Failure prediction and analysis techniques
Techniques for implementing Bayesian model for refining and petrochemical wall thickness monitoring are presented. The construction of a model may be automated for each piping circuit or piece of major fixed equipment, utilizing, for example, specific component data, historical thickness measurements, inspection practices and related inspection program information. The model contains nodes describing the most significant sources of variability, namely component original thicknesses, wall thickness degradation over time, corrosion rates and thickness measurement error. Bayesian prior distributions are assigned using readily available inspection program information, including assigned damage mechanisms, inspector and industry experience regarding the expected range of corrosion rates and degree of non-uniform corrosion, thickness monitoring practices, including surface preparation and instrument calibration, thickness data recording practices and component original thicknesses based on applicable industry specifications, typical values for size/component combinations or detailed ultrasonic thickness scanning data generated specifically for this purpose.
1. A method for implementing a hierarchical Bayesian model for pipe wall thickness monitoring, comprising automating construction of the model nodes for a particular equipment, utilizing component data, historical thickness measurements and related inspection information and;
where the hierarchical Bayesian model utilizes an overarching corrosion rate distribution and hierarchal shrinkage to account for the potential for differences in condition monitoring location (CML) rates.
2. The method of claim 1 , with nodes including component original thicknesses, wall thickness loss over time, calibration error and measurement location repeatability error.
3. The method of claim 1 , with a hierarchical Bayesian model prior corrosion rate distribution obtained by transforming the expected mean and maximum rates plus the confidence in this maximum rate for the particular equipment, into a continuous distribution.
4. The method of claim 1 , with a hierarchical Bayesian model prior corrosion rate distribution obtained by transforming the limiting internal thinning damage mechanisms and a measure of corrosion environment severity for the particular equipment, into a continuous distribution.
5. The method of claim 1 , further comprising utilizing a hierarchical Bayesian model prior distributions for calibration error obtained by assessing facility-specific thickness monitoring practices, including surface preparation and instrument calibration.
6. The method of claim 1 , further comprising producing quantitative probability of failure information through estimation of the intersection of a future, projected thickness distribution lower tail with the established minimum thickness value.
7. The method of claim 1 , with time-dependent corrosion rates, utilizing prior distributions that reflect estimates of relative corrosion rate mean and variance since unit startup.
8. The method of claim 1 , further comprising utilizing an overarching corrosion rate distribution and hierarchal shrinkage to account for the potential for differences in condition monitoring location (CML) rates.
9. The method of claim 1 , with grouping by CML corrosion rate clusters, to identify zones of potential localized corrosion.
10. The method of claim 1 , with CML or examination points (EP) grouping consistent with localized corrosion trends frequently experienced, based on a circuit's assigned internal thinning damage mechanisms (DMs), to identify areas of potential localized corrosion.
11. A series of models of the type in claim 1 , which are evaluated to identify the best-fitting model using a leave-one-out cross validation or similar technique, in order to determine the nature of localized corrosion in the specific equipment.
12. The method of claim 1 , further comprising utilizing credible or highest density intervals for historical thicknesses to validate the design of the specific equipment.
13. The method of claim 1 , further comprising employing credible or highest density intervals for historical thicknesses to identify data anomalies or locations warranting follow-up evaluation.
14. The method of claim 1 , further comprising employing credible or highest density intervals for historical thicknesses to identify locations with potential for localized corrosion thereby identifying follow-up evaluation.
15. The method of claim 1 , where posterior corrosion rate distribution mean and/or variance parameters are scaled to account for anticipated, future operating conditions.
16. The method of claim 1 , further comprising utilizing hierarchical Bayesian model credible intervals for probability of failure used to optimize the number of CMLs or EPs.
17. The method of claim 1 , with thickness measurements treated as right-censored data, to accommodate inspection programs where multiple EP thicknesses are taken per CML, yet only the minimum thickness for the CML is recorded.
18. A simulation process utilizing parameter values from the method in claim 1 to quantitatively perform CML optimization, by controlling risk uncertainty as a function of CML and EP quantities.
19. A risk matrix employing risk uncertainty for a specified point in time, based on credible or highest density intervals from the method in claim 1 .
20. The method of claim 1 , wherein the particular equipment includes one or more of a particular piping circuit or piece of major fixed equipment, including towers, drums, storage tanks, heat exchangers, pressure vessels, and fired heaters.