Reducing carbon emissions from collaborative distribution: a case study of urban express in China

Abstract

The booming development of e-commerce has brought about rapid growth in the express delivery industry in China. However, urban express distribution is increasingly difficult and seriously affecting the traffic, safety, and environmental conditions of cities due to small, scattered end points, unreasonable allocation of resources, and seriously repeated resource waste. Therefore, there is an urgent need to solve the problems associated with the unreasonable resource allocation of express distribution. In the context of green logistics, a new mode of collaborative distribution based on intelligent end service station (IESS) is proposed. Following the measurement models of carbon emissions, before and after collaborative distributions are provided to prove the environmental benefits of the new mode. The influencing factors considered in the models are the average daily distribution volume, number of distribution sections, vehicle ownership of various types, and their capacity, use, fuel, and power consumption. To verify the models’ validity, we conduct an empirical study of five express enterprises in China and make a comparative analysis on the results, which show that the implementation of collaborative distribution can extremely reduce carbon emissions and improve the overall load rate of vehicles. Specially, the use of new energy vehicles can contribute significantly to energy conservation and emissions reduction.

Introduction

Modern logistics industry belongs to the service sector of the tertiary industry, which makes great contribution to economic growth for the country, and it is positively correlated to regional economy (Jara 2000; Juskowiak and Wharton 2004; Sun 2017a; Khan and Dong 2017). Meanwhile, the development of logistics industry is also an important symbol to measure the optimization of industrial structure (Sun 2017a). Urban logistics distribution sector develops rapidly particularly in large cities in China due to strong supports from all levels of government (Yang et al. 2016). With the rapid development of e-commerce, logistics becomes the support for economic development and well-being promotion (Macharis et al. 2014). The booming development of e-commerce in China has brought about rapid growth in the express delivery industry. In 2019, express distribution volume exceeded 63.5 billion pieces, around 24.3% more than in 2018, the revenue of express business reached 749.8 billion yuan (approximately USD1 = RMB7.0), around 24.2% more than in 2018 (State Post Bureau of China 2020), revealing that China’s express delivery industry had entered a new period of development with an annual growth of more than 10 billion pieces, thereby becoming an important growth point in its national economy.

The modern logistics industry not only promotes economic growth overall but is also causing a series of economic and environmental problems, such as high logistics cost, excessive energy consumption, ecological environment damage, land redundancy and waste, and traffic congestion (Dai and Gao 2016; Xie et al. 2016; Fan et al. 2017). Problems with urban distribution in China are becoming increasingly prominent: there is low efficiency and high costs in the industry and a serious waste of resources (Zhang and Figliozzi 2010); the logistics market is highly fragmented and lack of third-party logistics (Chu et al. 2018; Gong et al. 2019); there are tricycles all over the streets, which seriously affect the traffic, safety, and environmental conditions of cities; the upgrading of logistics industry has led to the increase of carbon emissions and caused the attention of cities (Guo et al. 2016); and it is difficult to cope with the increasing number of express distributors and meet the increasingly high demand for express service quality. To sum up, the express logistics industry is mainly facing two problems to be improved: (1) high carbon emissions of express delivery and (2) low efficiency and high resource waste of disordered operation. Therefore, this study intends to optimize the current distribution mode from the perspective of green logistics and collaborative distribution.

The concept of “green logistics” can be traced back to 1990 (Srivastava 2007). Subsequently, green development in the logistics industry has been reviewed extensively by researchers, enterprises, decision-makers, and other stakeholders in supply chain management (Liu et al. 2018). Today, the development of green logistics has taken hold, involving green logistics packaging, green goods transportation, green logistics storage, and green logistics resource management (Khan and Dong 2017; Wang et al. 2018). Murphy and Poist (2000), starting from the strategic level of green logistics, pointed out that logistics have a significant impact on the environment and environmental problems seriously affect the logistics decisions in the value chain. Sun (2017a) pointed out that the development of green logistics should be supported and strengthened to minimize the energy consumption of the logistics industry by enhancing the technical level of enterprises. As an important part of the green economy, green logistics has become not only the mainstream direction of the modern logistics industry, but also an important means to optimize industrial structure, transform economic development, and promote regional green development.

In the logistics industry, the carbon emissions caused by logistics mainly focus on road transport, i.e., the fuel consumption of gasoline, diesel, and oil of transportation vehicles, and affect the country’s green sustainable development to a large extent (Guo et al. 2016; Talbi 2017; Alshehry and Belloumi 2017; Danish et al. 2018; Khan 2019). Shouket et al. (2019) pointed that China’s road transportation will steadily become the biggest oil consumer in China in the following two decades, and the annual oil demand for road vehicles in China will reach 363 million tons by 2030 if no control measures are executed. Sun (2017b) explained that the adjustment of energy structure, the optimization of logistics industrial structure, and the use of new energy can effectively reduce carbon emissions. Khan (2019) pointed out that green logistics is positively related to renewable energy; however, there is a lack of fuel-efficient/green vehicles use and green practices in logistical operations. Alshehry and Belloumi (2017) concluded that electric vehicles are required to achieve high-quality carbon emissions. To contribute to this field of research, this study intends to discuss carbon emissions of distribution vehicles and contribution of new energy vehicles to energy conservation and emission reduction.

