Analysis of environmental impacts of drone delivery on an online shopping system

Jarotwan KOIWANIT

Faculty of Engineering,King Mongkut's Institute of Technology Ladkrabang,Bangkok 10520,Thailand

Abstract In rural areas,drones are designed to replace road deliveries so as to overcome infrastructure challenges;though drones notably consume less fuel and consequently have a smaller impact on the environment,their full life cycle assessment should still be evaluated to comprehensively understand their environmental impact.This study presents a life cycle assessment study on drone delivery in Thailand using CML2001,the life cycle impact assessment(LCIA)method,to convert life cycle inventory data into environmental impacts.The observed results show that an online shopping system using drone delivery is one of the most environmentally friendly transportation options throughout a wide range of scenarios.However,the parts production contributed to signifi cant impacts on environmental issues while the drone operation showed the least impact to all impact categories.The dominant contributors to global warming,abiotic depletion(ADP elements and fossil),acidifi cation air,eutrophication,ozone layer depletion,and photochemical ozone creation impact categories were the coal mining and electricity generating station operation.However,the carbon fi bers and the battery,are the main contributors to other impact categories,which include the human toxicity,freshwater aquatic ecotoxicity,marine aquatic ecotoxicity,and terrestric ecotoxicity.

Keywords:LCA;GHG emissions;CML2001;Online shopping;Drone delivery

1.Introduction

With the advent of the internet and increasingly challenging competition in thedynamic businessworld of today,the online shopping system in Thailand has accelerated to a point where the Thai e-commerce market is expected to triple in growth,from THB47 billion in 2015 to THB139 billion in 2020(Chan,2016).New transportation technologies have been introduced to serve customers'requirements,save time and money,and deliver better information while providing companies’high sale numbers and supporting their environmentally friendly programs(Shrivastava,2013).Transportation is considered as one of the main contributors to CO2emissions which contribute to greenhouse gas(GHG)emissions and has posed a seriousimpact to natural systems(IPCC,2014;Koiwanit et al.,2014a).Because of traffi c congestion in urban areas in Thailand,transportation was named as the second largest source of CO2emissions emitted into the air(Mangmeechai,2016).Even though there is not much traffi c congestion in rural areasin Thailand,road infrastructureposesachallengeto retailersin completing each delivery.Unmanned aerial vehicles(UAVs)or drones have been introduced and announced alongside the successes of online shopping systems by a number of companies,such as Amazon,DHL,Google,UPS,etc(Stolaroff,2014;Heutger and Ku¨ckelhaus,2014;CBC News,2013;Davidson,2013).According to DHL,2014,electrical drones appear to be the most promising type of drone within short distancesfor online shopping systemseven though their use cases are still in the early stages.The carbon footprint calculation of the U.S.online shopping system using different delivery options,which include cars,buses,parcel carriers,road trucks,and airplanes,have been evaluated together along with electricity generation,the consumption of natural gas,and packaging materials.GWPwasevaluated via theproduction of Li-ion batteries for electric vehicles.The power,energy and capabilitiesof Li-ion batteriesfor different passenger carswere also summarized(Miller,2015).Belmonte et al.(2017)evaluated the GWPusing LCA through energy-storage systems on battery-based mobile systems.In contrast to their studies on onlineshopping systemsand energy-storagesystems,thisstudy attempts to estimate the environmental impacts of drone deliveries as there have been very limited studies on the creation of their database and their impact on the environment.This study is a cradle-to-gate study focusing mainly on culminating in the creation of a database for drone delivery in Chiang Mai,Thailand,using the LCA methodology.GHG emissions will not only be evaluated,but also other associated emissions of the drone delivery system.

2.Methodology:LCA study of drone delivery services

2.1.Goal and scope definition

This study aims to illustrate the environmental impacts of a drone delivery system in Chiang Mai,Thailand.The home delivery system in rural areas in Chiang Mai,Thailand was chosen since the city is the largest city in northern Thailand,providing a significant increase in its economic growth while still containing rural settlements.Asaresult,thiscity wasused as the case study.

Again,the study takes a cradle-to-gate approach,starting from the coal mining,electrical generating station operations,drone raw materials,drone productions,and drone use phase.

2.2.Functional unit

The functional unit of this study is chosen as one package delivered per kilometer.Thisisafunctional unit for comparing different forms of transportation that aim to serve homedelivery services.This unit allows significant comparisons to be made since other transportation systems consume different amounts of energy to produce and operate compared to drones but provide the same service within the same distances(Koiwanit,2015).

