Environmental benefits of taxi ride sharing in Beijing
Introduction
With over half of the global population living in urban areas, urban infrastructure is increasingly under pressure [1]. Vehicle transportation is a critical component of urban sustainability because it contributes significantly to energy consumption and emissions generation [2]. With accelerating economies and rising populations in urban centers, especially in developing countries, improving efficiencies in public transportation services and passenger automobile uses can provide more cost-effective and environmentally-friendly transportation solutions.
Sharing rides (combining two or more rider groups to travel together in the same car) as a way to reduce transportation energy consumption is not new. As early as the 19th century, the U.S. government has implemented policies to organize ride sharing (Car-Sharing Club) to conserve transportation fuel during World War II [3]. In addition to the societal benefits of reducing congestion, alleviating emissions, and conserving energy, ride sharing also offers benefits to the participants, including lowered travel (e.g., car ownership) cost, access to high occupancy vehicle (HOV) lanes, and elimination of the search for parking. However, due to the lack of attractive market mechanisms, difficulties of arrangement and logistics, and safety concerns (i.e. ride with strangers [3,4]), ride sharing has largely been constrained to families, friends, or colleagues, and is mostly prearranged (e.g., airport shuttles, vanpools) [5].
The recent developments in information and communications technologies (ICT), such as smartphones and various apps, have enabled users to exchange information in real-time and have facilitated participation in the “sharing economy”, both on a technical and a social level. Technically, the availability of real-time rider travel information, such as trip origin, trip destination, and desired departure and arrival time, has made it possible to develop a dynamic ride sharing (a.k.a., real-time ride sharing) system which only requires a minimal amount of lead time to identify sharing matches. Additional riders can also be picked up and dropped off along the travel trajectory, instead of requiring the ride sharing participants to have the same trip origins and/or destinations [6]. The involvement of social networks and reputation systems has helped build trust to share with strangers (e.g., Uber, Sidecar, Lyft, Airbnb) [7,8]. Therefore, ICT-enabled real-time ride sharing presents unprecedented opportunities to improve urban transportation efficiency. The technology and cyberinfrastructure for dynamic ride sharing at the large-scale is emerging. Several startup companies have already started to provide dynamic ride sharing services (e.g., Uberpool,1 Split,2 Lyft line3).
The current literature on ride sharing mainly focuses on the development of efficient algorithms for rides matching and recommender systems [[9], [10], [11], [12], [13], [14], [15]]. Limited attention has been paid to quantifying the “share-ability” of travel demands at the city level, which is important to evaluate the feasibility and impacts of ride sharing and persuade investors and stakeholders to invest in and promote such systems [16]. This research aims to fill this gap to quantify the environmental benefits of ride sharing in urban cities, taking into account the heterogeneous individual travel demands.
Four types of data are currently used to study ride sharing: travel survey data, cellphone traces, geo-tagged social media data, and trip origin and destination data captured by GPS devices. Travel surveys are conducted in many countries at different scales to understand national or regional travel and transportation patterns. In the form of questionnaire or travel dairy, travel surveys collect information such as the travel purpose, transportation mode, travel distance etc. [17] For example, based on commuting survey data, Amey (2010) estimated that sharing rides can reduce commuting vehicle-miles-traveled (VMT) by 6%–19% for the Massachusetts Institute of Technology (MIT) communities [5]. Although travel survey data are best suited to serve the purpose of analyzing ride sharing at the small scale (e.g., commuting within the MIT community), the information provided by survey data is static and cannot be used to study dynamic ride sharing at the larger geographic scale. Mobile phone trace data are collected by mobile network carriers for billing and operational purposes. It records the date, time, phone number (anonymized) of each cellphone activity (making or receiving a phone call or text message), and the coordinates of the cellphone tower routing the communication [18]. Compared to travel survey data, mobile phone traces have a much larger sample size and a broader spatial and temporal coverage. In addition, because the data are routinely collected for business operation purposes, the data collection cost is low [19]. Geo-tagged social media data are publicly shared information (e.g., tweets, photos, and check-ins) on different social media sites (e.g., Twitter, Facebook, Google+, Flickr, and Foursquare) with location data associated (typically as GPS coordinates). Depending on each social media site's policy, large-scale social media data could be hard to obtain [20]. Most of the large-scale geo-tagged social media data used for research are streamed from Twitter API [21]. Using cellphone records and geo-tagged tweets, Cici et al. (2014) estimated that ride sharing with friends' friends can reduce the number of cars in a city by 31% [21]. However, cellphone traces and geo-tagged social media data have very coarse granularity because the geolocation data of a user are only recorded when the user makes a phone call or posts a tweet. Trips that occur between two consecutive phone calls or tweets cannot be captured and may lead to inaccurate travel demand inference. Trip origin and destination data captured by GPS devices are location data collected by GPS devices equipped on vehicles [[22], [23], [24], [25], [26]], bikes [27], or individuals [28,29]. Because the data are collected passively and do not require active participation of the user, in contrast to other types of data mentioned above, GPS traces normally have finer granularity both spatially (more accurate location information) and temporally (high frequency of sampling) [26]. Analyzing trip origins and destinations of taxi trips in New York City, Santi et al. (2014) concluded that sharing taxi trips can cut trip length by 40% or more [16], and Lokhandwala et al. (2018) showed that the average vehicle occupancy can be increased from 1.2 to 3 [6]. Trip origins and destinations data can more accurately describe the travel demand of each traveler and therefore can better support large-scale dynamic ride sharing analysis.
