This section firsts introduce the problem description of CEVRPTW and CVRPTW, then analyzes the corresponding sub-costs of these two models. Finally, the specific formulations of CEVRPTW and CVRPTW are presented.
3.1. Problem Description
For CEVRPTW, there is a distribution center with a certain number of electric vehicles, and a set of customers to be served. The locations of the center, charging stations, and customers are known, and the demands of each customer are also known. When the transportation tasks are completed, all the vehicles must return to the distribution center. In traditional VRP network, the node set is a collection of customer nodes and distribution center, and all customer nodes can only be visited once. However, in EVRP network, the node set also includes charging station nodes which could be visited twice or more. The schematic diagram is shown in
Figure 1. The main purpose is to find an optimal solution considering factors of cost and carbon emission and make a comparison with fuel vehicles scheduling scheme under carbon trading market. The detailed assumptions are given as follows:
- (1)
The electric vehicles are homogeneous and the battery is fully charged when departing from the distribution center;
- (2)
Each electric vehicle has a limited load capacity and the total demand of the customers cannot exceed the total load of the fleet;
- (3)
The electric vehicles can only charge when arriving at the distribution center or charging stations;
- (4)
The electric vehicles will go to the charging stations only when they do not have enough energy to reach next customer;
- (5)
When recharging is undertaken, the batteries are filled to capacity and the charging time of electric vehicles is a fixed value.
For CVRPTW, the main differences about problem description compared with CEVRPTW is that the refueling process of fuel vehicles is not considered in the distribution process, hence the charging station nodes are not included in the note set.
The assumptions of CVRPTW are the same with CEVRPTW after relaxing the battery capacity constraint, such as, the vehicles need to be homogeneous and go back to the distribution center after the task is completed. In short, CEVRPTW is the same with CVRPTW if the battery capacity is a great positive value.
3.3. CEVRPTW Model
The CEVRPTW model in this paper is proposed to realize the optimal path selection from the perspective of both economic and social benefit. Carbon emissions generate from the thermal power plant to produce electricity, in other words, the lowest carbon emissions scheduling scheme is the least energy consumption. However, if only the minimum carbon emissions or energy consumption is set as the optimization goal, it is pointless for logistics enterprises. Moreover, with the rapid improvement of national carbon trading market, the land transportation companies are concerned about whether they need to use new energy vehicles under carbon trading market. Therefore, the objective function in this paper is not to minimize carbon emissions or electricity consumption, but to minimize total costs.
The total costs include fixed costs of electric vehicles, depreciation costs, electricity consumption costs, and penalty costs for violating customer time window and carbon emission trading costs.
(1) The fixed costs of electric vehicles
When dispatching the vehicles to carry out the distribution task, some fixed costs must be paid, including the drivers’ salary, the vehicle wear and tear and road maintenance fees. Thus, the fixed cost
can be expressed as:
(2) The depreciation costs of electric vehicles
Because the direct government subsidies for purchasing electric vehicles are decreasing, the cost of purchasing electric vehicles is relatively high. Hence, the depreciation costs of electric vehicles in distribution process cannot be neglected. By using the mileage depreciation method, the depreciation costs can be expressed as:
(3) The electricity consumption costs
As recent researches paid little attention to the impact of vehicle load and vehicle speed on energy consumption, the energy consumption of electric vehicles is a fundamental issue in how to dispatch electric vehicles for distribution. Therefore, after reading the relevant literatures about electric vehicle design [
27,
41], a calculation method of energy consumption considering travel speed and cargo load is presented. In addition, the energy recovery in regenerative braking is also described in the books. Hence, the energy consumption for propulsion is taken into consideration. The power consumption formula is given as:
where
is the rolling resistance of tires on the ground,
is the grading resistance,
is the aerodynamic drag and
is the acceleration force.
The relevant parameters are shown in
Table 2.
