Since it was first mentioned at the World Economic Forum in 2016, increased attention has focused on the fourth industrial revolution, which involves maximizing production capacity and efficiency by adding new technologies to existing industries [1
]. One innovative technology is adding the use of drones (or unmanned aerial vehicles, UAVs) [3
]. The global drone market reached $
552 million in 2014 and is expected to grow every year by 16.9% [5
]. Unlike conventional drones that have been widely used for military purposes, the private sector will lead the drone market in the future [6
]. Although drones were developed for military purposes and were mainly used as reconnaissance and assault weapons, their use has greatly expanded due to the advantage of collecting a wide range of information as they move rapidly through the sky [7
]. For example, images taken by drones are used in sports broadcasts, documentaries, and media reports. They are even commonly used for humanitarian purposes, such as transporting vaccines in low- and middle-income countries and helping refugees migrate [10
Drone use in logistics services has also become common, taking advantage of drones’ sensing abilities using cameras and sensors and their accessibility to difficult terrains [13
]. For example, Amazon, the largest online retailer in the US, introduced “Amazon Prime Air” in December 2013, which uses self-developed drones called “Octocopters” to deliver goods up to 55 pounds within 30 min to a customer within a radius of 16 km [15
]. Amazon has already succeeded in testing the drones on several occasions and has acquired a patent for a UAV delivery system [17
]. Another global logistics company, DHL, developed third-generation drones and has assessed the feasibility of use for the entire delivery system, including flight technology, accurate shipping and storage, flight performance, and autonomous flight [13
]. Drone use for delivery has also been tested for food delivery services, where delivery speed is critical. For example, in San Francisco, delivering tacos by drones was piloted, and Domino’s succeeded in developing a commercial delivery of pizza by drone in November 2016 [18
To apply drone capabilities to actual logistics services, research has been conducted on how drones can be combined with existing truck delivery services. Murray and Chu [6
] and Agatz, et al. [21
] proposed a model to solve the travelling salesman problem in the context of drone delivery. Carlsson and Song [22
] also claimed that using drones along with the current delivery system could improve the efficiency of travel distance by a square root of the ratio of the speeds of a truck and UAV. Based on these assumptions, scholars have conducted research on the economic efficiency of drone use. Choi and Schonfeld [23
] calculated the necessary size of a drone fleet to minimize delivery costs and found that the cost varies depending on the working period, operating speed, demand density, and battery capacity. In addition, Welch [15
] conducted a cost-benefit analysis of the Amazon Prime Air service, and it was predicted that Amazon would gain a competitive advantage over other companies through the introduction of drone delivery and that the benefits would be greater than the initial cost for system construction.
As the problems related to the drones such as safety and privacy issues were raised, countries established regulations for the commercial use of drones, and some studies have analyzed and compared the regulations of each country [24
]. In February 2015, the US Federal Aviation Authority (FAA) established an approval process for companies trying to use the drones commercially. According to this law, operators must pass a written exam developed for a private pilot, which is administered by the Transportation Security Administration. Also, operators must observe the safety regulations restricting the flight area of drones to be away from private property, and at least six miles away from the airport [26
]. Not only the government, but also the drone industry, has made efforts to set up regulations and follow them. Unmanned Vehicle Systems International, which represents the US drones industry, has released a Code of Conduct to protect privacy and liberty rights. Also, DJI Innovations, a drone manufacturer, recently applied a GPS system to prevent the drones from entering restricted areas or flying higher than their permitted height.
Other countries also have their own laws regarding the commercial use of drones. According to Jones, the commercial use of drones is prohibited or practically prohibited due to strict requirements in 16 countries, and 18 countries, including Korea, allow drones to be operated only within visual line of sight. On the other hand, 27 countries, including the United States and Japan, have made an exception to the visual line of sight restriction to allow the commercial use of drones if they have been approved in advance or the weight or altitude of the drones is within the permitted range. Finally, six countries have made the drones commercially viable if they meet the simple guidelines such as obtaining licenses. As the technological and social situations of drones are constantly changing, the laws regarding drones are also constantly being updated, usually toward permitting drone use [27
In addition to studies on costs and regulations, studies on environmental impacts should also be conducted prior to the introduction of this new technology. Stolaroff [28
] called for a life cycle assessment (LCA) of drone delivery to establish rules and regulations; however, to date, limited research has been done on the relative merit of drone delivery in terms of environmental impacts, although some studies have analyzed the energy use and CO2
emissions of drones. In particular, Goodchild and Toy [29
] showed that the relative CO2
emissions of drones compared to trucks vary according to the drones’ energy requirements, travel distances, and number of recipients. To increase the advantages of drone delivery in terms of CO2
emissions, energy requirements should be low, the delivery distance should be short, and the number of recipients should be small. Figliozzi [30
] also calculated the life cycle CO2
emissions of drones and conventional diesel vans. While the CO2
emissions per unit distance were much lower for drones, the relative advantage of drones was lost when the customers were grouped because drones can only deliver to one place at a time. Lohn [31
] identified the same weakness of drones and recommended additional drone centers to conserve energy.