In addition to the impact of carbon emissions caused by energy consumption of distribution vehicles in logistics activities, green logistics activities also include measuring the environmental impact of different distribution strategies (Kucukoglu et al. 2015). Studies have indicated that working alone is not enough to achieve a more environmentally friendly supply chain, and the logistics system can become greener only by improving the level of cooperation (Ramanathan et al. 2014). “Collaborative distribution” originated in Japan. Taniguchi et al. (1999) first proposed the concept of urban logistics and explained the important role of urban public logistics nodes in easing traffic congestion, saving energy, and reducing labor costs. Urban collaborative distribution refers to the intensive integration of resources and the unified arrangement of distribution activities among transportation enterprises within the city. At the macro level, collaborative distribution has produced several environmental benefits and social benefits. Firstly, it has been verified that the intensification of collaborative distribution can reduce the carbon emissions and alleviate the pollution (Kishino and Ikarashi 2010; Zhao et al. 2017). Secondly, through the standardization of vehicles, collaborative distribution has enabled traffic departments to control vehicles and alleviate the pressure of urban traffic (Wang et al. 2019b). At the micro level, collaborative distribution has also produced economic benefits for enterprises. On the one hand, due to the realization of the scale of distribution, enterprises have been able to save distribution costs. The total distribution cost in a logistics alliance is generally 9 to 30% lower than that for each partner separately (Dai et al. 2017). On the other hand, due to the large amount of distribution, an optimal allocation and loading of vehicles can be achieved, cross transport can be eliminated, and the distribution can be efficiently completed (Wang et al. 2017a, b).

Many scholars have measured the emissions of logistics activities from the perspective of the supply chain up till now. Benjaafar et al. (2013) investigated how collaboration among enterprises within the same supply chain affects their costs and carbon emissions. Lee and Cheong (2011) and Lee (2011) identified and measured carbon emissions across the supply chain and integrated the issue of carbon footprint into supply chain management. Shaw et al. (2013) introduced indirect carbon emissions in a network optimization model for sustainable supply chain, which considers dissimilar trucks used for transportation according to their carbon emissions. Sadrnia et al. (2013) proposed a multiobjective optimization model which takes the total logistics cost and carbon emissions into account to design the supply chain network.

Prior studies have primarily targeted at measuring and comparing the carbon emissions of logistics activities from the perspective of supply chain, and some of them considered supply chain collaboration. However, few scholars have paid attention to the impact of operational decision-making on the emissions of logistics activities between different express delivery enterprises. Meanwhile, few scholars have studied the impact of vehicle arrangement and vehicle types on carbon emissions in express delivery activities. This study intends to analyze carbon emissions of distribution activities by comprehensively considering collaborative distribution among express enterprises and vehicle arrangement (including different types, capacity, energy consumption, and use of new energy vehicles).

Faced with the strategic demand of green logistics development, how to realize the coordinated allocation of resources and optimize the distribution network has become a key problem to be solved urgently. In this context, this study proposes a new collaborative distribution mode for urban express delivery and puts forward a new concept of intelligent end service station (IESS), which provides end expresses collaborative delivery service for enterprises. Then, based on the theory of collaborative distribution, we investigate how to control carbon emissions to achieve environmental benefits from the new perspective of vehicle arrangement. We build mathematical models to measure carbon emissions before and after collaborative distribution by optimizing existing vehicle resources, including the selection and quantity of vehicles. The empirical study is carried out to verify the benefits of the new mode. This study would be helpful for governmental officials, express enterprises, and environmentalists to formulate policies and develop green logistics.

The remainder of this paper is organized as follows. The proposal of new distribution mode is presented in “Proposal of new distribution mode.” In “Model development,” the mathematical models of carbon emissions measurement before and after collaborative distribution are described. The empirical study and results are discussed in “Empirical study.” Finally, in “Conclusion,” we conclude with a discussion.

Proposal of new distribution mode

In the traditional mode for urban express delivery, each express enterprise delivered their daily packages from the logistics center to their own distribution outlets in the front-end distribution process, and then the expresses will be delivered to the end points after being sorted in the distribution outlets. There are some disadvantages in the traditional distribution mode due to the enterprises delivering separately for their own business, such as:

• 1. Serious waste of space, vehicles, personnel and other resources

• 2. High operation cost of express delivery enterprises

• 3. Low efficiency of urban overall distribution

• 4. Urban traffic, safety, and environmental conditions are seriously affected

Nowadays, the development of green logistics has become the consensus of the industry, which not only needs green packaging, green transportation, and green storage but also green intensive management of logistics resources. Some scholars have compared and analyzed the environmental benefits brought about by different logistics distribution modes. Montoya-Torres et al. (2016) compared the collaborative distribution and non-collaborative distribution and clarified that the implementation of collaborative distribution could reduce the transport cost and the impact of logistics on the environment. Guo and Ma (2017) conducted a comparative analysis on the comprehensive environmental impact of urban logistics systems in a self-operation mode, a collaborative distribution mode, and third-party logistics. Existing studies on the environmental comparison of different logistics modes have shown that collaborative distribution is an environmentally friendly distribution method. Thus, we start with collaborative distribution to optimize the current express end distribution network, propose a new collaborative distribution mode for urban express delivery, and explore the environmental benefits brought about by collaborative distribution.

