Method for use of in service inspection data in predicting future inspection intervals for seam welded steam pipes

Abstract

The dominant crack initiation and growth mechanisms for reheat steam pipes include creep and creep–fatigue, and the seam welds are particularly vulnerable. With historical data consisting of operating temperatures, pressures, average operating time between starts and stops, etc., and an extensive database of relevant material properties, an accurate model can be created that can be used to assess the risk of fracture in critical components and subsequent inspection strategies that mitigate the risks and lower them to acceptable levels. Physics based model predictions for material degradation combined with service experience based predictive approaches that constantly correct the model predictions using the experience base on specific machines can be used to create an approach that can produce highly reliable predictions. Such an approach requires the capability to solve inverse problems. This paper presents the capabilities of such a hybrid approach using reheat seam welded steam pipes as an example. It is demonstrated how the service database from successive previous inspections are used to reduce the variability encountered in weldment material properties to make more accurate inspection interval predictions.

Keywords:

• ARTIFICIAL NEURAL NETWORK
• CREEP CRACK GROWTH
• WELDED STEAM PIPES
• MATERIAL DEGRADATION
• SERVICE INSPECTION DATA