M. Eng. project teams are working with old and new clients
Since the end of Cornell’s spring break, ORIE’s Master of Engineering (M.Eng.) students are working especially hard to complete their projects before the semester ends. Students in the Financial Engineering concentration will work on their projects next fall, at Cornell Financial Engineering Manhattan in New York City.
This year’s Ithaca project slate has brought several earlier clients back for more. They are hospital services provider Weill Cornell Medical College, polyurethane foam products manufacturer FXI, medical transport organization Ornge, industrial filter manufacturer Pall Trinity Micro, online retailer Walmart.com, Canadian National Railway (CN), and accounting firm Ernst & Young.
Three new project sponsors have added variety to the project slate:
- The Houston Rockets and Toyota Center organization has asked an M.Eng. team to forecast demand for Rockets basketball games as a function of such factors as day of week, time of year, depth of season and win percentages, with a view towards helping the team optimize ticket sale prices. M.Eng. team member Erika Hopper reports that it has been challenging to develop accurate predictions of ticket demand as a function of many variables, but “it is gratifying to see how excited our client contact is with the analysis we have done so far.” The project is being advised by Professor Chris Anderson of Cornell’s School of Hotel Management, where he teaches data mining techniques.
- General Electric’s Energy Management business wanted a team to use their electricity production simulation software as the core of an approach to optimizing generation capacity in New York State. GE’s software is used by most New York State utilities and the New York Independent Systems Operator in planning the operation of the state’s power grid. The team is using the software to determine an optimal profile of generating capabilities, taking into account the integration of renewable sources such as wind and solar. According to team member Charles Hernandez, determining an effective approach to this optimization problem is a challenge, since each candidate profile takes hours to evaluate using GE’s software.
- Servigistics provides companies with software to manage the servicing of products throughout their life cycle, during which time spare parts may no longer be manufactured. Servigistics has presented a team with the task of analyzing alternative approaches to forecasting the demand for parts needed to service warranty claims that will arise after the parts are no longer made. This will help their service business customers make a wise “Last Time Buy” to avoid breaching the warranty. The team is making use of research with Servigistics by their advisor, Professor Peter Jackson, which showed that future demand for service parts can be based on the demand characteristics of clusters of similar parts. This approach overcomes the scarcity of data about the number of parts already installed under warranty at when the Last Time Buy must be made.
Weill Cornell Medical College
Two teams are working with Weill Cornell Medical College to follow up on last year’s successful effort that found ways to increase utilization and efficiency of urological surgery suites at New York-Presbyterian Hospital. One team will assist in the implementation of the earlier recommendations, while the other will adapt the approach used for the urology suites to endoscopy surgery suites at the hospital.
Last year’s urology team used simulation and other methods to find ways to increase the effective capacity of a three room operating suite at the hospital. This year’s follow-on team is working closely with hospital personnel to put the resulting recommendations into action. In so doing they are immersed in the human side of operations research. They frequently “scrub up” to observe procedures and processes, interview staff, and gather data.
"What makes this project really rewarding is the fact that we are working with not one, but multiple clients at once," said Weilin Meng. "It is fascinating seeing how different perspectives and objectives are. The great challenge is figuring out how to resolve these differences and find a common ground that everybody can agree on."
The team working with the nurses, doctors, patients, anesthesiologists and schedulers associated with the endoscopy operating suite is also enjoying the challenge of taking different viewpoints into account, according to team member Chelsea Feldman. The team is developing a simulation and is preparing recommendations to reduce patient time in the hospital, increase the number of procedures that are performed, and reduce overtime. Fortunately they are finding that that the hospital staff members “have been extremely helpful, and are very excited to see our findings,” Feldman said.
This year’s team working with Polyurethane foam manufacturer FXI is responding to the classic question of determining which of the company’s many products should be produced at which of their 18 plants to be shipped to which customers. The team has collected extensive data on production costs, yields and capacity as well as on shipping costs, which can be high because charges are based on volume, not weight. Their work is benefiting from the results of a project done last year to determine the impact of lot sizes on production costs.
