PREDICTIVE ANALYSIS OF WORK INCIDENT TROUGH SAFETY BEHAVIOR AND SAFETY CAPACITY

• Main Author: Adi Winarto, Ponco
• Format: Theses
• Language: Indonesia
• Subjects: Manajemen umum
• Online Access: https://digilib.itb.ac.id/gdl/view/47029
• Institution: Institut Teknologi Bandung

Oil and gas energy still demanding today and keep as the most energy resources to support world activities. Maintaining top safety performance is one of key factor to keep company reputation. In Rokan Petroleum, it this still challenges to achieve incident free operation in the past 5 years, since 2015. The incident rate in Hole and Flow department showed declining trend by year but start to steadily increase in March 2019. Introducing new approach by analyze safety behavior on site expected could understand team safety performance to predict incident or nearmiss event in the future. This research uses some methods to elaborate and find out the relationship among safety behavior observations to forecast the nearmiss-incident case in Hole and Flow department. The incident analysis use associative approach, logistic regression, to test relationship of aggregate difference cumulative safety capacity to nearmiss-incident. It is also use action research method to shape the best model as ideal condition which has high trustworthiness. Further, this research develop probability logistic regression model to predict the nearmiss-incident event in the following time with consider gap of the safety capacity. Time series forecasting, weighted moving average, is used to predict the number cumulative safety capacity difference as independent data to predict incident case. By using action research data iteration, this project could define and find a model to represent the best safety capacity baseline. As result, the nearmiss-incident prediction model has high validity with Pr(>|z|) value 0.00868 after 18 weeks implementation. It is believed that the model have strong trustworthiness to predict number of nearmiss-incident case at Hole and Flow operation. The implementation plan of this research is very positive with consider some points such as, not require new safety program, logic, iteration method which allows revisiting the prediction model, participative and do not need cutting-edge technology. Introducing new approach to predict possibility of nearmiss-incident event will increase awareness and sense of vigilance Hole and Flow department from incident. Further, address appropriate safety program expected could prevent the incident in operation.