An Intelligent Agent To Optimize and Control Inspection Programs: A Case Study on Data-Driven Integrity Analysis

Credit: Martin Lindland/Equinor.
The Gjøa platform in the North Sea.

A method of data-driven risk analysis has been developed and applied to improve the efficiency and effectiveness of inspecting surface corrosion of SS316 piping on the Gjøa platform on the Norwegian Continental Shelf. Using a probabilistic statistics method, in conjunction with a new digitized tool for inspection, reporting, and evaluation, inspection data on external pitting and crevice corrosion on each line were analyzed and used as the basis for estimating corrosion development rates to be expected for the coming years. The method uses advanced software and interative computation to calculate the risk of leakage of each line, which, in combination with the criticality of the pipe, provides a target schedule for the next inspection of each line. In effect, the data-driven model is an intelligent agent that generates a statistically based inspection program used to manage the risks of line failure while inspecting lines only when they need to be inspected.

In the case study presented here, the asset owner was able to implement a schedule to reduce SS316 piping inspection by 70% over 4 years compared with a conventional risk-based inspection calling for more frequent inspection of the SS316 lines.

The Gjøa platform platform is a semisubmersible floating production facility located approximately 70 km from the Troll field. Operated by Engie, the platform came on stream in 2010. The SS316 topside piping includes a total of 1,957 lines, with a combined length of 27 km, and as various utility systems for the platform carrying water, compressed air, and hydraulic fluids, typically at low pressures and moderate temperatures. A review of visual-inspection data collected up to 2015 documented that external pitting and crevice corrosion were present on one out of four of the SS316 lines. Development of external pitting and crevice corrosion is notoriously difficult to understand and predict, and numerous publications exist that describe the phenomena and recommending actions to mitigate the problems.  Faced with the prospect of inspecting several kilometers of piping every year, the asset owners decided to use a statistical method to assess risk and perform more targeted inspections.

Data-Driven Optimized Risk Management
Optimized risk management aims to provide operators with a risk-management strategy that is as efficient as possible. This is done by supplementing exisiting methods for risk management—such as reliability-centered maintenance (RCM) and risked-based inspection (RBI)—with statistical methods using available data and optimizing activity scheduling to achieve the most-efficient activity schedules possible. This approach removes three major causes of underperformance in maintenance and inspection scheduling: imprecise probability of failure estimates, underutilization of data, and lack of differentiating in scheduling.

RCM considers each piece of equipment to be inspected or maintained along with the functions of the equipment, performance standards, functional failures, failure modes, and the effects and consequences of failures, as well as activities to predict and prevent failures. The RCM process also maps relationships between these factors.

RBI also attaches a similar variety of failure-related factors to each piece of equipment, including deterioration mechanisms, and most often is concerned with loss of containment failures.

Several of the deterioration mechanisms are in some way rate-based, in the sense that there is an onset of a mechanism followed by a detectable increase in the deterioration caused by the mechanism before failures occur. This indicates that modeling the rates can be useful, and the historical data available from an installation often supports such modeling. The rate does not need to be assumed to be constant; it can be piecewise linear with unpredictable rate changes or move in jumps.

After examing a large number of maintenance and inspection schedules from several operators, the authors concluded that, in practice, only a weak correlation exists between the risk posed by a failure mechanism and the effort spent to mitigate that risk. This poor correlation is the result of imprecise probability of failure estimates, as well as the inability to use historical data to adjust the effort level and to produce sufficiently differentiated plans.

All of this can be addressed with a data-driven optimization approach to scheduling activities. The task is, as precisely as possible, to schedule risk-reducing activities with respect to adhering to an established risk threshold or to acheiving the lowest possible life cycle cost.

The data-driven optimization approach is used to calculate the best possible estimate of probability for failure for one or several pieces of equipment based on historical plant and equipment specific data (e.g., for each SS316 line on the Gjøa platform). Then, the calculated probability of failure and associated uncertainty are used to calculate the time to the next activity precisely. Because the time to the next activity is calculated, there are no limitations with respect to how many variables can be taken into account or how many schedules can be created. The main objective is to achieve the highest possible precision in determining the timing of activities. The scheduling of an activity is based on keeping risk, as precisely as possible, within acceptable limits, or onminimizing expediture.

Find the paper on the HSE Technical Discipline Page free for a limited time.



Don't miss our latest HSE content, delivered to your inbox twice monthly. Sign up for the HSE Now newsletter.  If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.





HSE Now is a source for news and technical information affecting the health, safety, security, environment, and social responsibility discipline of the upstream oil and gas industry.