Optimization of municipal solid waste collection system: systematic review with bibliometric literature analysis

For this study, the Web of Science and Scopus databases were used. The definition of scientific publications was performed through the systematization of the methodology known as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), which is divided into four steps: identification, selection, eligibility and inclusion [22].

The research was conducted during March 2023, and included publications between the years of 2006 and 2023. In the first stage, identification, the keywords "Municipal Solid Waste" OR "Urban Solid Waste" AND "Collection" AND "Logistic Operation" were defined for selection of the publications in the databases. Then, in the second stage, selection, an exclusion criterion was applied, excluding the publications that had no DOI and those which had no full PDF version available.

Subsequently, in the third stage, eligibility, the titles and abstracts of the publications were read. In the last stage, inclusion, the publications were read entirely to exclude those that did not fit the thematic axis addressed or which content did not match the abstract and fell outside the scope of the study. Afterwards, the list of bibliographical references within the selected publications was analysed, which led to the inclusion of six more publications that were part of the thematic axis of the research. Finally, the bibliometric data of the research were exported in a format (.bib and.xls) compatible with bibliometric software.

The bibliometric data of the research were subjected to a meta-analysis, considering a set of parameters from the relevant publications, namely: year of publication, authors, journal, citations, keywords, and considering also the study approach: research objective, methodology, and technology applied. The meta-analysis of these data allowed grouping the publications and enabled subsequent analyses.

The bibliometric analysis was carried out through qualitative and quantitative indicators. The Microsoft Office Excel and R Studio (Bibliometrix) software programs were used to produce graphs, tables and figures that synthesize the information necessary for understanding the analysis.

To explore the key problem of identify and understand the methodologies and tools which have been applied for optimization of MSW collection systems, the analysis of publications was carried out, enabling the preparation of summary tables to synthesize the information obtained.

Among the methods used, three publications used software to optimize MSWM. Negreiros et al. [27] used the software SisRot ®LIX to optimize the MSW collection routes in the city of Campo Grande, Brazil. They divided the assessments into two scenarios. In scenario 1 they analysed the sectors and routes with 9-tonne vehicles, and in scenario 2 with 12-tonne vehicles. After optimisation, a reduction in the total distance travelled by the vehicles occurred compared to the original situation. In scenario 1, the current sector geometry was maintained, and SisRot® LIX created only the optimised routes for the vehicles. The total distance saved during the operation was 13.7%. In scenario 2, both sectors and routes were optimised for 12-tonne collection vehicle loads, with an expected total saving in operating distance of 29.2%. Scenario 2 was selected for implementation in the field. After the practical implementation of scenario 2, the actual saving in distance was 28.6% on Monday and 28.5% on Friday, matching the expected simulated saving.

Similarly, Sahib et al. [12] conducted a truck route optimisation for the city of Karbala, Iraq, using a QSB system software, which treats the data mathematically to find three solutions, each solution being a suggested route for collection vehicles according to a certain objective function. The authors conclude that the basic principle adopted in the study was to reduce the length of roads, which means reducing costs. In addition, they reinforce the idea that a more efficient route for solid waste collection vehicles would save money by reducing the time spent on waste collection.

In a related paper, Badran et al. [28] used Mixed Modelling System (linear and non-linear) to simulate the optimization of MSWM in Port Said, Egypt. The system employed was the MPL V4.2 software. To this effect, the authors needed to create some assumptions to be followed, after defining their assumptions and inserting data into the software, 12 scenario models were generated taking into account the combination of data and assumptions employed. Amongst the 12 scenarios, the ideal scenario was selected using as definition criterion the optimal combination, being the minimum value of the function, which corresponds to the minimum cost. Subsequently, the authors carried out a sensitive analysis of the scenarios, where they verified the possibility of reducing landfill capacity by 57.16 ton/day. They conclude that the use of the proposed model and the optimal scenario results in a profit of 49,655.8 LE/day (US$ 8,418.23/day). However, authors report the difficulties in obtaining accurate data as to the distribution of waste sources and their maps, in addition to the fact that the routing of collection and transhipment vehicles within neighbourhoods or the periodicity of collection were not taken into account; only the regional optimization of MSWM was analysed.

Three studies used a GIS tool to optimize MSW collection. Ghose et al. [29] conducted a case study for Asansol, India. They used a proposed GIS model for solid waste disposal, a model that includes planning of dumpsite locations, vehicles, and optimal routing. The collection routes were planned using GIS, with the NETWORK tool of ArcGis software, in this way the shortest or minimum path through an infrastructure network is calculated. The total cost of the newly designed collection systems is estimated at about 80 million rupees for the fixed cost of storage containers, collection vehicles and landfill and about 8.4 million rupees for the annual operational cost of crews, vehicles and landfill maintenance. The current collection system of Asansol is approximately 25 million rupees annually in operating costs for collection only, without any proper storage/collection and landfill system. The authors conclude that the proposed model can be used as a decision support tool for municipal authorities for efficient MSWM, as well as load balancing on vehicles, managing fuel consumption and generating work schedules for workers and vehicles.

