With the economic development and the population increase, the urbanization process accelerates rapidly, but it also leads to a much larger amount of waste (Chen et al., 2018). The cities all over the world produce 1.47 billion tons of domestic waste every year, of which only 15% is recycled, unfortunately. The remaining 85% is disposed of in landfills, which causes “waste siege” and pollution (Pietzsch et al., 2017). The rapid growth of waste has presented huge challenges to waste management (Hua et al., 2017; Pérez-López et al., 2016; Zhang and Huang, 2014). Waste management has already become a global concern because of its direct impacts on the environment in the 21st century (Wilson and Velis, 2015). Based on the premise of the increasing amount and various kinds of waste, the waste classified collection system is an important link connecting the front and the rear ends of waste management, which can optimize the process. Waste classified recycling can accelerate the flow of waste and reduce the secondary pollution in the process, so as to reduce the adverse impact of waste on our life and the harm to the environment. The pressure and cost of works at the back end can be reduced, and the efficiency of the whole waste management process can be improved. The process of collection and transportation mostly adopts mixed loading and transportation, which brings pressure to the terminal classification and screening treatment. It increases the operation cost and waste disposal cost. At the same time, a large number of recyclable resources are polluted and wasted due to mixed loading and transportation. The proposal of vehicle with multi-compartment for co-collection can effectively solve the problems caused by mixed transport. A vehicle with multi-compartment allow it to transport multiple types of waste at the same time while ensuring isolation (Muyldermans and Pang, 2010). The waste classified collection system can improve the operational efficiency of environmental protection departments and save the waste disposal cost.
Waste classified collection can alleviate the problems of “waste siege” and pollution, which has been accounted for worldwide. Many countries and regions have set ambitious goals for waste classification and recycling. The European Union has set a common goal of recycling 65% of municipal waste and 75% of packaging waste by 2030, while limiting the use of landfills (Geissdoerfer et al., 2017). Some developed countries have gradually implemented waste classification collection since the 1970s, and their waste collection systems have reached a very high level (Tyson et al., 1996). Sweden has a high-level urban waste classification system, where the waste is divided into newspapers and recyclables, metal, glass, food waste, nonburnable garbage, and so on. There are waste classification and recycling rooms in residential areas, and the waste collection service providers are only responsible for food waste, while collection centers dispose of other types of waste by category (Dahlén and Lagerkvist, 2010). The United States has established five systems, including paid recycling, roadside recycling, and so on, to realize the classified collection and transportation of waste (Eriksson and Bisaillon, 2011). As early as 2000, China began to carry out waste classification and recycling pilots in eight cities, including Beijing and Shanghai (Meng et al., 2018). As of 2019, 237 cities in China have started waste classification, which has risen to an unprecedented level. Other developing countries are also implementing waste classification and recycling strategies (Troschinetz and Mihelcic, 2009). It is feasible and necessary to conduct waste classified collection.
Among the existing studies about waste collection, the waste collection vehicle routing problem that is designed on the basis of the vehicle routing problem (VRP) (Dantzig and Ramser, 1959) is an important issue, which seriously affects the economic, social, and environmental benefits of waste management (Kanchanabhan et al., 2011). In large cities of developed countries, more than 50% of total waste management expenditures is related to transportation, and the combined cost of collection and transportation accounts for approximately 85% of the expenditures (Sanjeevi and Shahabudeen, 2016). The collection route optimization can save the driven distance, which in turn can reduce the total logistics cost, operation time, and carbon emission. Babaee Tirkolaee et al. (2019) studied a multi-trip VRP with time windows related to urban waste collection, with the aim of minimizing the total cost, and designed an efficient simulated annealing algorithm to solve the problem. Hannan et al. (2018) designed a modified particle swarm optimization algorithm to solve the capacitated VRP model for scheduled solid waste collection and route optimization. With the purpose of the shortest travel distance, Akhtar et al. (2017) proposed a modified backtracking search algorithm in capacitated VRP models with the smart bin concept to find the best optimized waste collection route solutions. De Bruecker et al. (2018) studied an integrated shift scheduling and waste collection VRP. They presented a model enhancement approach, where they developed mixed-integer linear programming for the multi-trip capacitated arc routing problem to minimize the total cost of urban waste collection and disposal activities. Tirkolaee et al. (2019) developed a hybrid algorithm using the Taguchi parameter design method to solve the benchmark test instances. Raucq, Sörensen, and Cattrysse (2019) studied a complex, real-life VRP in the waste management sector, developed a novel column generation scheme and tested it on four real-life instances. Molina, Eguia, and Racero (2019) focused on environmental issues in waste collection, proposed a mathematical model with an eco-efficient objective, and then developed an improved variable neighborhood tabu search algorithm to solve the problem on a real instance in Seville, Spain.
