Tuesday, February 17, 2015

Read First: A brief history of cycleUshare

The cycleUshare pilot test was removed from the University of Tennessee (UT) Knoxville Campus in December 2014. We get frequent requests for information from interested parties thoughout the world on this concept and how it can be replicated. This document is meant to be a brief history of the pilot test, how it was born, how it was operated, how it died, and ultimately how it will continue to contribute to the evolution of e-bike sharing. There is a FAQ at the end of the post. 

First, the cycleUshare system was motivated by the parallel emergence of bikeshare and e-bike technology, from about 2005 onward. While both technologies rapidly evolved, the industries were doing everything they could to grow their existing business models without much (revealed) interest in trying to integrate the technologies. In the meantime, a few in the academic community were curious if the two technologies could co-exist and I (Chris Cherry) decided to see if the technology could work and if it could add value to existing bikeshare models. The cycleUshare system was a research and education experiment designed to test e-bikes in a bikeshare system and to provide many the opportunity to ride and see e-bikes and bikeshare, which was the first exposure of this technology to many of the observers. I wanted to make sure we also included regular bicycles, so people would have a chance to use those if desired, which are better on most metrics. 


System Development



The first seed to get this project started was a College of Engineering research equipment grant. As a transportation researcher, my research work is generally done in the field, so traditional lab equipment was of little use to me. In 2010, I proposed to use an equipment grant to fund the development of an e-bikeshare station. I was fortunate to win this grant ($30,000) which allowed me to buy 15 e-bikes (modified Currie Technologies I-Zip Trekking Enlighted) and 6 regular bikes (modified Marin Larkspur). The concept was for two 10 dock stations, each with seven e-bikes and three bicycles, and no distribution between stations, i.e., bikeshare trips must be “round trips” back to the origin station. The bikes had combinations locks with them to park away from the station. We also relied on a battery swap system to assure that when bikes were in the rack, they always had a charged battery available, and this reduced some of the challenges with recharging on the bike. We always had batteries on the charger, even when all the bikes were out. We were also able to use this seed funding to start developing a racking system, fundamentally based on very robust car-door latch technology, and a user kiosk that also included a battery charging bank. The software/hardware controls were originally developed using National Instruments hardware and Labview software. We also developed custom GPS data loggers for the bikes to support the research questions.

The initial bit of money was earmarked for equipment and hardware only. Since there were no vendors on the market, we built everything (except the bikes) from the ground up. The main partners were our own University of Tennessee Civil Engineering shop staff and research staff from the Department of Biosystems Engineering. They put in a lot of sweat equity. We also had plenty of students who volunteered­­–painting, welding, and doing coursework that fit and so on.

One of the bigger lessons learned was to include some padding in the schedule to let the lawyers have their way. UT is self-insured as a State institution and we had to make sure we did some due diligence to minimize risk and make sure that UT was as shielded as much as possible from the bad behavior of our bike riders. We were able to secure permission to run the program for 18-24 months as a research and education experiment.


We built the first station in Spring 2011 and installed it at Presidential Court in the heart of UT campus in the July 2011. We ran our initial trial during the academic school year of 2011-2012 and pushed to finish our second station, which was a solar-powered, off-grid station by spring 2012. We needed a financial boost from UT and the student sustainability fee was able to give us a $15,000 grant to finish the second solar-powered station. We were a little late with that station, finishing in the March/April period, just in time for students to disappear for the summer. The original intention was to tie that station into the power grid so that we could net-meter the solar and grid electricity. In the end, it would have been more expensive to tie into the grid than to just run the system off-grid. This made the system truly zero-emission (use phase), meaning we didn’t draw from any coal power plants to charge our batteries, contrasting with the other slightly more symbolic solar EV charging installations.



Operating the System



Once we had a good start, Tennessee DOT was interested in some of the research questions that the hardware enabled and we received a $50,000 grant to explore safety, health, and behavior related research. TDOT at the time was investing in bikeshare in Chattanooga and later Nashville, and most recently Franklin. In addition to the TDOT grant, the Southeastern Transportation Center, a Federal University Transportation Center, supported graduate students on the project, focusing on safety analysis. The nice thing about those grants are that they were for people (not hardware), so we were able to pay graduate students and staff to work on developing the system and running the experiments. This really helped Casey Langford and his dissertation work.


