Transportation Deployment Casebook/Bike-sharing Systems

Bike-sharing has emerged at the beginning of the 21th century as a way for cities to increase alternative, active transportation modes, reduce congestion and improve air quality (ITDP, 2013). Although the idea emerged in the 1960s in Amsterdam with the White Bikes program, it wasn’t until the early 2000s that the system became a worldwide phenomenon. The first generation of bike-sharing systems was located in Amsterdam and was very primitive: unmonitored, common bikes left at random locations for their use by the community. Today most cities have either 3rd or 4th generation systems. They are characterized by designated distribution points, centralized monitoring, payment and security systems, and especially designed and branded bikes

Definition

Bike-sharing systems are a type of transportation service which provides users a nonmotorized point-to-point travel option between a network of interconnected key destinations . Users can pick-up specially designated bikes at any station in the network and return them to any station located near their destination. Bike-sharing does not require the user either to wait for a vehicle (as buses or trains do) or waste time looking for parking (as private cars do). As such, it’s an option envisioned and encouraged for short (less than 3 miles), spontaneous trips due to its relatively smaller travel times.

Bike-sharing History

The White Bicycles Plan

Amsterdammers!

The asphalt terror of the motorized bourgeoisie has lasted long enough. Human sacrifices are made daily to this latest idol of the idiots: car power. Choking carbon monoxide in its incense, its image contaminates thousands of canals and streets.

Provo’s bicycle plan will liberate us from the car monster. Provo introduces the WHITE BICYCLE, a piece of public property.

The white bicycle is never locked. The white bicycle is the first free communal transport. The white bicycle is a provocation against capitalist private property, for the white bicycle is anarchistic.

The white bicycle can be used by anyone who needs it and the must be left for someone else. There will be more and more white bicycles until everyone can use white transport and car peril is past. The white bicycle is a symbol of simplicity and cleanliness in contrast to the vanity and foulness of the authoritarian car. In other words: A BIKE IS SOMETHING, BUT ALMOST NOTHING!

—Provo’s Provokatie (Pamphlet) #5 - Witte Fietsenplan (White Bicycle Plan)

The proto system for bike-sharing was proposed in the 1960s by a Dutch counterculture movement called Provo. The Provo (short term for Provotariat - proletariat in Dutch) nucleated several political activists who pronounced themselves against the capitalist, car-dominated culture that had pervaded throughout the previous decades.

The Netherlands has historically held a deep relationship with the bicycling culture, to the point of being identified as a national cultural symbol . Cycling in Europe fell dramatically during the 1950s and 1960s, when motorization rates began rising rapidly and urbanization started spreading out. During that period, many European cities focused on facilitating car use by expanding roadway capacity and parking supply, while systematically ignoring the needs of pedestrians and cyclists.

The White Bicycle Plan, created by industrial designer and Provo activist Luud Schimmelpennink, was one of the various “White Plans”, which advocated for limited commercial development and heavy taxation for polluters, among others. The bicycle plan was posed as a pragmatic response to traffic congestion in Amsterdam while also constituting a political critique, linking automobile development to unjust capitalist infrastructures, environmental pollution and consumer ideology .

The Plan proposed the closing of central Amsterdam to all motorised traffic to increase transit mode share to at least 40 percent. Provos also urged the City of Amsterdam to annually purchase 20,000 bicycles to be used freely by the residents, to supplement public transportation. The City rejected the Plan, but the Provos still insisted on the system, and began leaving white-painted bikes around the city, unlocked and free of charge. However the police impounded the bikes as they violated municipal law forbidding citizens to leave private property without securing it .

Schimmelpennink was a visionary, and his ideas of shared personal transportation systems, like the White Bike and the White Car (similar to today’s car-sharing scheme) became realities in the following decades. The White Bicycles Plan spurred an international interest, and similar White Bikes programs began appearing in some European cities, like Velos Jaunes in 1973 (La Rochelle, France), in the US with the Yellow Bike Project in 1994 (Portland, Oregon). However, due to the lack of security and system formality, most bikes were stolen and the system collapsed quickly.

Differences with traditional bike-rental schemes

Bike-sharing systems differ from traditional rental services in that they’re intended for short-distance and short-term duration trips. Bike-rental services tend to favor leisure users, like tourists. In contrast, bike-sharing systems target local residents and are usually devised within a larger transportation network, usually as complements for multi-modal trips at the beginning and end of bus or subway trip (what is usually called first-mile/last-mile legs), or as replacement for shorter trips (usually less than 3 miles). In addition, traditional bike-rental systems require users to return the bikes to the same station from where they were picked up, whereas bike-sharing allows users to pick up and drop off at different points within a larger, interconnected system. The use of special docking systems decreases user responsibility on providing security for the bicycle when the trip is finished, which minimizes user liability.

