Two-Stage AI Piping Wall Thinning Prediction From Fluid Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting wall thinning in piping due to flow-accelerated corrosion and liquid droplet impingement erosion are inaccurate and time-consuming, particularly in complex piping systems, due to the difficulty in accurately modeling diverse piping shapes and the need for extensive learning data.
Innovation Solution
A two-stage AI evaluation system is employed, where the first stage uses computational fluid dynamics to generate fluid characteristic values from piping shapes, and the second stage uses limited actual measurements to predict wall thinning and related indices, such as amount, rate, and remaining life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical formulas are used to predict wall thinning, then prediction accuracy can be improved, but calculation time becomes excessively long
Solution Approach 1:
The system pre-calculates and stores fluid characteristic values (flow velocity, turbulence intensity, shear stress) for various piping configurations before actual wall thinning prediction is needed. When predicting wall thinning, the system retrieves these pre-calculated values instead of performing time-consuming CFD simulations, thus maintaining accuracy while dramatically reducing calculation time.
Solution Approach 2:
The system creates a simplified digital model that copies the essential characteristics of complex piping systems. Instead of analyzing every geometric detail of actual piping, the system uses representative models with key parameters (piping shape, fluid conditions) that capture the dominant wall thinning mechanisms, enabling fast predictions without sacrificing critical accuracy.
2Productivity
If AI methods are used for high-speed evaluation, then calculation speed improves, but large amounts of learning data are required
Solution Approach 1:
The system introduces fluid characteristic values as intermediary parameters between piping geometry and wall thinning predictions. Instead of directly training AI on complex piping images and wall thinning outcomes, the system first calculates simplified fluid characteristics (flow velocity, turbulence, shear stress) that mediate the relationship, reducing the dimensionality and complexity of the learning problem.
Solution Approach 2:
The system transforms the wall thinning prediction problem from analyzing complex multi-dimensional piping geometries to analyzing key fluid dynamic parameters (flow velocity, turbulence intensity, shear stress). By changing the parameter space from geometric complexity to fluid dynamic characteristics, the AI model can achieve high-speed evaluation with limited learning data.
3Measurement precision
If wall thickness is measured at a huge number of locations, then wall thinning detection accuracy improves, but inspection time and operational disruption increase
Solution Approach 1:
The system identifies and focuses measurement efforts on specific high-risk locations within piping systems where wall thinning is most likely to occur based on fluid dynamic conditions. Instead of uniform measurement across all piping, the system applies localized assessment to regions with high flow velocity, turbulence, or shear stress, achieving accurate detection with fewer measurements.
Solution Approach 2:
The system divides the piping system into segments based on fluid dynamic characteristics and wall thinning risk. Each segment is evaluated independently using the prediction model, allowing the system to prioritize inspection resources on high-risk segments while reducing or eliminating measurements in low-risk segments, thus reducing total inspection time while maintaining detection accuracy.
Data Source
AI summary
The invention provides a piping wall thinning prediction system, a piping soundness evaluation system, and a method which can evaluate a wall thinning amount, a wall thinning rate, and a remaining life at a high speed based on limited actual measurement results, and identify a location requiring monitoring in a piping in a plant. The wall thinning prediction system includes: a first AI evaluation unit configured to evaluate a fluid characteristic value of a piping based on an input value including piping shape information and information of a fluid in the piping; and a second AI evaluation unit configured to evaluate an index related to wall thinning of the piping based on an input value including the fluid characteristic value. The wall thinning prediction system is configured to predict the index related to the wall thinning of the piping.