In another application of 3D tools in GIS, Tavares et al. [2] performed a case study for the Cape Verde islands, where they determined an optimal routing network that minimizes fuel consumption and associated emissions for the transportation of MSW from a network of collection points to a treatment plant. The authors based their work in 3 steps: 1 definition of the 3D road network using the ArcGIS 3D Analyst extension; 2 calculations of the fuel consumption per segment along the entire road network; and 3 optimizations of the MSW collection for minimum fuel consumption applying the ArcGIS Network Analyst extension. They subsequently compared the results with a 2D simulation. The authors conclude that using the 3D model is more efficient since it takes into account the elevation and gradients of the road network and that optimising the MSW collection vehicle route for minimum fuel consumption rather than the shortest distance or time results in greater benefits and savings.

Similar to previous studies, Kinobe et al. [30] had the objective of optimizing the waste collection and disposal in Kampala city. They used a GPS to collect data on dumpsites, routes, collection and waste storage, which were then inserted into ArcGIS software, using “Network Analyst” tool. Next, an optimization model was developed to calculate the shortest distances, considering the location of dumpsites, truck capacities, road network, and the volumes of waste generation. The “Network Analyst” tool was used in vehicle routing to determine the shortest route from the collection points to the landfill. The recorded distance and travel time of the routes before the optimization were 624.2 km and 1407 min, after the optimization and change of the collection orders, it was 506.9 km and 1161 min, providing a benefit of 19% and 17% of distance and time.

Ramos et al. [31] used IoT with 3 approaches: 1) a limited approach based on a Cluster First-Route Second heuristic in which a minimum fill level rule is imposed to select the bins to be serviced each day; 2) an intelligent collection approach in which a mathematical model is proposed to decide which bins should be visited each day and their order; and 3) a more intelligent collection approach in which a heuristic method is proposed to decide the best days to perform the collection operation according to the service level established in approach 2. They then compared the results with data from a company responsible for waste collection in 14 municipalities in Portugal but chose to use recyclable waste as a comparative parameter, After simulating the scenarios, the authors conclude that the use of sensors in the dumpsites combined with the operational management approach described in scenario 3 (which was the best alternative approach), lead to a 20% improvement in the kg per km ratio, as well as a 7% increase in profit, while the distance travelled decreased by 33%. These improvements are due to the increase in the amount of waste collected (as the bins are collected as late as possible, so waste may accumulate in some bins), and at the same time empty bins not being visited.

In another implementation of optimized MSWM using IoT, Idwan et al. [32] performed a simulation model for a random location in Islamabad, Pakistan. They determined the schedule and routes of the collection trucks. They simulated the waste collection operation in the sectors and compared the results using the smart bins with the conventional ones, creating two scenarios: (a) traditional scenario where IoT is not applied and (b) smart bins scenario where IoT is applied. After the simulations the authors conclude that the results were satisfactory and showed a great improvement in saving the total time to collect the waste and total distance travelled. The total average time for the traditional scenario was 5.6 h and the total average distance travelled is 344.3 km, whereas using the IoT technology the average time was 4.365 h and the total average distance travelled is 234.7 km. The authors also reinforce the idea that the model they created was extended to multiple dimensions and can be applied to large metropolitan areas.

Alternatively, Wan et al. [33] performed a case analysis of a central street (Gongbei Street) in the city of Zhuhai, China, but using storage capacity sensors introduced on the dumpsites, which were connected to the transport vehicles and the waste management central, to achieve online management of the MSW collection and transportation system. First, the collection point managers make the collection request. The waste management central determines the departure time, the driving route and the location of the transfer station for unloading. This information is issued to the transport sector through an application. When the collection vehicle arrives on site, the empty bins are unloaded and forwarded to the site with a portable mobile sensor. The full bin is placed on the vehicle, weighed and the collection point information is recorded on the vehicle. Once the vehicle is full, it drives to the nearest transfer station to unload. After unloading, the waste is weighed and the information from the dump is recorded. Afterwards, the next collection cycle is started. With the new MSW collection and transportation process, the operational efficiency of the system, with the same number of vehicles, has tripled. The annual collection and transportation cost was CNY 6.24 million, and the operation and management cost of the transfer station was CNY 2.7 million, after the implementation, the annual collection and transportation cost was CNY 5.232 million, with a year-on-year decrease of 16.15% and the annual operation and management cost of the transfer station was CNY 2 million, with a year-on-year decrease of 25.93%. The total saving was CNY 1.708 million per year. The authors conclude that the initial investment cost is relatively high, as it is necessary to invest in a large number of high-end computing devices. However, in the future, as the state-of-the-art computing technology develops and matures, the cost will gradually decrease.