Because this paper assumes the distance between each node is Euclidean distance, the road gradient is not taken into consideration and the vehicle speed in the distribution process is constant. That means
and
are equal to 0. Thus, the power consumption can be simplified as:
The energy consumption of electric vehicle
k travelling from note
i to note
j can be expressed as:
After multiplying the electricity price per kWh, the total electricity consumption costs
in the distribution can be expressed as follows:
(4) The penalty costs for violating customer time window
In the distribution process, the customers generally have a requirement for the expected delivery time. If the cargoes are not delivered to the customer within the time window required. The customer will be unsatisfied. Thus, some penalty costs should be paid. The time when vehicle
k reaches the customer
i can be expressed as:
Hence, the penalty costs
can be calculated as:
(5) The carbon emission trading costs
Although electric vehicles do not emit any carbon emissions during their journey, the power generation in China is mainly based on thermal power plant, accounting for nearly 70 percentages. In addition, according to China’s land transport enterprises greenhouse gas emissions calculation methods released by the government, the electricity consumption needs to be considered when calculating enterprises’ total carbon emissions. Thus, this paper uses Equation (9) to calculate the amount of carbon dioxide emissions from electricity consumption. The indirect carbon emissions
generated when the vehicle
k travels between note
i and note
j can be expressed as:
Based on the research about the impact of carbon trading mechanism on VRP [
34,
37,
39], when the actual carbon emissions are lower the carbon quota allocated, the enterprises can sell the rest quota to those whose actual carbon emissions are higher than the quota to gain profit. In the similar way, if the enterprises emit the carbon emissions exceeding the upper limit, they must purchase additional carbon quota to make up for the excess. Therefore, the carbon trading costs
can be expressed as:
On the basis of the above analysis, the MTCCEVRPTW model is expressed as follows:
The objective function of the model is to minimize the total costs shown in Equation (11). Equation (12) indicates that each customer must be served once by a vehicle. Equation (13) ensures that the number of entering the charging station equals the number of leaving the charging station. These two equations also indicates that the electric vehicles must return to the distribution center when the distribution tasks are completed. Equation (14) represents that the total load on each path do not pass the maximum load of vehicle. Equation (15) indicates that the load of electric vehicles will be reduced correspondingly after leaving the customer note. Equation (16) imposes the minimum number of electric vehicles for completing the distribution tasks. Equation (17) ensures the battery is full when leaving from the distribution center or charging stations. Equations (18) and (19) indicate the electric vehicles’ energy consumption and it would not break down in the distribution process. Equations (20) and (21) ensure the continuity of the travel time of electric vehicles. Equation (22) ensures the length of each route do not exceed the mileage limit of electric vehicle per day.
3.4. CVRPTW Model
Based on the recent researches about traditional fuel vehicle routing problem combining carbon trading [
7,
35,
36,
37,
38,
39,
40], this paper defines the sub-costs and constructs the CVRPTW model as follows.
(1) The fixed costs of fuel vehicles
(2) The depreciation costs of fuel vehicles
(3) The fuel consumption costs
There are some scholars have come up with a linear function for fuel consumption [
37,
39]. The linear function is presented as follows:
Hence, the fuel consumption of vehicle
k travelling form note
i to note
j can be expressed as:
The fuel consumption costs
are expressed as:
(4) The penalty costs
The time when vehicle reaches the customer
i is
, thus the penalty costs can be expressed as:
(5) The carbon trading costs
Different with carbon emissions from using electricity, the fuel vehicles will emit carbon dioxide in the distribution process directly. This paper gives the expression as: . There is fuel emission factor, such as gasoline and diesel, which is different with the grid emission factor mentioned above.
The fuel consumption
of vehicle
k travelling from note
i to note
j can be expressed as:
Hence, the carbon trading costs
is expressed as:
Based on the analysis above, the MTCCVRPTW model is constructed as follows:
Equation (32) presents that all the vehicles must return to the distribution center when the task is completed. Equation (33) indicates that each customer is only be visited once. Equation (34) shows that the vehicle cannot be overload. Equation (35) indicates that the load of electric vehicles will be reduced correspondingly after leaving the customer note. Equation (36) shows the minimum number of vehicles for completing the distribution tasks. Equation (37) indicates the service continuity for two customer nodes. Equation (38) ensures the length of each route do not exceed the mileage limit of vehicles per day.