Industrial Projects

1996-97 Projects

Robust Statistical Sampling to Improve the Effectiveness and Reliability of On-Line Manufacturing

Sponsor: Tambrands Incorporated
Advisor: Prof. Ming Chen
M.S. Thesis, 1 student involved

This project was concerned with statistical quality control. The advisor of the project, Professor Ming Chen, received a letter from the Quality Engineer of that company on December 20, 1996 stating that the agreement between WPI and the company on this project has been fulfilled. The letter went on to say that "... The combination of bridging the needs of industry with academia, and coming out in a win-win situation, has been very satisfying."

Mathematical Modeling in Metal Processing

Sponsor: Morgan Construction Company
Advisor: Prof. Bogdan Vernescu
Students: Forest Lee-Elkin, William Montbleau, Stanislav Oks

This project focuses on the wear of a mechanism called a Laying Pipe which coils steel rod in rolling mills. A mathematical model for the mechanism's wear is developed through a system of differential equations. Strong correlation is shown with measured data. To improve the wear distribution, an optimal shape problem is numerically solved using minimization techniques and finite difference methods. A new shape is achieved through the improvement of the wear distribution.

Statistical Consulting

Sponsor: Bose Incorporated
Advisor: Prof. Joseph Petruccelli
Student: Jason D. Wilbur

The report summarizes four statistical studies conducted by a leading audio speaker manufacturer. During a three month period, the author served as a consultant to the company. There, he was involved with each of the four studies to varying degrees. Statistical topics discussed in this report include experimental design, data collection, analysis of variance, analysis of covariance, pairwise comparisons of proportions, aliasing structures, confounding patterns, testing for significance, and non-parametric methods of analysis.

Development of a Test Algorithm for Cession Strategies

Sponsor: Premier Insurance Company
Advisor: Prof. Arthur C. Heinricher, Prof. Ansu Bagchi, Prof. Ann Wiedie
Students: Kristen Magnifico, Jeremy Olszewski, Daniele Recore

New methods for computing and predicting the net gain or loss for a certain cession strategy were explored for the Premier Insurance Company of Massachusetts. Various one and two stage strategies were developed and compared using different variables. The methods used and the programs developed can be used by Premier to test future strategies. They can also be used to investigate the sensitivity of the payoff and the strategies to fluctuations in certain parameters, including CAR's deficit.

Control of Simultaneous Motion in Shoulder Prostheses

Sponsor: Liberty Mutual Insurance Group
Advisors: Prof. Looft, F. J. (ECE) and Prof. P.W. Davis
Students: Jason Wening, Nicole Robert, Colleen Fox

This project developed a prototype electronic control unit to simultaneously execute motions in more than one joint of a prosthesis. The first phase established design parameters of the control unit by modeling the movements of a prosthetic arm and predicted the inertial forces acting on a moving prosthesis. The second phase built the motion control system and proved that digital signal processing can be used with analog technology to control simultaneous motions and counteract the predicted inertial forces.

Measures of Credibility

Sponsor: Allmerica Financial
Advisor: Prof. Ansu Bagchi, Prof. Ann Wiedie
Students: Nicole Burns, Richard Colby, Scott Lucarelli, Melissa Pitzen

This project investigated new measures of credibility for Allmerica Financial, a Worcester based insurance company. The current credibility measure has not been updated in the recent past and a critical analysis was required to measure its validity. In particular, the method of predicting future claims was analyzed using three consecutive years of data from the recent past. The project incorporated several new ideas into modifying Allmerica's current practices. These resulted in better predictions for Allmerica's future claims values.

A Bayesian Test for Trend in Case-Fatality Rate

Sponsor: University of Massachusetts Medical Center
Advisor: Prof. Balgodin Nandram
Students: Kathryn Sullivan, Timothy Doherty

The Cochran-Armitage test is frequently used to test for trends among binomial counts. The nonparametric order-restricted test is less common. We construct a Bayesian test for a trend which removes some restrictions associated with the Cochran-Armitage test and the order-restricted test We apply these tests to the Worcester Heart Attack Study to assess decreasing trends in case-fatality rates for patients hospitalized with acute myocardial infraction.

Experience Rating

Sponsor: Aetna UHSC
Advisor: Prof. Ansuman Bagchi
Student: William Cho

Mathematical Software Usability Testing

Sponsors: Macsyma, Incorporated, Wolfram Research (Mathematica), Waterloo Maple, and Softwarehouse (Derive)
Advisors: Prof. Joseph Fehribach and Prof. Ansuman Bagchi
IQP, 3 students involved

This project compared the ease of use of the mathematical software systems Macsyma, Derive, Mathematica, and Maple for first and second year college students.

Maintained by webmaster@wpi.edu
Last modified: Jun 20, 2010, 13:03 UTC