Bayesian Model of Risk Informed Scheduling

• Published in: Proceedings. Annual Reliability and Maintainability Symposium pp. 1 - 4
• Main Authors: Berlin, Geoffrey, Franklin, Paul
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.01.2025
• Subjects: Accuracy; Bayes methods; Bayesian Statistics; Delays; Maintenance; Predictive analytics; Rails; Random access memory; Resource management; Risk management; Schedules; Statistical Inference
• ISSN: 2577-0993
• DOI: 10.1109/RAMS48127.2025.10935170

The Metropolitan Atlanta Rapid Transit Authority (MARTA) is presently engaged in an aggressive program to upgrade the appearance and functionality of its heavy rail stations. Often this work is done on the weekends and during the night to minimize the inconvenience to passengers. To shut down a particular section of track requires a six-week advance request to permit the necessary arrangements for flaggers, staging of materials and equipment, and coordination with adjacent stations. Not receiving the requested track allocation can delay projects by many months and thereby increasing project costs. The goal of this effort is to assist the Track Allocation group in assessing of the likelihood that contractors will be prepared to take advantage of their assigned "track time" by using as much information about expected contractor performance as possible. Firstly, this risk assessment reflects the natural variability of construction projects in addition to other circumstances. These circumstances may reflect last minute cancellations of track requests for any number of reasons including unavailability of flaggers in addition to emergency maintenance. Secondly, anticipated contractor performance is included employing the Bayes Rule to improve the risk assessment. The proposed Bayesian analysis combines statistical estimates of anticipated contractor performance for completing preparatory tasks. As these tasks are completed the likelihood of success is updated, and thus, improving the accuracy of the risk assessment and confidence in using allocated track time.