Projects for Walmart.com, Xerox, and Canadian National Railway Win Top Materials Handling Awards
A major feature of each year's ORIE ceremony for Masters and Ph.D. graduates is the announcement of the Silent Hoist and Crane Materials Handling awards. This competition, open to all student teams in the College of Engineering, recognizes exceptional project work that promotes the advancement of materials handling technology, whether the materials be physical or information.
The three awards this year all went to Master of Engineering project teams in ORIE. All three projects were included in a diverse slate of projects carried out by students in the Applied Operations Research, Information Technology and Data Analytics concentrations in ORIE's program.
|Giancarlo Soto Del Rosario and Kenneth Chu receive the Andrew W. Schultz, Jr., award from M. Eng. Director Kathryn Caggiano
At the graduation ceremony, M. Eng. Director Kathryn Caggiano also announced recipients of an award, in honor of the late Dean of Engineering Andrew W. Schultz, Jr., that recognizes the most outstanding M. Eng. students as evidenced by high academic achievement, exceptional teamwork, willingness to encourage others, and demonstrated potential to become exemplary professional citizens. This year's award went to Kenneth Yeng-Kang Chu and Giancarlo Soto Del Rosario. Chu, from East Brunswick, NJ, was an undergraduate in ORIE and served as Vice President of the student INFORMS chapter. He will work for Barclays Capital. Soto Del Rosario is from the Dominican Republic, where he did his undergraduate work. He will work for Cervecería Nacional Dominicana, the largest brewer in the Dominican Republic.
Reverse Supply Chain Forecasting for Walmart.com
First place in the Silent Hoist and Crane competition went to a team that investigated the 'reverse supply chain' at Walmart.com that deals with returned merchandise. In order to run its operations efficiently, Walmart.com needs accurate forecasts of the rate at which various kinds of merchandise will be returned. The project team conducted an extensive analysis of return rates and found that a forecasting method due to Charles Holt and his student Peter Winters in the 1950's was the best of those tested, because of its ability to incorporate both variability by season and trends over time. The team calibrated the method so as to minimize the difference between forecast and actual return rates, and made their findings available to operations staff by building a software package that Walmart.com will use regularly in their business.
|Walmart team members with Prof. Jack Muckstadt and M. Eng. Director Caggiano. From left, Aazam Vishram, Aukrit Unahalekhaha, Prof. Muckstadt, Wenfeng Li, Eddie Hsiao, Tian Yu, Xiaole Wang, and M. Eng. Director Caggiano. Professor Peter Frazier, team advisor, was unable to attend the ceremony.
Walmart.com's senior operations manager David Win, a client for the work, commended the team for their innovative and collaborative work. "I am most impressed by how NIMBLE the team was," Win wrote. During weekly teleconferences various scenarios were raised, and the team "listened, assessed, and figured out a way to incorporate the scenarios into the model," which was "a demonstration of their focus on the customer," according to Win. The team also made recommendations for follow-up activities outside the scope of this year's work. The winning team, advised by Assistant Professor Peter Frazier, consisted of Eddie Hsiao, Wenfeng Li, Aukrit Unahalekhaka, Aazam Vishram, Xiaole Wang, and Tian Yu.
Workforce Scheduling for Xerox
In the United States and Canada, Xerox customer service engineers travel to respond to about 10,000 service calls every day. Determining which engineer should respond to each call is a complex scheduling problem, for which Xerox now uses software that takes a minimum of 4 hours to set up and to run a single scenario. An ORIE M.Eng. team won second place in the Silent Hoist and Crane competition by providing scheduling software that is much easier to set up, runs very quickly, and achieves better results than the current software.
|Team members Nathan Holley, Andy Sheung and Yu Zhang receive the prize from M. Eng. Director Caggiano, as team advisor Prof. Muckstadt looks on. Christine Barnett was unable to attend the ceremony.
The team, comprised of Christine Barnett, Nate Holley, Andy Sheung and Yu Zhang, used an array of OR techniques to come up with their solution. They built an optimization model and a rule-based heuristic model to devise schedules, and a simulator to test performance.
The optimization model calculates a best possible schedule when quantities such as service time and travel time, unknown until service is complete, are held at values based on past data. Calculations for the heuristic model, incorporating ideas of client sponsor Xerox Systems Strategy and Technology Manager Andy Huber, are incorporated in the team's scheduling software. The results of the calculations for the two models were then compared to determine the extent to which the heuristic, which takes much less time to compute, falls short of achieving optimality. Finally, the simulator was used to compare the results of the two models with those achieved by the current system, against actual detailed data for three Xerox service areas.
As Huber reported, the final project presentation resembled a boxing match between an established contender (the currently used software) and the student team. "The Cornell team types a few values in a spreadsheet and runs a month of calls in under one second. No special training is needed. Results average 10% better" than the current software. "Cornell wins!" He also commended the team for being "quick to understand the balance between the needs of the employees, the customers and the financial performance of the business. They effectively incorporated a means to incorporate and adjust the balance in the model."
Determining a Forecast Horizon for Canadian National Railway
For the fourth year in a row, a team of ORIE M.Eng. students worked with Canadian National Rail (CN) management and consultants from HCL-Axon on enterprise resource planning software for managing and scheduling railroad crews. This year's project, which placed third in the Silent Hoist and Crane competition, focused on a key parameter in the software, namely the planning horizon, i.e. the time span over which forecast data are used to create schedules.
Forecasts for demand for train crews evolve over time, especially since freight trains are not dispatched on a fixed schedule. Forecasts of the supply of crew workers also varies over time, especially since information is not readily available about people who become available after they finish other assignments. Using too short a planning horizon can result in inferior schedules that do not take adequate account of what is known with reasonable accuracy about future supply and demand. Using too long a horizon can result in inferior schedules based on incomplete data. The team analyzed data extracted from CN's crew planning process and determined that an eight hour planning horizon would be an effective balance between the two extremes. They also presented insights, acquired in the process of their analysis, that might lead to enhancements of the crew planning software.
|Giancarlo Soto Del Rosario and Viral Malnika, representing the CN team, with Professor Peter Jackson and M. Eng. Director Caggiano
Client representatives Marcel Dupere, Jean-Philippe Provencher and Len Podgurny commented favorably on the team's work: "It was impressive, right from the start, to see the ease with which the students understood a complex business like the railroad." They noted that the results "will immediately be used to begin discussion with end users and operational management in order to confirm findings and identify areas of required improvement."
Team members were Ruobing Han, James Hu, Viral Malnika, Giancarlo Soto Del Rosario and Systems Engineering M.Eng. student Tushar Madaan, under the guidance of Professor Peter Jackson.