In order to promote the collaborative development of express logistics, and improve environmental benefits effectively, here, we propose a collaborative distribution mode for urban express delivery, which is aiming at the “last mile” distribution problem. In the new mode, we set intelligent end service stations (IESSs) which provide end express delivery service for enterprises. The station can be operated by a third-party agent or an express enterprise alliance. The enterprises only need to deliver their packages to the IESSs instead of their own distribution outlets. The scale of the station is determined by the total volume of express business in the responsible area. Firstly, according to the characteristics of regional population distribution and the demand of express business, the specific radiation range of the IESS can be preliminarily determined, and then establish IESS at the upper level of the end points. Secondly, professional end distribution enterprises deploy intelligent express cabinets in the end points with large express delivery volume to relieve the pressure of end distribution. Finally, through the integration of the expresses to be delivered, the collaborative distribution from the IESS to the end points can be realized. The comparison of end distribution network under traditional mode and collaborative distribution mode is shown in Fig. 1. It can be seen that the establishment of IESS has a significant optimization effect on the end distribution network of urban express delivery.

Fig. 1

End distribution network under traditional mode and collaborative distribution mode

Through the implementation of the collaborative distribution mode for urban express delivery, the collaborative allocation of resources and the green delivery can be realized, and the end distribution network can be optimized. The advantages of the new mode are mainly reflected in the following four aspects. (1) The end distribution adopts the combination of door-to-door delivery and self-pick through intelligent express cabinets, avoiding inefficient point-to-point distribution. This mode can achieve a win-win situation: for consumers, there is a place to store the express, so they can arrange their free time to pick up the express and can also avoid the disclosure of address information. For express enterprises, they can improve the delivery efficiency and avoid repeated delivery. (2) The end delivery adopts collaborative distribution mode to avoid separate operation. The basic idea of urban collaborative distribution is after the expresses delivered into the destination city, each logistics enterprise will no longer deliver their own expresses, but the professional end third-party distribution enterprise will coordinate and integrate the expresses to be delivered in the whole city to realize collaborative distribution by setting IESSs, so as to reduce the overall cost of express enterprises and alleviate the traffic pressure and environmental pollution caused by the independent distribution of each express enterprise. (3) Establish a public information service platform to realize the docking and integration of logistics delivery information. Based on the principle of government guiding and market leading, the government should encourage the capable third-party distribution enterprises to build a third-party distribution and sorting information platform based on IESSs to provide public services for all express enterprises; thus, the problems of confusion, low efficiency, high cost, and low level of information at the end of express logistics can be solved. (4) Through the establishment of IESSs to achieve overall arrangement of distribution time, times, routes, and quantity of goods for distribution, delivery costs generated by traditional scattered transportation can be greatly reduced; the scale benefit of distribution can be realized; the utilization rate of logistics resources such as personnel, materials, and time can be improved; and the overall cost of logistics enterprises can be reduced.

In order to prove the environmental benefits of the new collaborative distribution mode for urban express delivery, we develop mathematical models to measure carbon emissions before and after the implementation of collaborative distribution, examining the daily distribution volume of several express enterprises in a specific region. Express distribution is divided into two processes: front-end and back-end. The process of express distribution is shown in Fig. 2.

Fig. 2

Work flow change of express distribution under collaborative distribution

For simplification, before the implementation of collaborative distribution, we assume that in the front-end process, daily packages are delivered from one logistics center (LC) in the city to the distribution outlets (DOs) of several express enterprises in the region by vans. In the back-end process, once the packages are delivered to the DOs of each enterprise, minivans and electric tricycles are used to deliver the packages to the end points (EPs). After the implementation of collaborative distribution, all express enterprises in this particular region integrate vehicle resources and collaboratively arrange vehicles for express distribution. Daily packages are delivered from LC to the IESS, and then delivered to EPs. Except that vehicle resources are shared and new energy vehicles are adopted in the back-end for distribution, other conditions remain unchanged.

Model development

Data collection

The data collection in this study includes the real business data of express DOs, including the average daily distribution volume, the number of distribution sections, and the vehicle ownership of various types, as well as their capacity, use, fuel, and power consumption. Details are as follows:

• 1. The number of DOs in the particular region where we measure carbon emissions

• 2. The average daily distribution volume of each DO

• 3. The number of distribution sections of each DO

• 4. Vehicle types used in the front-end and back-end distribution

• 5. The number of various vehicle types owned by each DO

• 6. The capacity of delivery per unit of each vehicle type

• 7. The fuel consumption per 100 km of each fuel-consuming vehicle type

• 8. The power consumption per 100 km of each power-consuming vehicle type

When it comes to measuring carbon emissions, some scholars have measured carbon emissions based on fuel consumption. Ramos et al. (2014) pointed out that greenhouse gas emissions relate mainly to carbon dioxide and used carbon emissions to measure environmental targets. Franceschetti et al. (2013) and Wang et al. (2017a) indicated that total carbon emissions are proportional to fuel consumption. Li et al. (2016) constructed the optimal vehicle routing model based on carbon emissions reduction, and it is assumed that during transportation, carbon emissions come from the complete combustion of fuel. Therefore, the carbon emissions can be measured through the usage of the distribution vehicles owned by the express enterprises in the distribution process. According to Ubeda et al. (2011), the conversion coefficient of carbon dioxide(e) is set as 2.61 kg/L. Using the fuel combustion chemical reaction C₁₃H₂₈ + 20O₂ → 13CO₂ + 14H₂O, according to mass conservation, CO₂ produced by fuel exhaust per kilogram is 1344/184 = 3.11 kg. Then, knowing the density of fuel is 0.84 kg/L, the carbon emission coefficient for power-consuming vehicles is set to 3.11 × 0.84 = 2.61. According to Zhang et al. (2019), producing 1 kW h of electricity will generate 0.9 kg of carbon emissions, so in this paper, the carbon emission coefficient for power-consuming vehicles is set to 0.90. The formulas for measuring carbon emissions are as follows:

$$ {\mathrm{CO}}_2\ \mathrm{from}\ \mathrm{fuel}-\mathrm{consuming}\ \mathrm{vehicles}\left(\mathrm{kg}\right)=\mathrm{Fuel}\ \mathrm{consumption}\ \mathrm{liter}\times 2.61 $$

