Physics-Based Failure Prognostics for Preventing Pressure-System Degradation and Unplanned Industrial Shutdowns

Abstract

This study investigates the continuing industrial problem of pressure-system degradation and unplanned shutdowns arising when corrosion, fatigue, cracking, creep, wall-thickness loss, and other deterioration mechanisms are not identified, modeled, and translated into timely maintenance interventions. The purpose of the research is to determine how physics-based failure prognostic capabilities contribute to improved pressure-system integrity and shutdown prevention within industrial enterprise environments. A quantitative, cross-sectional, explanatory case-based design was employed using purposive sampling and a structured five-point Likert questionnaire administered to technically qualified professionals with direct experience in industrial pressure systems. The study distributed 320 questionnaires, received 298 responses, and retained 286 valid cases for analysis, representing an effective response rate of 89.4%. The principal independent variables were Physics-Based Degradation Modeling Capability (PBDMC), Sensor-Integrated Condition State Estimation (SICSE), Remaining Useful Life and Failure Threshold Prediction (RULFTP), and Prognostic-Guided Maintenance Decision Integration (PGMDI), while Pressure-System Degradation and Unplanned Shutdown Prevention (PSDUSP) constituted the dependent variable. Descriptive statistics, reliability and validity testing, Pearson correlation analysis, and multiple linear regression were applied in IBM SPSS, with statistical significance evaluated at p < .05. The results showed a high mean for PSDUSP (M = 4.20, SD = 0.52), while PGMDI, SICSE, RULFTP, and PBDMC recorded means of 4.18, 4.14, 4.12, and 4.08, respectively. All predictors were positively correlated with PSDUSP, including PGMDI (r = .77), SICSE (r = .74), RULFTP (r = .72), and PBDMC (r = .68), all p < .001. The integrated regression model explained 72.4% of the variance in PSDUSP (R² = .724, adjusted R² = .720, F (4,281) = 184.30, p < .001). PGMDI produced the strongest standardized effect (β = .34), followed by SICSE (β = .28), RULFTP (β = .25), and PBDMC (β = .19), with all four hypotheses supported at p < .001. These findings imply that preventing degradation-related shutdowns requires an integrated reliability-oriented approach in which physical deterioration modeling, sensor-based condition assessment, failure-threshold forecasting, and prognostic-guided maintenance decisions operate collectively to support earlier intervention, stronger asset integrity, and improved operational continuity.