- Email: [email protected]
Municipal solid waste management via mathematical modeling: A case study in İstanbul, Turkey
Journal of Environmental Management 244 (2019) 362–369
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Research article
Municipal solid waste management via mathematical modeling: A case study in İstanbul, Turkey
T
Nur Ayvaz-Cavdaroglua, Asli Cobanb, Irem Firtina-Ertisc,∗ Kadir Has University Business Administration Department, İstanbul, 34083, Turkey İzmir University of Economics Faculty of Engineering, Civil Engineering Department, İzmir, 35330, Turkey c Bahçeşehir University Faculty of Engineering and Natural Science, Energy Systems Engineering Department, İstanbul, 34353, Turkey a
b
A R T I C LE I N FO
A B S T R A C T
Keywords: Municipal solid waste management Optimization Mathematical programming
The prominence of managing municipal solid waste (MSW) in an efficient and effective manner is increasing from day to day. In this paper, the solid waste management (SWM) system of İstanbul is analyzed by applying the techniques from mathematical programming methodology. In this manner, the solutions of the two optimization problems which aim to minimize the total cost and the environmental effects of SWM, respectively, are presented in this study. Additionally, a sensitivity analysis is performed and a multi-objective problem that combines two problems is presented. In this regard, the application of five MSW management technologies which are currently in use in İstanbul on six waste components is analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides, this optimal solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments for the policymakers. The results of the study emphasize the importance of material recovery and incineration facilities to improve profitability and to minimize environmental side effects. In particular, material recovery facility (MRF) should be expanded to be able to treat all of metal, paper and plastic from a cost management perspective. Incineration (INC) facility should also be expanded in order to treat plastics or organic waste from a Greenhouse Gas (GHG) minimization perspective. In addition to this, landfill appears to be the most prominent treatment technique according to the current problem parameters. However, regarding the waste composition, the amount of organic waste must be decreased by more than 37% for other waste streams to be treated in different facilities other than landfill. Anaerobic digestion and composting facilities need to be more cost-effective for becoming economically feasible. The methodology represented in this study can be extended and generalized to other cities around the world once the correct problem parameters are specified.
1. Introduction The prominence of managing municipal solid waste (MSW) in an efficient and effective manner is increasing from day to day. The growing numbers of urban residents and the increasing amount of waste generated per person are the two main reasons for the growing need of better solid waste management (SWM) (Hoornweg and BhadaTata, 2012). Poorly managed MSW has considerable adverse effects on human health and the economy. Moreover, poorly managed waste contributes to greenhouse gas (GHG) emissions increasing the global warming effect. In this vein, designing the optimal MSW system seems imperative for all public authorities. Mathematical programming (MP) models are broadly used in the design of optimal SWM systems in the past (Dai et al., 2011). Several
∗
case studies in the literature display the effectiveness of the solution produced by using mathematical tools. For instance, Benítez et al. (2008) establish mathematical models that correlate the generation of residential solid waste per capita to the variables such as education, income per household, and the number of residents in a Mexican city. Similar studies span various regions in the world, like e.g. Beijing-China (Xi et al., 2010); Foshan–China (Jing et al., 2009); Port Said–Egypt (Badran and El-Haggar, 2006); Alleghany County–the USA (Louis and Shih, 2007); and Hong Kong (Lee et al., 2016). Most of these papers involve mixed integer programming formulations, which is a prominently utilized method to deal with the complex MSW problem. Costi et al. (2004) use a mixed integer nonlinear programming decision support model, which aims to find the optimal number of facilities and locations to minimize the waste management cost. Similarly, Jing et al.
Corresponding author. E-mail addresses: [email protected] (N. Ayvaz-Cavdaroglu), [email protected] (A. Coban), irem.fi[email protected] (I. Firtina-Ertis).
https://doi.org/10.1016/j.jenvman.2019.05.065 Received 4 February 2019; Received in revised form 26 April 2019; Accepted 17 May 2019 Available online 23 May 2019 0301-4797/ © 2019 Elsevier Ltd. All rights reserved.
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
this optimal solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments for the policymakers. The problem is modeled as a multi-objective optimization model and solved by the commercial software MS Excel Solver. Our methodology can be extended and generalized to other cities around the world once the correct problem parameters are specified. The rest of the paper is organized as follows. In Section 2, the details of the SWM problem of İstanbul and the methodology used to analyze and solve this problem from a mathematical programming perspective are represented. The results are presented in Section 3, including the two optimization problems which aim to minimize the total cost of SWM and the environmental effects of SWM. The result of the multiobjective problem combines the objectives of these two problems. In Section 4, the findings of the study are discussed and recommendations are developed for policymakers. Finally, the importance of the findings is summarized in Section 5.
(2009) develop a two-stage chance-constraint mixed integer linear programming model for solid waste management in Foshan and find that the centralized composting and incinerating facilities are desired. However, most of the studies only consider the problem from an economic perspective and aim to minimize the total cost of the system. There appears to be limited work on the application of mathematical optimization to design SWM systems with the objective of minimizing GHG emissions. Lu et al. (2006) develop an inexact dynamic optimization model (IDOM) for MSW-management systems under uncertainty. This model integrates GHG components into the modeling framework treating them as constraints. Recent scientific literature presents a large number of studies related to the analysis of GHG emissions from MSW management mainly developed by Life Cycle Assessment (LCA) (Kirkeby et al., 2006; Thorneloe et al., 2002; Eschenroeder, 2001). These studies show how recycling can allow a reduction of GHG emissions. Another example, Levis et al. (2013), present the first lifecycle based framework to optimize the collection and treatment of all waste materials from the curb to final disposal by minimizing cost or environmental impacts. Similarly, Pressley et al. (2015), Münster et al. (2015), Calabrò (2009) present a detailed and novel life-cycle modeling of SWM systems, and apply different optimization objectives such as minimizing costs or greenhouse gas emissions. The research involving a multi-objective optimization approach that simultaneously considers the environmental and economic objectives is quite rare. An example of this line of research is the work of Mavrotas et al. (2013), who formulate a general problem and show the application of the general model on a Greek case by developing four scenarios (waste management options). On the contrary, our model does not require defining possible scenarios to be applied in a pilot region. Yu and Solvang (2017) develop a multi-objective location-allocation optimization model to balance the trade-off among system operating costs, GHG emissions and environmental impact. Their main model is conceptual while they also present a numerical experiment with hypothetical values. They do not model specific treatment facilities as we do. Another important difference between our work and these two papers is that we measure the robustness of the proposed solution via sensitivity analysis. There are also several studies that consider the MSW system of İstanbul. Berkun et al. (2005) forecast the overall solid waste production in İstanbul as 4.75 millions of tons annually and 0.63 kg/day/capita. Besides, a yearly solid waste constituent variation in Istanbul (% by weight) is given. In our paper, the flow chart of İstanbul Municipal Solid Waste Composting and Recycling Plant is adapted from Kanat et al. (2006). The information about waste collection practices, waste amounts, transfer stations and routes, number of trips, truck capacity and GHG emissions is gathered from the studies of Korkut et al. (2018) and Demir et al. (2017). Karadag and Sakar (2003) state the costs of SWM and collection-transfer services in İstanbul, which are also used in our study. Kanat (2010) represents the MSW quantity (including recyclables), waste components (% wet weight), and total MSW management cost in İstanbul. In addition, estimated costs for MSW disposal alternatives and capacities of SWM plants are given. The composition of MSW and recyclable wastes in İstanbul, as well as the cost estimation for a recycling program, are listed in Metin et al. (2003). Incineration and landfill cost parameters are estimated based on the study of Assamoi and Lawryshyn (2012). In addition, the anaerobic digestion cost parameter is estimated based on the study of Hochman et al. (2015). In this paper, the SWM practices of İstanbul are analyzed by applying the techniques from mathematical programming methodology. In this regard, the application of five MSW management technologies currently in use in İstanbul (namely; landfilling, composting, recycling; with an exception of incineration and anaerobic digestion) on six waste components (namely, organic, paper, plastic, glass, metal, and other) are analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides,
2. Methodology İstanbul is the biggest city in Turkey with a population nearing 15 million. In this huge city, there are two prominent sanitary landfill sites (stations), one in Asian (Kömürcüoda) and one in European (Odayeri) side. The European Environment Agency (EEA) disseminates the MSW composition in Turkey and disposal routes including controlled landfills, dumpsites, and composting. Composting facility (CF) capacities of İstanbul and incineration facility capacity of Kocaeli (İZAYDAŞ) are also mentioned in this study. The incineration plant capacity is assumed to be equal to İZAYDAŞ plant (located in Kocaeli, Turkey) at a capacity value of 35,000 tons/year and assumed to be located near one of the sanitary landfill sites of İstanbul. However, the incineration plant of İstanbul which is going to be located in the Fatih district is under construction, and is expected to be completed in 2020. Another assumption is made considering the anaerobic digestion plant of İstanbul. It is expected to be constructed in the coming years; however, in this study, it is assumed to be located near one of the sanitary landfill sites. All these MSW facilities of İstanbul, which are either constructed or to be constructed, are under the administration of İSTAÇ Co. İSTAÇ is the government-funded company that handles the MSW management in İstanbul with its more than 4000 personnel operating in more than 40 facilities. For our analysis, the first piece of information required is the waste composition in İstanbul. The typical physical composition of MSW in İstanbul and the other necessary information regarding the methane and energy producing potential of each component is presented in Table 1. The typical integrated SWM process in İstanbul (outer dashed line in Fig. 1) involves the following steps: First, all components of the waste are transferred to eight waste transfer stations (WTS) within the city. The majority of the waste remains non-separated at this point. We assume that this is the case for all the waste collected. All this collected Table 1 Energy production potential of each component in MSW. Component
Composition (%)a
Lower heating value (kWh/wet ton)b
Methane Potential (L CH4/wet kg)c
Paper Organic Plastics Metals Glass Other
13.3 50.22 14.39 1.63 5.82 14.64
3930 1193 10,880 0 0 0
145.8 300.7 0 0 0 0
a b c
363
Adapted from Kanat (2010). Adapted from Minoglou and Komilis (2013). Adapted from Minoglou and Komilis (2013).
