Final Results
Where Do You Want These Bikes?
modeling, simulating, and optimizing a rent-a-bike system
Advisor: Jeff Ondich
Background
Nice Ride Minnesota maintains a fleet of green bicycles throughout Minneapolis and much of St. Paul.
You can pick up or drop off a Nice Ride bike at any of several dozen automated stations. The bikes are heavily used, the stations are well maintained, and the program is growing after three successful years. And although most of the capital expenditures for building and expanding the system came from a variety of grants and donations, operating costs have been covered by usage fees (between 50% and 80% of operations) and station sponsorships (most of the balance). See section 6 of the 2012 mid-season annual report. So far, the Nice Ride system shows signs of being financially sustainable without large public subsidies, which is very promising.
A system like Nice Ride raises a lot of tough system design questions. Where should you build the stations? How big should each station be? How much should you charge per ride, per hour, per season, or per some other unit of measure? How do you balance support for high-traffic areas against equitable availability for as many citizens as possible? Another interesting problem to consider: every evening, a truck drives around, taking bikes away from some stations and dropping bikes off at others. How do you minimize the net flow of bicycles from some locations to others during a typical day, or should you even try?
These issues are clearly on the minds of people running Nice Ride. In their recently posted draft report System Optimization and Next Steps, they discuss where to build new stations, which stations they might want to move, questions of equitable access, and how net profits at popular station locations can be used to spread the system more widely to increase overall usage.
The Project
If I were designing a Nice Ride-like system from scratch or tweaking an existing system, I would want more than a spreadsheet of usage numbers. I would want an interactive simulation tool to let me try out various scenarios under assumptions of traffic patterns, usage rates, station locations and sizes, etc. This project's goal is to build such a tool.
Some steps you might take on this project:
- Research the general areas of simulation and system modeling.
- Study Nice Ride and similar systems to help prepare you for making a suitably detailed (but also suitably simple) model of such systems.
- Think about how you can measure the effectiveness of a simulation model. Could real data from Nice Ride help you evaluate the quality of your models?
- Start with a very simple (and undoubtedly unrealistic) model of a bike sharing system, and write a simulation engine for your model. Collect and analyze some data by running lots of simulations.
- Write an interactive front-end for your simulation engine, with the goal of allowing people to play with the model and see what happens in something like real time.
- Gradually increase the sophistication of your models and your tools.
- Measure the effectiveness of your models.
Good thinking early in this project could lead to models and tools effective enough that they could be useful for Nice Ride or similar organizations. That's a goal to keep in mind throughout.
(Want a little speculation to go with your comps? Look forward a few years: what happens if you combine these bicycles with a fleet of self-driving cars and legislation that allows those cars to drive themselves from place to place?)