Multi-Objective Approach for Optimization of City Logistics Considering Energy E ﬃ ciency

: Urban population increase results in more supply chain operations in these areas, which leads to increased energy consumption and environmental pollution. City logistics represents a strategy of e ﬃ cient freight transportation and material handling to fulﬁll customer and business demands. Within the frame of this paper, the authors describe an optimization model of a multi-echelon collection and distribution system, focusing on downtown areas and energy e ﬃ ciency, sustainability, and emission reduction. After a systematic literature review, this paper introduces a mathematical model of collection and distribution problems, including package delivery, municipal waste collection, home delivery services, and supply of supermarkets and o ﬃ ces. The object of the optimization model is twofold: ﬁrstly, to design the optimal structure of the multi-echelon collection and distribution system, including layout planning and the determination of required transportation resources, like e-cars, e-bikes, and the use of public transportation; and secondly, to optimize the operation strategy of the multi-echelon supply chain, including resource allocation and scheduling problems. Next, a heuristic approach is described, whose performance is validated with common benchmark functions, such as metaheuristic evaluation. The scenario analysis demonstrates the application of the described model and shows the optimal layout, resource allocation, and operation strategy focusing on energy e ﬃ ciency.


Introduction
City logistics field has become more forked with numerous available solutions within the last few years due to recent successive innovations in transportation and Industry 4.0. The pace of renewable energy developments in transportation such as e-cars and e-bikes are increasing and it has opened wide scope to the possibility of using them, next to the Industry 4.0 technologies that rely on the Internet of Things and artificial intelligence, which help to innovate smart solutions support the aim of shortening the needed time and route distance with collecting and saving information, giving the ability to analyze them. Furthermore, sustainability is an important topic that is taking priority within current world development goals. Working on reducing spent energy, emissions, and pollution is always highly recommended for its positive impact on the environment and climate, converting the existing world into a more sustainable one. Researching those modern solutions has dramatically increased in different aspects, releasing the importance of applying them to reality in a positive way. However, city logistics include different sections in urban areas, such as goods storage, waste collection, and home delivery service. It has been worked on optimizing the inventory and distribution logistics to give the maximum advantage. For instance, a multi-echelon inventory that raised the efficiency of collecting and distribution logistics by considering the entire supply network and managing the inventory in that network.

•
The articles that addressed the city logistics from a sustainable point of view are focusing on conventional supply chain solutions.

•
Few of the articles have aimed to provide an approach or to optimize the design of logistics networks within the urban areas while considering energy efficiency.

•
The number of articles regarding city logistics has dramatically increased in the last few years. • Energy efficiency becomes more and more important in the field of city logistics, while sustainability aspects are also taken into consideration.

•
Multi-echelon solutions can improve energy efficiency and the sustainability of supply chains, and especially city logistics.
These findings are linked with previous review articles [61,62]; therefore, the optimization of city logistics considering energy efficiency still needs more attention and research.
After this introduction, the remaining parts of the paper are divided into five sections. Section 2 presents the model of a conventional city logistics solution and an evaluation methodology to create reference parameters for further comparison. Section 3 presents a methodology for the modeling and optimization of multi-echelon supply in downtown areas based on e-vehicles. Conclusions and future research directions are discussed in the last section.
The main contribution of this article includes: (1) a methodology to evaluate conventional city logistics solutions from time, distance, energy consumption and emission point of view; (2) a methodology to design and optimize multi-echelon city logistics solution, where capacity, timeliness, suitability, availability, and energy-related constraints are taken into consideration; (3) comparative analysis of conventional and multi-echelon e-vehicle-based city logistics solutions focusing on energy efficiency and emission reduction; and (4) computational results of different scenarios to validate the developed methodology.