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Landfill Gas (CH4 ) to Energy (Electricity) Recovered (Granulated, PET, RDF)
Trucks
Municipalities
Waste Transfer Station (WTS)
1
Bigger Trucks
Waste Input
Material Recovery Facility (MRF)
Unrecovered
Landfill (LF) Leachate
2 Composting Facility (CF)
Compost Non-Compostable
3 Incineration Facility (INC)
Ashes Heat to Energy
4 Anaerobic Digestion Facility (ANB)
Waste CH4 to Energy
Fig. 1. Integrated SWM system in general (Outer dashed line shows the İstanbul case).
potential. Thus, the power generation can only be possible on the unprocessed waste and the non-recoverable recycling components accepted to the landfill. As apparent from Table 2, for example, it does not make sense to send glass or metal to an incineration facility since a significant portion (97%–98%) of these components will remain unprocessed in this facility. Similarly, composting can be possible on organic or paper waste, but not on glass, metal, or plastic. Finally, the cost functions for each type of waste management facility and the other parameters required for the problem formulation are presented in Table 3. These numbers are adapted from several sources in the literature, which are mentioned individually in Table 3. The values are converted to 2018 figures by taking into account the inflation rate and dollar/TL exchange rates. Note that only the operating cost of the facilities have been taken into consideration, since these facilities have already been in use for several years, and have reached the break-even point for the initial investment costs. Moreover, we only focus on the current (and assumed to be current) facilities and do not consider the problem regarding whether new facilities should be opened or not, since this dimension is going to complicate the current problem at hand. Hence, we can ignore the initial investment cost of the facilities.where: Ci : operating cost for facility i (per ton), (i = 1 WTS, i = 2 MRF, i = 3 INC, i = 4 CF, i = 5 ANB, i = 6 LF). Pj : production rate of the waste component j per day, (j = 1 Paper, j = 2 Organic, j = 3 Plastic, j = 4 Glass, j = 5 Metal, j = 6 Other). Pij : amount of waste component j that is sent from WTS to facility i per day, (i = 2 MRF, …j = 1 Paper, …) T1i : transportation cost of sending 1 ton of material from WTS to facility i, (i = 2 MRF, …j = 1 Paper, …) Ti6 : transportation cost of sending 1 ton of material from facility i to the LF, (i = 1 WTS, …) Rij : amount of residuals of waste component j that is sent from WTS to facility i per day, (i = 2 MRF, …j = 1 Paper, …) aij : recovery/processing rate of the waste component j in facility i (i = 1 WTS, i = 2 MRF, …j = 1 Paper, …) Mj : market value of the waste component j (j = 1 Paper, …) CMj : market value of the waste component j that is converted to compost (j = 1 Paper, …) MYj: methane yield of the waste component j (j = 1 Paper, …).
waste is transferred to the main waste collection facility in Kömürcüoda (Asian) and in Odayeri (European). The total waste content is sent to “MRF+CF” first. The recyclable fraction of the waste is sorted here, and sent to the material recovery facility (MRF) while the rest (organic fraction) is sent to the composting facility (CF). In İstanbul, as a special case, these two treatments are located within the same facility. The other alternative routes after the main waste input are Landfill (LF), Incineration (INC), and ANB (Anaerobic Digestion) facilities. All the residual materials and unrecovered part of recyclable materials are sent to the landfill (LF) as well. The schematic representation of the integrated solid waste management system in general is represented in Fig. 1. Four principal methods are numbered in Fig. 1, including Landfill (LF), Material Recovery and Composting (MRF+CF), Incineration (INC), and Anaerobic Digestion (ANB). The outer dashed line represents the boundaries of integrated SWM operations which are currently performed in İstanbul. Green and black lines represent the product and waste streams, respectively. The recovery rate for recyclable materials can vary from 0% to 100% depending on the source and condition of the material. Similarly, the residue amounts of different components vary across the treatment processes depending on the appropriateness (suitability) of the process for that particular waste component. The average rates for the efficiency of each process for each component are presented in Table 2 in the form of the percentage of residues. The unrecovered components are then hauled to the sanitary landfill. The sanitary landfill accepts unprocessed materials as well as the residues of the waste treated in one of the other facilities. We assume that the residues from the facilities of CF or INC do not have methane Table 2 The percentage (%) of residues after the treatment of different components in each treatment process. Component/Treatment
MRF
INC
CF
ANB
Paper Organic Plastics Metals Glass Other
30 N/A 30 30 30 N/A
11 1 2 97 98 25
65 35 100 100 100 100
75 45 100 100 100 100
364
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Table 3 Cost functions for each type of waste management facility and the other parameters required for the problem. 6
5
5
5
5
5
5
Min TC = C1 åPj + å å Ci Pij + å å T1i Pij + å å Ti6 Rij − å Mj P2j a2j − å CMj P4j a4j − å å 10.5MYj Pij BRECi EEConvi CostElec (1 − Rij ) − å Pij EEConv5 CostElecLHVj (1 − Rij ) j i=2
j
j i=1
j i=1
i=1
i=1
i = 3,6 j = 1
j=1
Rij = (1 − aij ) Pij, ∀ i , j; Pj = ∑ Pij, ∀ j; i
∑ Pij ≤ Capi ,
i = 2,3,4,5
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 j
j
i=1
Pij ≥ 0, ∀ i , j;
Cost Function
Value
Adapted from
Collection and transport cost of waste to WTS Operating cost for LF Operating cost for WTS Operating cost for CF Operating cost for ANB Operating cost for MRF Operating cost for INC Market price of recycled components (metal) Market price of recycled components (plastic) Market price of recycled components (glass) Market price of recycled components (paper) LF biogas recovery efficiency Revenue from the sale of 1 kWh of electrical energy ANB biogas recovery efficiency Electrical energy conversion efficiency (INC) Electrical energy conversion efficiency (ANB) Electrical energy conversion efficiency (LF) Transport cost of waste from INC, ANB, CF, MRF to LF Daily processing capacity of LF Daily processing capacity of WTS Daily processing capacity of CF (MRF included) Daily processing capacity of ANB Daily processing capacity of MRF (CF included) Daily processing capacity of INC GHG Emissions Global Warming Factor (GWF) GHG per capita
24.4 + 6 = 30.4 $/ton 20 $ 10% of 10 $/ton 30 $ 60 $ 20 $ 82 $ 120 $/ton 150 $/ton 35 $/ton 80 $/ton 30% 0.25 TL/kWh 100% 27% 33% 30% 0 16,000 tons/day 6000*2 = 12,000 tons/day (for two sides of Istanbul) 3000 tons/day 150 tons/day 3000 tons/day 35,000 tons/year = 97.2 tons/day 1189 ton CO2-eq 7.7 kg CO2-eq/ton 3.16 kg CO2-eq/ca
Karadag and Sakar (2003) Hochman et al. (2015) (10–30$) https://www.epa.gov/ Hochman et al. (2015) (30–60$/ton) Hochman et al. (2015) (60–100 $/ton) Pressley et al. (2015) (19.8–24.9$/ton) Hochman et al. (2015) (80–120$). Metin et al. (2003) Metin et al. (2003) Metin et al. (2003) Metin et al. (2003) Minoglou and Komilis (2013) https://enerjienstitusu.org/ Minoglou and Komilis (2013) Minoglou and Komilis (2013) Minoglou and Komilis (2013) Minoglou and Komilis (2013) Assumed as located at the same plant Yıldız et al. (2014) Demir et al. (2017) Kanat (2010) Yıldız et al. (2009) Kanat (2010) https://www.izaydas.com.tr/ Korkut et al. (2018) Korkut et al. (2018) Korkut et al. (2018)
Based on these cost functions and parameters, the cost minimization problem (OPT 1) is formulated as follows.
respectively. The seventh term corresponds to the revenue obtained by generating electricity in incineration (INC) and landfill (LF) facilities. The final term is again the revenue obtained from electricity generation by Anaerobic Digestion (ANB). Note that the revenue terms appear with a minus sign in front of them since the objective function represents the total costs. Regarding the constraints, the first term determines how much residual is generated by treating each component in each of the treatment facilities. The second constraint ensures that all waste components are treated somehow; and the third constraint ensures that the capacities of the treatment facilities (except the LF) are not violated. The fourth term states the capacity constraint for the specific case of LF, since one has to consider the amount of residuals from the other treatment facilities in the computation of the capacity usage in LF. In addition to cost minimization, one of the other concerns is the effect of waste treatment on the environment. One possible way to measure this effect is to measure the CO2 emissions of a combination of treatment methods. Our second optimization problem with the objective of minimizing the total amount of CO2 (OPT 2) will thus be defined as follows.
BRECi : biogas recovery efficiency in LF or ANB facility, (i = 3 ANB, i = 6 LF). EEConvi : electric energy conversion efficiency of facility i (i = 3 ANB, i = 5 INC, i = 6 LF). CostElec : Revenue from the sale of the electricity generated in facility LF, I or ANB. LHVj : Lower heating value of MSW component j (kWh/t) (j = 1 Paper, …), see Table 1. Capi : Capacity of facility i (i = 1 WTS, i = 2 MRF, …) 10.5: lower calorific value of methane (kWh/Nm3 CH4). In the above optimization problem (OPT 1), the main objective is to find the best combination of treatment methods in order to minimize the total cost of treating all waste components. The first component of the objective function is the total operating cost of WTS. The second component is the sum of the operating costs of all other treatment facilities (depending on the amount of waste sent to each one of these facilities). The third term indicates the total transfer cost of sending separated components from WTS to each one of the other treatment facilities while the fourth term corresponds to the transfer cost of residuals from each facility to the landfill (LF). The fifth and the sixth terms are the revenue generated from each component treated in the material recovery facility (MRF) and composting facility (CF), 365
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al. 6
6
the problem. Luckily, our optimization problems are simple in the sense that both the objective functions and the constraints are linear in decision variables. Therefore, we have a convex problem. Hence, a simple “weighing method”, which involves giving appropriate weights to both objective functions with their sum being equal to 1 will produce Paretooptimal solutions sufficient for our purpose (Gass and Saaty, 1955). It is well known that any Pareto optimal solution can be found by simply altering the weights of the objective functions for the convex optimization problems (Branke et al., 2008). Hence, the problem to be solved takes the following form:
∑ ∑ dij Pij DEm + ∑ ∑ CO2ij Pij
Min CO2 =
i=1
j
i
j=1
Rij = (1 − aij ) Pij, ∀ i , j Pj = ∑ Pij, ∀ j i
∑ Pij ≤ Capi ,
i = 2,3,4,5
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 j
MO = min{α (TC ) + (1 − α ) CO2} Pij
i=1
j
Subject to. Pij ∈ S for various values of. α ∈ {0,1}. The solutions of OPT 1, OPT 2 and the multi-objective optimization problem MO are presented in the next section and their implications are discussed in the following section.