We ran both of the stations throughout the 2012/2013 academic years. We didn’t charge users anything, instead focusing on the system as a transportation service to the community with zero marginal cost (similar to most bikeshare systems). We did explore conducting price experiments, but couldn’t get our software to behave. Through this time were able to gather a good amount of data from the platform that ultimately resulted in some publications that were the first of their kind focused on e-bike sharing. Of course, one of the big challenges was running the system while doing the research. My students and I spent a lot of time working with customers (UT students, faculty, and staff). We dealt with flat tires, dead batteries, lost bikes, and crashes; everything a bikeshare operator has to deal with. We never pretended to have a functioning customer service operation and our users knew this. We also had a fair share of software problems. Some of these problems were mysterious bugs in the software, others were simply design problems. We relied on users who could follow directions precisely. Our software was full of dead-ends, meaning that a user would make a wrong turn in the check-out/check-in prompts and would never be able to recover, leaving the system stalled for the next person. The good user would send a message to me or Taekwan and we’d respond, most users just walked away leaving the system frozen. More on that later. 

In terms of demand, we ran two stations that did not have much overlap in service, so it was a little bit difficult managing users. We usually had a lot of capacity on our solar station on Ag campus, but we regularly ran out of bikes at our main Presidential Court station. So when we would add users we would speculate on which station they would use. Our solar Ag campus station was close to a greenway and more often used for lunchtime excursions etc. Our Presidential Court station was used for getting around campus. We couldn’t open up the system for all users because we wanted to make sure there were bikes available. We ended up with about 100 active users between the two stations (20 bikes). We had about 700 on the waiting list. This worked out pretty well, but would create image problems. For example, when working on the station in the evening/morning, we would encounter people on the wait list and they’d ask why they weren’t able to use the system yet. We, with a straight face, would tell them that we had too much demand, while we were standing by a station full of bikes. We’d then go on to talk about peak availability and peak demand, but it would have been good to figure out how to accommodate more users. Pricing could have accomplished this.


The end of cycleUshare


When the summer of 2013 came, most of my graduate students and I left the country to do research overseas. The research funding ran out as well and we had arrived at the sunset of our two-year permission window with UT administrators. We had a couple of research grants that were under review, but we couldn’t justify spending time and expense running the system over the summer without the resources. We shut down the system for the summer and lent a few of our most active users bikes that they could use. When the 2013/2014 school year started, we approached the UT administration and asked if they were interested in continuing to support the bikeshare system on campus. We offered to run the system, still conducting research and pursuing grants for the open research questions, if they could give us support on the operations and maintenance. I also suggested that the system should expand if it were to really make a dent in campus transportation. After a few proposal iterations, the campus administration allowed us to continue running the system if we could find external funding, but UT (administration or student sustainability fee) would not provide any funds for the system.  At this point we waited on outstanding research proposals that were under review, but saw the end potentially coming near. In Fall 2014, we learned that we did not win the last research grant that was outstanding (focusing on fully integrating the bikeshare system with transit) and we decided that we would pull the system off campus, which we ultimately did in December 2014. We are in the process of repurposing much of the hardware and bikes for uses that can support the interesting things UT is doing and supporting our next steps. 


What’s next for cycleUshare



We found when we were operating our system that things we did for expediency, were not suitable for a commercial product. For example, our controls and user interface were created in Labview, which is not designed for this type of application. It was good in the sense that we could deploy it rapidly and it integrated nicely with existing (National Instruments) controls hardware. However, created some problems with a few of the research applications we were trying to explore (like pricing experiments). Also, when interested parties wanted what we created, it was difficult to hand over the code for them to evaluate, or find programmers to modify the system. One of the main frustrations I hear from municipalities and others is that existing bikeshare systems are single-design, proprietary systems that do not integrate well into existing system. Once expensive hardware systems are purchased, the customer is locked into that technology, a marriage that has little flexibility and is very expensive to break.


We decided to open up our system, hardware, software, and electronics. This is involving some careful redesign of our controls system, moving away from National Instruments control hardware to a more versatile open source input/output hardware system. We also are rewriting our software, in Java and Python, so that it will be accessible to the largest community of developers. It will be integrated into our new hardware system, and will be more robust and flexible than our previous version. This helps with some of the software bugs discussed above. We are finally developing physical design documents that could be applied or modified to fabricate stations. The idea is, for the first release, to provide everything needed to replicate the system we developed, warts and all. The placeholder website is currently www.openbikeshare.org, and all information will be linked or posted from there. We intend to provide all of software and system design under a non-commercial open source license in the near future.


So, cycleUshare is dead. It was, in my mind, a successful first attempt at integrating electric bikes into bikeshare. It was never meant to be run by me and my research team indefinitely, but we would have been excited to see it grow into something bigger than a small research experiment. We learned a lot from the project and hope that we can continue to build on the lessons to bring sustainable transportation to the forefront.  