Beijing (China), a city which has been called out in the media and academic research for its severe air pollution [30] and congestion [31], may particularly benefit from ride sharing to reduce total trip length and vehicular emissions. Taxi sharing can be the first step to implement ride sharing at the urban scale. Beijing has around 66,000 taxis, which serve over 2 million trips a day and contribute 1.08 million tons CO2 (7.2% of total transportation emission) each year based on 2012 data [32]. While some mobile phone apps (e.g., DiDi) are already in development to allow taxi sharing in Beijing [33], it is still unclear what the potential environmental benefits of taxi sharing are. To the best of our knowledge, there is only one study evaluated the environmental benefits of ride sharing in Beijing [34]. However, this study has a different scope, focusing on evaluating the environmental impacts of peer-to-peer car sharing trips from the perspective of transportation mode replacement, adoption of battery electric vehicles, and the potential changes of car ownership. Knowing the scale of trips that can be shared and the emissions that can be reduced through taxi sharing is critical to inform decision makers to design policies related to ridesharing.
Using real world trip origins and destinations extracted from the taxi trajectory data in Beijing, China as a case study, this research evaluates the environmental benefits of shared taxis (energy savings and emissions reduction of VOCs, NOx, PM10, PM2.5, and CO due to sharing). Although ride sharing using private vehicles may be different from shared taxi rides, the framework and methods developed in this research can be applied to private vehicles when trip origins and destinations using private vehicles become available at the large-scale. In addition, compared to ride sharing among private drivers, which requires more individual initiatives, shared taxi rides are more ready to be implemented. Compared to the existing literature, the unique contributions of this research are that (1) we proposed a framework to identify sharable trips and quantify the environmental benefits of ride sharing; (2) we analyzed the potential energy and emission reduction of taxi sharing in Beijing using real-world data; and (3) we evaluated the feasibility of ride sharing throughout the entire day, with different hourly travel demands.
Section snippets
Data
Data used in this study are vehicle trajectory data for 12,083 taxis (18.3% of the taxi fleet) in Beijing from November 1 to December 1 in 2012, which is the latest dataset we have access to. This dataset represents a typical winter month in Beijing, which tends to have poorer air quality [35]. We did not observe any spatial coverage (e.g., where the taxis were distributed) or travel pattern (e.g., the densities of trip origin and destinations) bias during this period compared to a smaller
Sharing benefits
Regardless of the variations in the total number of trips accrued during each day, the ride sharing benefits (miles saved and trips shared) are relatively stable (Fig. 3a). On average, about 77% of the trips can be shared, leading to 33% of the total VMT saved. This can be translated into saving 77,454 gallons of gasoline per day. The day-to-day variances of hourly sharing benefits are also relatively small, regardless of weekdays and weekends. These results indicate that the travel patterns of
Conclusion
In summary, using the shared taxi case study in Beijing as an example, we analyzed the environmental benefits of ride sharing. Shared taxis can provide stable sharing benefits in total VMT, energy, and emissions reduction, regardless of the travel volume and daily travel pattern variations. With a rider's tolerance level at 10 min, ride sharing can reduce fleet VMT by 33%. If implemented for the entire taxi fleet, shared taxis can save 28.3 million gallons of gasoline and reduce 186 tons VOC,
Acknowledgement
This material is based upon work partially supported by the Department of Energy under Award Number DE-PI0000012. HC thanks the support of the Dow Sustainability Fellows Program.
References (46)
Structure and environmental impact of global energy consumption
Renew Sustain Energy Rev
(2014)Ridesharing: the state-of-the-art and future directions
Transp Res Part B Methodol
(2013)The Share-a-Ride Problem: people and parcels sharing taxis
Eur J Oper Res
(2014)- et al.
Dynamic ride sharing using traditional taxis and shared autonomous taxis: a case study of NYC
Transport Res C Emerg Technol
(2018) Optimization for dynamic ride-sharing: a review
Eur J Oper Res
(2012)- et al.
An economic and operational evaluation of urban car-sharing
Transport Res Transport Environ
(2000) - et al.
Investigating ride sharing opportunities through mobility data analysis
Pervasive Mob Comput
(2014) - et al.
A real-time algorithm to solve the peer-to-peer ride-matching problem in a flexible ridesharing system
Transp Res Part B Methodol
(2017) Understanding individual mobility patterns from urban sensing data: a mobile phone trace example
Transport Res C Emerg Technol
(2013)- et al.
A statistical approach to estimating acceptance of electric vehicles and electrification of personal transportation
Transport Res C Emerg Technol
(2013)