(Monica Gaston, Niki Holland, Jana Lindley, Davis Peden, Jeff West, 2001)
As a rapidly growing electrical engineering firm, TestChip Technologies, Inc. felt that it was essential to find a more productive way to analyze and assign tasks to engineers. As client demand and their workforce grow, TestChip must have a program that will assign engineers to tasks systematically, rather than haphazardly. TestChip is also interested in a program that will forecast project length. This senior design project analyzes TestChip's current operations and ultimately provides TestChip with software that will fulfill these goals.
Excel spreadsheets are TestChip's main form of data organization. They currently track all projects, employee information and employee productivity in spreadsheets. Engineer job assignment is random. Although an engineer might not be the best one to perform a job, he or she will be assigned to the job, if he/she is available. As TestChip's workforce grows, they will no longer be able to rely on this nonsystematic assignment.
Another problem arises when TestChip needs to forecast the length of a given project. Currently, TestChip can only use intuition to forecast a project's length. After analyzing these current operations, two approaches were discussed to solve TestChip's problems.
The first approach to optimize Testchip's job assignment schedule was to use an integer linear program. Numerous problems were found when evaluating this approach. The huge number of variables that would be required to solve the problem would cause the program to take hours, possibly days to find an optimal solution. TestChip would also have to purchase a large, expensive super-computer as well as linear programming software that could solve a problem of this size. Clearly, this approach would not work.
A second approach is a modified job shop model. A program could be written to assign each job of each project to an engineer who is skilled to perform the job. This program could then calculate the expected completion date of any project after assigning all of the jobs to engineers. In this way, TestChip could forecast project lengths. The senior design group formulated two different algorithms: shortest processing first and longest processing first. Both of these algorithms were coded by the Java-gurus of our group and were analyzed and discussed to decide the practicality of each of the programs as we analyzed the algorithms.
After analyzing the algorithms, we found that the shortest processing first program provided a slightly superior solution. However, we feel that with the limited amount of project data that was supplied to us, we are unable to thoroughly analyze which algorithm is superior.
It is our conclusion that the optimization program and the Access database structure will undoubtedly benefit TestChip in the future. The following pages provide detailed explanations of the project, the solution, and the results.