Problem Description
In the case of conventional city logistics solutions, the supply of pick-up and delivery points (households, supermarkets, shops, etc.) is processed directly (Figure 1). However, more and more Sustainability 2020, 12, 7366 4 of 23 e-vehicles are adopted in supply chain solutions, but most of the cargo trucks are conventional diesel trucks. Their processes are optimized by the agents of each service provider, but the separated optimization without any cooperation leads to increased fuel consumption and emission. Within the frame of this section, an evaluation methodology is shown, which makes it possible to evaluate existing conventional city logistics solutions to define reference parameters for further comparison with the optimized system. Sustainability 2020, 12, x FOR PEER REVIEW 4 of 25 frame of this section, an evaluation methodology is shown, which makes it possible to evaluate existing conventional city logistics solutions to define reference parameters for further comparison with the optimized system. Without any cooperation of large service providers and self-employed truck drivers, it is not possible to optimize this conventional solution. The optimization of each service provider is great from their point of view, but it has no significant impact on the emission reduction target. Meeting the targets of zero-emission in urban centers by 2030 [63], the below-described methodology makes it possible to find the bottlenecks of the system, which can have a great impact on the emission of the urban area.
In the second scenario, an optimization methodology for a multi-echelon city logistics solution is described. The external logistics service providers are transporting goods to/from logistics centers located outside of the urban area (city border). The collection and distribution of goods to/from pickup and delivery points are processed from this intermediate storage directly by e-trucks and micromobility e-vehicles ( Figure 2). The optimization of the whole process is centralized. It means that in this case, there is strong cooperation among transportation resources and not only the fuel consumption but also the emission of various greenhouse gases can be reduced. Without any cooperation of large service providers and self-employed truck drivers, it is not possible to optimize this conventional solution. The optimization of each service provider is great from their point of view, but it has no significant impact on the emission reduction target. Meeting the targets of zero-emission in urban centers by 2030 [63], the below-described methodology makes it possible to find the bottlenecks of the system, which can have a great impact on the emission of the urban area.
In the second scenario, an optimization methodology for a multi-echelon city logistics solution is described. The external logistics service providers are transporting goods to/from logistics centers located outside of the urban area (city border). The collection and distribution of goods to/from pick-up and delivery points are processed from this intermediate storage directly by e-trucks and micro-mobility e-vehicles ( Figure 2). The optimization of the whole process is centralized. It means that in this case, there is strong cooperation among transportation resources and not only the fuel consumption but also the emission of various greenhouse gases can be reduced.
The intelligent agent optimizes scheduling, assignment, routing layout design, and controlling tasks, that focus on time, distance, energy consumption, and emission-related objective functions, while capacity, availability, suitability, time-window, energy, and service level related constraints can limit the optimal solution.
This scenario focuses on an e-vehicle-based solution, where the efficiency of the whole system can be increased by using existing Industry 4.0 technologies, like smart devices, radiofrequency identification, digital twin solutions, and cloud and fog computing to solve big data problems of a large scale system, including a wide range of users, transportation resources and goods. The intelligent agent optimizes scheduling, assignment, routing layout design, and controlling tasks, that focus on time, distance, energy consumption, and emission-related objective functions, while capacity, availability, suitability, time-window, energy, and service level related constraints can limit the optimal solution.
This scenario focuses on an e-vehicle-based solution, where the efficiency of the whole system can be increased by using existing Industry 4.0 technologies, like smart devices, radiofrequency identification, digital twin solutions, and cloud and fog computing to solve big data problems of a large scale system, including a wide range of users, transportation resources and goods.

Evaluation of a Conventional City Logistics Solution
The evaluation methodology focuses on time-, fuel-and emission-related objective functions, while no capacity, energy, availability, and time-related constraints are taken into consideration, because the system is in this case only evaluated and not optimized. Table A1 in Appendix A shows the nomenclature used in the mathematical model of the conventional city logistics solution.
The first parameter of the evaluation is the total length of transportation routes within the time span of analysis. The transportation is performed with conventional trucks, and no logistics center is taken into consideration for pick-up and delivery operations; all pick-up and delivery are performed by the trucks as direct supply. The parameter of the evaluation, in this case, can be written as follows: where is the total length of the transportation routes within the time span of optimization, is the number of delivery trucks, is the number of pick-up and delivery points assigned to collection route , , * is the ID number of pick-up and delivery task assigned to route as pickup or delivery task ,