By analyzing the existing research about waste collection, we find they have the following features: 1) Most studies examine the waste collection vehicle routing between the waste generation and the disposal center; however, to simplify, they seldom consider the transfer station. The waste transfer station is an important hub for urban waste collection and treatment and an indispensable link in the urban waste classification and disposal system. According to the best practices in the field of waste management, as shown in Fig. 1, waste is first collected from bins by small vehicles and brought to waste transfer stations for pretreatment and then carried to the different kinds of disposal centers. 2) The existing studies seldom consider the waste classified collection because it may greatly increase the difficulty of establishing the waste collection model and solving the problem. Actually, through compression treatment in the waste transfer station, the waste is classified and compressed into boxes to avoid the secondary pollution. 3) Most research aims to minimize the operation cost or the travel distance, while few studies pay attention to environmental impacts occurred by a great deal of carbon emission. During the waste collection activity, the increased carbon emission accelerates climate deterioration, which should also be considered.
In the past years, meta-heuristic algorithms have been successfully used to solve various operation optimization problems, and thus they are regarded as effectiveness. Habibi et al. (2019) proposed an adaptive two-phase iterative heuristic combined with the algorithmic framework of sample average approximation to deal with a collection-disassembly problem. To optimize collaborative transportation service, Zhang et al. (2017) designed a stochastic plant-pollinator algorithm and proved its effectiveness. Bożek and Werner (2018) designed a hybrid heuristic algorithm combined by tabu search and greedy search and applied it in flexible job shop scheduling problem. Inspired by combined scheduling and picking strategies, (Kong et al., 2016) proposed a heuristic solution for scheduling. In recent years, a novel meta-heuristic method was introduced by Shi (2011), namely brain storm optimization (BSO). BSO is based on human collective behavior of brainstorming process, its effectiveness and usefulness in optimization problems have been proven by Shi (2015a). Because BSO is easy to implement and has strong search performance, it is widely used to solve kinds of operation optimization problems. Yu et al. (2018) proposed an adaptive step-length structure, together with a success memory selection strategy, to improve the original BSO algorithm (ASBSO) and tested it on 57 benchmark functions and four actual problems. A multi-objective BSO was applied in a distributed manufacturing system by Fu, Wang and Huang (2019), and compared with two existing multi-objective algorithms it was proven as a good solver to achieve satisfactory solution. Aldhafeeri and Rahmat-Samii (2019) validated the effectiveness of BSO in electromagnetic applications, proposed a new binary version of BSO to deal with discrete problems, and applied it to an array thinning example and a pixelated patch antenna design. Fu et al. (2019) designed a BSO incorporating a stochastic simulation to copy with dual-objective energy-conscious flow shop scheduling with uncertainty, and they validated its effectiveness by performing experiments on benchmark instances. Cheng et al. (2017) summarized the development, implementation, variant, and future directions of BSO. From the existing literature, we can see BSO’s effectiveness in solving various kinds of optimization problems. However, it is seldom applied to copy with VRP. In this paper, BSO is further developed to solve the low-carbon waste classified collection problem that is abstracted to a bi-objective multi-depot two-echelon green VRP with various pickups (MD-2EGVRP-VP) model, which is combined with the rank method clustering strategy and differential mutation.
According to our analysis of these prior works, the actual waste collection is divided into two stages, where waste is collected from waste bins to transfer stations, and then conveyed to disposal centers. We study waste classification collection and its environmental impacts. Therefore, we deform the two-echelon VRP (2EVRP) (Crainic et al., 2009) to realize the waste classification collection. For the sake of simplicity, we divide the waste into recyclable and unrecyclable types. Therefore, we come up with the MD-2EGVRP-VP model applied in the waste classified recycling, whose objectives are to minimize the total driven distance during two echelons and minimize the carbon emission. To copy with the problem effectively, we design a multi-objective brain storm optimization (MOBSO) algorithm that uses the rank method clustering strategy and differential mutation. Compared with the existing research, our work makes the following contributions:
-
We abstract the problem of waste classified collection into an optimization MD-2EGVRP-VP model. Unlike the previous literatures without consideration of waste transfer station, our work takes it into consideration, making our model more realistic. And two objectives (i.e., to minimize driven distance and carbon emission) are considered, which fills the gap in terms of few multi-objective optimization researches on waste collection. Actually, the firstly proposed model can solve the VRP with multiple depots, multiple intermediate facilities, and various pickups.
-
To apply the proposed model, a novel MOBSO algorithm combining the rank method clustering strategy and differential mutation is developed. We conduct experiments on generated test instances and a real-world case. And via comparisons with classical multi-objective evolutionary optimization algorithms, that is, nondominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) (Deb et al., 2002) and multi-objective evolutionary algorithm based on decomposition (MOEA/D) (Zhang and Li, 2007), we validate the proposed MOBSO algorithm’s effectiveness and feasibility in solving the investigated problem.
-
Our work on models and methods in this paper can help the sanitation departments better fulfill its corporate social responsibility to protect the environment, while taking operational costs into account. Considering the benefits of classified collection to terminal waste treatment, the sanitation departments should replace mixed loading and transportation with multi-compartment co-collection in the process of waste collection, which makes the advantages of waste classified collection in the economic benefit of resource utilization more prominent.
This paper is organized as follows: The problem statement and model formulation are presented in Section 2. Section 3 introduces the MOBSO algorithm proposed for the low-carbon waste classified problem. The computational results of the numerical experiments are discussed in Section 4. Finally, the conclusion and our further work are provided in Section 5.