Several studies have suggested that drones have the potential to conserve energy and to reduce CO2
emissions; however, because there is no infrastructure for drone delivery yet, few scholarly analyses have been conducted to determine the actual CO2
reduction when drones are used in cities. Other environmental impacts must be considered as well. Many companies are striving to reduce environmental pollution as part of their corporate social responsibility, and the Korean government has implemented a certification system and subsidies for green logistics companies [32
]. Therefore, a comprehensive and specific analysis of the environmental impacts of drone delivery is needed in preparation for the application of drones in the logistics industry.
The purpose of this study was to evaluate and compare the environmental impacts of an existing motorcycle delivery system and a new drone delivery system to deliver food using the LCA. Regional variations of environmental improvement impacts after introducing drones were also evaluated based on the average delivery distance in urban and rural areas. In addition, other factors that could affect the results were examined, such as the introduction of electric motorcycles and changes in the way electricity is produced.
2. Materials and Methods
2.1. The Item to Be Shipped and the Shipping Method
Shipping distance and fuel efficiency were considered to evaluate the amount of pollutants and the environmental impacts of the delivery method. To focus on the delivery stage to compare a conventional and drone-based delivery system, information regarding some conditions that are unavailable for drones was excluded from the scope of this study, including the price and maintenance costs of drones, the current technology development stage, the delivery failure rate, and weather conditions.
Using drones for delivery facilitates access to areas that are inaccessible or difficult to reach by land, and delivery time is decreased by avoiding traffic congestion; however, a limited battery life and load capacity make drones unsuitable for long distances or large capacity freight. In addition, previous studies have shown that drones are inefficient compared to other modes of transportation when delivering to multiple destinations in the same area. Therefore, using drones for food delivery is more appropriate than parcel delivery because the weight and volume of the products are constant, the shipping distance is relatively short, and the recipients are much less likely to be absent. In addition, the advantage of drone delivery speed can be maximized for food delivery because a faster delivery speed prevents the food from becoming cool or spoiled. Thus, food delivery was selected as the delivery item for the current study.
Food delivery service is well developed in Korea and is expected to grow more due to structural changes of the population, such as the increase of single-person households, and the enhanced convenience of using delivery service according to technological development. Food delivery in Korea is mostly done by using a motorcycle to deliver to one house at a time, which is a suitable condition to change the delivery means to drones. Specific delivery distance was necessary in order to make an accurate comparison between drone and motorcycle delivery. Therefore, in this study, a specific food and brand was selected. The three most common delivery foods in Korea were chicken, pizza, and Chinese food [33
]. In this study, pizza was selected because there was little volume difference between each menu and the packaging was light. Also, Domino’s has already tested the use of drones to deliver pizza and has proven that it is feasible [19
Unlike packages, which are typically delivered by truck, pizza is mainly delivered by motorcycle in Korea. Therefore, the environmental impacts of pizza delivery by motorcycle and by drone were compared in this study. To evaluate the environmental impact of drone delivery, the environmental impact of 1 kWh of electricity was evaluated. Then, shipping distance and electricity consumption per unit distance were multiplied to calculate the environmental impact value for a single drone delivery. The electricity consumption per unit distance was obtained by referring to the product information of a specific drone model. The selected model was MD4-1000 (Microdrones GmbH: Siegen, Germany) from Microdrones, which was successfully used in DHL’s delivery test of drugs in 2014. MD4-1000 has an average flight time of 45 min and a flight speed of 12 m/s using a 22.2 V, 13,000 mAh battery (Microdrones). For the motorcycle test, the environmental impact of 1 L of gasoline was evaluated and then multiplied by the gasoline mileage and the shipping distance to calculate the environmental impact value for a single delivery. The gasoline mileage value used in this study was 63.5 km, which is the mileage of a Honda Super Cub, a motorcycle model for business purposes.