From Coins to IT: 2nd and 3rd Generation Systems

The first generation of bikesharing systems were mostly small, non-formal pilot projects from advocacy and political groups who sought to make bike travel in urban areas available to a wide range of people and detter car use.

It wasn’t until 1995 that bikesharing systems got a “second chance”. The Bycyklen (City Bikes) Program launched in Copenhagen in 1995, becoming Europe’s first large-scale bikeshare program . These new systems, also called Coin-Deposit Systems, addressed some of the main issues of the previous ones by requiring a 20 DKK (roughly equivalent to $3 USD) coin to unlock the bikes, which was refunded upon return at the end of the rental . Bikes were also designed for intense use, with solid rubber tires, and incorporated advertising on wheel plates to help with operating costs . Although more formalized than previous systems, theft was still prominent due to user anonymity and the relatively low value of the security deposit .

The end of the 20th Century sought to address these shortcomings with the use of smart technology and centralized systems (IT) to allow detailed user information and vehicle tracking . These changes constituted a new generation of bikesharing, also known as the “IT-based generation”. Technology-based systems allowed for a variety of improvements: user identification, electronically-locking racks, automated unmanned kiosks, electronic payment systems and electronic user interfaces . The first 3rd generation bike-sharing system was Bikeabout in 1996 in Portsmouth University (England), and began growing slowly in the following years, with programs in France (Vélo à la Carte, Rennes, 1998) and Germany (Call A Bike, Munich, 2000). Within the next few years, France launched then the first two major large-scale developments: Lyon’s Vélo'v in 2005 and Paris’ Vélib’ in 2007. Velib’ initially launched 7,000 bikes, which has increased to 23,900 and remains the largest bikesharing system in Europe . These programs generated noticeable repercussions within the transportation world, becoming milestones for bikeshare development. They also spurred the emergence of multiple new systems in Europe, Asia and America.

The introduction of RFID technology has also allowed existing programs to track bicycle usage and user information, helping curb bicycle theft which was a major drawback for prior system generations . The requirement of credit cards for signing up in third generation systems acts both as payment method and security deposit, further discouraging theft. In addition, a pricing structure is established to incentivize bicycle turnover, usually allowing the first 30 minutes for a fixed membership price, after which users are charged on a rising scale . Finally, some third generation systems include GPS technology which allows them to follow ridership patterns in real time, and in turn, drive daily redistribution efforts and provide useful data for planning system expansion .

The main challenges of third generation bikesharing systems are bicycle redistribution and integration with public transit systems Fleet rebalancing (moving bikes between high supply/low demand to low supply/high demand areas) can counteract the environmental benefits and car usage decrease by requiring providers to shuffle bikes around, consuming fuel and increasing car vehicle miles traveled (VMT). The lack of proper integration with other public transit systems is also a major deterrent: for example, there most bikeshare systems operate with their own payment and card system, and do not integrate with current transit payment systems, such as smart cards .

The rise of the Fourth Generation: Emerging trends

Since 2010, a new generation of bikesharing systems has been slowly emerging, often as a result of previous system improvements. The fourth generation of bikesharing is defined as a demand-responsive, multimodal scheme to create flexible, inclusive and integrated systems. The main areas for improvement have been :

• (a) Docking systems: Flexible docking would allow stations to be installed and redistributed with ease, acknowledging that usage patterns can change according to external factors (transit expansion, new urban features, etc.) and user demands. .
• (b) Rebalancing techniques: New rebalancing techniques such as incentivizing user-based redistribution would increase system efficiency and decrease operating costs .
• (c) Integration with public transit: A more seamless integration with transportation via smartcards that support multiple transit modes and other alternative sharing schemes, such as carsharing, would facilitate multimodal trips, further reducing auto usage.
• (d) Real-time tracking: Real time tracking through GPS could increase bicycle security and provide useful data for multimodal analysis .
• (e) Alternative energy sources: Electric bikes can increase ease of use in cities with varying topography and lessen physical exertion, thereby attracting users who otherwise wouldn’t be physically able to ride bikes , or commuters wearing business attire . Solar and Wi-Fi powered stations would ease installation costs and need for a support energy grid for operations
• (f) Improved business models: new models of provision would allow smaller populations to access these systems where advertising models would not be profitable.

