Multi-agent based simulation for household solid waste recycling behavior

G Model

ARTICLE IN PRESS

RECYCL-3383; No. of Pages 11

Resources, Conservation and Recycling xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Resources, Conservation and Recycling journal homepage: www.elsevier.com/locate/resconrec

Full length article

Multi-agent based simulation for household solid waste recycling behavior Xiaoyan Meng a,b , Zongguo Wen a,b,∗ , Yi Qian a a

State Key Joint Laboratory of Environment Simulation and Pollution Control (SKLESPC), School of Environment, Tsinghua University, Beijing 100084, China Key Laboratory for Solid Waste Management and Environment Safety (Tsinghua University), Ministry of Education of China, Tsinghua University, Beijing 100084, China b

a r t i c l e

i n f o

Article history: Received 31 May 2016 Received in revised form 26 September 2016 Accepted 26 September 2016 Available online xxx Keywords: Multi-agent based simulation (MABS) Household solid waste (HSW) management Agents’ behavior HSW source separation Complex adaptive system (CAS)

a b s t r a c t The urban household solid waste (HSW) classification and recycling system is a complex adaptive system (CAS), containing multiple agents and the behaviors between them are interactive. In order to figure out which is the most effective waste management policy for the HSW classification and recycling, this study tried to establish a simulation model combining multi-agent based simulation (MABS) techniques with a social survey questionnaire. The model can simulate the behavior change of the agents in the system under different policy scenarios. Then the proposed model is utilized in Suzhou city in eastern China. The system contains three main agents: (1) the resident agents that generate the HSW, (2) the recycling site and scavenger agents that collect the recyclable materials (paper, metal, plastics and etc.) in the household garbage, (3) the agents in the sanitation department (incineration power plants and landfills) that is responsible for the municipal solid waste collection and terminal garbage disposal. In addition, three waste charge policy scenarios are set and relevant simulation experiments are carried out. The results show that the specific charge policy can improve the performance of residents’ separation behavior, which is a more effective way to reduce the HSW and increase the collection rate of domestic recyclable resources (DRR). And there exists certain benefit conflicts between the environmental sanitation agent and the DRR recycling agent at the present stage in Suzhou city. To resolve the issue of the urban HSW management in China, it is necessary to balance the profit of charge point of agents in the system. In addition, it needs to combine the municipal garbage collection network and the renewable resource recycling network together. © 2016 Elsevier B.V. All rights reserved.

1. Introduction 1.1. Research background Urban household solid waste (HSW) management is one of the enormous challenges around the world. Especially China and most of other developing countries have been experiencing high-speed urbanization and industrialization, which leads to too much consumption of resources and rapid growth of urban solid waste (Mo et al., 2009; Meng et al., 2015; Dong et al., 2016). Waste sorting and resource recycling are considered to be the most efficient ways to

∗ Corresponding author at: State Key Joint Laboratory of Environment Simulation and Pollution Control (SKLESPC), School of Environment, Tsinghua University, Beijing 100084, China. E-mail addresses: [email protected] (X. Meng), [email protected] (Z. Wen).

solve the problem of “waste city” (Tong and Tao, 2016; Dong et al., 2013). Since problems of municipal solid waste are getting more serious, the Chinese government has currently made great efforts in the municipal solid waste management (MSWM), especially towards the urban household solid waste (HSW). Eight cities (i.e., Beijing, Shanghai, Guangzhou, Shenzhen, Hangzhou) have participated in the HSW source separation pilot program launched by the Ministry of Construction (MC) since 2000, and 26 cities was chosen as national HSW classification pilots by five ministries (Ministry of Housing and Urban-Rural Development, etc.) in 2015. But the program remains largely ineffective (Deng et al., 2013). The statistical data show that the amount of MSW clean-up in Chinese urban areas reached 179 million tons, and the amount of urban renewable resources recycling run up to 245 million tons. Among them, there were 60 million tons of domestic recyclable resources (DRR). In China, the DRR are the recycled parts of the valuable materials in the household garbage, which mainly include

http://dx.doi.org/10.1016/j.resconrec.2016.09.033 0921-3449/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

2

paper, metal, plastic and fabrics (Fei et al., 2016). The recovery rate of DRR was near 25%, which was far lower than that in developed countries (Matsumoto, 2011). Moreover, there is a prodigious disjunction between the construction of HSW source separation system and renewable resources recycling system in the current MSW management in China. On September 21st, 2015, the Central Committee of CPC (Communist Party of China) and State Council (2015) have announced the Overall Plan for the Reform of Eco-civilization System. The plan has stressed the importance and indispensability of waste recycling industry and renewable resources industry and proposed that China would accelerate the establishment of a mandatory HSW classification system and renewable resources recycling directory. Then, on March 17th, 2016, “The Thirteenth Five-Year Plan for National Economic and Social Development of the People’s Republic of China” was announced. The plan has put forward the primary tasks for the urban solid waste management with the aim to improve the recycle network of renewable resources and strengthen the connection of HSW separation and DRR recovery during the period of the thirteenth five-year. Therefore, all the above analysis has indicated that it is extremely practical significant to simulate the HSW separation and DRR recovery system and study the evolvement mechanism of the system for the decision making. 1.2. Review of the previous research There has been some research on the mechanism of recycling behavior (Boonrod et al., 2015). The previous studies have found that main factors influencing the behavior of separating waste and participation namely: (1) knowledge and perception (i.e., environmental awareness and recycling knowledge) (KarimGhani et al., 2013); (2) opportunity costs, such as time costs, distance to recycling facilities (Matsumoto, 2014); (3) economic and policy incentives, such as financial rewards, reward vouchers, laws, regulations and etc. (Timlet and Williams, 2008); (4) social norms such as neighborhoods, publicity and education from the radio and television and signs on buses (Chu et al., 2013). For the investigation on the cause-effect relationships, there are some commonly used methods such as Fuzzy Cognitive Map (Iakovidis and Papageorgiou, 2011; Kannappan et al., 2011), decision-making trial and evaluation laboratory (DEMATEL) (Ren and Benjamin, 2014; Liang et al., 2016; Ren et al., 2013), and etc. These methods can identify the critical factors and support the decision-makers/stakeholders in the selection of best scenario. However, it is a pity that these studies can neither reflect the dynamic information of space and time in the complex system nor quantitatively predict the performance of the policy implementation. In this study, the HSW separation and recyclable material collection system is considered as a complex adaptive system (CAS). The theory of CAS was first proposed by Holland in 1995 and the members of the system were considered to have the properties of autonomy, activity, adaptability, communication and reactivity (Holland, 1995). The basis of CAS theory can be generalized that agents’ adaptability brings up complexity. In this study, the HSW sorting and recyclable resources collection system is considered as a complex adaptive system (CAS), which is complicated, dynamic and open and contains different elements such as waste disposal management, economics, society and their interplay. Therefore, as a typical complex system, it contains multiple active agents and each agent’s behavior will influence others’ in the system. The structure of this complex system is sophisticated and contains a large number of interacting components and frequent interaction process. Therefore, it is difficult to study it well using the traditional analytical method, such as numerical methods