2.3.System boundaries

This study is a cradle to gate study that takes into account resource extraction to produce drone,fuel,electricity,and resource depletion through the drone use phase.The system boundary defi nes the unit processes to be included in the system(Koiwanit,2015).The system boundary of the drone is shown in Fig.1.

2.3.1.Temporal and geographical boundaries

The lifespan in this study is assumed to be 5000 h which will be around 250,000 km using the experts’estimates based on drone operation.This is because the major components of vehicles tend to last around 12 years depending on maintenance and economic conditions.However,it is possible that vehicles will be able to operate more than their designed lifespans(Koiwanit,2015).The processesand data included in the system boundaries come from Thailand,other Asian countries,the U.S.,and worldwide(Zhou et al.,2014).The efficiency and specifi cation of the drone delivery in this study can be used in Chiang Mai which has hills in rural areas;as a result,the data from Thailand is for generating the specifi cation of the drone system.However,the parts production is from different countries(Table 1).

2.3.2.Technological boundaries

This study evaluates drone delivery services and their emission impacts.One system in particular is analyzed,which is the dronedelivery system using Li-ion battery asasourceof energy.

3.Life cycle inventory data quality,sources,and assumptions

In this study,because of data limitation,the LCI data of drone composition by material type comes mainly from experts’estimates.The majority of the data used in this study is specifi c to Thailand.Theassumptionsabout emissionsemitted from the drone delivery processes are mainly derived from theEEA(2009)and the IVL Swedish Environmental Research Institute(SERI,2017).The main parts of the drone delivery system and their references are shown in Table 1.Because of thedata absence for conducting astudy of thewholelifecycle of the drone delivery system in Thailand together with the limitation of drone site-specifi c data sources,information from many sources was drawn(Koiwanit et al.,2014b).These sources include:

Fig.1.System boundaries for drone delivery system.

1)City of Toronto(www1.toronto.ca),

2)IVL Swedish Environmental Research Institute(www.ivl.se),

3)European Environment Agency (EEA)(www.eea.europa.eu),

4)U.S.Environmental Protection Agency(U.S.EPA)(www3.epa.gov),

5)Harper International(www.harperintl.com).

The drone specification and its key parameters are listed in Tables 2 and 3,respectively.

4.Model description

In rural transportation issues regarding the movement of retail goods,trucking companies have been providing freight transportation services;consequently,a truck is considered as one of the most frequently vehicles used which serve customers’shipping needs(IL,2015).The energy use for the truck isfrom the combustion of gasolinerefi ned from crude oil and thisresultsin an increasein emissionsemitted into the air.Trucks carrying packages for delivery require many more miles than drone delivery which is able to travel on individual direct routes(Stolaroff,2014).As trucks face the increasing challenge of transporting freight across remote supply chain,this brings about alternative transportation methods,one of thesebeing dronedelivery systems.Thisisbecausethe system is not only inexpensive and easily available but can also perform complex and dangerous tasks(Rao et al.,2016).

This study develops a drone model and evaluates its environmental impacts to see its environmental feasibility of replacing traditional freight transportation services using other transportation modes.There are two main types of engines used in non-military drones which include 1)the electric engine and 2)the internal-combustion engine.The internalcombustion engine is best used for longer distance fl ights but is expensive and noisy(Heutger and Ku¨ckelhaus,2014).Unliketheinternal-combustion engine,the electric enginewas chosen since it can be operated without much noise,it is inexpensive to charge its battery,and is more environmentally friendly.As a result,it appears to be the most promising choice for application in the logistics industry(Heutger and Ku¨ckelhaus,2014).Even though the limitation of this engine type lays in its limited battery capacity,the battery-powered drones can be recharged when they are docked(Heutger and Ku¨ckelhaus,2014;McFarland,2017).In this study,several assumptions have been investigated,which include 1)drone energy use,2)average number of packages,3)averageweightlimit,and 4)the average distance to be traveled.Since drones aredesigned for usein different environmentsand they vary in size depending on the number of packages and their total weight,this results in different energy usages and distances to be traveled.The basic designs in this study are as follows:

Table 2 Drone specifi cation(modifi ed from www.dji.com).