(1)

$$ {\mathrm{CO}}_2\ \mathrm{from}\ \mathrm{power}-\mathrm{consuming}\ \mathrm{vehicles}\left(\mathrm{kg}\right)=\mathrm{Power}\ \mathrm{consumption}\ \mathrm{degree}\times 0.90 $$

(2)

Measurement model of carbon emissions before collaborative distribution

Assumptions

Based on the actual requirements and distribution process of express distribution, we make several assumptions regarding some practical constraints to reduce the complexity of the problem.

• 1. In the front-end distribution process, we assume that all packages transported to the end DOs are from only one LC in the city.

• 2. In the front-end distribution process, we assume that each express enterprise uses fuel-consuming vans for transportation.

• 3. In the back-end distribution process, we assume that each express enterprise uses fuel-consuming minivans and electric tricycles for distribution.

• 4. We assume that the efficiency of fuel and power consumption remains the same during the whole distribution process for all vehicles.

• 5. We assume that according to the average daily distribution demand, the distribution operation is divided into two batches where the first distribution batch of packages accounts for about 2/3 of the deliveries.

The first assumption (1) is made to reduce the complexity of the problem. The second and third assumptions, (2) and (3), are reasonable based on the actual operating conditions of most express enterprises. The fourth (4) assumes that there are many factors affecting express vehicle consumption of fuel and power, and these are difficult to measure. The fifth assumption (5) considers that the capacities of the vehicles and the sorting areas of each express enterprise are unable to meet the daily distribution demand through a single batch and is therefore divided into two batches to complete distribution.

Indices and parameters

For the front-end distribution:

i:

is the index for express DOs, i ∈ I.

j:

is the index for vehicle type used in the front-end distribution, j ∈ J.

m:

is the number of DOs.

n:

is the number of vehicle types used in the front-end distribution.

O_j:

is the average fuel consumption per 100 km of vehicle type j.

D_j:

is the capacity of delivery per unit of vehicle type j.

S_i:

is the average daily distribution volume of DO i.

R_i:

is the number of distribution sections of DO i.

X_ij:

is the number of type j vehicles owned by DO i.

For back-end distribution:

b:

is the index for the vehicle type minivans.

c:

is the index for the vehicle type electric tricycles.

d:

is the index for the vehicle type new energy vehicles.

f_b:

is the average fuel consumption per 100 km of b.

f_c:

is the average power consumption per 100 km of c.

f_d:

is the average power consumption per 100 km of d.

g_b:

is the capacity of delivery per unit of b.

g_c:

is the capacity of delivery per unit of c.

g_d:

is the capacity of delivery per unit of d.

q_ib:

is the number of b owned by the DO i.

q_ic:

is the number of c owned by the DO i.

q_id:

is the number of d owned by the DO i.

Decisions variables

For front-end distribution:

Y_ij:

is the number of type j vehicles used by DO i for the first batch.

${Y}_{ij}^{\prime }$:

is the number of type j vehicles used by DO i for the second batch.

For back-end distribution:

t_ib:

is the number of b used by DO i for the first batch.

t_ic:

is the number of c used by DO i for the first batch.

t_id:

is the number of d used by DO i for the first batch.

${t}_{ib}^{\prime }$:

is the number of b used by DO i for the second batch.

${t}_{ic}^{\prime }$:

is the number of c used by DO i for the second batch.

${t}_{id}^{\prime }$:

is the number of d used by DO i for the second batch.

Objective function

Before collaborative distribution, the objective function of the carbon emissions measurement model of the urban express distribution network (the model hereafter) is established based on the minimum carbon emissions, given as follows.

For front-end distribution:

$$ \mathit{\min}{C}_1=\frac{L}{100}\times 2.61\times \left(\sum \limits_i^m\sum \limits_j^n{Y}_{ij}{O}_j+\sum \limits_i^m\sum \limits_j^n{Y}_{ij}^{\prime }{O}_j\right) $$

(3)

For back-end distribution:

$$ \mathit{\min}{C}_2=\frac{L^{\prime }}{100}\times \left(2.61\sum \limits_i^m{t}_{ib}{\mathrm{f}}_b+0.90\sum \limits_i^m{t}_{ic}{f}_c+2.61\sum \limits_i^m{t}_{ib}^{\prime }{f}_b+0.90\sum \limits_i^m{t}_{ic}^{\prime }{f}_c\right) $$

(4)

$$ minC=\mathit{\min}{C}_1+\mathit{\min}{C}_2 $$

(5)

The objective function is composed mainly of two parts. The first part (C₁) is the measurement of carbon emissions in the front-end distribution process, and the second part (C₂) is the measurement of carbon emissions in the back-end distribution process. The first part consists of two items, namely, the arrangement of the first batch and the second batch using fuel-consuming vans. The second part consists of four items, namely, the first batch with minivans and tricycles and the second batch with minivans and tricycles.