Evaluation of a Conventional City Logistics Solution
The evaluation methodology focuses on time-, fuel-and emission-related objective functions, while no capacity, energy, availability, and time-related constraints are taken into consideration, because the system is in this case only evaluated and not optimized. Table A1 in Appendix A shows the nomenclature used in the mathematical model of the conventional city logistics solution.
The first parameter of the evaluation is the total length of transportation routes within the time span of analysis. The transportation is performed with conventional trucks, and no logistics center is taken into consideration for pick-up and delivery operations; all pick-up and delivery are performed by the trucks as direct supply. The parameter of the evaluation, in this case, can be written as follows: where L is the total length of the transportation routes within the time span of optimization, α is the number of delivery trucks, β α max is the number of pick-up and delivery points assigned to collection route α, x * α,β is the ID number of pick-up and delivery task assigned to route α as pick-up or delivery task β, y x * α,β defines the ID of pick-up or delivery point, p y x * α,β is the position of pick-up or delivery point assigned to route α as pick-up or delivery task β and l is the length of transportation route as a function of positions of pick-up and delivery points.
The second parameter of the evaluation is the fuel consumption, which can be calculated depending on the length of transportation routes, required material handling operations (loading and unloading), and the specific fuel consumption rate: where C FUEL T is the fuel consumption of the whole transportation process without material handling (loading and unloading), c FT α,β is the specific fuel consumption of transportation, C FUEL MH is the fuel consumption of material handling operations at the pick-up and delivery points, c FMH α,β is the specific fuel consumption regarding material handling operations and v is the average speed of the truck.
The fuel consumption of the transportation process can be expressed as where q x α,β is the pick-up or delivery volume assigned to route α as pick-up or delivery task β.
The specific fuel consumption of the transportation process can be calculated as follows: where c FT α,min and c FT α,max are the lower and upper limits of fuel consumption of transportation depending on the weight of loading, and q TRANS αmax is the upper limit of the loading weight. The fuel consumption of the loading and unloading operations performed by the truck mounted crane can be given by The specific fuel consumption of material handling processes can be calculated as follows: where c FMH α,min and c FMH α,max are the lower and upper limit of fuel consumption of material handling depending on the weight of loading and q MH αmax is the upper limit of the material handling weight. The third parameter of the evaluation is the emission, which can be calculated depending on the fuel consumption: where E r is the total emission in the time span of the optimization for emission type r (CO 2 , NO x , CO, HC, PM, SO 2 ). The emission of the transportation and material handling process can be described by Equations (8) and (9): Within the frame of this scenario, the following conventional city logistics problem is analyzed and evaluated. There are 25 pick-up and delivery points in the downtown area, where five delivery trucks collect and distribute various types of goods (e.g., package delivery, waste collection). The positions of the delivery points, and weight and loading/unloading time of goods at all pick-up and delivery points are known (see Tables 1 and 2).  There are five delivery routes within the time span of analysis; the capacity of each delivery truck is 400 LU (loading unit). Each delivery route includes six pick-up or delivery points excluding the reference point. The fuel consumption of the trucks is between 41 and 52 L/km depending on the weight of the load, while we are calculating with an average speed of 25 km/h in the downtown area. The loading and unloading operations are processed by truck-mounted cranes, which have an energy consumption between 25 and 37 L/loading per hour, depending on the weight of loading.