or other formula and semi-formula approaches. Recent research has demonstrated that the simulation method is the most effective solution (Green, 2009). And the classic simulation techniques include system dynamics (SD), agent-based modeling (ABM), discrete event simulation (DES), and etc. SD is a methodology used to model and simulate the complex system. And it represents a system in the form of stocks, flows, time delays, variables and feedback loops (Reddi et al., 2013; Forrester, 1961; Mashayekhi, 1993). Karavezyris et al. (2002) presented a model of waste management systems using the SD to forecast the influence of environmental behavior of the households, and combined fuzzy logic to enhance confidence in the validity of the model. Ulli-Beer (2003) developed a SD model of recycling dynamics in a typical Swiss locality to simulate the interactions between citizen choices and preferences and public policy initiatives. The author emphasized the importance of both individual behavior and the environment. In addition, some scholars (Dyson and Chang, 2005; Antmann et al., 2013; Talyan et al., 2007) used the SD modeling to forecast municipal solid waste generation in a fast-growing urban region to quantitatively evaluate the importance of MSW management. In general, these SD models above have been very mature in forecasting the municipal solid waste management. However, there are some limitations of this modeling technology. For example, the key elements of the simulation system should be defined and quantified as variables, and their influences have to be formulated mathematically (Zhao et al., 2011). But not all the interactions in the complex system can be presented by formulations directly or indirectly. Agent-based modeling and simulation (ABMS) is a creative method to modeling systems composed of autonomous, interacting agents (Axelrod, 1997; Epstein, 1996). It is generally acknowledged that ABMS is the most significant and prospective technique for exploring dependencies among stakeholders involved in the complex system. This is because ABMS can connect the decisionmaking agents in microcosmic level together with the macroscopic phenomena of the system, which can help people to study the microscopic mechanism behind the macroscopic phenomena of the complex adaptive system (Wooldridge et al., 2000). ABMS approach and CAS theory have been successfully used in such fields, like United States power market and decision (Vladimir and Koritarov, 2004), economic and social systems (Farmer and Foley, 2009), supply chain research (Ioannis and Athanasiadis, 2005), river basin water resources allocation, carbon emissions trading (Liu, 2013; Li et al., 2014) and assessment of bioenergy systems (Bichraoui-Draper et al., 2015). To date, there are very few reports on comprehensive municipal solid waste management and urban renewable resources recycling system using the agent-based model. An agent-based simulation framework for collaborative decision making of an effective planning for single-stream recycling (SSR) programs was developed in Florida, and it is used by stakeholders for the evaluation of several “what-if” scenarios in their system before reaching a conclusion and making a decision (Shi et al., 2014). In general, the complexity of urban solid waste management systems has become a controversial issue, but the research on agents’ behavior of urban HSW separation and recycling system is limited, and the multi-agent based simulation model for this system is seldom reported in literature. In this paper, we have developed a multi-agent based model to simulate the policy impacts on the waste disposal behavior of three agents including residents, environmental sanitation department and DRR recycling sites. The goal of the model is to provide the decision makers a perspective of the overall HSW recycling and disposal system performance, especially the performance of each individual agent. We develop three policy scenarios: (1) BAU Scenario—Ration charge for the household waste; (2) Simulation Scenario S1 —Specific charge

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

3

Fig. 1. Geographic location of Suzhou urban area. Source: http://chinapage.com/main2.html

for all the discharged waste; (3) Simulation Scenario S2 —Specific charge for the waste burned or land filled but not for classified waste. And relevant policy experiments are carried out to simulate the behavior changes of agents under these scenarios.

2. Brief information on Suzhou MSWM status and research boundaries

Suzhou began to regulate the DRR recycling system after it was selected as one of the 29 cities to pilot renewable resource recycling system (bunch 2) in June 2009, but it developed slowly. Up to now, the DRR recycling is still relying on a large number of informal recyclers and scavengers, however, almost all of the recyclers sell DRR to the informal or formal recycling sites. In fact, there is no much difference in the function between the formal recycling sites and informal recycling sites (Fei et al., 2016).

2.1. The municipal solid waste management status in Suzhou

2.2. Research boundaries

Suzhou is located in Eastern China at 119◦ 55 E–121◦ 20 E, (see Fig. 1). It is one of the metropolitan and best developed cities in China (Wen and Meng, 2015; Zhang and Wen, 2014). In 2015, Suzhou’s gross domestic product (GDP) was 1450 billion RMB (232.8 billion USD) and ranked seventh in China, and its GDP per capital reached 126,300 RMB (U.S. $ 20,278) (Bureau, 2016). Suzhou central urban areas include five administrative districts, namely Gusu District, Wuzhong District, Xiangcheng District, Suzhou New District and Suzhou Industrial Park (Fig. 1.), with the total population of 9.534, 11.204, 7.280, 5.903 and 7.946 million (Bureau, 2015) respectively. Suzhou goverment started to make a brief exploration in the municipal solid waste (MSW) source separation from 2000 (Zhang and Wen, 2014). In addition, it drew up a series of supporting regulations and policies, such as “Domestic waste sorting collection implementation plan in suzhou city (2012–2016)”. In 2012, Suzhou city carried out a new round of MSW classification work and has now established a preliminary MSW special collection system (see Fig. 2). In the system, the MSW is diverged into five special branches including domestic waste, restaurant waste, construction waste, garden waste and organic waste; then the domestic waste can be further classified into hazardous waste, domestic recyclable resources and other waste. The waste after decomposition and classification is separately recycled with harmless treatment. In April 2015, Suzhou was chosen to be one of the 26 national domestic waste sorting collection pilot cities (bunch 1) by five ministries. By the end of 2015, more than 300 communities have participated in the domestic waste source separation pilot program. In this study, our research focus on the source classification of household solid waste, the recovery of domestic recyclable resources (DRR) and the disposal of clean-up waste, which is shown in the shadow parts of Fig. 2.