Frequently Asked Questions

1)   How much did it cost to build your system?
This is a tricky question, because we used a lot of “soft” resources like student and staff support, as well as access to fabrication facilities for both our stations and our electronic and controls systems. We also made a few mistakes along the R&D track, which we would not make again. We would also spend more money on some sensors to improve reliability. So, if we had to do it over again, each grid-tied 10-bike station would cost about $5,000-7,500 in materials. Add solar power and peripherals would be another $5,000 or so. Better electronics, etc. would cost marginally more. Our open source system should provide all that is needed to get good fabrication cost estimates. Our system can use any type of bike, so it is up to you how much you want to spend on bikes.

2)   If you could do three things different, what would they be?
1)   Step through frames for our bikes. Our one-size-fits-all bikes didn’t comfortably fit people in the low 5’ range, eliminating a group of users, mostly women. Step through frames would have solved this.
2)   Design more idiot-proof software, even if it takes longer. Our software was not robust enough to handle user mistakes. Never overestimate the human factors!
3)   For our system, on-bike battery charging. We didn’t have so much turnover that it justified battery swapping (see some of our papers). Onboard charging could have made the software a little simpler and more user-friendly for our demand levels. This might have creates some new challenges though associated with running higher voltage to the rack itself.

3)   If you deemed your system a success, why is it gone?
The primary goals of cycleUshare were to conduct research, really a proof of concept pilot test, on e-bikesharing and a few questions surrounding the system. It was also an education platform where we could share EV technology and bikeshare technology broadly with the UT community and beyond. We achieved most of those goals. Like all transportation systems, their benefits are “public” so they rely on some combination of user fees and public investments; transit, highways, and parking come to mind.  Bikeshare systems have been shown to be very cost-effective ways to serve short urban trips, but still require non-user-fee investment. Bikeshare in most cities relies on a mix of user fees, direct government support, and sponsorship (advertisement). UT’s administrators would not invest in something that was not “self sustaining”, which is impossible for nearly all transportation systems. The research team was not in a position to create a package of funding to support the system so we allowed it to die.

4)   Did you have any theft problems?
No, we did not have any explicit theft problems. We had a little bit of vandalism, where bikes would have brake cables ripped out, or people would use the leverage of the bike to twist on the bracket (theft attempt or not?). We never lost a bike. However, UT is a pretty low bike-crime type of area and there were times when I found our cycleUshare bikes unlocked, but unstolen.

5)   Did you have any safety problems?
We had three documented crashes in about two years, two car-bike crashes and one reported single bike crash. A regular bicycle crash resulted in a minor injury and a police report (car turned left in front of cyclist). One e-bike rider was struck in a crosswalk, without injury. Another e-bike rider hit a curb and crashed, without injury. Those three crashes resulted in minor damage to the bikes. There was some evidence of minor crashes (or abuse) on bikes that were not recorded, slight damage was found on a few bikes.

6)   What’s next for cycleUshare?
We get a lot of enquiries from organizations and individuals who want to replicate what we’ve done in some form or another. All requests have slightly different flavors, but unfortunately, most of the bikeshare industry sells only minor variations of the same product. We are hesitant to sell anyone the keys to our system (hardware/software/electronics) without the backend support to maintain it, so we’ve decided to provide all of our station designs, software, and electronics in an open source platform (www.openbikeshare.org). This way, the community can take our system, which is not elegant in many places, and improve it and modify it to meet their specific needs.

7)   Where can I get any of the literature you produced from this study?
Each of the articles below has a link to the publisher website. Most are locked behind a pay wall. If you can’t access the articles through your library subscriptions, email me (cherry@utk.edu) and I’ll find a way to share the results. We have at least three more working papers that are not yet published in journals.

1    1.    Langford, B.C., Chen, J. C. Cherry (2015) Risky riding: naturalistic methods comparing safety behavior from conventional bicycle riders and electric bike riders. Accident Analysis and Prevention, 82, 220-226. http://www.sciencedirect.com/science/article/pii/S0001457515001992


2.    Langford, B.C., C. Cherry, T. Yoon, S. Worley, D. Smith (2013) North America’s first electric bicycle share: A year of experience. Transportation Research Record: Journal of the Transportation Research Board. 2387. Pg 120-128. 10.3141/2387-14. http://dx.doi.org/10.3141/2387-14 

3.     Ji, S., C. Cherry, L. Han, D. Jordan (2014) Electric Bike Sharing: Simulation of User Demand and System Availability. Journal of Cleaner Production. 85. Pg. 250-257. http://dx.doi.org/10.1016/j.jclepro.2013.09.024

4.   Cherry, C., S. Worley, D. Jordan (2011) Electric Bike Sharing—System Requirements and Operational Concepts. The 90th Annual Meeting of the Transportation Research Board. Washington D.C. January 23-27, 2011. http://amonline.trb.org/2011-1.191976/t-11-030-1.205909/411-1.206120/11-0640-1.206127?qr=1