The pick-up and delivery routes are optimized by each service provider without any cooperation. It means that, within the frame of this scenario, there is no further optimization performed; the results of the analysis of this scenario are used as reference parameters for the later optimization.
As an example, the calculated parameters regarding transportation time, fuel consumption, and emission of route 1 are shown in Figures 3 and 4. The first service provider is a municipal waste collection provider using a garbage collection truck. It means that its route is a simple collection route with pick-up points. Its collection route is 19.04 km, and the total collection time is 0.97 h, while the energy consumption is 12.48 L fuel (see Figure 2). The emission of diesel consumption can be calculated by [64], as shown in Figure 5. In the case of the first collection route, the CO2 emission is 33,624 g, the NOx emission is 148 g, the CO emission is 37.5 g, the HC emission is 14.9 g, the PM emission is 1.25 g, and the SO2 emission is 0.99 g.   The emission of diesel consumption can be calculated by [64], as shown in Figure 5. In the case of the first collection route, the CO2 emission is 33,624 g, the NOx emission is 148 g, the CO emission is 37.5 g, the HC emission is 14.9 g, the PM emission is 1.25 g, and the SO2 emission is 0.99 g. The values of the parameters calculated for the other four routes (route 2-5) and the summarized values for Scenario 1 are shown in Table 3. Figures A1 and A2 show the detailed reference parameters. Within the frame of the next chapter, the improved city logistics system regarding the collection and distribution of goods such as communal waste, package delivery, or supply of warehouses and shops is described. The above-mentioned parameters will be used as reference parameters to evaluate the new multi-echelon cyber-physical city logistics system, including energy-efficient and The emission of diesel consumption can be calculated by [64], as shown in Figure 5. In the case of the first collection route, the CO 2 emission is 33,624 g, the NO x emission is 148 g, the CO emission is 37.5 g, the HC emission is 14.9 g, the PM emission is 1.25 g, and the SO 2 emission is 0.99 g. The virtual emission of energy consumption depending on the energy generation source can be calculated by [67]. We have compared two extreme cases. In the first case, the energy generation source is coal, and peak loads are met by natural gas turbines (see Figure 6). In the case of e-vehicle A, the CO2 emission is 3810 g, the NOx emission is 16.99 g, the CO emission is 2.77 g, the HC emission is 1.72 g, the PM emission is 0.13 g, and the SO2 emission is 0.12 g. The emission reduction is significant because more pick-up and delivery points were processed with this e-vehicle than in one route of Scenario 1. It would be expected that, in this case, the emissions would not be better than for diesel trucks, but in this case, two major influencing factors must be taken into consideration. The first one is the energy generation source. It does not have a significant impact on emission reduction. The second one is the structure of the supply chain; in the case of the multi-level solution, the total energy consumption of vehicles can be decreased, which leads to an emission reduction.
In the second case, the energy generation source is hydro. In the hydro-based energy generation, the CO2 emission is 122 g, the NOx emission is 0.56 g, the CO emission is 0.1 g, the HC emission is 0.06 g, the PM emission is 0.0047 g, and the SO2 emission is 0.002 g (see Figure 7). In this case, not The values of the parameters calculated for the other four routes (route 2-5) and the summarized values for Scenario 1 are shown in Table 3. Figures A1 and A2 show the detailed reference parameters. Within the frame of the next chapter, the improved city logistics system regarding the collection and distribution of goods such as communal waste, package delivery, or supply of warehouses and shops is described. The above-mentioned parameters will be used as reference parameters to evaluate the new multi-echelon cyber-physical city logistics system, including energy-efficient and environmentally friendly transportation, and material handling solutions. The results of the optimization will be validated by comparing the parameters of the new system with the reference parameters of the conventional system.