In this study, the field survey and research geographic boundary is the five administrative districts of Suzhou urban area shown in Fig. 1. And our research focus on the processes of source classification of domestic waste, the recovery of domestic recyclable resources (DRR) and the disposal of clean-up waste, which is shown in the shadow parts of Fig. 2.

30◦ 47 N–32◦ 2 N

3. Methods and data In this study, a computer modeling method namely multi-agent based simulation (MABS) is adopted for the household solid waste recycling and disposal system. And the main steps of model established and experiments simulation are presented in Table 1. Based on this process, the multi-agent based simulation model for the urban HSW separation and DRR recycling system is proposed. The framework of the model is presented in Fig. 3. The system includes four subsystems: waste produce and source classification subsystem, DRR recycling subsystem, garbage disposal subsystem and policy control subsystem. Accordingly, there are three main types of agents in the simulation model: resident agent, recycling site agent, environmental sanitation agent. Besides, the behavior change of the local policy-maker is taken as the exogenous variable of the model. The model was developed using Anylogic AB simulation tool. AnyLogic is a multi-method simulation modeling tool developed by the AnyLogic Company (https://en.wikipedia.org/wiki/AnyLogic). It supports agent-based, discrete-event, and system dynamics simulation methodologies (Nikolai and Madey, 2009). The version used in this study is AnyLogic 7.3.1 Professional. In addition, the initial data was obtained from the field survey, social questionnaire and face to face interview of residents, recycling sites, and government statistic data. Then we emphatically introduce the attributes, parameters,

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

4

Fig. 2. The municipal solid waste collection system in Suzhou.

Table 1 The main process of the MABS modeling method. Item

Steps

Details

1

Establish a conceptual model

2

Define the attributes and parameters of the agents

3 4

Build the behavior rule function for the agents Simulation model debugging

5

Scenarios and experiments

6

Results analysis and discussion

To define the research boundaries of time and space, make clear the research object, main agents, content, goals, etc. in the system. To define the property of the main agents in the system, set related parameters, variables and their initial values in the simulation model. To set the rules and decision function of each agent’s behavior, etc. To transfer the mathematical model into computer simulation language, implement and adjust the simulation model in the modeling platform. To adjust related parameters according to the requirements of policy scenarios, carry out experiment simulation. To analyze the policy experiment simulating results.

Fig. 3. Framework of the proposed multi-agent based simulation model for urban HSW separation and DRR recycling system.

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model

ARTICLE IN PRESS

RECYCL-3383; No. of Pages 11

X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

Others

5

second is the economical factor including perceived costs and profits for waste disposal.

8

Influence from neighbors

29

Perceived cost

45

Perceived profit

35

Time cost

56 0

10

20

30

40

50

60

(%)

Fig. 4. Reasons that influence residents to choose HSW disposal ways. Note: “Time cost” for “Residents’ time cost and perceived convenience for the waste disposal ways”; “Perceived profit” for “Residents’ perceived profit from recyclable resource(DRR) selling”; “Perceived cost” for “Residents’ perceived cost for garbage emission”; “Influence from neighbors” for “Received influence from the neighbors, family and friends”; “Others” for “Other factors like residents’ environmental awareness”.

3.1.2. Decision-making utility function for waste disposal behavior According to the above survey results, the study intends to form a hypothesis that the factors like the time costs, economic benefits and neighborhood participation would to a large extent influence the selection of residents on the HSW disposal in Suzhou. Therefore, the authors try their best to build the utility function to reflect the decision-making process for selecting waste disposal ways combined with the theory of planned behavior (TPB) (Zhang et al., 2015; Botetzagias et al., 2015; Nigbur et al., 2010). The decision-making utility function for waste disposal behavior of residents is proposed as follows (Eqs. (1)–(4)): Ures = Upro + Utim + Ufol + εoth



(1)

4

decision-making functions and data corresponding to each kind of agents in the proposed model. 3.1. Resident agents 3.1.1. Resident classes and their behavior Resident agents are the generation source of the HSW, and their waste separation and disposal behavior have important influence on the urban domestic waste management. There are four main different waste disposal options for residents. In the proposed model, the residents are divided into four classes according to their HSW disposal behavior, and they are encoded with 1–4 in the model to represent the attribute of the residents. Precisely, 1 for ‘No classifications mixed up with all the HSW together and discarded into dustbins’, 2 for “Classification deposition, classified and delivered to dustbins, knowing scavengers will pick the DRR up”, 3 for “Selling after classifications, separating out the DRR from the HSW at home and then selling them to recyclers”, 4 for “Other ways”. It should be clarified that the part of “others” was omitted in the model building later. In order to obtain the initial data and identify the main factors influencing residents’ waste disposal behavior, the authors have designed a survey questionnaire for this study. n, the necessary 2 (n − 1) × sample size, was calculated by the equation of n = Za/2 S 2 /2 , S and  are respectively the maximum standard deviation and tolerable error of the survey questionnaire (Zhang and Wen, 2014). Ultimately, 503 completed surveys were received, with 121, 135, 82, 75 and 90 surveys in Gusu, Wuzhong, Xiangcheng, Suzhou New District and Industrial Park. The initial percentage of four class residents in five administrative districts calculated by the survey results is shown in Table 2. In addition, the main factors influencing the residents’ selection of the HSW disposal ways were summarized in Fig. 4, according to the survey results. As shown in Fig. 4, we can see that the most important influential factor on the selection of residents by the waste disposal ways is time cost and perceived convenience of the disposal ways, the