The team has built a spreadsheet model to compute the optimal allocation of production requirements to plants. The model may point the way to consolidate the production of certain products at a smaller number of plants, if it shows that the savings in material costs justify the purchase of machinery to compress the foam products, which would bring shipping costs down.
“We have tried to communicate with FXI as much as possible since we, the students, do not understand the field as much as they do,” said team member Jung Ryul Kim. “Having the people working in the actual industry who find our early results valuable has been truly gratifying for us,” he added, “and might have raised a bar of expectation for us to deliver much more interesting stuff for them later on.”
This Canadian medical transport company is one of ORIE’s long term clients, with completed projects stretching back several years. This year’s project team is analyzing the base locations for the company’s fleet of fixed and rotary wing aircraft, examining for the first time urgent requests for fixed-wing transport. They are looking at two years’ worth of data to determine where aircraft should be based to handle urgent calls, and whether Ornge should consider redistributing aircraft during the day to better match demand.
The project entails a large mixed-integer linear programming formulation that is being solved with two different criteria – minimizing the total miles travelled, and minimizing response time – as well as a mix of the two criteria.
“We are considering more aspects of urgent requests than previous projects did,” said team member Mike Ingersoll, seen at near right with team members Katrina Mehringer and Jun Yang Chua during an audio conference with clients in Toronto. “We are taking variation in the availability of aircraft over the course of the day into account,” he said.
Examining the data in detail, they have found for example that aircraft are less available at some bases in the late afternoon because of demand earlier in the day. In addition to data analysis and optimization techniques, the team is also carrying out simulations of operations under different approaches to allocating aircraft to bases in order to confirm their predictions.
Pall Trinity Micro
In a plant near Ithaca, this affiliate of Pall Corporation manufactures devices, called coalescers, to separate immiscible liquids from a gas stream. The M.Eng. team is working to increase throughput of the cell that makes these industrial filters.
“It has been gratifying to see that the lessons we learned in our classes are directly applicable to real life,” said team member Drew Weirman, seen at left. The team has found it challenging to schedule plant visits, but “once we get in there, it’s just a matter of being observant and inquisitive, and lots of area for improvement just seem to pop up in front of your eyes. Many common industrial engineering concepts have come in handy,” Weirman said.
Some products offered by Walmart.com cannot be handled by conventional shipping methods because size or configuration prevents moving them on conveyor belts at the shipper’s facility. These products require special handling that adds cost. However some products that pass through special distribution centers can actually be handled by conventional methods.
This year’s Walmart team is working on an approach to identify characteristics of products that are unnecessarily sent to the special centers. They have been given access to data characterizing thousands of items that have been erroneously characterized as unsuitable to be moved on conveyor belts and are working on a spreadsheet model to predict whether an item with given characteristics should be considered for special handling.
Canadian National Railway (CN)
This is the sixth year that a team of Cornell M.Eng. students have worked to enhance the software used by CN to manage and schedule work crews on the freight railroad. The team and clients are seen at right with their advisor, Professor Peter Jackson.
Past projects have dealt with forecasting crew vacancies, determining an appropriate planning horizon, and predicting the excess of crew supply over crew demand.
The current project is improving the precision and accuracy of the forecasting model produced by earlier teams, and enhancing it to allow users to run ‘what if’ changes to assignments and to do so through an effective and intuitive user interface. The team is also using statistical methods to analyze the history of actual schedules in order to discover how crew supply relates to train delays.
Ernst and Young
As a global financial services company, Ernst & Young offers a variety of services to client companies on a fee basis. Following upon a project last year that focused on identifying future revenue from clients, this year’s M.Eng. project is using statistical methods to understand the margin obtainable from a service engagement. (Margin is defined as the difference between payments received and direct costs incurred, without regard to the allocation of indirect costs that are beyond control by the executive who manages the engagement).
According to team member Louis Segalini, the team is “working to identify the key factors driving the margin for service engagements and how their impact differs across market segments, account types, services and industries.” Segalini reports that they are assessing the extent of variation in margin within groups of similar engagements, and what this may imply in terms of appropriate goals for margin on specific engagements. They hope rigorous statistical analysis will answer questions their client has about its business model, including whether there is a relationship between engagement margin and repeat engagements.