Model of Multi-Echelon Collection and Distribution System in Downtown Areas Based on E-Vehicle Transportation
Within the frame of this scenario, the following parameters are taken into consideration as input parameters of the optimization task regarding the city area, including locations and tasks: location of pick-up and delivery points, the weight of pick-up and delivery tasks, upper and lower time limits for pick-up and delivery tasks. The following input parameters are linked to the logistics center: the capacity of loading devices, warehouse capacity, location of warehouses, available resources for transportation and materials handling, specific emission, and energy consumption of resources. These parameters are extensively discussed after the equations.
Within the frame of Scenario 2, we use the evaluation functions of Scenario 1 as objective functions. The first objective function is the minimization of the total length of transportation routes which can be based on Equation (1): where x α,β is the decision variable of the optimization problem.
The second objective function is the minimization of the fuel consumption, which can be given like Equation (2) by where C eFUEL is the energy consumption of e-trucks and micro-mobility vehicles in kWh.
The specific fuel consumption can be calculated by Equations (2) and (6). The third objective function is the minimization of CO 2 , NO x , CO, HC, PM, and SO 2 emission, which can be written like Equation (7): where c eFT α,β is the specific energy consumption of e-trucks and micro-mobility vehicles in kWh/LUkm (LUkm = loading unit kilometer) and the emissions depend on the e-fuel consumption: e r α,β = e r α,β c eFT α,β and e r α,β = e r α,β c eFMH It is important to mention the similarity between Equations (1) and (10), Equations (2) and (11), and Equations (7) and (12), but Equations (1), (2) and (7) are evaluation functions, where x * α,β is a given parameter, while Equations (10)- (12) are objective functions, where x α,β is the decision variable of the optimization problem.
The solutions of the above-mentioned optimization problem are limited by some constraints. The first constraint is a capacity-related constraint, which defines that it is not allowed to exceed the loading capacity of the available e-trucks and micro-mobility vehicles (e-cargo bikes, e-cargo scooters or cargo drones): where Q Tmax α is the loading capacity of vehicle α. The second constraint defines that all pick-up and delivery operations must be performed within a given time span: ∀k The third constraint defines that it is not allowed to exceed the capacity of the available loading resource (mounted loading crane or human resource): where Q Lmax α is the capacity of the available loading resource of transportation device α. The fourth constraint defines that it is not allowed to exceed the available energy of e-truck and micro-mobility vehicles: ∀α : C eFUEL where C eFUEL α,β α max is the energy consumption of e-truck α passing the last pick-up or delivery point assigned to route α and C eFUELmax α is the available energy of e-truck α. The fifth constraint defines that the utilization of available e-trucks and micro-mobility vehicles must be as equal as possible to increase the flexibility of the system: where η α is the utilization of the e-truck, which can be written as follows: and η is the average utilization of e-vehicles, which can be calculated by The sixth constraint defines that the pick-up and delivery tasks can be processed only with suitable vehicles: ∀k : s k,α = 0 → x α,β = 0 otherwise x α,β ∈ (0, 1) where s k,α is the suitability parameter; if s k,α = 1 then e-vehicle α is suitable to process pick-up or delivery task k, otherwise not. Within the frame of Scenario 2, the following multi-echelon city logistics problem is analyzed and evaluated. There is a logistics center outside the city border and e-vehicles are available to perform pick-up and delivery tasks. The 25 pick-up and delivery points in the downtown area and the 30 pick-up and delivery tasks are the same as in Scenario 1. The positions of the delivery points, the weight and loading/unloading time of goods at each pick-up and delivery points are known (see Tables 1 and 2). Table 4 shows the suitability matrix, which is an assignment matrix among e-vehicles and pick-up or delivery tasks.
Other input parameters of the optimization problem regarding the e-vehicles, like capacity, specific energy consumption, are shown in Table 5.
We have used the Excel evolutive Solver Tool to solve the above described NP-hard optimization problem, which includes the clustering of pick-up and delivery tasks, the assignment of pick-up and delivery tasks to the available and suitable e-vehicles, and the routing of e-vehicles.  PID ID 1  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15   GT 2  1  1  1  1  1  1  0  0  0  0  0  0  0  0  0  e-T A 3  0  0  0  0  0  0  1  0  1  1  1  0  0  1   The evolutive option of the Excel Solver represents a robust algorithm to solve NP-hard optimization problems. The evolutive option of the Solver is based on an evolutionary method, like the genetic algorithm or particle swarm optimization methods. The Solver starts with a random population and uses crossover and mutation operators to avoid local optimum, in order to find the near-global optimum solution of the NP-hard optimization problem [65]. Table 6 shows the results of the numerical experiments for algorithm evaluation with various benchmark functions [66].