Upro = (

Mi × Pbi × ˛1 − Cdis × ˛2 ) × pro

(2)

i=1

Utim = −(Tsep × ˛3 + Tdis × ˛4 ) × tim

(3)

Ufol = ˇfol × fol

(4)

Where, Ures , Upro, Utim , Ufol , and εoth respectively represent the total utility, cost-benefit utility, time utility, follow-up psychological utility and random utility; Mi represents the quantity of the DRR of type i (1 for waste paper, 2 for waste plastics, 3 for waste metals and 4 for other recyclable materials) which are recycled from the residents and calculated per ton; Pbi denotes the price of the recycled DRR of type i sold to the recycling sites in Chinese Yuan Renminbi; Cdis denotes the cost paid by the residents for processing the discarded garbage, measured in Yuan per week; Tsep denotes the residents’ time cost for the HSW separation, and Tdis denotes the residents time cost for the HSW disposal such as discarding into dustbins, selling the DRR to recyclers, calculated the minutes (for time unit) per week. ␣1 , ␣2 , ␣3 and ␣4 are the value coefficients, if those coefficients exist in the item, then the value will be 1. Otherwise, it will be 0. Especially, the follow-up psychological utility Ufol is mainly reflected by the proportion of some types of neighbors in the simulation model; the value of ˇfol is determined by the proportion of residents participating in the waste separation within a certain distance. In this study, it is assumed that if the proportion of residents participating in the HSW source separation within 2 square kilometers would be more than 60%, the value of ˇfol could be 1, otherwise the value would be 0, and the proportion is calculated by the computer programming.  pro ,  tim and  fol are random coefficients and respectively denote the weights of the indexes Ures , Upro, Utim and Ufol , which can respectively reflect the importance of the cost-benefit, time cost and follow-up psychological utility of the surrounding neighbors’ behavior on the selection of waste disposal. The initial value of related parameters (e.g. Tsep , Tdis ,  pro ,  tim and  fol ) in the model is obtained by the social questionnaires conducted

Table 2 Percentages of four class residents in five administrative districts in 2015. Class

Total

Gusu a

1 2 3 4 a b

b

Wuzhong

Xiangcheng

Suzhou New District

Industrial Park

Samp

PCT (%).

Samp

PCT (%)

Samp

PCT (%)

Samp

PCT (%)

Samp

PCT (%)

Samp

PCT (%)

193 171 130 9

38.37 33.99 25.84 1.79

40 48 31 2

33.06 39.66 25.61 1.65

53 42 37 3

39.26 31.11 27.40 2.22

36 24 21 1

43.90 29.27 25.61 1.22

31 25 19 0

41.33 33.33 25.33 0

33 32 22 3

36.67 35.56 24.44 3.33

Samp is the abbreviation of Sample. PCT is the abbreviation of Percentage.

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model

ARTICLE IN PRESS

RECYCL-3383; No. of Pages 11

X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

6

Table 3 The value and distribution of variables/parameters in the utility functions for the different classes of residents. Variable/Parameter

Tsep Tdis ˛1 ˛2 ˛3 ˛4 pro tim fol

Value/Range

Distribution

No-classification

Classification-deposition

Selling-after-classification

0 20–30 0 1 1 1 0–0.3 0.4–0.6 0.2–0.4

20–30 25–35 0 1 0 1 0.2–0.5 0.1–0.4 0.3–0.5

20–30 35–45 1 1 0 1 0.3–0.6 0–0.3 0.2–0.5

Random distribution Random distribution / / / / Random distribution Random distribution Random distribution

Note: The value range of  pro ,  tim and  fol were obtained according to the questionnaire results. The computer-assisted sampling process is randomized in the range for one class of residents, while the values of  pro ,  tim and  fol for one resident satisfy the constraint:  pro +  tim +  fol < 1.

among three classes of residents in the five districts of Suzhou city. We designed questions to ask the residents’ about the time cost of their waste separation and disposals in their daily lives, and also the questions for the importance (the score is graded from 1 to 9, the higher the score is, the more important the factor will be) of each key factor influencing the residents’ selection of the HSW disposal ways. Because of the different residents’ class status, the values of the related variables and parameters in the utility function are different. The survey results are shown in Table 3. Because the units of Upro, Utim and Ufol are different, they need to be normalized before calculating the sum of these three utility. In this study, we use Min-Max normalization method for mathematical expression. Min-Max normalization is the process of taking data measured in its engineering units and transforming into a value between 0.0 and 1.0. The lowest (min) value is set to 0.0 and the highest (max) value is set to 1.0. This provides an easy way to compare values that are measured using different scales or different units of measure (Mohamad and Usman, 2013). The normalized value zi can be calculated by Eq. (5) as follows. zi =

xi − min(x) max(x) − min(x)

DRR recycling agents mainly consist of recycling sites, including formal and informal recycling sites. Almost all of the DRR is traded by the recycling sites. The residents and scavengers sell the DRR to the recycling sites after some primary processing by the sites, and then the DRR is sold to the processing centers and recycling plants. By the end of year 2015, there are 587 recycling sites in the whole urban area of Suzhou city. Among them, there are 61 formal recycling sites and 526 informal recycling sites. The distribution of these sites is determined by the density of servicing population. Therefore, this study calculates the quantity of recycling sites in the five districts by their population proportion, and the results are shown in Table 4. The formal sites are established or remolded by the government, but in fact, there is a lack of governmental supervision on them now. Both the formal and informal recycling sites are managed by themselves and assume sole responsibility for their losses and profits. The management status of the recycling sites is evaluated via the profit function as described in Eq. (7).

(5)

The decision-making strategy for the residents’ HSW disposal behavior is stated by Eq. (6) as follows: Ui = Ures (Ti+1 ) − Ures (Ti )

3.2. Recycling site agents

(6)

Where, Ures (Ti ) and Ures (Ti + 1 ) respectively represent the total utility in the period of Ti and Ti + 1 . In this study, a week is set as one simulation cycle time. It is assumed that when Ui > 0 remained in eight continuous cycles, the residents HSW disposal behavior would be transferred to a higher level; when Ui < 0 maintained in eight continuous cycles, the residents HSW disposal behavior would be transferred to a lower level; otherwise, the residents’ behavior would stay the same. The state transition diagram for the residents’ HSW disposal behavior is shown in Fig. 5.