Bikesharing in the 2010s

As of 2014, there are more than 700 cities in 57 countries operating bike share programs , a growing rate of over 50 percent when compared to 2013 . Systems vary widely in size, from single-dock systems and a few bikes to 1,000 docking stations and 20,000 bikes . Each city has adapted the system to the local context, taking into consideration urban density and infrastructure, economic resources and the natural environment .

According to an Earth Police Institute (EPI) report , between 2000 and 2014 bikesharing worldwide has increased over 110 times in number of programs (from 6 to over 700) and close to 200 times in fleet size (from 4,000 to 800,000). The distribution of bike-share programs around the world, however, seems to focus heavily in Asia and Europe. Since the beginning of the first programs in 2008, Asia has seen a tremendous growth rate both in number of programs and fleet size, outpacing all other continents, even Europe. As of September 2014, China leads the bike-sharing world with 170 systems . China also has the largest bikesharing system in the world, the Hangzhou Public Bicycles system, comprising 78,000 bikes distributed across 3,113 service points . However, Europe still holds the highest number of programs (approx. over 400), with Italy and Spain having close to 130 each .

Business Model Provision

Current bikesharing systems are funded through a mixture of advertising, self-funding, user revenues, municipalities and public-private partnerships . First and second generations were mostly non-revenue generating systems administered through non-profit organizations and government funding . The success and further spread of third generation systems through the globe have originated a diverse range of nonprofit and for-profit business models.

Several models of provision are present throughout the different generations of bikesharing:

• (a) governments;
• (b) transport agencies;
• (c) nonprofits;
• (d) for-profits and
• (e) advertising companies

The government model designs and operates bikesharing systems as just another transit service, exerting full control over its deployment . These systems can be funded mostly by their own governments, but also through usage fees and advertising on bikes and stations . However, public bodies may not have the necessary expertise that existing bike-share operators do and this can be counterintuitive with the fact that they maintain full liability for its operation .

Quasi-governmental models provide bikesharing services under the guidance of a public authority, such as a public transit agency . Revenue is collected through a mixture of government subsidies, usage fees and advertising . Rather than generating revenue, the purpose of quasi-governmental models is to provide good service, with the added bonus of the expertise from bike-sharing service providers . However, it is the (lack of) expertise of the agency operator that can be a detriment to the quality of the service .

Non-profit models have been used since the early generations of bike-sharing programs. They can be specifically created or they can be an existent organization that incorporates the service into its interests . This business mode places liability on an organization much less likely to be sued, which can be attractive for governments . However, they can be heavily dependent on local funding .

For-profit models are usually provided outside government purview, which can be helpful for private sector entrepreneurs as they do not need to wait for the rise of new transportation strategies usually associated with the implementation of bikesharing programs . This of course means an overall lack of funding assistance from public bodies, using advertising and user fees as their main sources of revenue. This can also lead to an unbalanced system since private models will be more interested in deploying stations on heavily populated areas where revenue is more likely to be higher, regardless of transportation equity issues.

Advertising models are the most common model in Europe, where private companies offer bike-sharing programs to jurisdictions in exchange for the right to use public space for advertisement purposes . This can be very cost-effective for local governments, which has contributed to its popularity . However, since system revenues usually go to the jurisdiction, system efficiency and quality of service can be compromised. Advertising models are found more prominently on Europe, while North American cities have tended towards non-profit and quasi-governmental models. Public agencies are also becoming main providers in China .

Life-Cycle Analysis

Data Collection

Bike-sharing statistics are hard to find. Availability of data regarding number of trips, number of users, fleet size or number of stations varies greatly depending on the country, the city, the service provider and the generation of the service. Most bike-sharing systems currently deployed are 3rd generation systems, which means statistical data is collected through centralized systems. This is true for the majority of large scale systems with hundred or thousands of bikes, which are found on larger cities. However, smaller cities tend to have systems on the range of 25-150 bikes. In such cases, most of the time data is not collected, either because the required infrastructure costs are high for a small scale system, or because these systems are older generations which have not been upgraded.

Aggregate data on bike-sharing systems is sparse at best. Although there exists a number of organizations and agencies with a demonstrated interest in bike-sharing practices, the difficulty of obtaining detailed data on a regular basis makes it a time-consuming and unreliable process. JCDecaux, an private operator of some of the world’s largest systems in France such as Velib’, does not release user statistics because they are not obligated by law . As such, some researchers and organizations have obtained data by monitoring and collecting real-time information provided by 3rd generation’s online systems who allow third-party access through their APIs, such as The Bike-Sharing World Map and The Bike Share Map. All in all, this makes bikesharing analysis and system size prediction an estimate at best, with huge caveats regarding data reliability. As more systems are upgraded to 3rd+ generation systems with real-time online data updates, data will become more reliable.