·Drones can last up to 30 min(Pandit and Poojari,2014)

·Drones can carry products up to 5 kg

·Drones'cargo box dimension is 30.48 cm×30.48 cm×25.4 cm

5.Life cycle impact assessment(LCIA)

This study aims to determine the hotspots throughout the full life cycle of the drone delivery system.Towards this aim CML2001 was chosen,as this method is well-established,comprehensive,and commonly used.The CML2001 method has been acceptably used by the Thailand Environment Institute(TEI)and the Thai government(Lohsomboon et al.,2004;Wankanapon et al.,2013).Consequently,we can deem CML2001 to be best suited for this study.CML2001 was developed by the Institute of Environmental Sciencesat Leiden University in the Netherlands,Guin＇ee,2002).It is a midpoint-oriented method that includes characterization and normalization in the impact assessment process.In addition,GaBi 7,an LCA software product of PEInternational,Germany,is used in this study as this is the most frequently used software that provides a full range of analytical LCA tools(Koiwanit et al.,2014a,2014b;Koiwanit,2015).The study takes into account all associated emissions,wastes,and resource and energy consumption amounts during the life cycle stages using CML2001 methodology.The main focus of this research is on all 11 environmental impact categories,which include global warming potential(GWP),abiotic depletion(ADP elements),abiotic depletion(ADP fossil),acidifi cation potential(AP),eutrophication potential(EP),freshwater aquatic ecotoxicity potential(FAEP),human toxicity potential(HTP),marine aquatic ecotoxicity potential(MAEP),ozone layer depletion potential(OLDP),photochemical ozone creation potential(POCP),and terrestric ecotoxicity potential(TETP).

Table 3 Key parameters of the studied drone delivery.

6.Results

The LCIA results from the CML2001 methodology are shown in detail by impact category and summarized in the following sections.

6.1.Global warming potential(GWP)

The environmental impact in terms of global warming potential is0.079 kg CO2-eq in thedronedelivery system.The impact was primarily due to parts production,which accounted for 99.2%of the impact.CO2,CH4,and N2O were the main substances that contributed to global warming,which accounted for 80.88%,2.75%,and 0.89%,respectively.

6.2.Abiotic depletion(ADPelements)

The total abiotic depletion impact is 1.67×10-8kg SO2-eq for the drone delivery system.Part production and part transportation contributed 93.50%and 6.44%,respectively to the impact on the abiotic depletion(ADPelements)category.The main substances that contributed to the non-cancer air category include copper(Cu),silver(Ag),lithium(Li),and lead(Pb),which accounted for 20.47%,17.67%,15.72%,and 15.37%,respectively of the impact in this category.

6.3.Abiotic depletion(ADP fossil)

Thereisa total number in theenvironmental impactcategory of abiotic depletion considering ADPfossil of 5.16×10-5MJin thedronedelivery system.In theimpact to theabiotic depletion,around 98.12%of thecontributiontothiscategory wasgenerated from parts operation and 1.83%from the parts transportation.The main substances that contributed to the abiotic depletion were lignite,hard coal,and natural gas,which accounted for 43.62%,33.39%,and 18.40%,respectively.As the emissions were from the operation of the electric generating station,renewableenergy can beused to reducethisimpact category.

6.4.Acidification potential(AP)

The environmental outcome in the acidification air impact category is 5.16×10-5kg SO2-eq for the online shopping system using a drone.In this impact category,98.20%of the contribution was derived from parts production and 1.78%was from the parts transportation,and 0.02%from the drone operation.The dominant contributors to the acidifi cation air were SO2,NOx,and H2S,which accounted for 51.32%,34.91%,and 7.86%,of the contribution to thisimpact category,respectively.

6.5.Eutrophication potential(EP)

The CML2001 method represents eutrophication in different media(soil,air,or water)asone category of nutrient enrichment potential impact.The total result of this impact category is 8.65×10-6kg Phosphate-eq.This indicated that the potential for eutrophication was derived mainly by parts production and transportation,which accounted for 93.20%and 6.97%of the effect,respectively.The main contributors in the eutrophication air impact category calculated per 1-km traveled were fromnitrogen organic bound,and N2O,which accounted for 57.18%,8.99%,6.40%,and 4.82%,of the contribution in this impact category,respectively.

6.6.Freshwater aquatic ecotoxicity potential(FAEP)

The results from the CML2001 method indicated that the total impact is 1287.28 kg DCB-eq.The main processes that contributed to the freshwater aquatic ecotoxicity impact category was parts production,which accounted for almost 100%.There are many substances that cause freshwater aquatic ecotoxicity;some examples include chemical Cu,Co,Ni and Cd.These accounted for 43.15%,23.3%,19.6%,and 8.01%,respectively.

6.7.Human toxicity potential(HTP)

The dronedelivery system showsthe total impact in human toxicity potential which is 5.28×10-5kg DCB-eq.It was observed that parts production contributed almost 100%to the human toxicity category.NOx,NO,and NO2were the main contributors to the category of eutrophication,which accounted for 48.62%,42.65%,and 2.22%,of the contribution in this impact category respectively.The parts production was the source of heavy metals.As,Cd,Ni and Cu emissions were mainly emitted from carbon fi bers and Li-ion productions.