2.2. Target Area and Shipping Distance
Drone delivery was introduced to reduce shipping costs in countries with large land areas, a low population density, and high labor costs. They have been particularly suitable in rural areas where the population density and accessibility by land are lower than for cities. In addition, introducing drone delivery to rural areas could be more cost-effective and environmentally friendly. To compare the environmental improvement effects of using drone delivery in urban and rural areas, Yangcheon-gu, which has the highest population density (26,463.6 per km2) in Seoul, and Pyeongchang-gun, Gangwon-do, which has a low population density (27.6 per km2), were selected.
Since the environmental impacts of delivery increase proportionally with the distance traveled, it is important to estimate the appropriate shipping distance in each region. The location of elementary schools in each region was used to determine the shipping distance because the location of elementary schools can easily be determined in both urban and rural areas and because they reflect the actual residential area of the local residents. According to Rules on the Structure and the Establishment of Urban Facilities, schools are constructed to maintain proper distance intervals considering the population density and estimated student enrollment. One elementary school serves two neighboring residential districts, and the commute distance should be within 1500 m. In other words, the number and location of elementary schools are determined based on the population size, density, and residential areas of the region. Furthermore, according to a study in Pyeongchang-gun by [34
], town halls, which are the center of the living areas in rural regions, are typically within 5 km of convenient facilities, such as elementary schools, police stations, and public health clinics, suggesting that such facilities could represent the center of each administrative district. Therefore, for 30 elementary schools in Yangcheon-gu and 20 elementary schools in Pyeongchang-gun, the distance from each school to the nearest pizza restaurant was calculated, and the average was used to calculate the environmental impacts of the delivery method in each region (Figure 1
For drone delivery, the straight-line distance between the elementary school and the pizza restaurant was used. For motorcycle delivery, the actual travel distance was estimated by multiplying the bypass coefficient of Seoul and Gangwon-do to the straight line distance [35
]. Pizza restaurants that do not deliver were excluded. For Pyeongchang-gun, the stores located in a resort area were also excluded because they usually do not deliver outside the resort.
2.3. Life Cycle Analysis (LCA)
LCA quantifies all inputs and outputs to measure environmental impacts and compares the overall environmental impacts of particular products or services [36
]. “Life cycle” is a concept that includes all the necessary steps to consume a product, including raw material production, manufacturing, distribution, use, disposal, and transportation. According to ISO 14040, an international standard for environmental management, LCA has four stages: goal and scope definition, life cycle inventory analysis, life cycle impact assessment, and interpretation [37
The first stage of LCA is to define the purpose and scope of the research. At this stage, the background of the research should be presented. The functional unit, scope of analysis, and allocation method are also determined. A functional unit is a quantitative representation of what is to be analyzed specifically. Especially when comparing various products or services that can replace each other, setting an appropriate functional unit is crucial. The quantity or level of the desired function is defined first, and the number of products or services needed to achieve the function are then calculated. In this study, the functional unit was set as a single delivery of pizza in Yangcheon-gu, Seoul, and Pyeongchang-gun, Gangwon-do. The system boundary determines the scope of the process to be evaluated. The system boundary of this study is the delivery of pizza from the area of production to consumption, and drones and motorcycles were selected as the transportation methods (Figure 2
A list of inputs and outputs was created during the life cycle inventory (LCI) analysis step. The values of inputs and outputs were applied according to the functional unit determined in the goal and scope definition step. In this study, the LCI data on electricity from a national LCI database provided by the Korea Environmental Industry and Technology Institute (KEITI) were used for drones, and the gasoline LCI data from the same database were used for motorcycles. For gasoline usage, LCI data at the use stage were also included because many types of greenhouse gases and pollutants are emitted during the combustion process. The data related to gasoline combustion were obtained from the US LCI database provided by the National Renewable Energy Laboratory (NREL) [38
]. The sum of the environmental impacts from the production and combustion of gasoline was used as the environmental impact of motorcycle delivery.
