212-8271 © 2017 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 24th CIRP Conference on Life Cycle Engineering doi:10.1016/j.procir.2016.11.225

264

Benjamin Raabe et al. / Procedia CIRP 61 (2017) 263 – 268

One of the most successful and well-known cases of industrial symbiosis can be observed in the city of Kalundborg, Denmark [4]. Numerous projects have tried to replicate the case and since then, many tools have been developed for planning and enabling industrial symbiosis. In the review article by Grant et al. [5], the authors identified 17 such tools which were developed during the period of 1997 to 2009. Most of them were developed in an academic environment and hence, not deployed beyond the application on specific case studies. Moreover, of these 17, nine are already inactive and only one is publicly accessible. More recently, the idea of leveraging big data analytics to facilitate the enablement of industrial symbiosis was also explored [6]. Based on this and further review of the current state-of-theart, four tools stood out: Core Resource for Industrial Symbiosis Practitioners (CRISP), Bourse des résidus industriels du Québec (BRIQ), eSymbiosis and RecycleMatch. Most of them do not go beyond a simple input-output matching of different resource streams from industrial entities. More specifically, they do not provide decision support in terms of analysing the economic viability of potential symbiotic exchanges among different entities, which poses one of the biggest barriers in enabling industrial symbiosis [7]. Therefore, with this as the motivation, we are proposing the concept of a collaboration platform for enabling industrial symbiosis. In this paper, we introduce the system framework for the collaboration platform. Within this framework, the different subsystems and how they function together are explained with a focus on the by-product exchange network (BEN) model. Finally, using a case study of food waste in Singapore, we demonstrate the application of the model as a decision support tool for companies to evaluate the economic viability of industrial symbiosis. 2. System Architecture of Collaboration Platform The system architecture for the proposed concept of the collaboration platform for enabling industrial symbiosis

consists of two main modular subsystems: the waste-toresource matching subsystem and the industrial symbiosis simulation subsystem. Figure 1 is a schematic representation of the system architecture with the subsystems and their components segmented based on the three functional layers: data, logic and presentation. 2.1. Waste-to-Resource Matching Subsystem The main function of waste-to-resource matching subsystem is to determine technically feasible waste-to-resource matches, i.e. possible options for recovering value from wastes or byproducts regardless of economic viability. It maintains and updates information about wastes and by-products, and analyses the data to match the wastes and by-products to resources that they could possibly substitute. This subsystem consists of three major components: input interface, waste-toresource matches database and waste-to-resource matching engine. In the context of industrial symbiosis, wastes or by-products of one entity (factory or facility) can potentially be resources for another. Hence, without discriminating between virgin resources, by-products and wastes, a waste-to-resource database holds information about resources, their properties and possible substitutes. The information is structured based on a taxonomy which classifies and helps distinguish between different types of resources. The input interface of the subsystem allows maintenance and updates of the waste-toresource matches database to be carried out by qualified users such as a waste-to-resource knowledge engineer. Although the capability of the system largely depends on the scope and accuracy of the waste-to-resource matches database, the actual process of determining the possible waste-toresource matches is handled by a waste-to-resource matching engine. The engine first analyses the data about resources consumed, and by-products and wastes produced by an entity on the collaboration platform, which is stored in the network entity information database of the industrial symbiosis

Figure 1. System architecture of the collaboration platform for enabling industrial symbiosis.


265

simulation subsystem. It then uses the results to retrieve from the waste-to-resource database, a set of technically feasible options for substituting the entity’s virgin resources with wastes or by-products, and a set of technically feasible options for utilising its by-products and wastes as substitutes for raw materials. With these sets of technically feasible options, the engine searches in the network entity information database for other entities that can fulfil the requirements of these technically feasible options. The results of the search are then stored in the network entity matches database of the industrial symbiosis simulation subsystem. 2.2. Industrial Symbiosis Simulation Subsystem The main function of the industrial symbiosis subsystem is to provide the next-level decision support in terms of evaluating the economic viability of technically feasible waste-to-resource matches as determined by the waste-to-resource matching subsystem. It maintains and updates information about the entities on the collaboration platform, and analyses the data with the results from the waste-to-resource matching subsystem to help find economically viable and favourable symbiotic exchanges of by-products between the entities. This subsystem consists of five major components: input interface, network entity information database, network entity matches database, by-product exchange network (BEN) model and results interface. The network entity information database contains the name and location of entities on the collaboration platform, their inputs in terms of the types and quantities of resources consumed, and their outputs in terms of the types and quantities of wastes and by-products produced. The input interface of the subsystem allows entity information to be maintained and updated by authorised users on the collaboration platform such as an entity owner who is a member of the collaboration platform. As for the network entity matches database, it stores information about possible symbiotic exchanges of by-products between the entities based on technically feasible waste-toresource matches as determined by the waste-to-resource matching subsystem. This database is dynamically updated as new waste-to-resource matches are determined or when the network entity information database is updated with new information. Based on the information in the two databases, the BEN model tries to look for and simulates the most economically favourable exchanges of by-products between entities on the collaboration platform. This capability enables the what-if analysis and evaluation of different industrial symbiosis scenarios. The results of the simulation and analysis are presented to the user by the results interface of the industrial symbiosis simulation subsystem. A more detailed procedure showing how the industrial symbiosis simulation subsystem works is provided in Figure 2. Details of the BEN model development is provided in the following section.

