ADDRESSING INDEPENDENT NATURAL VARIABLES FOR PRODUCTION MODELING - A CASE STUDY INVOLVING LIGHTNING WARNINGS

摘要

This paper focuses on the man-nature interface -specifically, halting production operations when a thunderstorm approaches by relying on signals from weather monitoring equipment. Consequently, production managers may miss the delivery schedule when the clouds line up wrong. This paper explores two approaches to mitigating schedule impacts. The first calculates a buffer by month that can be incorporated as an expected loss into a production schedule. The effectiveness becomes constrained as month-end approaches. The second approach involves a regression analysis of 7-day weather forecasts versus observed warnings for a 6-month period to predict downtime. Success would minimize costly overtime pay and quality-of-work/life impacts associated with weekend work. Production manager's natural instincts proved to have a higher accuracy resulting in the model being binned as a fixes-that-fail archetype. The main benefit realized was the conditioning of production managers on interpreting statistical predictions for a common variable — the weather.