Powell Sum function Rastrigin function 6.22 × 10 −5 7.15 × 10 −5 3.28 × 10 −6 As the below described computational results and the comparison with the reference parameters of Scenario 1 show, it was not necessary to develop a special heuristic or metaheuristic algorithm to solve the problem. As an example, the parameters of the optimized system regarding transportation time, fuel consumption, and emission of route 1 are shown in Figures 5 and 6. The route sum of e-vehicle A is 19.28 km, the total time sum is 1.17 h, while the energy consumption is 4.7 kWh (see Figure 5). The virtual emission of energy consumption depending on the energy generation source can be calculated by [67]. We have compared two extreme cases. In the first case, the energy generation source is coal, and peak loads are met by natural gas turbines (see Figure 6). In the case of e-vehicle A, the CO2 emission is 3810 g, the NOx emission is 16.99 g, the CO emission is 2.77 g, the HC emission is 1.72 g, the PM emission is 0.13 g, and the SO2 emission is 0.12 g. The emission reduction is significant because more pick-up and delivery points were processed with this e-vehicle than in one route of Scenario 1. It would be expected that, in this case, the emissions would not be better than for diesel trucks, but in this case, two major influencing factors must be taken into consideration. The first one is the energy generation source. It does not have a significant impact on emission reduction. The second one is the structure of the supply chain; in the case of the multi-level solution, the total energy consumption of vehicles can be decreased, which leads to an emission reduction.
In the second case, the energy generation source is hydro. In the hydro-based energy generation, the CO2 emission is 122 g, the NOx emission is 0.56 g, the CO emission is 0.1 g, the HC emission is 0.06 g, the PM emission is 0.0047 g, and the SO2 emission is 0.002 g (see Figure 7). In this case, not only the multi-level structure of the supply chain, but also the energy generation source, has a great impact on the virtual emission reduction. The virtual emission of energy consumption depending on the energy generation source can be calculated by [67]. We have compared two extreme cases. In the first case, the energy generation source is coal, and peak loads are met by natural gas turbines (see Figure 6).
In the case of e-vehicle A, the CO 2 emission is 3810 g, the NO x emission is 16.99 g, the CO emission is 2.77 g, the HC emission is 1.72 g, the PM emission is 0.13 g, and the SO 2 emission is 0.12 g. The emission reduction is significant because more pick-up and delivery points were processed with this e-vehicle than in one route of Scenario 1. It would be expected that, in this case, the emissions would not be better than for diesel trucks, but in this case, two major influencing factors must be taken into consideration. The first one is the energy generation source. It does not have a significant impact on emission reduction. The second one is the structure of the supply chain; in the case of the multi-level solution, the total energy consumption of vehicles can be decreased, which leads to an emission reduction.
In the second case, the energy generation source is hydro. In the hydro-based energy generation, the CO 2 emission is 122 g, the NO x emission is 0.56 g, the CO emission is 0.1 g, the HC emission is 0.06 g, the PM emission is 0.0047 g, and the SO 2 emission is 0.002 g (see Figure 7). In this case, not only the multi-level structure of the supply chain, but also the energy generation source, has a great impact on the virtual emission reduction.  Figure 8 shows the summarized results of the optimization. Four pick-up and delivery routes were organized, depending on the capacity of e-vehicles, the suitability of e-vehicles and loading equipment, the weight of pick-up and delivery tasks, the position of pick-up, and delivery points.
The waste collection route was not changed, because municipal waste collection needs special garbage collection vehicles. This means that the collection of municipal waste cannot be integrated with other collection or distribution processes. The remaining 24 pick-up and delivery tasks were organized into three additional routes. The comparison of the conventional city-logistics solution and the multi-echelon, e-vehicle based system shows that the transformation of a conventional city   The waste collection route was not changed, because municipal waste collection needs special garbage collection vehicles. This means that the collection of municipal waste cannot be integrated with other collection or distribution processes. The remaining 24 pick-up and delivery tasks were organized into three additional routes. The comparison of the conventional city-logistics solution and the multi-echelon, e-vehicle based system shows that the transformation of a conventional city logistics solution into a multi-echelon supply based on a logistics center outside of the city border (or downtown area) and the application of e-trucks and other e-vehicles can significantly reduce energy consumption and the emission.
The transformation of a conventional city logistics solution into the suggested e-vehicle-based multi-echelon system led to a decrease in the required time, transportation distances, energy consumption, and emission. Depending on the energy generation source, it is possible to reduce the emission with significant values. As Table 7 shows, in the case of coal-based e-energy generation (peaks are based on natural gas), this emission reduction is about 90%, but in the case of photovoltaic, wind, or water source of energy, the reduction can be increased compared with fuel-based emission. Figures A3 and A4 show the detailed parameters of the optimized routes. The next scenario shows the solution of a real-world size problem in the shopping area of the Miskolc City Centre West, where 35 last-mile objects (banks, stores, supermarkets, pharmacies, restaurants) were taken into consideration. We have compared the conventional material supply (supply chain) solution and the e-vehicle based solution in this area. As Table 8 shows, the e-vehicle-based solution resulted in a 92% emission reduction (average). Figures A5-A7 in Appendix E show the optimized routes of the e-vehicle-based solution.
The upgrade of the conventional fleet to e-vehicles has a great impact on energy demand, sustainability, and the environment. It means that the future of city logistics is the use of e-vehicles and the transformation of conventional solutions into cyber-physical systems using Industry 4.0 technologies.