Figure 2. Procedure of the industrial symbiosis simulation subsystem.

3. By-product Exchange Network (BEN) Model Development The BEN model is developed using an agent-based modelling approach. In this model, entities (factories or facilities) in the industrial symbiosis network are represented by agents that are programmed based on simple rules to actively consume and/or produce resources. Resources are represented by agents that passively change their states (i.e. quantities and locations) in response to them being produced and consumed by entities, and transported between entities. The modes in which resources are transported (e.g. trucks and pipelines) are represented by agents that are imbued with the motivation or behaviour of the entities they are acting for, i.e. their owners. For the work in this paper, the behaviour of the transportation agents’ owners is assumed to be driven financially. This means that transportation agents are programmed with the objective function to maximise the value of transactions when exchanging resources for their owners. The objective function can be written as:

266


Maximise: (1)

cost is taken into account whenever any resource is exchanged between entities. From equations (1) to (6), the economic performance of an entity in the industrial symbiosis network can be computed as

Subject to: xijm,r,t t 0

(2)

GMi ,t

0 d yijm,r,t d 1

(3)

where GMi,t is the gross margin (measure of economic

(4)

performance) of entity i at time t. OCi,t is the overhead cost of

¦ n

TVim,r ,t

v j ,r ,t xijm,r ,t

c

m

d ijm

yijm,r ,t

j 1

n

¦

xijm,r ,t

(7)

r ,m

Qim,r ,t

j 1

Qim,r ,t d Lm

¦Tim,r,t OCi,t

(5)

(6) xijm,r,t 1 yijm,r,t 0 where i is the entity of interest (owner), j the entity which entity i can exchange resources with, m the identifier for the transportation agent, n the number of entities which entity i can exchange resources with, r the resource type, and t the time. TVim,r,t is the transaction value of exchanging resource r at time t via transportation (agent) m for entity i (owner). v j,r,t is the value per unit of resource r that can be exchanged at entity j at time t. xijm,r,t is the quantity of resource r to be purchased (negative value indicating cost incurred from purchase of resource) or sold (positive value indicating revenue from sale of resource) by entity i from entity j via transportation m. c m is the cost per distance of using transportation m and d ijm the distance that needs to be travelled by transportation m between entities i and j. yijm,r,t is the binary decision variable indicating if resource r is exchanged between entities i and j at time t via transportation m. Qim,r,t is the order quantity of resource r to be purchased or sold by entity i at time t via transportation m. Lm is the load (carrying) capacity of transportation m. The objective function incorporates the economic value of the resource to be exchanged and the relevant transportation cost. Constraint (4) ensures that sufficient quantities of a resource are exchanged by the transportation agent in order to meet its order quantity designated to it by its owner. Constraint (5) limits the order quantity and thus, the overall quantity of a resource to be exchanged by the transportation agent to its carrying capacity. Constraint (6) ensures that the transportation

(operating) entity i at time t. At this point, it is worth noting that although the model attempts to optimise the industrial symbiosis network for resource exchange, this is not its main focus. There are existing models that are more specialised in optimising such networks. In this paper, the aim of the BEN model is to simulate the behaviour of the entities participating in resource exchanges within the network. Figure 3 is the geographic information system (GIS) implementation of the BEN model using the AnyLogic (Personal Learning Edition) software. 4. Case Study In order to demonstrate the application of the by-product exchange network (BEN) model within the system architecture of the collaboration platform for enabling industrial symbiosis, a case study of biodiesel production from waste cooking oil (WCO) in Singapore is used. This case study together with the data used is based on a producer of WCO-derived biodiesel in Singapore [8]. This is complemented with data from available information on the web and interviews with the company [9]. Currently, they have one big refinery at the city border of Singapore. Their WCO feedstock is supplied by their commercial (industrial symbiosis) partners, including schools and restaurants, who are more centrally located in Singapore. In addition, there are several community collection points from which they collect WCO once or several times a week. This current scenario will be used as the base case for analysis in the case study. Using the BEN model, there are different types of what-if scenarios that could be simulated. For example, simulating the

