Application of Two-Parameter Weibull Distribution for Predictive Maintenance: A Case Study
Stefania Ferrisi, Paolo Cappellari, Rosita Guido, Domenico Umbrello, Giuseppina Ambrogio
Abstract
The maintenance of facilities, machinery, and infrastructure is a fundamental aspect of any organization. The modern approach of Predictive Maintenance (PM) allows for predicting and preventing asset failures. This leads to several advantages, including cost reduction and reduced machine downtime. However, this approach is typically based on sensor data, and a gap in the literature exists concerning the applications of PM to older machines that lack sensors. This paper demonstrates the potential for developing a PM model using statistical techniques that can assist the company’s maintenance department in optimizing maintenance plans even in the absence of sensor data. In particular, the model is developed by applying the Weibull Distribution to define the probability of a failure occurring within a future time period. The historical data used for predictions concern the maintenance records of a real company that manufactures materials and equipment of excellence for the oil and gas sector.