6.8.Marine aquatic ecotoxicity potential(MAEP)

In themarine aquatic ecotoxicity potential impact category,the total impact is 6.9×106kg DCB-eq.Parts production processes contributed almost 100%.Heavy metals,which include Cu,Co,Ni were the main substances that contributed to marine aquatic ecotoxicity,which accounted for 32.43%,31.09%,and 25.94%,respectively.

6.9.Ozone layer depletion potential(OLDP)

The total potential for ozone layer depletion in the drone delivery system were 2.14×10-12kg R11-eq.This number was derived from parts production and transportation,which accounted for 98.17%and 1.83%of thetotal,respectively.The main substances that caused ozone layer depletion included dichlorotetrafl uoroethane(R114)and chlorodifl uoromethane(R22),which accounted for 97.03%and 2.97%,respectively,of the impact in this category.

6.10.Photochemical ozone creation potential(POCP)

The online shopping system using a drone for delivery showed a total photochemical ozone creation potential impact of 3.82×10-6kg Ethene-eq.Production and transportation of the drone for delivery contributed 98.22% and 1.78%respectively to the photochemical ozone creation potential impact category.NMVOC,SO2,and NOxaccounted for 29.78%,27.61%,and 26.32%,respectively of the contribution in this impact category.

6.11.Terrestric ecotoxicity potential(TEP)

The total number in the environmental impact category of terrestric ecotoxicity was 3.85×103kg DCB-eq in the drone delivery system.This impact was due to the emissions mainly derived from the parts operations,which accounted for almost 100%.The dominant contributors to the terrestric ecotoxicity impact category were As,Cr,and Pb,which accounted for 52.8%,40.8%,and 2.1%,of the contribution in this impact category respectively.

7.Discussion

The results from the LCA of the drone delivery system shows that the dominant contributor to all environmental impact categories is the partsoperation.Partsoperation can be divided into three main categories,which include coal mining,electrical generating station operation,and parts production.Coal mining and electricity generating station operation were the main contributors to global warming,abiotic depletion(ADP elements and fossil),acidifi cation air,eutrophication,ozone layer depletion,and photochemical ozone creation impact categories.Parts production,especially the carbon fibersproduction which isthe raw material for the cargo box and Li-ion production which isthe main input for the battery,isthe main contributor to the human toxicity,freshwater aquatic ecotoxicity,marine aquatic ecotoxicity,and terrestric ecotoxicity impact categories.Droneoperation showstheleast impact to all impact categories as seen in Fig.2.According to Park et al.(2018),the global warming per 1-km delivery by drone was 0.004 kg CO2-eq.However,the global warming potential from this study was 0.079 kg CO2-eq,which shows higher impact than that of(Park et al.,2018).In addition,the global warming potential of 1 km traveled by motorcycle and electric motor was 0.028 and 0.018 kg CO2-eq,respectively(Park et al.,2018).This shows that drone delivery for online shopping systems is environmentally friendly compared to other delivery systems especially when they are under operation(Gulden,2017).However,drones are suitable for short trips with light-weight items while ground vehicles are suitable for carrying heavier products in long distance(Tseng et al.,2017).

Fig.2.Contributions of each process toward all impact categories.

8.Conclusions

This study proposes an alternative implementation of online shopping delivery systems in which customers can get their groceries at their doorstep within a short period of time.The environmental performance of the drone delivery system was analyzed and the results were generated using CML2001.The results showed that emissions were mainly from parts production,which include coal mining,electrical generating station operation,and parts production while the drone operation showed theleastimpacttoall of theimpactcategories.Themain contributorsthatcaused globalwarming,abiotic depletion(ADP elements and fossil),acidifi cation air,eutrophication,ozone layer depletion,and photochemical ozone creation include coal mining and electricity generating station operation.The carbon fi bersand Li-ionarethemaincontributorstothehumantoxicity,freshwater aquatic ecotoxicity,marine aquatic ecotoxicity,and terrestric ecotoxicity impact categories.However,therearestill some other raw materials or unavailable data which have not been taken into account which will result in different environmentalimpacts.Inthefuture,thisLCAstudy shouldbeextended to include other raw materials.In addition,it is recommended thatfor futurework other delivery systemssuchastrucksor cars should be evaluated to compare their environmental performancewiththeresultsof thisstudy.Thiswill show whichsystem ismoreenvironmentally friendly.

Acknowledgments

The author is grateful for the fi nancial support from Research Seed Grant for New Lecturer from KMITL Research Fund and the technical support from Mr.Junraprach Petchang,a Chulalongkorn University's graduate student.