Josh Jacobs, LGO Director of Operations and Partner Integration
An unexpectedly complex modeling challenge during an LGO internship with LGO industry partner National Grid wound up benefiting the company and the researchers.
The prototype model that Sid Balwani (LGO '12) created with his advisor, Professor Georgia Perakis, and Joline Uichanco, a Ph.D. candidate in operations research at the MIT Operations Research Center, is being piloted to help schedule repair crews at National Grid yards in New England. And the technical innovations that the MIT team put into place to solve National Grid's challenges have been recognized by the acceptance of their paper at two major operations conferences this year.
Balwani was the first LGO student to do his internship at National Grid after the company joined LGO as an industry partner in 2010. For Mallik Angalakudati, National Grid's Director of Operations Performance for U.S. Gas Distribution, the LGO project was a chance to optimize how the company deploys its repair teams and resources at their local service yards.
"We face significant uncertainty in terms of work that needs to be completed on a given day as well as the actual field capacity to complete the work," said Angalakudati. "Our 2011 LGO project with Sid was aimed at developing a stochastic optimization resource model to assist in daily operations at the yard level." Balwani conducted the bulk of his research at National Grid's Waltham headquarters, with the pilot effort for the resource model targeted at the Beverly yard.
As the project started, Balwani, Perakis and Uichanco realized that the modeling effort was more complex than anyone had realized, requiring additional technical skills beyond those expected of LGO students. The close-working team MIT team began to meet frequently with each other and their National Grid collaborators, and they were ultimately able to combine their skills to address the company's challenges.
As Perakis noted, this teamwork leveraged the capacities of both Balwani and Uichanco. "Sid came to LGO from Intel and is strong in database and software design, while Joline has the modeling skills required for the huge regression and optimization analysis that was needed for this project," she said.
The result of the team's hard work was a prototype model that Angalakudati expects will eventually be rolled out across National Grid's gas business and potentially their electric business. The model was, he says, "a deciding factor in our desire to fund two LGO projects in 2012." For the MIT researchers, the modeling innovations they devised to solve National Grid's challenges led to a paper that will be presented at this year's MSOM Conference at Columbia University and the INFORMS Conference in Phoenix, two of the most important gatherings of operations researchers in the world. Perakis said the team also plans to submit this work to the flagship journal in their field.
Perakis, the MIT Sloan faculty co-director for the LGO program, is working on another LGO internship (in this case with Spanish fashion company Zara) that involves another of her Ph.D. students. She is optimistic that involving Ph.D. students with LGO projects will become a model that will benefit all the participants.
"The Ph.D. students get real-world experience that they see as valuable for their careers, and as researchers, we are encouraged by the fact that this first LGO project has already led to significant conference presentations," said Perakis. "The LGO students involved in these projects, as well as their sponsoring companies, can rely on a more robust research team that will generate more comprehensive solutions to the internship challenges."