Pipeline risk assessment is a crucial aspect in maintaining the integrity of the pipeline system and ensuring the continued operation of the oil and gas industry. Pipeline failure can result in technical, economic, and reputational losses for a company. Therefore, a risk assessment method capable of accurately and data-drivenly describing pipeline conditions is necessary. This study aims to develop a pipeline risk assessment model using a fuzzy logic-based risk assessment approach utilizing actual inspection data.
The data used in this study include thickness, corrosion rate, remaining life, and the visual condition of the pipeline. The method used is a Mamdani-type fuzzy inference system with a membership function based on the pipeline's technical condition, a rule base structured based on relationships between variables, and a defuzzification process using the centroid method to generate a Probability of Failure (PoF) value on a scale of 0 to 100. The PoF value is then combined with the Consequence of Failure (CoF) to obtain a risk score.
The results showed that the PoF value was able to represent the condition of the distribution pipeline in stages. High PoF values were generated by segments with low thickness, high corrosion rates, short remaining life, and poor visual condition. The CoF value was 3, derived from four parameters: population safety, affected infrastructure and environment, business loss, and company reputation. The analysis results were categorized as low. Therefore, variations in risk values were primarily influenced by the PoF value. The risk distribution showed that approximately 74.83% of distribution pipeline segments were in the low-risk category, 18.18% in the medium-risk category, and 6.99% in the high-risk category. High-risk segments were identified as critical areas requiring priority monitoring, inspection, maintenance, and prevention.
This research demonstrated that the fuzzy logic approach yields more flexible and realistic risk assessments than conventional methods and can be used as a basis for more effective, data-driven technical decision-making in the management and maintenance of distribution pipelines.