Orion Controls (B)

Abstract

Orion Controls is a leading designer of industrial valve systems. Avion, a chemical company, is interested in a new, improved valve to manage the highly volatile phenol in its production process. Avion and Orion strike a deal for Orion to try to produce these valves. However, there are time constraints, engineering constraints, and software constraints that make the valve-redesign project a risky endeavor. The A case (UVA-QA-0602) consists of a basic decision tree that has been used everywhere from an introductory case in decision analysis to the core material for an exam (with additional complicating factors introduced in questions). The A case is a rewrite of two earlier cases (UVA-QA-0480 and UVA-QA-0481). In those cases, the key issues surround the scheduling of the projects, leading to insights about which to schedule first. Here, the scheduling is fixed and the issues are less calculation-intensive and more interesting. In the A case, there are two options: one the students usually see and one they don't. The former is the abort option that occurs if the software-redesign shortcut is unsuccessful. The one they tend to miss (even if the shortcut is successful) is if the valve improvement is only modest, it is better to deliver the old valve rather than the new one. The B case consists of a quick series of changes in the situation that leads to a totally different decision structure and requirement for simulation modeling (or some other technique to handle the introduction of continuous uncertainties). This case also has a wonderful real option embedded in the problem around which an entire class can be built. Because the software redesign will not occur until after Engineering completes its work, Engineering can only estimate the final performance of the valve. Based on this estimate, however, Orion can decide to abort the project. Because the estimate is continuous, the option problem becomes one of finding the optimal estimated-performance value at which one aborts the software redesign.

Keywords: decision analysis, simulation, decision trees