Pij ≥ 0, ∀ i , j where CO2ij is the amount of CO2 gas emitted during the treatment of component j (j = 1 Paper, j = 2 Organic, j = 3 Plastic, j = 4 Glass, j = 5 Metal, j = 6 Other) in the treatment facility i (i = 1 WTS, i = 2 MRF, i = 3 INC, i = 4 CF, i = 5 ANB, i = 6 LF). The rest of the problem is similar to the first optimization problem formulation; i.e. the same constraints ensuring that all waste is treated, and the capacities are not violated apply here, too. In order to estimate the appropriate values of COij2 , several references were utilized. First, the average CO2 emission rates of various components were noted (Korkut et al., 2018). Next, CO2 emission amounts from various treatments in the Turkish case were gathered (European Environment Agency, EEA). Considering the total amount of waste and the composition of waste treated in each method, the approximate values of CO2ij were estimated. These values are represented in Table 4.
3. Results First, we have solved OPT 1 and OPT 2 separately. The percentage of each waste component that needs to be treated in each facility according to the optimal solution of OPT 1 is presented in Table 5. The results represented in Table 5 are mostly expected except for a few surprises. For instance, recycling is recommended for paper, plastics and metals, but is not profitable for glass. Moreover, composting and anaerobic digestion facilities are not efficient options for treating any of the waste components due to high operating costs and relatively low returns. Incineration is a different case although only 5.7% of the plastic waste is recommended to be sent to this facility. The reason for this low rate is the limited capacity of INC, not the cost efficiency of the process. Finally, all organic, glass and other waste and a part of the plastic waste are advised to be sent to LF. The main reason for this is the relatively low operating costs for the LF facility and the moderately high returns obtained by the generation of electricity. Moreover, we have generated a “sensitivity analysis”, namely a post-optimization analysis to understand how robust the results are to the problem parameter values. Some insights obtained from this analysis are stated below.
Note that we did not take into consideration the total amount of CH4 produced during the waste treatment since this value is always 25 times the amount of CO2 emission regardless of the type of component or the type of treatment method (Minoglu and Komilis, 2013). After solving OPT 1 and OPT 2 separately, we would like to solve both problems simultaneously by using the multi-objective optimization. In the multi-objective optimization, the problem to be solved is formulated (MO) as follows:
MO = min{TC , CO2} Pij
Subject to. Pij ∈ S where the feasible region S is defined by the constraints of OPT 1 and OPT 2, i.e.
S = {Pij | Pij = (1 − aij ) Pij, ∀ i, j;
∑ Pij ≤ Capi ,
• The return rate for the recycled glass was estimated as 70%*(glass
i = 2,3,4,5;
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 ; j
j
i=1
Pij ≥ 0, ∀ i, j}
•
Our objective in this problem is to identify the “Pareto optimal solutions”, i.e. the solutions for which the cost cannot be improved further without increasing the CO2 emission rate and vice versa. In order to attain this set of solutions, we will use “scalarization”. Scalarization means the problem involving multiple objectives is converted into an optimization problem with a single objective function (Branke et al., 2008). Scalarization can be quite complicated depending on the form of
recovery rate in MRF facility)*35$. It is found that, including the operating cost of MRF facility and the transportation costs, if this rate was only 12$ higher, the optimal solution would advocate sending some glass to the MRF facility, too. Hence, if the glass recovery rate is improved or if the market value of glass is increased by approximately $17, it would be profitable to recycle glass waste as well. The operating costs of CF and ANB are too high to allow any profitable operation. In fact, it would be intuitive to send organic waste to these facilities. However, neither the revenue generated from compost commercially nor producing electricity in ANB justify treating these components in CF and ANB facilities. For these
Table 5 The percentage (%) of waste components treated in each facility according to the optimal solution of OPT 1.
Table 4 Tons of CO2 produced by treating 1 ton of each component under each method. Component/Treatment
MRF
INC
CF
ANB
LF
Component/Treatment
MRF
INC
CF
ANB
LF
Paper Organic Plastics Metals Glass Other
0 N/A 0 0 0 N/A
0.043 0.39 0.294 0 0 0.019
0.001 0.014 N/A N/A N/A N/A
0 0 N/A N/A N/A N/A
0.086 0.78 0.587 0 0 0.039
Paper Organic Plastics Metals Glass Other
74.5% N/A 94.3% 100% 0 N/A
0 0 5.7% 0 0 0
0 0 N/A N/A N/A N/A
0 0 N/A N/A N/A N/A
25.5% 100% 0 0 100% 100%
366
Journal of Environmental Management 244 (2019) 362–369
367
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
3057.79 −702,125
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF
to MRF, 25.5% to
100% to LF
94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC
to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
2760.74 100% to LF
0.1
•
and ANB would be helpful in reducing the total CO2 emissions. In fact, increasing the capacity of INC and ANB by 1 ton per day would decrease the CO2 emissions rate by 0.39 and 0.78 tons/day, respectively, since more organic waste can be sent to either of these facilities in that case. Increasing the capacity of CF by itself does not help because of the special case in İstanbul that composting and recycling have a shared capacity. If the capacity of composting and the total shared capacity is increased by 1 ton, the CO2 emissions rate will improve by 0.77 tons/day. On the other hand, if the capacity of CF is held constant while the total shared capacity is increased by 1 ton, the CO2 emissions rate will improve by 0.5 tons/ day because now some plastics can be sent to the MRF facility. We have already pointed out that the main reason for the above results is the fact that organic waste produces very high CO2 emissions rates compared to other types of waste. According to the sensitivity analysis, it is found that even if the amount of organic waste was 37% lower than the current values estimated, nothing changes. However, if it is 37.05% lower, then some plastics waste can be treated in the MRF. Hence, if more of other waste types are aimed to be treated in facilities other than LF, either the capacity of
100% to MRF
• Since their capacities are used fully, increasing the capacity of INC
0.001
Next, we solve OPT 2. The following Table displays the percentage of each waste component that needs to be treated in each facility according to the optimal solution of OPT 2. The results in Table 6 appear to be more surprising than the solution of OPT 1. For instance, one would intuitively expect some material to be sent to the MRF facility to minimize CO2 emission. However, the main reason underlying these results is the high CO2 emission rate of organic waste, especially if treated in LF. To avoid this, organic waste is sent to the incineration, composting and anaerobic digestion facilities up to their limits. Moreover, due to the fact that the composting and the recycling facilities are combined in İstanbul sharing the same processing capacity, none of the other waste components could be sent to MRF facility although this would be a good option to decrease the total CO2 emissions. Finally, rest of the materials that could not be treated elsewhere due to capacity issues are sent to LF. Again, performing a sensitivity analysis on this solution helps us identify several key insights. The following bullet points discuss the main insights we observe.
100% to LF
Other
Table 7 The percentage (%) of waste components treated in each facility according to the optimal solution of MO for different values of ∝ .
•
processes to be commercially feasible, their operating costs must be decreased at least by 90%; and the recovery rates of paper and organic waste in these processes must be improved. Both MRF and INC facilities operate up to their maximum limits. Hence, it will be much more profitable if their capacities can be increased. For instance, increasing the capacity of the MRF facility by 1 ton would improve the costs by 160 TL per day. In this way, firstly, more paper can be sent to the MRF facility; and after the extra capacity in MRF is sufficient to treat all paper, some glass can be treated in the MRF facility. Similarly, if the capacity of the incineration facility can be improved, an additional cost saving of approximately 40 TL per ton per day is obtained and more plastic can be treated here.
100% to LF
0 37.1% 100% 100% 100% 100%
100% to LF
0 2.9% N/A N/A N/A N/A
1.9% to INC, 58.1% to CF, 2.9% to ANB, 37.1% to LF 27%CF, 70% to LF, 3% to ANB
0 58.1% N/A N/A N/A N/A
100% to LF
0 1.9% 0 0 0 0
0
0 0 0 0 0 N/A
total cost (transportation not included) (TL/day)
Paper Organic Plastics Metals Glass Other
Metal
LF
Glass
ANB
Plastics
CF
Organic
INC
Paper
MRF
α
Component/Treatment
CO2 emission (tons/day)
Table 6 The percentage (%) of waste components treated in each facility according to the optimal solution of OPT 2.
525,729
N. Ayvaz-Cavdaroglu, et al.
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
note several interesting points. For instance, Minoglou and Komilis (2013) find ANB and INC as more desirable treatment strategies. They find LF as the least desirable due to the high costs and high GHG emissions. However, in our case, from a cost perspective, LF is among the desirable treatments. This is mainly because operating costs in LF are low and the capacity of MRF and INC are not sufficient currently. Mavrotas et al. (2013) also find MRF, INC and ANB among the desirable treatments. Jing et al. (2009) find that the centralized composting and incinerating facilities are desired for treating the organic waste flows. Levis et al. (2013) again develop scenario-based solutions and find that in the MinCost scenario, all of the generated waste was landfilled, resulting in the lowest cost, the greatest GHG emissions, and zero diversion. In the GHG scenario, MRF was utilized as much as possible leading to the least GHG emissions. Overall, our results are mostly aligned with the literature such that LF, MRF and INC are the most desirable treatment facilities to minimize the cost and reduce GHG emissions.
treatment facilities must be increased or the organic waste amount should be decreased. After solving OPT 1 and OPT 2 separately, we next solve both problems simultaneously by using the multi-objective optimization. According to the formulation given in MO, the problem is solved by giving various weights, ∝ and 1 − ∝ , respectively, to the two objective functions. The solutions are presented in Table 7 for different values of ∝. The results in Table 7 display that the solution very quickly converges to the optimal solution of OPT 1. That is, for even small values of ∝, the optimal solution advocates for sending 74.5% of paper; 94.3% of plastics; and all the metal to MRF instead of LF; organic waste and glass to LF, allocating the INC capacity to treat the remaining part of plastic, and not using CF or ANB at all. The reason is simply the fact that if environmental concerns are the main priority, treating organic waste in other treatment facilities while sending all the other components to LF would result in escalating costs; and the revenue-generating possibilities cannot be utilized at all.