Constraints

During the optimization process of the model, the following constraints should be satisfied in both processes as follows:

For front-end distribution:

$$ 0\le {Y}_{ij}\le {X}_{ij},\forall i\in I,\forall j\in J $$

(6)

$$ 0\le {Y}_{ij}^{\prime}\le {X}_{ij},\forall i\in I,\forall j\in J $$

(7)

$$ \sum \limits_j^m{Y}_{ij}{D}_j\ge \frac{2}{3}{S}_i,\forall i\in I $$

(8)

$$ \sum \limits_j^m{Y}_{ij}^{\prime }{D}_j\ge \frac{1}{3}{S}_i,\forall i\in I $$

(9)

For back-end distribution:

$$ 0\le {t}_{ib}\le {q}_{ib},\forall i\in I $$

(10)

$$ 0\le {t}_{ic}\le {q}_{ic},\forall i\in I $$

(11)

$$ 0\le {t}_{ib}^{\prime}\le {q}_{ib},\forall i\in I $$

(12)

$$ 0\le {t}_{ic}^{\prime}\le {q}_{ic},\forall i\in I $$

(13)

$$ {t}_{ib}{g}_b+{t}_{ic}{g}_c\ge \frac{2}{3}{S}_i,\forall i\in I $$

(14)

$$ {t}_{ib}^{\prime }{g}_b+{t}_{ic}^{\prime }{g}_c\ge \frac{1}{3}{S}_i,\forall i\in I $$

(15)

$$ {t}_{ic}\ge {R}_{i,\forall i\in I} $$

(16)

$$ {t}_{ic}^{\prime}\ge {R}_i,\forall i\in I $$

(17)

Constraints (6), (7), (10), (11), (12), and (13) indicate that the express distribution vehicles of each express enterprise before the implementation of collaborative distribution will be selected from vehicles already owned by each enterprise, that is, the number of express vehicles used will be less than or equal to the number owned by each enterprise. Constraints (8), (9), (14), and (15) indicate that the total capacity of the express vehicles of each enterprise before collaborative distribution will meet the average daily distribution volume of each batch, that is, the total capacity of the vehicles used will be greater than or equal to the distribution demand of each batch. Constraints (16) and (17) indicate that before collaborative distribution, considering that tricycles may be required in some special areas for distribution, each enterprise will reserve a tricycle for each distribution section, that is, the number of tricycles arranged by each enterprise will be greater than or equal to the total number of distribution sections of each enterprise.

Measurement model of carbon emissions after collaborative distribution

Assumptions

After the implementation of collaborative distribution, express enterprises integrate existing vehicle resources and carry out unified distribution according to the daily distribution volume. Moreover, they adopt 4.2 m new energy vans to assist back-end distribution. Based on the actual requirements and process of express distribution, we make several assumptions regarding some practical constraints to reduce the complexity of the problem.

• 1. In the front-end distribution process, we assume that all packages are transported to the DOs from only one LC in the city.

• 2. In the front-end distribution process, we assume that each express enterprise uses fuel-consuming vans for transportation.

• 3. In the back-end distribution process, we assume that each express enterprise uses fuel-consuming minivans, power-consuming electric tricycles, and 4.2 m new energy vans for distribution.

• 4. We assume that the efficiency of fuel and power consumption remains the same during the whole distribution process for all vehicles.

• 5. We assume that according to the average daily distribution demand, the distribution operation is divided into two batches, where the first batch accounts for about 2/3 of the deliveries.

The first assumption (1) is made to reduce the complexity of the problem. The second and third assumptions, (2) and (3), are reasonable based on the actual operating conditions of most express enterprises. Assumption four (4) considers that there are many factors affecting express vehicle consumption of fuel and power and that they are difficult to measure. Assumption (5) considers that the capacities of the vehicles and the sorting areas of each express enterprise are unable to meet the daily distribution demand through a single batch; therefore, they are divided into two batches to complete the distribution.

Decisions variables

For front-end distribution:

P_j:

is the number of type j vehicles used for the first batch.

${P}_j^{\prime }$:

is the number of type j vehicles used for the second batch.

For back-end distribution:

h_b:

is the number of b used for the first batch.

h_c:

is the number of c used for the first batch.

h_d:

is the number of d used for the first batch.

Objective function

After collaborative distribution, the objective function of the carbon emissions measurement model of the urban express distribution network (the model hereafter) is based on the minimum carbon emissions, given as follows.

For front-end distribution:

$$ \mathit{\min}{C}_1^{\prime }=\frac{L}{100}\times 2.61\times \left(\sum \limits_j^n{P}_j{O}_j+\sum \limits_j^n{P}_j^{\prime }{O}_j\right) $$

(18)

For back-end distribution:

$$ \mathit{\min}{C}_2^{\prime }=\frac{L^{\prime }}{100}\times \left(2.61{h}_b{f}_b+0.90{h}_c{f}_c+0.90{h}_d{f}_d+2.61{h}_b^{\prime }{f}_b+0.90{h}_c^{\prime }{f}_c+0.90{h}_d^{\prime }{f}_d\right) $$

(19)

$$ \mathit{\min}{C}^{\prime }=\mathit{\min}{C}_1^{\prime }+\mathit{\min}{C}_2^{\prime } $$

(20)

Similar to the model before the implementation of collaborative distribution in “Measurement model of carbon emissions before collaborative distribution,” the objective function consists mainly of two parts: the front-end (${C}_1^{\prime }$) and back-end (${C}_2^{\prime }$) distribution processes. The composition of the first part is the same as that of the first part of the objective function in “Objective function.” The composition of the second part is different from the second part of the objective function in “Objective function”; that is, after collaborative distribution is implemented, to reduce carbon emissions further and realize green and intensive urban express distribution, 4.2 m new energy vehicles are included in the back-end distribution process.