Figure 3. Geographic information system (GIS) implementation of the by-exchange network (BEN) model using the AnyLogic (Personal Learning Edition) software.


what-if scenario of a new competitor entering the market and outbidding the company’s bidding price for WCO. Or simulating the what-if scenario of a cheaper source or even substitute for WCO. For the scope of this paper, the case study is focused on simulating the what-if scenario of decentralising the big biodiesel refinery at the outskirt of the city into two smaller identical refineries closer to the city centre and thus, closer to the WCO suppliers and collection points. This what-if scenario has two main effects. First, the distances to and from the WCO suppliers and collection points are reduced since the two smaller refineries are located nearer to them. This means the company could benefit from an improvement in lead time and transportation efficiency. However, it is worth noting again that the aim here is not to optimise the WCO collection route and biodiesel delivery route. Rather, the motivation is to understand the effects of the change in physical setup and locations of biodiesel refineries in the what-if scenario. Second, instead of one big refinery, there are now two smaller refineries consuming WCO to produce biodiesel. This means higher operating costs due to the extra overheads incurred from operation of two separate sites. For this, the economic performance of the company is measured by addition of the gross margins of the two smaller refineries based on equation (7). The parameters which are common across the base case and what-if scenarios are summarised in Table 1. The parameters which are dependent on the different scenarios are summarised in Table 2.

To further analyse the difference in monetary terms, Figure 4 is a sensitivity analysis of the transportation cost savings in the what-if-scenario relative to the base case scenario. As transportation cost is also affected by the load (weight), we assume that it is proportional to the quantity of resources exchanged by the transportation agent, which in turn generates revenues for the entities when they are sold. Therefore, as indicated in Table 2, the transportation cost per 1,000 km is calculated as a percentage of the monthly revenue of the company and used as the input for c m in equation (1). From the sensitivity analysis, we can see that regardless of the range of transportation cost that is within the 0.5% to 1.5% of the company’s monthly revenue, there is savings in the what-if scenario, i.e. decentralisation of the big biodiesel refinery into two smaller centrally-located refineries so that they are closer to the WCO suppliers and collection points. And based on the graph’s linear upward trend, it is clear that the what-if scenario will be more favourable the higher the transportation cost per distance is. Besides the transportation cost, another effect that is to be investigated is the higher operating costs due to the extra overheads incurred from the operation of two separate sites in the what-if scenario. Figure 5 is a sensitivity analysis of the additional overhead cost in the what-if-scenario relative to the

Table 1. Common parameters across the base case and what-if scenarios. Parameter

Value

Number of commercial partners Number of community collection points Total monthly WCO collection WCO price Total monthly biodiesel production output Diesel price Monthly revenue Total number of trucks (transportation) Truck capacity

57 7 54,000 litres S$0.30 per litre 54,000 litres S$1.70 per litre S$91,800.00 2 500 litres

Figure 4. Sensitivity analysis of transportation cost savings in the what-ifscenario relative to the base case scenario.

Table 2. Parameters dependent on the base case and what-if scenarios.

Parameter Number of refineries Average distance to collection points Transportation cost per 1,000 km as a percentage of monthly revenue Overhead cost as percentage of monthly revenue

Base Case Scenario

What-if Scenario

1 27.6 km

2 11.17 km

0.5 to 1.5%

0.5% to 1.5%

7%

7% to 17%

For this case study, some results are reported in the form of relative differences due to the sensitivity of the company’s financial information. From the simulation, the most apparent difference is in the total monthly distances covered by the trucks. In the what-if scenario, the company’s trucks covered a total distance of 2,028 km, which is less than half of the 4,407 km covered by the trucks in the base case scenario.

Figure 5. Sensitivity analysis of additional overhead cost in the what-if scenario relative to the base case scenario.

267

268


base case scenario. To calculate overheads, we assume that the cost is a variable component of the operating cost, which is proportional to revenue generation. In other words, increased revenue generation requires increased operating expenses to drive up biodiesel production. Therefore, like the transportation cost and as indicated in Table 2, the overhead cost is calculated as a percentage of the monthly revenue of the company and used as the input for OCi,t in equation (7). From the sensitivity analysis, we can see that at the baseline transportation cost of 0.5% of the company’s revenue, if the overhead cost in the what-if scenario does not exceed 8% of the company’s revenue (1 percentage point more than the overhead cost in the base case scenario), then the what-if scenario is favourable. At the extreme end of the transportation cost’s spectrum, which is at 1,5% of the company’s revenue, the overhead cost can go up to 11% (4 percentage points more than the overhead cost in the base case scenario) before the what-if scenario becomes unfavourable against the base case scenario. The analyses above illustrates that decisions, such as whether to decentralise the big biodiesel refinery into two smaller centrally-located refineries so that they are closer to the WCO suppliers and collection points, may not be as straightforward as they seem. Therefore, as demonstrated by the case study, the capability of the BEN model to simulate possible outcomes of decisions before they are made, will be a useful functionality for users of the collaboration platform. More broadly, the case study showcases the potential of the collaboration platform in providing users, who could be current or future participating companies of industrial symbiosis, with the necessary visibility and decision support to establish symbiotic waste-to-resource exchanges with one another.