5. Conclusion In this study, the SWM practices of İstanbul are analyzed by applying the techniques from MP methodology. In this regard, the application of five MSW management technologies which are currently in use in İstanbul on six waste components is analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides, this solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments. Our problem does not involve facility location selection or construction costs of new facilities. Moreover, due to the fact that all facilities are in the same place, no transportation costs are included between facilities. The solutions of OPT 1 (cost-effective option), OPT 2 (environmentally friendly option), and MO (the multi-objective optimization problem), and the sensitivity analysis were performed to be able to evaluate the integrated SWM system of İstanbul. The results of the study emphasize the importance of material recovery and incineration facilities. In particular, MRF should be expanded to be able to treat all of metal, paper and plastic from a cost management perspective. INC should be expanded in order to treat plastics or organic waste from a GHG minimization perspective. For ANB and CF, the current parameters do not enable a commercially feasible operation; their operating costs must be decreased at least by 90% for these processes to be economically feasible. Regarding the waste composition, the amount of organic waste must be decreased by more than 37% for other waste streams to be treated in different facilities other than landfill. Our results are also mostly aligned with the findings in the literature. The featuring of LF is mainly due to low capacity of the other facilities and low operating costs of the landfill. However, these results may differ according to problem parameters. The sensitivity analysis we performed gives key insights regarding how the solutions can change with respect to the waste composition, costs of operation and facility capacities. All this said, the methodology represented in this study can be extended and generalized to other cities around the world once the correct problem parameters are specified.
4. Discussion The solutions of OPT 1, OPT 2 and MO, and the results of the sensitivity analysis lead to several insights for the waste management problem of İstanbul. First of all, we found that, from a cost management perspective, it is desirable to recycle all the metal, most of the paper and most of the plastics. Currently the return rate of glass is too low to make the recycle process profitable. If the recovery efficiency of glass improves, or if its profit per unit increases by around 50%, recycling glass could also become a profitable option. Moreover, currently CF and ANB do not seem to be efficient waste treatment alternatives. New technology enabling higher efficiency levels of the ANB or CF processes or decreasing the operating costs in these facilities are required for these facilities to be considered as profitable options for treating organic waste. Finally, the capacities of INC and MRF are currently not sufficient. Regarding an environmental perspective, handling organic waste seems to be the main problem. Again, the limited capacity of INC seems to be a problem here, too. Increasing the capacity of INC and MRF would be helpful in reducing the total CO2 emissions. Moreover, the special case of the shared capacity of CF and MRF facilities in İstanbul emerges as a big problem here, since the environmentally-friendly and cost-efficient (revenue generating) process of MRF cannot be utilized due to the fact that the shared capacity is used to treat organic waste in CF. Hence, the capacity issue of MRF emerges once again. Observing the Pareto efficient solutions of the multi-objective optimization problem, it is evident that the main obstacle in front of minimizing the CO2 emissions is the quickly escalating costs. The main reason for this is again the limited capacity of INC and MRF facilities. Hence, our main recommendation for İstanbul would be to expand the capacity of MRF facility as soon as possible and to build an INC facility with the highest possible capacity (at least with a larger capacity than İZAYDAŞ). Comparing our results with the current MSW operations in İstanbul, we can note the following differences: First, the recycling process occurring in MRF is not sufficient. A big portion of the waste input directly goes to the LF. However, according to our results, it is very desirable to recycle all metal, and paper and plastic as much as possible. CF is heavily used to produce compost material to be used in landscaping activities of the municipality. However, our results suggest that CF and ANB are not efficient facilities at their current cost rates. Finally, INC facility of İstanbul city is still being constructed. However, our results suggest that the planned capacity might not be enough to serve the entire metropole of İstanbul and the decisionmakers should better think of planning for larger capacities. Comparing our results with the results of relevant literature, we can
Declarations of interest None. References Assamoi, B., Lawryshyn, Y., 2012. The environmental comparison of landfilling vs. incineration of MSW accounting for waste diversion. Waste Manag. 32 (5), 1019–1030. Badran, M.F., El-Haggar, S.M., 2006. Optimization of municipal solid waste management in Port Said – Egypt. Waste Manag. 26 (5), 534–545. Benítez, S.O., Lozano-Olvera, G., Morelos, R.A., de Vega, C.A., 2008. Mathematical modeling to predict residential solid waste generation. Waste Manag. 28, 7–13.
368
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Manag. Res. 1–9. Lee, C.K.M., Yeung, C.L., Xiong, Z.R., Chung, S.H., 2016. A mathematical model for municipal solid waste management–A case study in Hong Kong. Waste Manag. 58, 430–441. Levis, J.W., Barlaz, M.A., DeCarolis, J.F., Ranjithan, S.R., 2013. A generalized multistage optimization modeling framework for life cycle assessment-based integrated solid waste management. Environ. Model. Softw 50, 51–65. Louis, G., Shih, J.-S., 2007. A flexible inventory model for municipal solid waste recycling. Soc. Econ. Plann. Sci. 41, 61–89. Mavrotas, G., Skoulaxinou, S., Gakis, N., Katsouros, V., Georgopoulou, E., 2013. A multiobjective programming model for assessment the GHG emissions in MSW management. Waste Manag. 33, 1934–1949. Metin, E., Erozturk, A., Neyim, C., 2003. Solid waste management practices and review of recovery and recycling operations in Turkey. Waste Manag. 23, 425–432. Minoglou, Komilis, 2013. Optimizing the treatment and disposal of municipal solid wastes using mathematical programming—a case study in a Greek region. Resour. Conserv. Recycl. 80, 46–57. Münster, M., Ravn, H., Hedegaard, K., Juul, N., Söderman, M.L., 2015. Economic and environmental optimization of waste treatment. Waste Manag. 38, 486–495. Pressley, P.N., Levis, J.W., Damgaard, A., Barlaz, M.A., DeCarolis, J.F., 2015. Analysis of material recovery facilities for use in life-cycle assessment. Waste Manag. 35, 307–317. Thorneloe, S.A., Weits, K.A., Nishtala, S.R., Yarkosky, S., Zannes, M., 2002. The impact of municipal solid waste management on greenhouse gas emissions in the United States. J. Air Waste Manag. Assoc. 52, 1000–1011. Xi, B.D., et al., 2010. An integrated optimization approach and multi-criteria decision analysis for supporting the waste-management system of the City of Beijing, China. Eng. Appl. Artif. Intell. 23, 620–631. Yıldız, Ş., Saltabaş, F., Balahorli, V., Sezer, K., 2009. Organik atıklardan biyogaz üretimi (biyometanizasyon) projesi – İstanbul örneği. TÜRKAY . İstanbul 1–8. Yıldız, Ş., Balahorli, V., Hoşoğlu, F., Sezer, K., 2014. Waste Management in the Metropolis of Istanbul. Yu, H., Solvang, W.D., 2017. A multi-objective location-allocation optimization for sustainable management of municipal solid waste. Environ. Syst. Decisions 37 (3), 289–308.
Berkun, M., Aras, E., Nemlioglu, S., 2005. Disposal of solid waste in Istanbul and along the black sea coast of Turkey. Waste Manag. 25, 847–855. Branke, J., Deb, K., Miettinen, K., R. (Eds.), 2008. Slowiński. Multiobjective Optimization: Interactive and Evolutionary Approaches., vol. 5252 Springer Science & Business Media. Calabrò, P.S., 2009. Greenhouse gases emission from municipal waste management: the role of separate collection. Waste Manag. 29, 2178–2187. Dai, C., Li, Y.P., Huang, G.H., 2011. A two-stage support-vector-regression optimization model for municipal solid waste management – a case study of Beijing, China. J. Environ. Manag. 92 (12), 3023–3037. Demir, G., Ozcan, H.K., Karakus, P.K., Bakis, Y., 2017. Solid waste collection route optimisation by geographical information system (GIS) in fatih-istanbul, Turkey. Int. J. Glob. Warming 11 (3), 263–272. Eschenroeder, A., 2001. Greenhouse gas dynamics of municipal solid waste alternatives. J. Air Waste Manag. Assoc. (51), 1423–1427. Gass, S., Saaty, T., 1955. The computational algorithm for the parametric objective function. Nav. Res. Logist. Q. 2, 39–45. Hochman, G., Wang, S., Li, Q., Gottlieb, P.D., Xu, F., Y, Li, 2015. Cost of organic waste technologies: a case study for New Jersey. AIMS Energy 3 (3), 450–462. Hoornweg, D., Bhada-Tata, P., 2012. WHAT A WASTE A Global Review of Solid. Urban Development & Local Government Unit World Bank, Washington, DC 20433 USA. Jing, S., Huang, G.H., Xi, B.D., Li, Y.P., Qin, X.S., Huo, S.L., Jiang, Y., 2009. A hybrid inexact optimization approach for solid waste management in the city of Foshan, China. J. Environ. Manag. 91, 389–402. Kanat, G., 2010. Municipal solid-waste management in Istanbul. Waste Manag. 30, 1737–1745. Kanat, G., Demir, A., Ozkaya, B., Bilgili, M.S., 2006. Addressing the operational problems in a composting and recycling plant. Waste Manag. 26, 1384–1391. Karadag, D., Sakar, S., 2003. Cost and financing of municipal solid waste collection services in İstanbul. Waste Manag. Res. 21, 480–485. Kirkeby, J.T., Birgisdottir, H., Hansen, T.L., Christensen, T.H., Bhander, G.S., Hauschild, M., 2006. Evaluation of environmental impacts from municipal solid waste management in the municipality of Aarhus, Denmark (EASEWASTE). Waste Manag. Res. 24, 16–26. Korkut, N.E., Yaman, C., Küçükağa, Y., Jaunich, M.K., 2018. Greenhouse gas contribution of municipal solid waste collection: a case study in the city of Istanbul, Turkey. Waste
369
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Research article
Municipal solid waste management via mathematical modeling: A case study in İstanbul, Turkey
T
Nur Ayvaz-Cavdaroglua, Asli Cobanb, Irem Firtina-Ertisc,∗ Kadir Has University Business Administration Department, İstanbul, 34083, Turkey İzmir University of Economics Faculty of Engineering, Civil Engineering Department, İzmir, 35330, Turkey c Bahçeşehir University Faculty of Engineering and Natural Science, Energy Systems Engineering Department, İstanbul, 34353, Turkey a
b
A R T I C LE I N FO
A B S T R A C T
Keywords: Municipal solid waste management Optimization Mathematical programming
The prominence of managing municipal solid waste (MSW) in an efficient and effective manner is increasing from day to day. In this paper, the solid waste management (SWM) system of İstanbul is analyzed by applying the techniques from mathematical programming methodology. In this manner, the solutions of the two optimization problems which aim to minimize the total cost and the environmental effects of SWM, respectively, are presented in this study. Additionally, a sensitivity analysis is performed and a multi-objective problem that combines two problems is presented. In this regard, the application of five MSW management technologies which are currently in use in İstanbul on six waste components is analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides, this optimal solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments for the policymakers. The results of the study emphasize the importance of material recovery and incineration facilities to improve profitability and to minimize environmental side effects. In particular, material recovery facility (MRF) should be expanded to be able to treat all of metal, paper and plastic from a cost management perspective. Incineration (INC) facility should also be expanded in order to treat plastics or organic waste from a Greenhouse Gas (GHG) minimization perspective. In addition to this, landfill appears to be the most prominent treatment technique according to the current problem parameters. However, regarding the waste composition, the amount of organic waste must be decreased by more than 37% for other waste streams to be treated in different facilities other than landfill. Anaerobic digestion and composting facilities need to be more cost-effective for becoming economically feasible. The methodology represented in this study can be extended and generalized to other cities around the world once the correct problem parameters are specified.