In a similar approach, Martikkala et al. [34] developed and used the IoT system, which was based on low-cost tools and open-source code. They developed and improved sensors, as well as tools for interoperability between the different parts of the architecture of the IoT system, with the aim of demonstrating that the use of smart bins for textile waste collection can be linked to a dynamic route optimization system to improve the overall performance of the system. To this end, the authors developed the sensors from a suitable hardware platform that enables wireless data transfers. Consequently, the sensors are based on low-cost Heltec CubeCell development boards with LoRa communication capabilities, the CubeCell boards could be easily programmed with Arduino. Subsequently, they carried out the comparison between two different collection schemes called fixed and dynamic; in the fixed scheme, the textile collectors are emptied with fixed scheduled trips, while in the dynamic scheme, the collection is done based on the filling of the bins. The two schemes were compared using Open Door Logistics (ODL) software. The authors simulated the efficiency of the system with data from the city of Seinäjoki, located in the southern province of Ostrobothnia in Finland. They were able to demonstrate that the conventional scheme resulted in a greater collection distance compared to the dynamic scheme, besides that, since in the dynamic system each trip emptied different dumpsters, the routes became dynamic and optimized at each collection. The authors conclude their work by stating that the dynamic collection had shorter travel distances and fewer stops, which translates into reduced operation time, fuel consumption and cost. Furthermore, the cost of collection per kg of waste for the dynamic scheme was around 7.4% less than for the conventional scheme and CO2 emissions were reduced by 10.2%. They also stated that in a fully operational intelligent waste collection system, the additional costs of the sensors are negligible compared to the other costs related to textile collection.

For authors Rossit et al. [35], one of the biggest problems in relation to MSWM is the location of the dumps. The authors studied an exact multi-objective approach to optimise the problem of dumpsite location, where they used as optimisation criteria the accessibility to the system, the investment cost and the necessary frequency of waste removal. For this, they analysed the real scenarios of Montevideo in Uruguay and Bahía Blanca in Argentina. They used a mathematical approach focused on minimizing the average frequency necessary to collect the dumpsters: minimizing the cost of installing the dumpsters and minimizing the average distance between the generators and the dumpsters. After computational simulations and use of mathematical criterion, the authors compared the multi-objective solutions and simulations of the real distribution of the dumpsites in Montevideo, obtaining improvements up to 51.14% in terms of cost, up to 37.76% in terms of accessibility to the system and up to 12.15% in terms of collection frequency.

Similar to the previous study, Mahdavi et al. [36] applied a sustainable multi-trip periodic redesign-routing problem (SMTP–reLRP) for the district 8, region 2, Haft Howz neighborhood, of the city of Tehran, capital of Iran (divided into 22 districts), which discharges the MSW at the Beyhaghi intermediate transfer station. A collection in this region is done by two vehicles, which repeat their trips at least twice every 24 h. Mathematical models were used to define the key MSW container nodes for the routes of the two vehicles. The capacity of the semi-trailer to transfer MSW from the transfer station to the processing plants is 22 tons with a variable cost of 5,6*10e-7 UC (Unit cost) per kilogram and meter. The cost saving for route redesign was found to be 7474 UC. The new routing improved the number of generation nodes visited in a trip for the vehicles and the number of trips assigned to each vehicle in a day. Regarding the cost, after the new routing the weekly cost decreased by 2,348,916 UC, which is equivalent to 122,143,679 UC in 1 year. A cost saving of 66% was obtained via finding the optimal weekly collection frequency and finding the best redesign.

In summary, results show that all publications analysed in the present work were able to optimize the respective MSWM process. The results were divided and analysed into three categories: waste generation, collection and transportation, and solid waste disposal. Each category considered different parameters, as shown in Table 4. Only Badran et al. [28] managed to optimise at least one item of the three defined categories. Out of the analysed publications, 82% managed to optimize the waste collection and transportation route, being this the main objective of their studies. The optimization of the waste final disposal was the stage that generated less interest in optimization, and only two publications proposed the optimization of either the capacity or the location. In the waste generation stage, the optimisation of the location of bins was the main focus of interest, with 55% of the authors managing to optimise the location of bins to decrease the costs of the process.