Constraints

During the optimization process of the model, the following constraints should be satisfied for the two distribution processes:

For front-end distribution:

$$ 0\le {P}_j\le \sum \limits_i^m{X}_{ij},\forall j\in J $$

(21)

$$ 0\le {P}_j^{\prime}\le \sum \limits_i^m{X}_{ij},\forall j\in J $$

(22)

$$ \sum \limits_j^n{P}_j{D}_j\ge \frac{2}{3}\sum \limits_i^m{S}_i $$

(23)

$$ \sum \limits_j^n{P}_j^{\prime }{D}_j\ge \frac{1}{3}\sum \limits_i^m{S}_i $$

(24)

For back-end distribution:

$$ 0\le {h}_b\le \sum \limits_i^m{q}_{ib} $$

(25)

$$ 0\le {h}_c\le \sum \limits_i^m{q}_{ic} $$

(26)

$$ 0\le {h}_b^{\prime}\le \sum \limits_i^m{q}_{ib} $$

(27)

$$ 0\le {h}_c^{\prime}\le \sum \limits_i^m{q}_{ic} $$

(28)

$$ {h}_d\ge 0 $$

(29)

$$ {h}_d^{\prime}\ge 0 $$

(30)

$$ {h}_b{g}_b+{h}_c{g}_c+{h}_d{g}_d\ge \frac{2}{3}\sum \limits_i^m{S}_i $$

(31)

$$ {h}_b^{\prime }{g}_b+{h}_c^{\prime }{g}_c+{h}_d^{\prime }{g}_d\ge \frac{1}{3}\sum \limits_i^m{S}_i $$

(32)

$$ {h}_c\ge \frac{1}{m}\sum \limits_i^m{R}_i $$

(33)

$$ {h}_c^{\prime}\ge \frac{1}{m}\sum \limits_i^m{R}_i $$

(34)

Constraints (21), (22), (25), (26), (27), and (28) indicate that the express distribution vehicles used by each express enterprise after the implementation of collaborative distribution will be selected from the vehicles owned by each enterprise, that is, the number of express vehicles used will be less than or equal to the number owned by each enterprise. Constraints (25) and (30) indicate the inclusion of new energy vehicles. Constraints (23), (24), (31), and (32) indicate that the total capacity of the express vehicles used by each enterprise after collaborative distribution will meet the average daily distribution volume of each batch, that is, the total capacity of the vehicles used will be greater than or equal to the distribution demand of each batch. Constraints (33) and (34) indicate that after collaborative distribution, considering that tricycles may be required in some special areas for distribution, each enterprise will reserve a tricycle for each distribution section, that is, the number of tricycles used by each enterprise will be greater than or equal to the total number of distribution sections of each enterprise.

Empirical study

Case study background and data

Here, we investigate an actual case of carbon emissions reduction by assessing the collaborative distribution mode in Beijing. Using five express DOs on College Road, Haidian District, in Beijing as an example, we estimate the environmental benefits by measuring the changes in carbon emissions before and after the implementation of collaborative distribution by the five express enterprises. The locations of the five express distributors on the map are shown in Fig. 3.

Fig. 3

Address of express DOs on College Road in Haidian District, Beijing

Among them, C3, C4, and C5 are in the same position; C1 has two outlets and the data of the two outlets will be combined in the measurement process. We conducted field surveys on the real business data of the five express DOs on College Road, including the average daily distribution volume, the number of distribution sections, the vehicle ownership of various types, and their capacity, use, fuel, and power consumption. According to the investigation, the average daily distribution volume and distribution sections of the five express enterprises on College Road are shown in Table 1.

Table 1 Average daily distribution volume and distribution sections of express enterprises (Source: Beijing Municipal Postal Administration System)

Our assessment indicates that the main types of vehicles used by the five express enterprises are vans, minivans, and electric tricycles, among which vans are used mainly for express transportation between city distribution centers and network outlets; minivans are used for major account work and large business volume distribution; and electric tricycles are used for distribution to end points such as universities, communities, and office buildings. The information on the express distribution vehicles for each type owned by the five express enterprises is shown in Table 2.

Table 2 Vehicles owned by express enterprises

To make the best use of existing resources, we assume that after the establishment of IESS, the number of express vehicles in each express DO remains unchanged to maintain the current status of express transportation from the urban distribution center. Since the IESSs have been set up at the back-end, the couriers only need to deliver the packages from the regional LC to IESSs; therefore, the back-end process of distribution can be unified by distribution vehicles. In theory, the larger the vehicle size, the higher the load capacity is and the more significant the scale benefit is. However, considering the poor flexibility of large vans, we use the 4.2 m new energy electric vans as the collaborative distribution vehicles. The basic parameters of each vehicle are shown in Tables 3 and 4.