1. Introduction The prominence of managing municipal solid waste (MSW) in an efficient and effective manner is increasing from day to day. The growing numbers of urban residents and the increasing amount of waste generated per person are the two main reasons for the growing need of better solid waste management (SWM) (Hoornweg and BhadaTata, 2012). Poorly managed MSW has considerable adverse effects on human health and the economy. Moreover, poorly managed waste contributes to greenhouse gas (GHG) emissions increasing the global warming effect. In this vein, designing the optimal MSW system seems imperative for all public authorities. Mathematical programming (MP) models are broadly used in the design of optimal SWM systems in the past (Dai et al., 2011). Several
∗
case studies in the literature display the effectiveness of the solution produced by using mathematical tools. For instance, Benítez et al. (2008) establish mathematical models that correlate the generation of residential solid waste per capita to the variables such as education, income per household, and the number of residents in a Mexican city. Similar studies span various regions in the world, like e.g. Beijing-China (Xi et al., 2010); Foshan–China (Jing et al., 2009); Port Said–Egypt (Badran and El-Haggar, 2006); Alleghany County–the USA (Louis and Shih, 2007); and Hong Kong (Lee et al., 2016). Most of these papers involve mixed integer programming formulations, which is a prominently utilized method to deal with the complex MSW problem. Costi et al. (2004) use a mixed integer nonlinear programming decision support model, which aims to find the optimal number of facilities and locations to minimize the waste management cost. Similarly, Jing et al.
Corresponding author. E-mail addresses: [email protected] (N. Ayvaz-Cavdaroglu), [email protected] (A. Coban), irem.fi[email protected] (I. Firtina-Ertis).
https://doi.org/10.1016/j.jenvman.2019.05.065 Received 4 February 2019; Received in revised form 26 April 2019; Accepted 17 May 2019 Available online 23 May 2019 0301-4797/ © 2019 Elsevier Ltd. All rights reserved.
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
this optimal solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments for the policymakers. The problem is modeled as a multi-objective optimization model and solved by the commercial software MS Excel Solver. Our methodology can be extended and generalized to other cities around the world once the correct problem parameters are specified. The rest of the paper is organized as follows. In Section 2, the details of the SWM problem of İstanbul and the methodology used to analyze and solve this problem from a mathematical programming perspective are represented. The results are presented in Section 3, including the two optimization problems which aim to minimize the total cost of SWM and the environmental effects of SWM. The result of the multiobjective problem combines the objectives of these two problems. In Section 4, the findings of the study are discussed and recommendations are developed for policymakers. Finally, the importance of the findings is summarized in Section 5.
(2009) develop a two-stage chance-constraint mixed integer linear programming model for solid waste management in Foshan and find that the centralized composting and incinerating facilities are desired. However, most of the studies only consider the problem from an economic perspective and aim to minimize the total cost of the system. There appears to be limited work on the application of mathematical optimization to design SWM systems with the objective of minimizing GHG emissions. Lu et al. (2006) develop an inexact dynamic optimization model (IDOM) for MSW-management systems under uncertainty. This model integrates GHG components into the modeling framework treating them as constraints. Recent scientific literature presents a large number of studies related to the analysis of GHG emissions from MSW management mainly developed by Life Cycle Assessment (LCA) (Kirkeby et al., 2006; Thorneloe et al., 2002; Eschenroeder, 2001). These studies show how recycling can allow a reduction of GHG emissions. Another example, Levis et al. (2013), present the first lifecycle based framework to optimize the collection and treatment of all waste materials from the curb to final disposal by minimizing cost or environmental impacts. Similarly, Pressley et al. (2015), Münster et al. (2015), Calabrò (2009) present a detailed and novel life-cycle modeling of SWM systems, and apply different optimization objectives such as minimizing costs or greenhouse gas emissions. The research involving a multi-objective optimization approach that simultaneously considers the environmental and economic objectives is quite rare. An example of this line of research is the work of Mavrotas et al. (2013), who formulate a general problem and show the application of the general model on a Greek case by developing four scenarios (waste management options). On the contrary, our model does not require defining possible scenarios to be applied in a pilot region. Yu and Solvang (2017) develop a multi-objective location-allocation optimization model to balance the trade-off among system operating costs, GHG emissions and environmental impact. Their main model is conceptual while they also present a numerical experiment with hypothetical values. They do not model specific treatment facilities as we do. Another important difference between our work and these two papers is that we measure the robustness of the proposed solution via sensitivity analysis. There are also several studies that consider the MSW system of İstanbul. Berkun et al. (2005) forecast the overall solid waste production in İstanbul as 4.75 millions of tons annually and 0.63 kg/day/capita. Besides, a yearly solid waste constituent variation in Istanbul (% by weight) is given. In our paper, the flow chart of İstanbul Municipal Solid Waste Composting and Recycling Plant is adapted from Kanat et al. (2006). The information about waste collection practices, waste amounts, transfer stations and routes, number of trips, truck capacity and GHG emissions is gathered from the studies of Korkut et al. (2018) and Demir et al. (2017). Karadag and Sakar (2003) state the costs of SWM and collection-transfer services in İstanbul, which are also used in our study. Kanat (2010) represents the MSW quantity (including recyclables), waste components (% wet weight), and total MSW management cost in İstanbul. In addition, estimated costs for MSW disposal alternatives and capacities of SWM plants are given. The composition of MSW and recyclable wastes in İstanbul, as well as the cost estimation for a recycling program, are listed in Metin et al. (2003). Incineration and landfill cost parameters are estimated based on the study of Assamoi and Lawryshyn (2012). In addition, the anaerobic digestion cost parameter is estimated based on the study of Hochman et al. (2015). In this paper, the SWM practices of İstanbul are analyzed by applying the techniques from mathematical programming methodology. In this regard, the application of five MSW management technologies currently in use in İstanbul (namely; landfilling, composting, recycling; with an exception of incineration and anaerobic digestion) on six waste components (namely, organic, paper, plastic, glass, metal, and other) are analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides,
2. Methodology İstanbul is the biggest city in Turkey with a population nearing 15 million. In this huge city, there are two prominent sanitary landfill sites (stations), one in Asian (Kömürcüoda) and one in European (Odayeri) side. The European Environment Agency (EEA) disseminates the MSW composition in Turkey and disposal routes including controlled landfills, dumpsites, and composting. Composting facility (CF) capacities of İstanbul and incineration facility capacity of Kocaeli (İZAYDAŞ) are also mentioned in this study. The incineration plant capacity is assumed to be equal to İZAYDAŞ plant (located in Kocaeli, Turkey) at a capacity value of 35,000 tons/year and assumed to be located near one of the sanitary landfill sites of İstanbul. However, the incineration plant of İstanbul which is going to be located in the Fatih district is under construction, and is expected to be completed in 2020. Another assumption is made considering the anaerobic digestion plant of İstanbul. It is expected to be constructed in the coming years; however, in this study, it is assumed to be located near one of the sanitary landfill sites. All these MSW facilities of İstanbul, which are either constructed or to be constructed, are under the administration of İSTAÇ Co. İSTAÇ is the government-funded company that handles the MSW management in İstanbul with its more than 4000 personnel operating in more than 40 facilities. For our analysis, the first piece of information required is the waste composition in İstanbul. The typical physical composition of MSW in İstanbul and the other necessary information regarding the methane and energy producing potential of each component is presented in Table 1. The typical integrated SWM process in İstanbul (outer dashed line in Fig. 1) involves the following steps: First, all components of the waste are transferred to eight waste transfer stations (WTS) within the city. The majority of the waste remains non-separated at this point. We assume that this is the case for all the waste collected. All this collected Table 1 Energy production potential of each component in MSW. Component
Composition (%)a
Lower heating value (kWh/wet ton)b
Methane Potential (L CH4/wet kg)c
Paper Organic Plastics Metals Glass Other
13.3 50.22 14.39 1.63 5.82 14.64
3930 1193 10,880 0 0 0
145.8 300.7 0 0 0 0
a b c
363
Adapted from Kanat (2010). Adapted from Minoglou and Komilis (2013). Adapted from Minoglou and Komilis (2013).