The purpose of the third step, which is the impact assessment, is to evaluate the actual damage to the environment based on the result of the inventory analysis. The impact assessment step should include characterization, which distributes output materials to impact categories and converts them into a single unit. The normalization and weighting steps may then optionally be performed. Weighting is used to determine the relative importance of each impact category considering its social meaning. The environmental impact value of a product or service can be derived by summing the products of the weighting factors and the normalized value of each category. In this study, TOTAL (a tool for type III labeling and LCA), developed by KEITI, was used for the impact assessment. TOTAL is an LCA software developed by the Korean government. Since the “Enforcement Decree of the Environmental Technology and Industry Support Act” came into effect in 1995, the Korean government has encouraged the use of LCA. As there was no LCA software developed in Korea, software from Europe such as SimaPro and GaBi were used. However, they were not compatible due to different data formats and they were not easy for beginners to access. This led to the development of the government-led LCA software. After analyzing the strengths and weaknesses of various LCA software, and gathering the opinions from various experts, TOTAL was born in 2005. TOTAL provides a national database in the form of an Eco-Spold based on ISO TS 14048, and also provides public databases such as APME and IISI, so that it can be compatible with other LCA software [39
]. The results from TOTAL were compared in terms of particulates and the 10 categories selected by the Korean Ministry of Trade, Industry, and Energy (Table 1
At the interpretation stage, the main issues are summarized, the results are evaluated, and the conclusions are drawn. In this study, the results were compared to a third type of transportation, the electric motorcycle, since it is another environmentally friendly means of delivery. The environmental impact of 1 km of delivery distance by electric motorcycle and the environmental impacts of all three modes of transportation were compared. In addition, the US LCI database was used to compare the reduction effect of drones based on the national power generation schemes.
Finally, the environmental impact of the increased use of renewable energy was estimated.
In this study, the environmental impact of drones, a new means of delivery, was compared with the environmental impact of conventional delivery vehicles using the LCA. Based on the gasoline and electricity production data of the national LCI DB, the environmental impacts of a 1-km delivery distance using each type of vehicle were evaluated. In addition, the environmental improvement effects of using drones in urban and rural areas were compared based on the actual delivery distances in the residential areas of Yangcheon-gu, Seoul, and Pyeongchang-gun, Gangwon-do, based on the locations of elementary schools.
The LCA results showed that the GWP generated by 1 km of drone delivery distance (4.41 × 10−3 kg CO2-eq.) was one-sixth that of the GWP generated by 1 km of motorcycle delivery distance (2.85 × 10−2 kg CO2-eq.). The particulate emissions of drones (9.65 × 10−7 kg PM2.5-eq.) was estimated to be about half that of motorcycles (2.09 × 10−6 kg PM2.5-eq.). When the comprehensive environmental impact was evaluated after adding nine impact categories and normalizing and weighting the data, the environmental impact of drones was found to be one-twelfth that of the motorcycle.
The reduction of pollution by utilizing drones was more effective in rural areas than in cities. In Yangcheon-gu, the GWP reduction per delivery was 2.41 × 10−2 kg CO2-eq., and the particulate reduction was 1.56 × 10−6 kg PM2.5-eq. In Pyeongchang-gun, the GWP reduction was 0.323 kg CO2-eq., and the particulate reduction was 2.20 × 10−5 kg PM2.5-eq. The environmental improvement effect was higher in Pyeongchang-gun for both GWP and particulates. Pyeongchang-gun had a 13.2 times higher environmental impact reduction effect than Yangcheon-gu when the ten categories were considered together because it had a longer delivery distance.
In addition, in preparation for the emergence of new eco-friendly vehicles, electric motorcycles were added to the comparison to determine the environmental impact per distance. The GWP was higher for gasoline motorcycles because more greenhouse gases were generated in the combustion stage of gasoline than in the production stage, but electric motorcycles produced more particulates because particulates are mostly generated during electricity production. Therefore, the environmental impacts of various impact categories, as well as the GWP, should be analyzed and compared before introducing electric vehicles.
Another important consideration is that Korea’s electricity production method generates 2.49 × 10−3 kg CO2-eq. more GWP and 1.88 × 10−7 kg PM2.5-eq. more particulates than US electricity production. As mentioned in the 4th Basic Renewable Energy Plan, if 13.4% of energy could be supplied as renewable energy such as solar and wind power by 2035, 10~14% of greenhouse gases could be reduced compared to current emissions. Therefore, if the environmental impact per unit of electricity could be reduced by the increased use of renewable energy, the environmental improvement effect of drone use would be higher.
In this study, the environmental impacts of each delivery vehicle type were calculated using the national LCI database and the US LCI database; however, the baseline year of the LCI data was 2000 for the national LCI database and 2008 for the US LCI database, so the changes in environmental impacts due to technological developments and changes in the electricity generation plan have not been adequately considered. Therefore, if the databases were upgraded in the future and the data on various electricity generation methods were added, the accuracy of the results would improve.