symbiosis lack the functionality to help companies evaluate the economic viability of participating in industrial symbiosis, a specific research gap is being tackled by the work in this paper. However, the work is still in a preliminary stage. Since the platform will not only be able to evaluate, but also to identify and advise on possible waste-to-resource exchanges, the database needs to be further populated with more information about resources and potential substitutes. Also, the BEN model needs to be further developed in order to take into account the environmental impacts of industrial symbiosis. With the planned future work abovementioned, more case studies need to be conducted to further verify the system and validate the effectiveness of the collaboration platform. Nevertheless, the work in this paper has provided an early proof-of-concept and showcases the potential use as well as capability of the collaboration platform for enabling industrial symbiosis. Acknowledgement We would like to thank Mr. Allan Lim, Mr. Jack Ling and Mr. Chan Huan Yong of Alpha Biofuels Pte Ltd for rendering their support, and providing us with valuable data and feedback for the case study. References [1]

[2]

5. Conclusion One of the biggest barriers in enabling industrial symbiosis is that companies’ lack the understanding of the economic viability of participating in industrial symbiosis. Therefore, we proposed the concept of a collaboration platform for enabling industrial symbiosis and introduced the system framework undergirding the platform. Within the system framework and focused on the by-product exchange network (BEN) model, we used a case study of a waste cooking oil (WCO)-derived biodiesel producer in Singapore to demonstrate the decision support capability and potential use of the collaboration platform. In the case study, a what-if scenario was simulated and compared with the actual (base case) scenario. The simulation was able to provide insights on whether two smaller biodiesel refineries, which are more expensive to operate but located more centrally and closer to the WCO collection points, will be more economically advantageous as compared to one big biodiesel refinery, which is currently located at the outskirt of the city and cheaper to operate. So far, the case study only demonstrated one aspect of how the collaboration platform can be used. But there are many other possible applications. For instance, fluctuations in resource prices, dynamics of new participants in an industrial symbiosis network, and the effects of discovering new substitutes for input resources are just some of the what-if scenarios that can be simulated by the agentbased BEN model. Since, existing tools for enabling industrial

[3]

[4] [5]

[6]

[7]

[8]

[9]

D. Hoornweg, P. Bhada-Tata, What a waste : a global review of solid waste management , World Bank, Washington, DC, 2012. http://documents.worldbank.org/curated/en/2012/03/16537275/wast e-global-review-solid-waste-management. D. Hoornweg, P. Bhada-tata, C. Kennedy, Waste production must peak this century., Nature. 502 (2013) 615–617. doi:10.1038/502615a. National Environmental Agency, Waste Statistics and Overall Recycling, (2016). http://www.nea.gov.sg/energy-waste/wastemanagement/waste-statistics-and-overall-recycling (accessed May 16, 2016). Kalundborg Symbiosis, (n.d.). http://www.symbiosis.dk/en (accessed July 13, 2016). G.B. Grant, T.P. Seager, G. Massard, L. Nies, Information and Communication Technology for Industrial Symbiosis, J. Ind. Ecol. (2010) 740–753. S. Bin, Y. Zhiquan, L.S.C. Jonathan, D.K. Jiewei, D. Kurle, F. Cerdas, et al., A Big Data Analytics Approach to Develop Industrial Symbioses in Large Cities, Procedia CIRP. 29 (2015) 450–455. doi:10.1016/j.procir.2015.01.066. J.O. Levänen, J.I. Hukkinen, A methodology for facilitating the feedback between mental models and institutional change in industrial ecosystem governance: A waste management case-study from northern Finland, Ecol. Econ. 87 (2013) 15–23. doi:http://dx.doi.org/10.1016/j.ecolecon.2012.12.001. C.B.H. Chua, H.M. Lee, J.S.C. Low, Life cycle emissions and energy study of biodiesel derived from waste cooking oil and diesel in Singapore, Int. J. Life Cycle Assess. 15 (2010) 417–423. Alpha Biofuels changing perceptions, (n.d.). http://alphabiofuels.sg/index.asp (accessed August 30, 2016).