Conclusions
The adoption of e-vehicles in city logistics solutions appears to be progressing faster than expected [68,69]. City logistics processes based on e-vehicles lead to decreased fuel consumption and emission, while the availability and flexibility can be increased. Energy efficiency, sustainability, and emission reduction have been extensively researched in all fields of logistics. Service processes are also discussed, but many of the articles are focusing on manufacturing-related systems. In order to start filling this gap, this work has developed a methodology to analyze existing conventional city logistics solutions and optimize new e-vehicle-based multi-echelon systems. The model includes a wide range of objective functions, like the minimization of transportation distance, energy consumption, and the emission of greenhouse gases, while various constraints like the capacity of resources, service level, availability and suitability of material handling resources, and available energy in high voltage batteries or fuel cells are taken into consideration. The described methodology shows that the transformation of conventional city logistics solutions into an e-vehicle based multi-echelon supply chain significantly decreases energy consumption and emission, while service level and flexibility are likely to be increased. Depending on the source of electric energy generation, different emission reduction can be realized. The article shows that by using oil-based energy generation sources, 88% emission reduction can be reached. Table A2 shows these reduced rates in the case of the same scenario taking other energy generation sources, like coal, photovoltaic, wind, or water into account.
As a managerial impact, we would like to mention that the application of the above-described methodology can support managerial decisions regarding the logistics center, the adoption of various e-vehicles, and micro-mobility vehicles, or the operation strategy of the whole supply chain. We can summarize the conclusions and research implications as follows: • The development of new city logistics solutions must be based on the performance evaluation of available conventional systems. We developed a new methodology for the evaluation of conventional city logistics solutions to calculate time-, distance-, energy consumption-, and emission-related performance parameters.

