Top Master of Engineering projects win awards

For sixty years, Cornell Master of Engineering projects have been awarded prizes established by the Silent Hoist and Crane company.  These prizes, intended to stimulate "the thinking of engineering students with aptitude or special interest in the Art and Science of Materials Handling," now recognize not only physical materials in manufacturing, but the handling of people, data and information.  The competition is open to all students and student teams in the College of Engineering. 

At the 2010 ORIE Graduation Ceremony, Associate Director Kathryn Caggiano announced the three winning projects, all in ORIE.  Winners were awarded a certificate of recognition and monetary prizes.  The selection of the winning projects was based on submissions by the teams, input from the project advisors, and detailed feedback from the organizations that were clients for the projects.

Washington D.C. Emergency Evacuation Response

First place was awarded to a project for MITRE, which is a not-for-profit organization based in Bedford, MA and McLean, VA that manages Federally Funded Research and Development Centers.  The project was part of an ongoing study to investigate the logistical challenges presented by an emergency evacuation of the Washington D.C. metropolitan area.  The team - Rob Guzman, Darren Hearn, Chris Mito and Sam Schmitz - investigated the potential for use of buses to speed the evacuation to a planned destination in West Virginia.  The team was advised by Professor Mark Lewis.

In the team's approach, buses pick up passengers at depots (where passengers may leave their cars), take the passengers to West Virginia, and return for more passengers.  The team identified twelve potential depot sites (seen at left) having capacity to hold the abandoned cars, worked out how buses should be optimally assigned to these depots, and simulated the flow of vehicles along a key corridor with the objective of determining the performance of the bus evacuation strategy.   

In addition to developing an integer programming optimization model the team integrated a Federal Highway Administration transportation-planning tool into a software application that automates the intermediate steps and handles the information required for individual runs.

Using the simulations, the team determined how many buses would be required to carry out the evacuation; measured the extent to which increasing the volume of traffic would have diminishing returns on the rate at which people could be evacuated; and measured the extent to which high traffic volumes greatly increased bus travel times.   They concluded that getting 25 % of drivers to abandon their cars in favor of buses resulted in the highest evacuation rate: raising the fraction to 40% or lowering it to 10% would reduce the throughput of the system.  

While ORIE has completed projects for MITRE for the past several years, sponsors Matt Olson and Eric Blair noted that "this is the best group and best output we have seen over the course of our involvement" with MEng projects.  They commented that "The team's product has contributed significant insight to our understanding of the problem addressed and will be used to guide future research in this area."  One representative told the team that the project "looks to be a crucial component in a larger framework for disaster planning awareness."

Delivering Paramedic Support in Rochester, New York

3rd Battalion, a volunteer emergency response vehicle organization in the Rochester, N.Y. area, uses both ambulances and so-called flycars to provide different services from different bases in southeastern Monroe County.  3rd Battalion's  ambulance service provides basic life support while transporting patients to full-service medical care emergency rooms.  Flycars are able to handle more serious medical problems since they carry paramedics who can provide patients with on-the-spot service and always are accompanied by an ambulance in case transport is needed.  An earlier project focused on the coverage of basic ambulance service; this year's project team, which won second place in the Silent Hoist and Crane competition, examined the response time pattern of 3rd Battalion's flycar operation as a function of equipment location. 

 Average response time with original base locations (in black).  Red squares show locations with time exceeding 11 minutes; yellow with time between 9 and 11 minutes, and blue with time between 7 and 9 minutes. The remaining were under 7 minutes.

This year's team, Esther Chiew, Aksara Sumethkul, Jonathan Timura, and Valerie Wilson,  set out to help 3rd Battalion meet a New York State requirement that 90% of emergency patients be reached from the base within nine minutes.  They developed a detailed simulation that accepts inputs - such as shift schedules, base locations, and number of fly cars - and outputs statistics on the time to reach patients.  Unlike most commercial emergency vehicle response software, the team's simulation took into account the random variability in service time within short-duration (e.g. quick procedures), medium-length (e.g. onsite procedures) and long-term (e.g. accompanying an ambulance to a hospital) services, since this can impact vehicle availability. The team was advised by Professor Eric Friedman. 

 While the team used the simulation primarily to test possible locations for a new base and to recommend the one that performs best, 3rd Battalion's Bill Stumbo, who heads the unit providing flycar service, notes that "being able to project changes in response time based on vehicle location and staffing levels will be useful.  Going forward we will be able to predict overall response rates and validate the results against actual data."   He said that "the team did a great job of taking our input and acting upon it in a meaningful way." 

Mutual Fund Replication

The client representative for the third place project - T.P. Enders of Goldman Sachs Asset Management - had the advantage of understanding the project process from a student perspective, having been an ORIE M.Eng. project team member himself as one of ORIE's first financial engineering M.Eng. students in 1996 as well as the sponsor of other projects in recent years.  Enders, a VP in Goldman's Portfolio Strategy Group, said that the project "was more thoroughly analyzed, better-organized and more compellingly presented than any of the other M.Eng. projects I've had the opportunity to sponsor or sit in on."  

The project, advised by Professor David Matteson, began with a hypothesis by Enders and others that it is possible to replicate the returns of certain managed U.S. mutual funds, with potential benefits to the investor.  Rather than replicate an arbitrarily chosen fund, the team's approach  was based on finding a strategy that has a good chance of identifying a set of outperforming funds, in the next month, quarter or even longer period.  They created an index, called the BRFFI, based on this set of funds, which was chosen from among those using the S&P 500 index as a benchmark. They then determined a weighted mix of passive investment vehicles that would replicate the returns of this index.

The team, consisting of Sharath Alampur, Brian Clapp, Kenneth Jim, Andrew Park, Pranav Sawjiany and Sharat Shekar, analyzed four strategies aimed at selecting funds that would outperform the S&P 500 index and an average of selected mutual funds in the next period.  Three of the four, suitably adapted, did a sufficiently good job that the team recommended using a replicating portfolio derived by risk-weighting each of the three strategies (one assuming that success persists, another that funds with a large difference between short-term returns and mid-term returns have success going forward, and a third with a high Sharpe Ratio - a measure of return per unit of risk).   

The team was able to develop a replication that had a correlation of .9086 with BRFFI and beat the average mutual fund performance, net of fees, by 3.4% per year. The Sharpe Ratio for the BRFFI replication was about twice the Sharpe Ratio of the S&P 500 in the same period of backtesting.  The replication outperformed the S&P 500 by an average of 4.4% per year, including all costs as well as a hypothetical management fee, but with only 0.7% more volatility (risk) than the S&P 500.