The data for this life-cycle analysis comes from the Earth Policy Institute’s 2013 and 2014 Data Center. This data corresponds with the 2000-2013 period. Additional data was obtained from individual reports to incorporate the first 3rd generation systems from 1995 to 1999.

Methodology

The Life-Cycle analysis uses a S-Curve prediction method to detect a system’s three main deployment periods: birth, growth and maturity. The model uses system size data (total number of bikes in bikesharing systems in the world) for the 1995-2013 period for its prediction.

The Life-Cycle model estimates the growth cycle of the system using a three-parameter logistic function:

$S(t)=K/[1+e^{(-b(t-t_{0})}]$

Where:

• $S(t)$: status measure (total number of bikes)
• $t$: time variable (years),
• $t_{0}$: inflection time (year in which half the saturation level is achieved),
• $K$: saturation status level (estimated),
• $b$ is a coefficient.

A regression model was used to find the saturation value (K) of the system, as the bikeshare deployment is still on its early stages. To find an accurate value for K, a statistical significant value with a high coefficient of determination (R-squared). During the curve-fitting phase, all values of K above 1 million had coefficients of determinations above 0.95. As a result, other variables were considered for K in order to achieve a saturation value, e.g. the feasibility of such number to happen, contemplating population size and bike modal share in transportation. Finally the selected K value was 6 million bikes worldwide.

Results

Measured vs estimated worldwide system size for 1995-2060.

Actual and Modeled System Size

Year	No. Bikes	Predicted No. Bikes	Year	No. Bikes	Predicted No. Bikes	Year	No. Bikes	Predicted No. Bikes
1995	1,305	634	2017		2,299,171	2039		5,998,358
1996	2,350	940	2018		2,877,839	2040		5,998,893
1997	1,645	1,395	2019		3,465,758	2041		5,999,254
1998	1,820	2,069	2020		4,019,169	2042		5,999,497
1999	1,820	3,070	2021		4,503,899	2043		5,999,661
2000	4,080	4,553	2022		4,902,405	2044		5,999,772
2001	3,980	6,753	2023		5,213,303	2045		5,999,846
2002	9,600	10,014	2024		5,446,089	2046		5,999,896
2003	10,740	14,846	2025		5,615,078	2047		5,999,930
2004	12,080	22,000	2026		5,735,020	2048		5,999,953
2005	17,140	32,583	2027		5,818,794	2049		5,999,968
2006	24,150	48,216	2028		5,876,653	2050		5,999,979
2007	58,130	71,260	2029		5,916,303	2051		5,999,986
2008	147,750	105,123	2030		5,943,330	2052		5,999,990
2009	261,800	154,657	2031		5,961,686	2053		5,999,993
2010	367,450	226,636	2032		5,974,123	2054		5,999,996
2011	449,840	330,221	2033		5,982,534	2055		5,999,997
2012	517,240	477,237	2034		5,988,217	2056		5,999,998
2013	800,000	681,834	2035		5,992,053	2057		5,999,999
2014		958,914	2036		5,994,641	2058		5,999,999
2015		1,320,632	2037		5,996,387	2059		5,999,999
2016		1,770,863	2038		5,997,565	2060		6,000,000

Regression Summary Output

Variable	Value
K	6,000,000
b	0.394522436
tnought	2018.206543

Regression Statistics	'
Multiple R	0.977101436
R Square	0.954727216
Adjusted R Square	0.952064111
Standard Error	0.497466249
Observations	19

'	Coefficients	Standard Error	t Stat	P-value	Lower 95%	Upper 95%	Lower 95.0%	Upper 95.0%
Intercept	-796.2277626	41.75663878	-19.06829156	6.51667E-13	-884.3265696	-708.1289557	-884.3265696	-708.1289557
X Variable 1	0.394522436	0.020836568	18.93413692	7.31023E-13	0.35056112	0.438483753	0.35056112	0.438483753

Conclusion

Bike-sharing systems seem to be leaving the birth phase, entering a period of sustained and incremental growth. All in all, the values seem to fit within the predicted S-Curve, except for the period between 2008-2012, where the actual system size is higher than expected. In 2012 the expected size fits again with the predicted system size. According to the S-Curve predictions, the system size will reach its inflection point in 2018 with a number of bikes of approximately 2.9 million. The maturity phase will be reached around the year 2025 with 5.6 million of bikes in the world, and by 2060 saturation (6 million of bikes) will be reached.