PRS =

4 

Mi × (Psi − Pbi ) − CRtra ×

i=1

4 

Mi

(7)

i=1

Where, PRS denotes the total economic profits of the recycling site agents, denominated in Yuan; Mi represents the quantity of the recycled DRR of type i (including paper, plastic, metal and others), denominated in tons; Pbi and Psi respectively represent the price of the recycled DRR of type i buying from the residents or scavengers and selling to the processing centers or recycling plants, denominated in Yuan/kg; CRtra represents the transportation cost of the sites for the DRR recycling activities, which is averagely 60 Yuan/ton according to the field survey. It is assumed that there were about 2 people in one recycling site and the salary were 2500 Yuan one month, so the labor cost would be 1250 Yuan of one cycle (a week). For each informal recycling site,

Fig. 5. The state transition diagram for residents’ HSW disposal behavior.

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model

ARTICLE IN PRESS

RECYCL-3383; No. of Pages 11

X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

7

Table 4 The quantity of the distribution of recycling sites in the five districts. Districts

Gusu

Wuzhong

Xiangcheng

Suzhou New District

Industrial Park

Number of formal sites Number of informal sites Total number

14 120 134

16 141 157

10 91 111

9 74 83

12 100 112

Table 5 The bid-ask spread of the recycled DRR of type i. Categories

Waste paper

Waste plastic

Waste metal

Simulation values

Coding Average profit Psi − Pbi (Yuan/kg)

1 0.5 Triangular (0.1,0.5,0.7)

2 0.3–0.6 Triangular (0.2,0.45,0.9)

3 0.3 Triangular (0.1,0.3,2.2)

/ / Random distribution

Note: It should be clarified that the part of ‘others’ was omitted here.

it will exit market when PRS < 0 remains in eight continuous cycles (two month). According to the sampling survey of both the formal and informal recycling sites in the five districts, the bid-ask spread of the recycled DRR of type i of 53 recycling sites were obtained. Then, the data were calculated in accordance with the triangular distribution of random sampling in each simulation period. The detailed parameters and data are shown in Tables 5–7 . 3.3. Environmental sanitation agents In this study, environmental sanitation agents refer to the environmental sanitation system, and it is responsible for the municipal solid waste collection and terminal garbage disposal. The agents contain an incineration power plant named Guangda and a landfill named Qizishan in Suzhou city. According to the statistics, nearly 71.2% of garbage was sent to the incineration power plant for burning and 28.8% of the garbage was sent to the landfill. The calculation method for the profits of the environmental sanitation agents is based on Eqs. (8)–(10). According to these equations, the weekly profits of waste disposal can be recorded and finally the predicted profits under the setting scenarios can be obtained by the computer simulation. PES = Sgov × Mdis + (Qgen − Qcon ) × Tfee × Minc − Mlan × Clan − Mdis × Ctra

(8)

Minc = Mdis × inc

(9)

Mlan = Mdis × lan

(10)

Where, PES denotes the total economic profits of the environmental sanitation agents, denominated in Yuan; Sgov represents the government subsidy for the garbage disposal with allowance of 150 Yuan per ton in Suzhou, Mdis , Minc , Mlan respectively denote the quantity of the garbage processed by both the incineration power

plant and the landfill, the quantity of the garbage burned and the quantity of the garbage landfilled, denominated in tons. the data change along with the transformation of the residents HSW disposal behavior, and they are obtained by the computer updating in each simulation period;inc and lan represent the rates of garbage burned and landfilled, which account for 71.2% and 28.8%respectively for the Suzhou case according to the statistical data; Qgen and Qcon respectively represent the quantity of electricity generated and consumed for one ton of garbage burnt, which are 380 kWh/ton and 100 kWh/ton in this case according to the field survey; Tfee is the feed-in tariff, which is 0.502 Yuan/kWh in Suzhou city; Clan is the landfill cost of the garbage, which is averagely 50 Yuan/ton according to the actual operation situation of Qizishan landfill; Ctra represents the transportation cost in the process of the garbage disposal, which is averagely 35 yuan/ton according to the field survey. It is assumed that the labor cost and the depreciation of capital investment were not taken into consideration for the profit calculation in this study.

4. Simulation experiments and results discussions Fig. 6 is a screenshot of the running urban HSW separation and disposal model on the Anylogic software platform. The model is running after the users input the initial parameter values. First, the resident agents judge the category they belong to, and determine their waste disposal behavior. Meantime, the amount of garbage burned and landfilled and the amount of DRR recycled are calculated, and then feedback to Environmental Sanitation agents and DRR recycling site agents for the calculations of their profit in the simulation cycle. The recycling site agents will choose to continue operating or exit market according to their profit change. In addition, the change in administration policy is as an external variable

Table 6 The amount of DRR recycled per capita in five districts in 2015. Categories

Waste paper Waste plastic Waste metal others

Per capita product DRR (tons/a)

Per capita recycled DRR(tons/a)

Gusu

Wuzhong

Xiangcheng

New & Hi-tech Zone

Industrial Park

Gusu

Wuzhong

Xiangcheng

New & Hi-tech Zone

Industrial Park

61.68 63.11 23.83 4.25

138.35 69.22 26.64 3.03

51.42 51.2 27.46 2.78

89.73 75.23 16.62 6.92

149.2 61.65 102.59 4.18

14.86 2.19 4.1 0.02

29.75 1.41 1.44 0

7.67 0.65 3.64 0

4.1 0.68 1.37 0.01

24.39 1.17 3.87 0.04

Table 7 Proportions of domestic recyclable resources of total recycled waste per capita for different classes of residents. Categories

No classification (%)

Classification deposition (%)

Selling after classification (%)