Table 3 Vehicle capacity, use, and fuel consumption

Table 4 Power-consuming vehicle parameters

Carbon emissions measurement before collaborative distribution

Front-end distribution

We assume that in the daily distribution of the five express enterprises, they are transporting packages from Beijing Shunyi Airport Logistics Base Service Area to College Road, Haidian District, Beijing. By using the Baidu Map intelligent navigation, we estimate that the transportation distance for an express vehicle is 35.6 km; thus, the round-trip distance is 35.6 × 2 = 71.2 km. As shown in Table 3, front-end distribution vehicles include mainly 4.2 m, 4.8 m, 5.4 m, and 7.5 m vans. Collecting the data on daily distribution volume of each express enterprise, number of vans owned by each express enterprise, and the fuel and power consumption of each type of vehicle as the input of the model, we express the data in the modeling tool and solve the measurement model for front-end carbon emissions before collaborative distribution using Python.

Tables 5 and 6 show the optimal vehicle arrangement for the two batches of average daily distribution volume on College Road and the carbon emissions under the optimal vehicle arrangement for the front-end distribution before collaborative distribution. In the front-end process, it is estimated that carbon emissions from the first batch are about 275.03 kg, and from the second batch about 163.53 kg, totaling about 438.56 kg.

Table 5 Carbon emissions in the front-end distribution process from the first batch

Table 6 Carbon emissions in the front-end distribution process from the second batch

Back-end distribution

According to the scope of the College Road, Haidian District, Beijing, as shown in Fig. 2, each enterprise has a large number of about 24 distribution sections, on average. Therefore, we assume that the round-trip distance of each vehicle in back-end distribution is 2.5 km, on average. As shown in Tables 2 and 3, back-end distribution vehicles include mainly minivans and electric tricycles.

In this study, the optimal vehicle arrangement is based on the data on the vehicles owned by each express enterprise as shown in Table 2. In the arrangement of vehicles in the back-end distribution process, according to the distribution sections shown in Table 1, considering that electric tricycles may be required in some special areas, one tricycle is reserved for each section. Taking the data of daily distribution volume of each express enterprise, the data on the minivans and electric tricycles owned by each express enterprise, and the fuel and power consumption data of the two types of vehicles as the input of the model, we use Python to solve the model to get the optimal vehicle arrangement. As shown in Tables 7 and 8, in the back-end process, it is estimated that carbon emissions from the first batch are about 14.81 kg, and from the second batch about 13.16 kg, totaling about 27.97 kg.

Table 7 Carbon emissions the back-end distribution process from the first batch

Table 8 Carbon emissions in the back-end distribution process from the second batch

Carbon emissions measurement after collaborative distribution

Front-end distribution

After collaborative distribution, five express enterprises integrate vehicle resources and distribute collaboratively two batches of packages. Taking the data of daily distribution volume of each express enterprise and the data of all types of vans and the fuel and power consumption of all vans as the input of the model, we use Python to solve the model to get the optimal vehicle arrangement. As shown in Table 9, in the front-end process, it is estimated that carbon emissions from the first batch are about 228.57 kg, and from the second batch about 117.07 kg, totaling about 345.65 kg.

Table 9 Front-end carbon emissions after collaborative distribution

Back-end distribution

According to the number of distribution sections of each enterprise shown in Table 1, considering that electric tricycles may be needed in some special areas for distribution, one tricycle is reserved for each section, and the average is 24 tricycles per batch after collaborative distribution. The optimal vehicle arrangement is shown in Table 10. In the back-end distribution process, it is estimated that carbon emissions from the first batch are about 5.18 kg, from the second batch about 3.71 kg, totaling about 8.89 kg.

Table 10 Back-end carbon emissions after collaborative distribution

Results

The results of carbon emissions before and after the implementation of collaborative distribution are sorted according to the daily distribution demand of a total of 29,250 pieces from five express enterprises on College Road, Haidian District, Beijing, as shown in Table 11.

Table 11 Comparison of carbon emissions before and after collaborative distribution

Comparing the environmental benefits of front-end and back-end distribution: Before collaborative distribution, carbon emissions from the front-end are about 438.56 kg, accounting for 94.00%, and those from the back-end are about 27.97 kg, accounting for 6.00%, as shown in Fig. 4. After collaborative distribution, it is estimated that carbon emissions from the front-end are about 345.65 kg, accounting for 97.49%, and from the back-end about 8.89 kg, accounting for 2.51%, as shown in Fig. 5. We can see that the environmental benefits generated by the back-end are more prominent than those from the front-end. The reason for this is that 4.2 m new energy vehicles, which are power consuming, are used in the back-end to some extent, instead of minivans, which are fuel consuming. As Wang et al. (2019a) have said, the combination of power-consuming and fuel-consuming vehicles can effectively achieve a win-win scenario with low carbon emissions and distribution costs, which could be ideal for enterprises. Therefore, promoting the application of new energy vehicles in express distribution will reduce carbon emissions and promote the development of low-carbon logistics distribution.