5.     Campbell, A.A. (2012) Factors Influencing the Choice of Shared Bicycles and Electric Bicycles in Beijing – A Stated Preference Approach. Masters Thesis. Civil and Environmental Engineering, University of Tennessee. http://trace.tennessee.edu/utk_gradthes/1364/

6.     Brian Casey Langford (2013) A comparative health and safety analysis of electric-assist and regular bicycles in an on-campus bicycle sharing system. Doctoral Dissertation. Civil and Environmental Engineering, University of Tennessee. http://trace.tennessee.edu/utk_graddiss/2445


8)   Who gets credit for working on the system and associated research?
Apologies if I miss people here, but the main players who made contributions to either building systems or research were:

Levon Brassfield (BS Student EECS)
Andrew Campbell (MS Student CEE)
Phillip Goldfarb (BS Student EECS)
Shuguang Ji (PhD Student CEE)
David Jordan (MS Student CEE)
Brian Casey Langford (PhD Student CEE)
Ryan Overton (PhD Student CEE)
Larry Roberts (CEE Shop Fabricator)
David Smith (Research Associate Biosystems Engineering)
Ken Thomas (CEE Shop Machinist)
Stacy Worley (Research Associate Biosystems Engineering)
John Wilkerson (Professor Biosystems Engineering)
Taekwan Yoon  (PhD Student CEE)

Really big thanks to Larry Pizzi at Currie Technologies and his staff for a coming in with a very good price on e-bikes and batteries and years of free technical support. Also thanks to SRAM/Quarq for some other enabling hardware.

9)   I’m interested in information that I could not find here. Who do I contact? Chris Cherry, Associate Professor, Civil and Environmental Engineering University of Tennessee, Knoxville

Monday, November 4, 2013

Journal of Cleaner Production publication: Simulation of e-bike systems.

Journal of Cleaner ProductionWe just published another article on simulation of e-bike sharing station, focusing on simulating demand parameters and optimizing station design (e.g., number of bikes, batteries etc). Our TRR paper is still in press too and can be downloaded here

Electric bike sharing: simulation of user demand and system availability  


Shuguang Ji, Christopher R. Cherry, Lee D. Han, David A. Jordan

Suggested Citation:
Ji, S., Cherry, C.R., Han, L.D., Jordan, D.A. (2013) Electric bike sharing: simulation of user demand and system availability. Journal of Cleaner Production. DOI:10.1016/j.jclepro.2013.09.024 (In Press).
Abstract: 
This paper describes the operational concepts and system requirements of a fully automated electric bike (e-bike) sharing system demonstrated through a pilot project at the University of Tennessee, Knoxville (UTK) campus (deployed in September 2011). This project is part of a movement to develop a sustainable transportation system, and is one of the green initiatives on UTK campus. E-bikes are more energy efficient and produce fewer greenhouse gas (GHG) emissions per person compared to other transport modes such as car, bus, and motorcycle. Without empirical demand information for an e-bike sharing system, we simulated the operations of such a system to gain insights during the design process before field deployment.  The simulation exercise focused on three critical demand parameters – distributions of trip rates, trip lengths, and trip durations – and coupled them with supply parameters – number of e-bikes, number of swappable batteries, and battery recharging profiles. The primary purpose of these simulations is to evaluate the efficiency of an off-board battery recharging system, where the depleted battery is removed from an e-bike upon its return and inserted into one of the charging slots at the kiosk. We tested various scenarios with different number of batteries always maintaining an initial condition with the battery to e-bike ratio greater or equal to 1.0 to ensure battery availability. We applied empirical battery recharging rates and system operations rules to determine the number of e-bikes and batteries available under different potential demand situations, with a focus on optimizing the number of batteries to meet user demands. By adjusting input parameters, numerous scenarios were simulated for sensitivity analysis. Based on the results of the simulation, this paper presents a cost constrained e-bike sharing system design that can maintain a high level of system reliability (e-bike and battery availability) through optimal battery charging and distribution management. We found that high demand scenarios require multiple swappable batteries per e-bike to reasonably meet the maximum demand. Trip duration has the most influence on e-bike and battery availability, followed by trip rate, and then trip length.

Thursday, September 12, 2013

Report: Findings from first year

There has been a flurry of interest in our program from around the world over the past several weeks. It seems that e-bike sharing is getting more mainstream. Many have asked for data or information on our system use. We have one published paper that describes some of our main findings from our first year, to be published in the Transportation Research Record soon. Instead of waiting, below is a link to the peer-reviewed preprint. We will update the file when the final paper is ultimately published. We have a couple of other interesting papers coming out of this system so stay tuned.

 North America’s first e-bike share: A year of experience