Simulation values

Waste paper Waste plastic Waste metal

uniform(0.15,0.25) uniform(0.3,0.4) uniform(0.15,0.25)

uniform(0.3,0.4) uniform(0.45,0.55) uniform(0.3,0.4)

uniform(0.75,0.85) uniform(0.85,0.95) uniform(0.75,0.85)

Random distribution Random distribution Random distribution

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11 8

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

Fig. 6. A screenshot of the running urban HSW source separation and recycling model.

of the model, it will result in the changes of the decision function values and eventually cause the changes of agents’ behavior. This proposed model can be served as a laboratory to solve the “what-if” question: What will happen if some policies have been chosen? In order to find out which domestic waste charge policy is more effective to the urban HSW separation and management, three policy scenarios are designed: (1) BAU Scenario (S0 ). Recently, ration charge mode is widely used in most of cities in China. According to the existing Solid Waste Disposal Fee Collection and Management Approach in Suzhou City, the charge is standardized with 4 Yuan/month per household, and it is charged with water fee. As there are averagely 2.5 residents per household, the initial value of Cdis here is about 0.4 Yuan/week per resident. We set ration charge policy scenario as BAU Scenario; (2) Simulation Scenarios (S1 and S2 ). According to the specific charge policy implemented from 2000 in Taipei, the residents must use special garbage bag to store rubbish, and have been charged 0.42 TWD (about 0.09 CNY) per kilogram of the garbage disposal fee (Qiu, 2008). In this study, S1 is set as the scenario of specific charge for all the discharged waste, S2 is set as the scenario of specific charge for the waste burned or landfilled but not for classified waste. If these two polices are taken in Suzhou city, how will the agents’ behavior change? In order to answer this question, relative policy experiments are carried out. The simulating timing is that it runs four cycles under BAU Scenario then respectively introduces the policy stimulus of S1 and S2 scenarios. The simulation cycle T in this study is one week, and the total duration is from 1 January 2016 to 31 December 2020 with 260 simulation cycles.

4.1. Simulation results 4.1.1. Experiment 1: specific charge for all the discharged waste If the specific charge policy is taken in Suzhou city in accordance with the charge standard in Taipei, how will the agents’ behavior change? Figs. 7 and 8 show the simulation results under this scenario. Fig. 7 shows the change in the number of three classes of residents. The number of residents began to decrease, the number of residents increased, and the number of residents decreased at the beginning and increased at the end. Since the policy have changed from the ration charge to the specific charge for all discharged waste, more garbage has been generated and more higher economic cost residents have to pay to the environment

Fig. 7. Three classes of residents’ HSW source separation and disposal behavior change under S1 .

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

9

Fig. 8. Profit change of environment sanitation and DRR recycling sites under S1 .

sanitation for the garbage disposal. In the beginning, more and more residents will participate in HSW source separation to reduce waste discharge and the number of residents decrease rapidly. For the residents of getting used to the source separation, most of them have become willing to sell the DRR to the scavengers or the recycling sites after classification, and transform into residents. Since this scenario is specifically charged for all the discharged waste, that is to say, the cost-benefit utility (Upro ) makes no difference to residents and residents. After a period of new charge policy implementation, in order to reduce their time costs (Utim ) and then reduce the total utility, some of residents will become unwilling to classify before discharging waste, then change into residents. Therefore, the number of residents rebound after 500 days. Fig. 8 shows the profit change of environment sanitation and DRR recycling sites during the simulation period. Since the residents’ HSW separation and disposal behavior changed, the fraction of classified and recycled waste has increased and the fraction of

garbage for burning and landfill has decreased. As a consequence the profits of the DRR recycling sites have continuously accelerated but the profits of environment sanitation are in the opposite direction. The profit of environment sanitation agents will be reduced by 5%, which indicates that the amount of the garbage disposal has been reduced by about 5%.

4.1.2. Experiment 2: specific charge for the waste burned or landfilled but not for classified waste The simulation results are shown in Figs. 9 and 10. Under this scenario, the number of residents began to increase in the first half of the simulation period and decreased at the end, the number of residents is the same trend. And the number of residents continued to decline (See Fig. 9). It indicates that more and more residents become willing to participate in HSW source separation under this policy scenario. Fig. 10 shows the profit change of Environment Sanitation and DRR recycling sites during the simulation period under S2 . Since the residents participated in the HSW source separation increases significantly, the amount of DRR recycled increases and the garbage for final disposal decreases. As a consequence the profits of the DRR recycling sites have continuously accelerated but the profits of environment sanitation are in the opposite direction. Moreover, it can be seen that there exists certain benefit conflicts between the environmental sanitation agents and the DRR recycling agents at the present stage. That is mainly explained in details: the sanitation system gains profits from the garbage processing, and the DRR recycling system gains profits from the DRR separated from the HSW and the business activity of the recycling system will reduce the quantity of the garbage processed.

4.2. Policy implications

Fig. 9. Three classes of residents’ HSW source separation and disposal behavior change under S2 .

Urban HSW management is a systematic project, which includes many stakeholders: residents, enterprises like environmental sanitation and recycling enterprises, governments, and etc. We can see that the behaviors of agents are interacted on each other, so it

Fig. 10. Profit change of environment sanitation and DRR recycling sites under S2 .

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11 10

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

is necessary to balance the profit points of agents in urban HSW management. Based on the results above, the specific charge system can stimulate more residents to participate in HSW separation and recycling and the residents unaware of the garbage classification will pay more waste disposal fees. This policy not only improves the HSW separation and DRR recycling rate, but also reflects the principle of fairness. Although the waste classification will reduce the amount of garbage disposals and the profit of environmental sanitation within a short period accordingly, it is not ideal for incineration plant but good the landfill site in the long term since the lifetime of landfill site will be extended. Then the profit loss of treatment sites will actually be paid by residents. At the present stage, there exists a profit conflict between environmental sanitation and DRR recycling sites in many Chinese cities like Suzhou, which has suspended the HSW classification work. To solve this problem, we put forward constructive suggestion that the municipal solid waste collection network and the renewable resource recycling network could be integrated together. After the merger of these two networks, there will be no profit conflicts, and the operational efficiency will be improved. At the same time, the labor and operating costs and the expenses on recycling facilities will be reduced subsequently.