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Landfill Gas (CH4 ) to Energy (Electricity) Recovered (Granulated, PET, RDF)
Trucks
Municipalities
Waste Transfer Station (WTS)
1
Bigger Trucks
Waste Input
Material Recovery Facility (MRF)
Unrecovered
Landfill (LF) Leachate
2 Composting Facility (CF)
Compost Non-Compostable
3 Incineration Facility (INC)
Ashes Heat to Energy
4 Anaerobic Digestion Facility (ANB)
Waste CH4 to Energy
Fig. 1. Integrated SWM system in general (Outer dashed line shows the İstanbul case).
potential. Thus, the power generation can only be possible on the unprocessed waste and the non-recoverable recycling components accepted to the landfill. As apparent from Table 2, for example, it does not make sense to send glass or metal to an incineration facility since a significant portion (97%–98%) of these components will remain unprocessed in this facility. Similarly, composting can be possible on organic or paper waste, but not on glass, metal, or plastic. Finally, the cost functions for each type of waste management facility and the other parameters required for the problem formulation are presented in Table 3. These numbers are adapted from several sources in the literature, which are mentioned individually in Table 3. The values are converted to 2018 figures by taking into account the inflation rate and dollar/TL exchange rates. Note that only the operating cost of the facilities have been taken into consideration, since these facilities have already been in use for several years, and have reached the break-even point for the initial investment costs. Moreover, we only focus on the current (and assumed to be current) facilities and do not consider the problem regarding whether new facilities should be opened or not, since this dimension is going to complicate the current problem at hand. Hence, we can ignore the initial investment cost of the facilities.where: Ci : operating cost for facility i (per ton), (i = 1 WTS, i = 2 MRF, i = 3 INC, i = 4 CF, i = 5 ANB, i = 6 LF). Pj : production rate of the waste component j per day, (j = 1 Paper, j = 2 Organic, j = 3 Plastic, j = 4 Glass, j = 5 Metal, j = 6 Other). Pij : amount of waste component j that is sent from WTS to facility i per day, (i = 2 MRF, …j = 1 Paper, …) T1i : transportation cost of sending 1 ton of material from WTS to facility i, (i = 2 MRF, …j = 1 Paper, …) Ti6 : transportation cost of sending 1 ton of material from facility i to the LF, (i = 1 WTS, …) Rij : amount of residuals of waste component j that is sent from WTS to facility i per day, (i = 2 MRF, …j = 1 Paper, …) aij : recovery/processing rate of the waste component j in facility i (i = 1 WTS, i = 2 MRF, …j = 1 Paper, …) Mj : market value of the waste component j (j = 1 Paper, …) CMj : market value of the waste component j that is converted to compost (j = 1 Paper, …) MYj: methane yield of the waste component j (j = 1 Paper, …).
waste is transferred to the main waste collection facility in Kömürcüoda (Asian) and in Odayeri (European). The total waste content is sent to “MRF+CF” first. The recyclable fraction of the waste is sorted here, and sent to the material recovery facility (MRF) while the rest (organic fraction) is sent to the composting facility (CF). In İstanbul, as a special case, these two treatments are located within the same facility. The other alternative routes after the main waste input are Landfill (LF), Incineration (INC), and ANB (Anaerobic Digestion) facilities. All the residual materials and unrecovered part of recyclable materials are sent to the landfill (LF) as well. The schematic representation of the integrated solid waste management system in general is represented in Fig. 1. Four principal methods are numbered in Fig. 1, including Landfill (LF), Material Recovery and Composting (MRF+CF), Incineration (INC), and Anaerobic Digestion (ANB). The outer dashed line represents the boundaries of integrated SWM operations which are currently performed in İstanbul. Green and black lines represent the product and waste streams, respectively. The recovery rate for recyclable materials can vary from 0% to 100% depending on the source and condition of the material. Similarly, the residue amounts of different components vary across the treatment processes depending on the appropriateness (suitability) of the process for that particular waste component. The average rates for the efficiency of each process for each component are presented in Table 2 in the form of the percentage of residues. The unrecovered components are then hauled to the sanitary landfill. The sanitary landfill accepts unprocessed materials as well as the residues of the waste treated in one of the other facilities. We assume that the residues from the facilities of CF or INC do not have methane Table 2 The percentage (%) of residues after the treatment of different components in each treatment process. Component/Treatment
MRF
INC
CF
ANB
Paper Organic Plastics Metals Glass Other
30 N/A 30 30 30 N/A
11 1 2 97 98 25
65 35 100 100 100 100
75 45 100 100 100 100
364
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Table 3 Cost functions for each type of waste management facility and the other parameters required for the problem. 6
5
5
5
5
5
5
Min TC = C1 åPj + å å Ci Pij + å å T1i Pij + å å Ti6 Rij − å Mj P2j a2j − å CMj P4j a4j − å å 10.5MYj Pij BRECi EEConvi CostElec (1 − Rij ) − å Pij EEConv5 CostElecLHVj (1 − Rij ) j i=2
j
j i=1
j i=1
i=1
i=1
i = 3,6 j = 1
j=1
Rij = (1 − aij ) Pij, ∀ i , j; Pj = ∑ Pij, ∀ j; i
∑ Pij ≤ Capi ,
i = 2,3,4,5
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 j
j
i=1
Pij ≥ 0, ∀ i , j;
Cost Function
Value
Adapted from
Collection and transport cost of waste to WTS Operating cost for LF Operating cost for WTS Operating cost for CF Operating cost for ANB Operating cost for MRF Operating cost for INC Market price of recycled components (metal) Market price of recycled components (plastic) Market price of recycled components (glass) Market price of recycled components (paper) LF biogas recovery efficiency Revenue from the sale of 1 kWh of electrical energy ANB biogas recovery efficiency Electrical energy conversion efficiency (INC) Electrical energy conversion efficiency (ANB) Electrical energy conversion efficiency (LF) Transport cost of waste from INC, ANB, CF, MRF to LF Daily processing capacity of LF Daily processing capacity of WTS Daily processing capacity of CF (MRF included) Daily processing capacity of ANB Daily processing capacity of MRF (CF included) Daily processing capacity of INC GHG Emissions Global Warming Factor (GWF) GHG per capita
24.4 + 6 = 30.4 $/ton 20 $ 10% of 10 $/ton 30 $ 60 $ 20 $ 82 $ 120 $/ton 150 $/ton 35 $/ton 80 $/ton 30% 0.25 TL/kWh 100% 27% 33% 30% 0 16,000 tons/day 6000*2 = 12,000 tons/day (for two sides of Istanbul) 3000 tons/day 150 tons/day 3000 tons/day 35,000 tons/year = 97.2 tons/day 1189 ton CO2-eq 7.7 kg CO2-eq/ton 3.16 kg CO2-eq/ca
Karadag and Sakar (2003) Hochman et al. (2015) (10–30$) https://www.epa.gov/ Hochman et al. (2015) (30–60$/ton) Hochman et al. (2015) (60–100 $/ton) Pressley et al. (2015) (19.8–24.9$/ton) Hochman et al. (2015) (80–120$). Metin et al. (2003) Metin et al. (2003) Metin et al. (2003) Metin et al. (2003) Minoglou and Komilis (2013) https://enerjienstitusu.org/ Minoglou and Komilis (2013) Minoglou and Komilis (2013) Minoglou and Komilis (2013) Minoglou and Komilis (2013) Assumed as located at the same plant Yıldız et al. (2014) Demir et al. (2017) Kanat (2010) Yıldız et al. (2009) Kanat (2010) https://www.izaydas.com.tr/ Korkut et al. (2018) Korkut et al. (2018) Korkut et al. (2018)
Based on these cost functions and parameters, the cost minimization problem (OPT 1) is formulated as follows.
respectively. The seventh term corresponds to the revenue obtained by generating electricity in incineration (INC) and landfill (LF) facilities. The final term is again the revenue obtained from electricity generation by Anaerobic Digestion (ANB). Note that the revenue terms appear with a minus sign in front of them since the objective function represents the total costs. Regarding the constraints, the first term determines how much residual is generated by treating each component in each of the treatment facilities. The second constraint ensures that all waste components are treated somehow; and the third constraint ensures that the capacities of the treatment facilities (except the LF) are not violated. The fourth term states the capacity constraint for the specific case of LF, since one has to consider the amount of residuals from the other treatment facilities in the computation of the capacity usage in LF. In addition to cost minimization, one of the other concerns is the effect of waste treatment on the environment. One possible way to measure this effect is to measure the CO2 emissions of a combination of treatment methods. Our second optimization problem with the objective of minimizing the total amount of CO2 (OPT 2) will thus be defined as follows.
BRECi : biogas recovery efficiency in LF or ANB facility, (i = 3 ANB, i = 6 LF). EEConvi : electric energy conversion efficiency of facility i (i = 3 ANB, i = 5 INC, i = 6 LF). CostElec : Revenue from the sale of the electricity generated in facility LF, I or ANB. LHVj : Lower heating value of MSW component j (kWh/t) (j = 1 Paper, …), see Table 1. Capi : Capacity of facility i (i = 1 WTS, i = 2 MRF, …) 10.5: lower calorific value of methane (kWh/Nm3 CH4). In the above optimization problem (OPT 1), the main objective is to find the best combination of treatment methods in order to minimize the total cost of treating all waste components. The first component of the objective function is the total operating cost of WTS. The second component is the sum of the operating costs of all other treatment facilities (depending on the amount of waste sent to each one of these facilities). The third term indicates the total transfer cost of sending separated components from WTS to each one of the other treatment facilities while the fourth term corresponds to the transfer cost of residuals from each facility to the landfill (LF). The fifth and the sixth terms are the revenue generated from each component treated in the material recovery facility (MRF) and composting facility (CF), 365
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al. 6
6
the problem. Luckily, our optimization problems are simple in the sense that both the objective functions and the constraints are linear in decision variables. Therefore, we have a convex problem. Hence, a simple “weighing method”, which involves giving appropriate weights to both objective functions with their sum being equal to 1 will produce Paretooptimal solutions sufficient for our purpose (Gass and Saaty, 1955). It is well known that any Pareto optimal solution can be found by simply altering the weights of the objective functions for the convex optimization problems (Branke et al., 2008). Hence, the problem to be solved takes the following form:
∑ ∑ dij Pij DEm + ∑ ∑ CO2ij Pij
Min CO2 =
i=1
j
i
j=1
Rij = (1 − aij ) Pij, ∀ i , j Pj = ∑ Pij, ∀ j i
∑ Pij ≤ Capi ,
i = 2,3,4,5
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 j
MO = min{α (TC ) + (1 − α ) CO2} Pij
i=1
j
Subject to. Pij ∈ S for various values of. α ∈ {0,1}. The solutions of OPT 1, OPT 2 and the multi-objective optimization problem MO are presented in the next section and their implications are discussed in the following section.