Fig. 4

CO₂ emissions before collaborative distribution

Fig. 5

CO₂ emissions after collaborative distribution

Comparing the environmental benefits before and after collaborative distribution: As shown in Table 11 and Fig. 6, for the front-end, carbon emissions before and after collaborative distribution are about 438.56 kg before and 345.65 kg after, with emissions reduced by 92.91 kg, accounting for a 21.19% reduction, representing significant environmental benefits. For the back-end, it is estimated that carbon emissions are about 27.97 kg before and about 8.89 kg after, with emissions reduced by 19.08 kg, accounting for a 68.21% reduction. For the whole distribution process, carbon emissions are about 466.53 kg before and 354.54 kg after, with emissions reduced by 111.99 kg, accounting for a 24.00% reduction. As we can see from Fig. 6, compared with the 21.19% emissions reduction in the front-end, the 68.21% emissions reduction through collaborative distribution in the back-end are nearly three times greater, with particularly more significant environmental benefits generated. According to the arrangement of vehicles, it is the contribution of 4.2 m new energy vehicles.

Fig. 6

Comparison of carbon emissions before and after collaborative distribution

In urban logistics networks, vans are a major source of greenhouse gas emissions (Gan et al. 2018). Sim and Sim (2017) found that vans had a significant impact on potential global warming and acidification. According to Lin et al. (2014), a better utilization of vehicles and energy would contribute to sustainable transportation schemes more directly such as the promotion of alternative fuels and the use of next-generation electronic vehicles.

As a next step, we analyze the loading rate of vehicles before and after collaborative distribution and explore the benefits brought about by collaborative distribution from the perspective of vehicle usage. We collect statistics on the overall vehicle arrangement before and after the implementation of collaborative distribution in the two-batch daily distribution volume on College Road, as shown in Table 12.

Table 12 The number of vehicles before and after collaborative distribution

The capacity of each vehicle type is shown in Tables 3 and 4. According to the number of vehicles in Table 12 and the capacity of each vehicle, we can get the overall loading rate of the vehicles by comparing the distribution volume, as shown in Table 13. We can see that before collaborative distribution, the overall vehicle loading rate for the first batch is 45.09% and that for the second batch is 30.81%. After collaborative distribution, the overall vehicle loading rate for the first batch is 49.00% and that for the second batch is 48.15%. The implementation of collaborative distribution increases the overall loading rate for the two batches by 21.25%, which shows that collaborative distribution can contribute significantly to the reduction in the number of express vehicles needed and improve the loading rate.

Table 13 Total vehicle loading rate before and after collaborative distribution

Finally, we estimate the environmental benefits of collaborative distribution to the region. The average daily distribution demand in the Haidian District, Beijing, was about 718,300 pieces in 2016, sourced from Beijing Municipal Postal Administration System Database. According to the proportion of express distribution volume, we estimate that the total carbon emissions can be reduced by 2.75 t for the daily distribution demand of the Haidian District, with an annual decrease of 1003.82 t. According to the survey of express distribution volume in Beijing, the average daily distribution volume is about 5,814,100 pieces (Beijing Post Administration 2016). We estimate that after the implementation of collaborative distribution, total carbon emissions can be reduced by 22,26 t for the daily distribution demand in Beijing, with an annual decrease of 8125.10 t. To sum up, in the long run, the new collaborative distribution mode for urban express delivery among express enterprises can integrate logistics resources to a certain extent, significantly reduce carbon emissions in the process of distribution, and promote green and orderly development of urban logistics.

Conclusion

Considering the urgent need for green logistics development with low carbon emissions and the green intensive management of logistics resources, the main contribution of this study is to propose a new mode of collaborative distribution for urban express delivery which is aiming at the “last mile” distribution problem, and put forward a new model of intelligent end service station (IESS) which provides collaborative express delivery for express enterprises. Since the carbon emissions contributed by vehicles make up an enormous portion of the total environmental impact of express distribution, we propose mathematical models to measure carbon emissions before and after collaborative distribution by optimizing vehicle arrangement of urban express distribution network and prove the environmental benefits of the new mode.

This study investigates the basic information of five express enterprises on College Road in the Haidian District in Beijing and conducts a comparative analysis of the results. Among the main findings, we estimate that carbon emissions can be reduced on College Road, for an average daily distribution volume of 29,250 pieces, from 466.53 to 354.54 kg. We also find that the overall loading rate of vehicles can be increased, and especially the use of new energy vehicles instead of minivans can contribute to emissions reduction. Obviously, the implementation of collaborative distribution will promote green development in the urban express distribution service system and can be significant for the sustainable development of the urban economy and environment.

Our findings will be encouraging and significant for the local government to implement more effective policies and regulations in the future. (1) It is suggested that the express enterprises should avoid vicious competition and implement collaborative distribution to reduce the waste of resources. (2) The professional third-party enterprises should be supported to provide unified standard service for end consumers. The government should provide tax and land occupation preferential policies for the IESSs, and vigorously promote the development of IESSs as public infrastructures. (3) The government should standardize and unify the express vehicles, actively promote the new energy express vehicles, and accelerate the construction of supporting infrastructures such as charging stations and maintenance departments.

While this study proposes a new mode of collaborative distribution based on IESS, and identifies the environmental benefits by a case study in Beijing, there are still some limitations. First, as it is difficult to obtain transportation distance data of each vehicle, we assume that all expresses come from one logistics center. Second, we do not consider the differences in fuel and power consumption of the same type of vehicles with different load and brands. The future study may include but is not limited to (1) the actual trajectory data of each vehicle can be obtained through the positioning device of express vehicles to estimate carbon emissions more accurately and (2) the mathematical relationship between carbon emissions and the load of express vehicles will be further studied, and the measurement models of carbon emissions will be improved.