5. Conclusion This research tried to apply the multi-agent based simulation method to establish a model for the urban HSW separation and disposal system with the purpose to predict the behavior change of the main agents in the system under different policy scenarios. The authors carried out extensive social survey to obtain the initial data for the simulation model. The main findings of our research are that specific charge policy can improve the performance of residents’ separation behavior, and it simulates how much the stakeholders would be influenced by the policy, and there exists certain benefit conflicts between the environmental sanitation agents and the DRR recycling agents at the present stage in Suzhou city. In addition, the established model can tell how much the stakeholders would be influenced by the policy change. This study intends to quantitatively evaluate the costs and benefits of the waste management policy in a dynamic way to impel better decision-making in China. In recent years, although China has implemented the HSW reduction management, compared to other developed countries, China is lack of intensive study on the solid waste reduction policy from the perspective of systems engineering, and the most important reason lies in the limitation of the policy experience. To this end, the model proposed in this study will serve as an effective alternative to the simulation of such complex system. It is an ideal choice to establish a virtual society using the multi-agent simulation method, and then some mechanism research can be done to predict the implementation effects of various policies. In addition, the policy experiments can help get more confidence in the model and test correlation, which is very helpful to both researchers and policy makers. In the long term, the specific charge policy is an inevitable trend, it is quite effective to promote the resources recovery and quantitative reduction of urban domestic garbage, and it is also the primary prerequisite for the marketization and industrialization of Chinese urban HSW management. In a few words, there are still some limitations and shortcomings in the present study. As we have combined the computer simulation technology and the social survey together to conduct the policy experiments, these results can’t be verified by the comparison with historical data, and there may be some uncertainties in parameter setting and data processing. Further study is needed to improve the measures and verify the feasibility.

Acknowledgements The project was supported by the National Key Research and Development Program of China (2016YFC0502802) and the National Natural Science Fund for Outstanding Young Scholars of China (71522011). In the research, we have received a lot of support from the Department of Environmental Sanitation and Supply and Marketing Cooperative of Suzhou city, without which it would have been impossible to complete this study.

References Antmann, E.D., Shi, X., Celik, N., Dai, Y., 2013. Continuous-discrete simulation-based decision making framework for solid waste management and recycling programs. Comput. Ind. Eng. 65 (3), 438–454. Axelrod, R., 1997. The Complexity of Cooperation: Agent-Based Models of Competition and Collaboration. Princeton. Bichraoui-Draper, N., Xu, M., Miller, S.A., Guillaume, B., 2015. Agent-based life cycle assessment for switch grass-based bioenergy systems. Resour. Conserv. Recycl. 103, 171–178. Boonrod, K., Towprayoon, S., Bonnet, S., Tripetchkul, S., 2015. Enhancing organic waste separation at the source behavior: a case study of the application of motivation mechanisms in communities in Thailand. Resour. Conserv. Recycl. 95, 77–90. Botetzagias, I., Dima, A., Malesios, C., 2015. Extending the theory of planned behavior in the context of recycling: the role of moral norms and of demographic predictors. Resour. Conserv. Recycl. 95, 58–67. Bureau, S.S., 2015. Suzhou Statistic Year Book. China Statistics Press, Suzhou, China. Bureau, S.S., 2016. The Statistical Bulletin of the Economic and Social Development of Suzhou City in 2015, Available online: http://www.sztjj.gov.cn/Info Detail. asp?id=22727 (accessed 22.02.16). In Chinese. CPC Central Committee and State Council, 2015. Overall Plan for the Reform of Eco-Civilization System, In Chinese. Chu, Z., Xi, B., Song, Y., Crampton, E., 2013. Taking out the trash household preferences over municipal solid waste collection in Harbin, China. Habitat Int. 40, 194–200. Deng, J., Xu, W.Y., Zhou, C.B., 2013. Investigation of waste classification and collection actual effect and the study of long acting management in the community of Beijing. Chin. J. Environ. Sci. 34, 395–400. Dong, L., Fujita, T., Zhang, H., Dai, M., Fujii, M., Ohnishi, S., Geng, Y., Liu, Z., 2013. Promoting low-carbon city through industrial symbiosis: a case in China by applying HPIMO model. Energy Policy 61, 864–873. Dong, L., Fujita, T., Dai, M., Geng, Y., Ren, J., Fujii, M., Wang, Y., Ohnishi, S., 2016. Towards preventative eco-industrial development: an industrial and urban symbiosis case in one typical industrial city in China. J. Clean. Prod. 114, 387–400. Dyson, B., Chang, N., 2005. Forecasting municipal solid waste generation in a fast-growing urban region with system dynamics modeling. Waste Manag. 25 (7), 669–679. Epstein, J.M., 1996. Growing Artificial Societies: Social Science from the Bottom Up. Brookings Institution Press. Farmer, J.D., Foley, D., 2009. The economy needs agent-based modeling. Nature 460, 685–686. Fei, F., Qu, L., Wen, Z., Xue, Y., Zhang, H., 2016. How to integrate the informal recycling system into municipal solid waste management in developing countries: based on a China’s case in Suzhou urban area. Resour. Conserv. Recycl. 110, 74–86. Forrester, J.W., 1961. Industrial Dynamics. Productivity Press, USA. Green, D.G., 2009. Hierarchy, complexity and agent based models. In: Our Fragile World: Challenges and Opportunities for Sustainable Development. UNESCO, Paris. Holland, J., 1995. Hidden Order How Adaptation Builds Complexity. Addison—Wesley Publishing Company Inc. Iakovidis, D.K., Papageorgiou, E., 2011. Intuitionistic fuzzy cognitive maps for medical decision making. IEEE Trans. Inf. Technol. Biomed. 15 (1), 100–107. Ioannis, N., Athanasiadis, R.A., 2005. Hybrid agent-based model for estimating residential water demand. Simulation 81 (3), 175–187. Kannappan, A., Tamilarasi, A., Papageorgiou, E.I., 2011. Analyzing the performance of fuzzy cognitive maps with non-linear hebbian learning algorithm in predicting autistic disorder. Expert Syst. Appl. 38 (3), 1282–1292. Karavezyris, V., Timpe, K., Marzi, R., 2002. Application of system dynamics and fuzzy logic to forecasting of municipal solid waste. Math. Comput. Simul. 60 (3), 149–158. KarimGhani, W.A.W.A., Rusli, I.F., Biak, D.R.A., Idris, A., 2013. An application of the theory of planned behaviour to study the influencing factors of participation in source separation of food waste. Waste Manag. 33 (5), 1276–1281. Li, Y., Huang, G., Li, M., 2014. An integrated optimization modeling approach for planning emission trading and clean-energy development under uncertainty. Renew. Energy 62, 31–46. Liang, H., Ren, J., Gao, Z., Gao, S., Luo, X., Dong, L., Scipioni, A., 2016. Identification of critical success factors for sustainable development of biofuel industry in