Pij ≥ 0, ∀ i , j where CO2ij is the amount of CO2 gas emitted during the treatment of component j (j = 1 Paper, j = 2 Organic, j = 3 Plastic, j = 4 Glass, j = 5 Metal, j = 6 Other) in the treatment facility i (i = 1 WTS, i = 2 MRF, i = 3 INC, i = 4 CF, i = 5 ANB, i = 6 LF). The rest of the problem is similar to the first optimization problem formulation; i.e. the same constraints ensuring that all waste is treated, and the capacities are not violated apply here, too. In order to estimate the appropriate values of COij2 , several references were utilized. First, the average CO2 emission rates of various components were noted (Korkut et al., 2018). Next, CO2 emission amounts from various treatments in the Turkish case were gathered (European Environment Agency, EEA). Considering the total amount of waste and the composition of waste treated in each method, the approximate values of CO2ij were estimated. These values are represented in Table 4.
3. Results First, we have solved OPT 1 and OPT 2 separately. The percentage of each waste component that needs to be treated in each facility according to the optimal solution of OPT 1 is presented in Table 5. The results represented in Table 5 are mostly expected except for a few surprises. For instance, recycling is recommended for paper, plastics and metals, but is not profitable for glass. Moreover, composting and anaerobic digestion facilities are not efficient options for treating any of the waste components due to high operating costs and relatively low returns. Incineration is a different case although only 5.7% of the plastic waste is recommended to be sent to this facility. The reason for this low rate is the limited capacity of INC, not the cost efficiency of the process. Finally, all organic, glass and other waste and a part of the plastic waste are advised to be sent to LF. The main reason for this is the relatively low operating costs for the LF facility and the moderately high returns obtained by the generation of electricity. Moreover, we have generated a “sensitivity analysis”, namely a post-optimization analysis to understand how robust the results are to the problem parameter values. Some insights obtained from this analysis are stated below.
Note that we did not take into consideration the total amount of CH4 produced during the waste treatment since this value is always 25 times the amount of CO2 emission regardless of the type of component or the type of treatment method (Minoglu and Komilis, 2013). After solving OPT 1 and OPT 2 separately, we would like to solve both problems simultaneously by using the multi-objective optimization. In the multi-objective optimization, the problem to be solved is formulated (MO) as follows:
MO = min{TC , CO2} Pij
Subject to. Pij ∈ S where the feasible region S is defined by the constraints of OPT 1 and OPT 2, i.e.
S = {Pij | Pij = (1 − aij ) Pij, ∀ i, j;
∑ Pij ≤ Capi ,
• The return rate for the recycled glass was estimated as 70%*(glass
i = 2,3,4,5;
j 5
∑ P6j + ∑ ∑ Rij ≤ Cap6 ; j
j
i=1
Pij ≥ 0, ∀ i, j}
•
Our objective in this problem is to identify the “Pareto optimal solutions”, i.e. the solutions for which the cost cannot be improved further without increasing the CO2 emission rate and vice versa. In order to attain this set of solutions, we will use “scalarization”. Scalarization means the problem involving multiple objectives is converted into an optimization problem with a single objective function (Branke et al., 2008). Scalarization can be quite complicated depending on the form of
recovery rate in MRF facility)*35$. It is found that, including the operating cost of MRF facility and the transportation costs, if this rate was only 12$ higher, the optimal solution would advocate sending some glass to the MRF facility, too. Hence, if the glass recovery rate is improved or if the market value of glass is increased by approximately $17, it would be profitable to recycle glass waste as well. The operating costs of CF and ANB are too high to allow any profitable operation. In fact, it would be intuitive to send organic waste to these facilities. However, neither the revenue generated from compost commercially nor producing electricity in ANB justify treating these components in CF and ANB facilities. For these
Table 5 The percentage (%) of waste components treated in each facility according to the optimal solution of OPT 1.
Table 4 Tons of CO2 produced by treating 1 ton of each component under each method. Component/Treatment
MRF
INC
CF
ANB
LF
Component/Treatment
MRF
INC
CF
ANB
LF
Paper Organic Plastics Metals Glass Other
0 N/A 0 0 0 N/A
0.043 0.39 0.294 0 0 0.019
0.001 0.014 N/A N/A N/A N/A
0 0 N/A N/A N/A N/A
0.086 0.78 0.587 0 0 0.039
Paper Organic Plastics Metals Glass Other
74.5% N/A 94.3% 100% 0 N/A
0 0 5.7% 0 0 0
0 0 N/A N/A N/A N/A
0 0 N/A N/A N/A N/A
25.5% 100% 0 0 100% 100%
366
Journal of Environmental Management 244 (2019) 362–369
367
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
4182.02 −1,417,222 to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
3057.79 −702,125
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
100% to LF to MRF, 25.5% to
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF 74.5% LF
to MRF, 25.5% to
100% to LF
94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC 94.3% INC
to MRF, 5.7% to
100% to LF
100% to MRF
100% to LF
2760.74 100% to LF
0.1
•
and ANB would be helpful in reducing the total CO2 emissions. In fact, increasing the capacity of INC and ANB by 1 ton per day would decrease the CO2 emissions rate by 0.39 and 0.78 tons/day, respectively, since more organic waste can be sent to either of these facilities in that case. Increasing the capacity of CF by itself does not help because of the special case in İstanbul that composting and recycling have a shared capacity. If the capacity of composting and the total shared capacity is increased by 1 ton, the CO2 emissions rate will improve by 0.77 tons/day. On the other hand, if the capacity of CF is held constant while the total shared capacity is increased by 1 ton, the CO2 emissions rate will improve by 0.5 tons/ day because now some plastics can be sent to the MRF facility. We have already pointed out that the main reason for the above results is the fact that organic waste produces very high CO2 emissions rates compared to other types of waste. According to the sensitivity analysis, it is found that even if the amount of organic waste was 37% lower than the current values estimated, nothing changes. However, if it is 37.05% lower, then some plastics waste can be treated in the MRF. Hence, if more of other waste types are aimed to be treated in facilities other than LF, either the capacity of
100% to MRF
• Since their capacities are used fully, increasing the capacity of INC
0.001
Next, we solve OPT 2. The following Table displays the percentage of each waste component that needs to be treated in each facility according to the optimal solution of OPT 2. The results in Table 6 appear to be more surprising than the solution of OPT 1. For instance, one would intuitively expect some material to be sent to the MRF facility to minimize CO2 emission. However, the main reason underlying these results is the high CO2 emission rate of organic waste, especially if treated in LF. To avoid this, organic waste is sent to the incineration, composting and anaerobic digestion facilities up to their limits. Moreover, due to the fact that the composting and the recycling facilities are combined in İstanbul sharing the same processing capacity, none of the other waste components could be sent to MRF facility although this would be a good option to decrease the total CO2 emissions. Finally, rest of the materials that could not be treated elsewhere due to capacity issues are sent to LF. Again, performing a sensitivity analysis on this solution helps us identify several key insights. The following bullet points discuss the main insights we observe.
100% to LF
Other
Table 7 The percentage (%) of waste components treated in each facility according to the optimal solution of MO for different values of ∝ .
•
processes to be commercially feasible, their operating costs must be decreased at least by 90%; and the recovery rates of paper and organic waste in these processes must be improved. Both MRF and INC facilities operate up to their maximum limits. Hence, it will be much more profitable if their capacities can be increased. For instance, increasing the capacity of the MRF facility by 1 ton would improve the costs by 160 TL per day. In this way, firstly, more paper can be sent to the MRF facility; and after the extra capacity in MRF is sufficient to treat all paper, some glass can be treated in the MRF facility. Similarly, if the capacity of the incineration facility can be improved, an additional cost saving of approximately 40 TL per ton per day is obtained and more plastic can be treated here.
100% to LF
0 37.1% 100% 100% 100% 100%
100% to LF
0 2.9% N/A N/A N/A N/A
1.9% to INC, 58.1% to CF, 2.9% to ANB, 37.1% to LF 27%CF, 70% to LF, 3% to ANB
0 58.1% N/A N/A N/A N/A
100% to LF
0 1.9% 0 0 0 0
0
0 0 0 0 0 N/A
total cost (transportation not included) (TL/day)
Paper Organic Plastics Metals Glass Other
Metal
LF
Glass
ANB
Plastics
CF
Organic
INC
Paper
MRF
α
Component/Treatment
CO2 emission (tons/day)
Table 6 The percentage (%) of waste components treated in each facility according to the optimal solution of OPT 2.
525,729
N. Ayvaz-Cavdaroglu, et al.
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
note several interesting points. For instance, Minoglou and Komilis (2013) find ANB and INC as more desirable treatment strategies. They find LF as the least desirable due to the high costs and high GHG emissions. However, in our case, from a cost perspective, LF is among the desirable treatments. This is mainly because operating costs in LF are low and the capacity of MRF and INC are not sufficient currently. Mavrotas et al. (2013) also find MRF, INC and ANB among the desirable treatments. Jing et al. (2009) find that the centralized composting and incinerating facilities are desired for treating the organic waste flows. Levis et al. (2013) again develop scenario-based solutions and find that in the MinCost scenario, all of the generated waste was landfilled, resulting in the lowest cost, the greatest GHG emissions, and zero diversion. In the GHG scenario, MRF was utilized as much as possible leading to the least GHG emissions. Overall, our results are mostly aligned with the literature such that LF, MRF and INC are the most desirable treatment facilities to minimize the cost and reduce GHG emissions.
treatment facilities must be increased or the organic waste amount should be decreased. After solving OPT 1 and OPT 2 separately, we next solve both problems simultaneously by using the multi-objective optimization. According to the formulation given in MO, the problem is solved by giving various weights, ∝ and 1 − ∝ , respectively, to the two objective functions. The solutions are presented in Table 7 for different values of ∝. The results in Table 7 display that the solution very quickly converges to the optimal solution of OPT 1. That is, for even small values of ∝, the optimal solution advocates for sending 74.5% of paper; 94.3% of plastics; and all the metal to MRF instead of LF; organic waste and glass to LF, allocating the INC capacity to treat the remaining part of plastic, and not using CF or ANB at all. The reason is simply the fact that if environmental concerns are the main priority, treating organic waste in other treatment facilities while sending all the other components to LF would result in escalating costs; and the revenue-generating possibilities cannot be utilized at all.