•
Designing and operating sustainable city logistics systems are great challenges for researchers because of the complexity of city logistics solutions, especially in the case of cyber-physical systems, led to NP-hard optimization problems, where the application of heuristic and metaheuristic solutions is unavoidable. We developed a mathematical model to support the design and optimization of a multi-echelon city logistics solution. The model takes capacity, timeliness, suitability, availability, and energy-related constraints into consideration.

•
The comparison and the computational results of conventional and multi-echelon e-vehicle-based city logistics solutions show that the multi-level supply chain and the application of e-vehicles have a great impact on costs, energy efficiency, emission, and service level.

•
The emission rates are based on well-to-wheel analysis, where the production and transportation of primary fuel, production and transportation, and road fuel are taken into consideration [70].
A further study of the proposed work would be the integration of the above-mentioned methodology with a digital twin solution. Another research direction is to develop a more robust and faster tailored computational method and algorithm to solve a real city logistics network. By using the digital twin process, it would be possible to extend the optimization range of the proposed model to real-time scheduling of new pick-up or delivery tasks into the existing processes. An additional aspect of optimization would be the application of the lean paradigm [71], simulation [72] and heuristic optimization [73], to improve the efficiency of the existing conventional system or the transformed system. The emissions of first mile supply to the hub are not taken into account, because the conventional and e-vehicle based systems are the same from the first mile suppliers to the hub (if the hub is near the check-in point of the downtown area). A potential future research direction is to improve the model with first mile options to create a combined first-mile/last-mile model. The emissions of the hub can also be taken into consideration. The impact of the emission regarding heating, water, air-conditioning, and electricity can be analyzed. Another future research direction is to check our model with the Fiori model [47], which is a complex model for the evaluation of energy consumption of electric freight vehicles in urban pickup/delivery operations.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1 shows the nomenclature used in the mathematical models. Table A1. Nomenclature used in the mathematical model.

Appendix B
Appendix B includes some analysis details regarding Scenario 1. Figure A1 shows the calculated length of transportation routes and the required times to perform the whole collection/distribution process including transportation and materials handling.  Figure A2 demonstrates the emission values for each route, including CO, CO2, NOx, SO2, PM, and HC. Figures A1 and A2 are based on the calculation results summarized in Table 3.

Appendix C
Appendix C includes some analysis details regarding Scenario 2. Figure A3 shows the calculated length of transportation routes and the required times to perform the whole collection/distribution process, including transportation and materials handling. Figure A4 demonstrates the emission values for each route, including CO, CO2, NOx, SO2, PM, and HC. Figures A3 and A4 are based on the calculation results summarized in Table 6.   Figure A2 demonstrates the emission values for each route, including CO, CO 2 , NO x , SO 2 , PM, and HC. Figures A1 and A2 are based on the calculation results summarized in Table 3.  Figure A2 demonstrates the emission values for each route, including CO, CO2, NOx, SO2, PM, and HC. Figures A1 and A2 are based on the calculation results summarized in Table 3.

Appendix C
Appendix C includes some analysis details regarding Scenario 2. Figure A3 shows the calculated length of transportation routes and the required times to perform the whole collection/distribution process, including transportation and materials handling. Figure A4 demonstrates the emission values for each route, including CO, CO2, NOx, SO2, PM, and HC. Figures A3 and A4 are based on the calculation results summarized in Table 6. Figure A3. Length of transportation routes, required times, and fuel consumption of transportation and materials handling in Scenario 2. Energy generation source is coal and peak loads are met by natural gas turbines.

Appendix C
Appendix C includes some analysis details regarding Scenario 2. Figure A3 shows the calculated length of transportation routes and the required times to perform the whole collection/distribution process, including transportation and materials handling. Figure A4 demonstrates the emission values for each route, including CO, CO 2 , NO x , SO 2 , PM, and HC. Figures A3 and A4 are based on the calculation results summarized in Table 6.   Table 3.

Appendix C
Appendix C includes some analysis details regarding Scenario 2. Figure A3 shows the calculated length of transportation routes and the required times to perform the whole collection/distribution process, including transportation and materials handling. Figure A4 demonstrates the emission values for each route, including CO, CO2, NOx, SO2, PM, and HC. Figures A3 and A4 are based on the calculation results summarized in Table 6. Figure A3. Length of transportation routes, required times, and fuel consumption of transportation and materials handling in Scenario 2. Energy generation source is coal and peak loads are met by natural gas turbines. Figure A3. Length of transportation routes, required times, and fuel consumption of transportation and materials handling in Scenario 2. Energy generation source is coal and peak loads are met by natural gas turbines.

Appendix D
Appendix D includes some analysis details regarding emissions in the case of various energy generation sources.

Appendix D
Appendix D includes some analysis details regarding emissions in the case of various energy generation sources.

Appendix E
Appendix E includes the results of a numerical experiment for algorithm performance for a realworld size problem.

Appendix E
Appendix E includes the results of a numerical experiment for algorithm performance for a realworld size problem.