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033

G Model RECYCL-3383; No. of Pages 11

ARTICLE IN PRESS X. Meng et al. / Resources, Conservation and Recycling xxx (2016) xxx–xxx

China based on grey decision-making trial and evaluation laboratory (DEMATEL). J. Clean. Prod. 131, 500–508. Liu, Y., 2013. Relationship between industrial firms, high-carbon and low-carbon energy: an agent-based simulation approach. Appl. Math. Comput. 219 (14), 7472–7479. Mashayekhi, A.N., 1993. Transition in the New York State solid waste system: a dynamic analysis. Syst. Dyn. Rev. 9 (1), 23–47. Matsumoto, S., 2011. Waste separation at home: are Japanese municipal curbside recycling policies efficient? Resour. Conserv. Recycl. 55, 325–334. Matsumoto, S., 2014. The opportunity cost of pro-environmental activities: separating time to promote the environment or earning more money? J. Fam. Econ. Issues 35, 119–130. Meng, X., Wen, Z., Qian, Y., Yu, H., 2015. Evaluation of cleaner production technology integration for the Chinese herbal medicine industry using carbon flow analysis. J. Clean. Prod., http://dx.doi.org/10.1016/j.jclepro.2015.10.067 (in press). Mo, H., Wen, Z., Chen, J., 2009. China’s recyclable resources recycling system and policy: a case study in Suzhou. Resour. Conserv. Recycl. 53, 409–419. Mohamad, I.B., Usman, D., 2013. Standardization and its effects on K-means clustering algorithm. Res. J. Appl. Sci. Eng. Technol. 6 (17), 3299–3303. Nigbur, D., Lyons, E., Uzzell, D., 2010. Attitudes, norms, identity and environmental behaviour: using an expanded theory of planned behaviour to predict participation in a kerbside recycling programme. Br. J. Soc. Psychol. 49 (2), 259–284. Nikolai, C., Madey, G., 2009. Tools of the trade: a survey of various agent based modeling platforms. J. Artif. Soc. Soc. Simul. 12 (22), 126–132. Qiu, S., 2008. Comparisons between domestic waste charge system in Fuzhou and Taibei City. Environ. Sanit. Eng. 6, 18–21, In Chinese. Reddi, K.R., Li, W., Wang, B., Moon, Y., 2013. System dynamics modelling of hybrid renewable energy systems and combined heating and power generator. Int. J. Sustain. Eng. 6 (1), 31–47. Ren, J., Benjamin, K.S., 2014. Quantifying, measuring, and strategizing energy security: determining the most meaningful dimensions and metrics. Energy 76, 838–849. Ren, J., Manzardo, A., Toniolo, S., Scipioni, A., 2013. Sustainability of hydrogen supply chain. Part I: identification of critical criteria and cause-effect analysis

11

for enhancing the sustainability using DEMATEL. Int. J. Hydrogen Energy 38 (33), 14159–14171. Shi, X., Thanos, A.E., Celik, N., 2014. Multi-objective agent-based modeling of single-stream recycling programs. Resour. Conserv. Recycl. 92, 190–205. Talyan, V., Dahiya, R.P., Anand, S., Sreekrishnan, T.R., 2007. Quantification of methane emission from municipal solid waste disposal in Delhi. Resour. Conserv. Recycl. 50 (3), 240–259. Timlet, R.E., Williams, I.D., 2008. Public participation and recycling performance in England: a comparison of tools for behavior change. Resour. Conserv. Recycl. 52, 622–634. Tong, X., Tao, D., 2016. The rise and fall of a waste city in the construction of an urban circular economic system: the changing landscape of waste in Beijing. Resour. Conserv. Recycl. 107, 10–17. Ulli-Beer, S., 2003. Dynamic interactions between citizen choice and preferences and public policy initiatives. A System Dynamics Model of Recycling Dynamics in a Typical Swiss Locality in 21st International Conference of the System Dynamics Society. Vladimir, S., Koritarov, N., 2004. Modeling the electricity market as a complex adaptive system with an agent-based approach. IEEE Power Energy Mag. 23 (4), 39–46. Wen, Z., Meng, X., 2015. Quantitative assessment of industrial symbiosis for the promotion of circular economy: a case study of the printed circuit boards industry in China’s Suzhou New District. J. Clean. Prod. 90, 211–219. Wooldridge, M., Jennings, N.R., Kinny, D., 2000. The Gaia methodology for agent-oriented analysis and design. Auton. Agents Multi-Agent Syst. 3, 285–312. Zhang, H., Wen, Z., 2014. Residents’ household solid waste (HSW) source separation activities: a case study in Suzhou, China. Sustainability 6, 6446–6466. Zhang, D., Huang, G., Yin, X., Gong, Q., 2015. Residents’ waste separation behaviors at the source: using SEM with the theory of planned behavior in Guangzhou, China. Int. J. Environ. Res. Publ. Health 12 (8), 9475–9491. Zhao, W., Ren, H., Rotter, V.S., 2011. A system dynamics model for evaluating the alternative of type in construction and demolition waste recycling center—the case of Chongqing, China. Resour. Conserv. Recycl. 55 (11), 933–944.

Please cite this article in press as: Meng, X., et al., Multi-agent based simulation for household solid waste recycling behavior. Resour Conserv Recy (2016), http://dx.doi.org/10.1016/j.resconrec.2016.09.033