5. Conclusion In this study, the SWM practices of İstanbul are analyzed by applying the techniques from MP methodology. In this regard, the application of five MSW management technologies which are currently in use in İstanbul on six waste components is analyzed; and the optimal solution regarding the best mixture of these technologies is developed on a given waste composition. Besides, this solution is compared with the current practice in İstanbul; and recommendations are presented about possible future investments. Our problem does not involve facility location selection or construction costs of new facilities. Moreover, due to the fact that all facilities are in the same place, no transportation costs are included between facilities. The solutions of OPT 1 (cost-effective option), OPT 2 (environmentally friendly option), and MO (the multi-objective optimization problem), and the sensitivity analysis were performed to be able to evaluate the integrated SWM system of İstanbul. The results of the study emphasize the importance of material recovery and incineration facilities. In particular, MRF should be expanded to be able to treat all of metal, paper and plastic from a cost management perspective. INC should be expanded in order to treat plastics or organic waste from a GHG minimization perspective. For ANB and CF, the current parameters do not enable a commercially feasible operation; their operating costs must be decreased at least by 90% for these processes to be economically feasible. Regarding the waste composition, the amount of organic waste must be decreased by more than 37% for other waste streams to be treated in different facilities other than landfill. Our results are also mostly aligned with the findings in the literature. The featuring of LF is mainly due to low capacity of the other facilities and low operating costs of the landfill. However, these results may differ according to problem parameters. The sensitivity analysis we performed gives key insights regarding how the solutions can change with respect to the waste composition, costs of operation and facility capacities. All this said, the methodology represented in this study can be extended and generalized to other cities around the world once the correct problem parameters are specified.
4. Discussion The solutions of OPT 1, OPT 2 and MO, and the results of the sensitivity analysis lead to several insights for the waste management problem of İstanbul. First of all, we found that, from a cost management perspective, it is desirable to recycle all the metal, most of the paper and most of the plastics. Currently the return rate of glass is too low to make the recycle process profitable. If the recovery efficiency of glass improves, or if its profit per unit increases by around 50%, recycling glass could also become a profitable option. Moreover, currently CF and ANB do not seem to be efficient waste treatment alternatives. New technology enabling higher efficiency levels of the ANB or CF processes or decreasing the operating costs in these facilities are required for these facilities to be considered as profitable options for treating organic waste. Finally, the capacities of INC and MRF are currently not sufficient. Regarding an environmental perspective, handling organic waste seems to be the main problem. Again, the limited capacity of INC seems to be a problem here, too. Increasing the capacity of INC and MRF would be helpful in reducing the total CO2 emissions. Moreover, the special case of the shared capacity of CF and MRF facilities in İstanbul emerges as a big problem here, since the environmentally-friendly and cost-efficient (revenue generating) process of MRF cannot be utilized due to the fact that the shared capacity is used to treat organic waste in CF. Hence, the capacity issue of MRF emerges once again. Observing the Pareto efficient solutions of the multi-objective optimization problem, it is evident that the main obstacle in front of minimizing the CO2 emissions is the quickly escalating costs. The main reason for this is again the limited capacity of INC and MRF facilities. Hence, our main recommendation for İstanbul would be to expand the capacity of MRF facility as soon as possible and to build an INC facility with the highest possible capacity (at least with a larger capacity than İZAYDAŞ). Comparing our results with the current MSW operations in İstanbul, we can note the following differences: First, the recycling process occurring in MRF is not sufficient. A big portion of the waste input directly goes to the LF. However, according to our results, it is very desirable to recycle all metal, and paper and plastic as much as possible. CF is heavily used to produce compost material to be used in landscaping activities of the municipality. However, our results suggest that CF and ANB are not efficient facilities at their current cost rates. Finally, INC facility of İstanbul city is still being constructed. However, our results suggest that the planned capacity might not be enough to serve the entire metropole of İstanbul and the decisionmakers should better think of planning for larger capacities. Comparing our results with the results of relevant literature, we can
Declarations of interest None. References Assamoi, B., Lawryshyn, Y., 2012. The environmental comparison of landfilling vs. incineration of MSW accounting for waste diversion. Waste Manag. 32 (5), 1019–1030. Badran, M.F., El-Haggar, S.M., 2006. Optimization of municipal solid waste management in Port Said – Egypt. Waste Manag. 26 (5), 534–545. Benítez, S.O., Lozano-Olvera, G., Morelos, R.A., de Vega, C.A., 2008. Mathematical modeling to predict residential solid waste generation. Waste Manag. 28, 7–13.
368
Journal of Environmental Management 244 (2019) 362–369
N. Ayvaz-Cavdaroglu, et al.
Manag. Res. 1–9. Lee, C.K.M., Yeung, C.L., Xiong, Z.R., Chung, S.H., 2016. A mathematical model for municipal solid waste management–A case study in Hong Kong. Waste Manag. 58, 430–441. Levis, J.W., Barlaz, M.A., DeCarolis, J.F., Ranjithan, S.R., 2013. A generalized multistage optimization modeling framework for life cycle assessment-based integrated solid waste management. Environ. Model. Softw 50, 51–65. Louis, G., Shih, J.-S., 2007. A flexible inventory model for municipal solid waste recycling. Soc. Econ. Plann. Sci. 41, 61–89. Mavrotas, G., Skoulaxinou, S., Gakis, N., Katsouros, V., Georgopoulou, E., 2013. A multiobjective programming model for assessment the GHG emissions in MSW management. Waste Manag. 33, 1934–1949. Metin, E., Erozturk, A., Neyim, C., 2003. Solid waste management practices and review of recovery and recycling operations in Turkey. Waste Manag. 23, 425–432. Minoglou, Komilis, 2013. Optimizing the treatment and disposal of municipal solid wastes using mathematical programming—a case study in a Greek region. Resour. Conserv. Recycl. 80, 46–57. Münster, M., Ravn, H., Hedegaard, K., Juul, N., Söderman, M.L., 2015. Economic and environmental optimization of waste treatment. Waste Manag. 38, 486–495. Pressley, P.N., Levis, J.W., Damgaard, A., Barlaz, M.A., DeCarolis, J.F., 2015. Analysis of material recovery facilities for use in life-cycle assessment. Waste Manag. 35, 307–317. Thorneloe, S.A., Weits, K.A., Nishtala, S.R., Yarkosky, S., Zannes, M., 2002. The impact of municipal solid waste management on greenhouse gas emissions in the United States. J. Air Waste Manag. Assoc. 52, 1000–1011. Xi, B.D., et al., 2010. An integrated optimization approach and multi-criteria decision analysis for supporting the waste-management system of the City of Beijing, China. Eng. Appl. Artif. Intell. 23, 620–631. Yıldız, Ş., Saltabaş, F., Balahorli, V., Sezer, K., 2009. Organik atıklardan biyogaz üretimi (biyometanizasyon) projesi – İstanbul örneği. TÜRKAY . İstanbul 1–8. Yıldız, Ş., Balahorli, V., Hoşoğlu, F., Sezer, K., 2014. Waste Management in the Metropolis of Istanbul. Yu, H., Solvang, W.D., 2017. A multi-objective location-allocation optimization for sustainable management of municipal solid waste. Environ. Syst. Decisions 37 (3), 289–308.
Berkun, M., Aras, E., Nemlioglu, S., 2005. Disposal of solid waste in Istanbul and along the black sea coast of Turkey. Waste Manag. 25, 847–855. Branke, J., Deb, K., Miettinen, K., R. (Eds.), 2008. Slowiński. Multiobjective Optimization: Interactive and Evolutionary Approaches., vol. 5252 Springer Science & Business Media. Calabrò, P.S., 2009. Greenhouse gases emission from municipal waste management: the role of separate collection. Waste Manag. 29, 2178–2187. Dai, C., Li, Y.P., Huang, G.H., 2011. A two-stage support-vector-regression optimization model for municipal solid waste management – a case study of Beijing, China. J. Environ. Manag. 92 (12), 3023–3037. Demir, G., Ozcan, H.K., Karakus, P.K., Bakis, Y., 2017. Solid waste collection route optimisation by geographical information system (GIS) in fatih-istanbul, Turkey. Int. J. Glob. Warming 11 (3), 263–272. Eschenroeder, A., 2001. Greenhouse gas dynamics of municipal solid waste alternatives. J. Air Waste Manag. Assoc. (51), 1423–1427. Gass, S., Saaty, T., 1955. The computational algorithm for the parametric objective function. Nav. Res. Logist. Q. 2, 39–45. Hochman, G., Wang, S., Li, Q., Gottlieb, P.D., Xu, F., Y, Li, 2015. Cost of organic waste technologies: a case study for New Jersey. AIMS Energy 3 (3), 450–462. Hoornweg, D., Bhada-Tata, P., 2012. WHAT A WASTE A Global Review of Solid. Urban Development & Local Government Unit World Bank, Washington, DC 20433 USA. Jing, S., Huang, G.H., Xi, B.D., Li, Y.P., Qin, X.S., Huo, S.L., Jiang, Y., 2009. A hybrid inexact optimization approach for solid waste management in the city of Foshan, China. J. Environ. Manag. 91, 389–402. Kanat, G., 2010. Municipal solid-waste management in Istanbul. Waste Manag. 30, 1737–1745. Kanat, G., Demir, A., Ozkaya, B., Bilgili, M.S., 2006. Addressing the operational problems in a composting and recycling plant. Waste Manag. 26, 1384–1391. Karadag, D., Sakar, S., 2003. Cost and financing of municipal solid waste collection services in İstanbul. Waste Manag. Res. 21, 480–485. Kirkeby, J.T., Birgisdottir, H., Hansen, T.L., Christensen, T.H., Bhander, G.S., Hauschild, M., 2006. Evaluation of environmental impacts from municipal solid waste management in the municipality of Aarhus, Denmark (EASEWASTE). Waste Manag. Res. 24, 16–26. Korkut, N.E., Yaman, C., Küçükağa, Y., Jaunich, M.K., 2018. Greenhouse gas contribution of municipal solid waste collection: a case study in the city of Istanbul, Turkey. Waste
369