Estimating Bending Strain at Geohazard Locations using Historical Inertial Measurement Unit Data

Thursday - January 28 | 11:00 · 5.2 Geohazards 3

Abrar Islam1, Mathew Bussiere1, Onyekachi Ndubuaku2, Nader Yoosef-Ghodsi2

1C-FER Technologies, Edmonton, Canada, 2Enbridge, Edmonton, Canada

Operators receive reports with numerous bending strain features from an IMU inspection run and some of the reported features may be located within flagged geohazard sites. Features reported at these locations must be reviewed to determine whether they may evolve into an integrity concern and whether the features warrant closer examination. For this purpose, operators can use bending strain measurements from IMU inspection logs to estimate strain demand. However, forecasting bending strain using past inspection data involves numerous assumptions, and quantifying the associated uncertainty can be challenging. Enbridge maintains two key datasets: (1) a geohazard database, which catalogs geohazard locations and severity ratings, and (2) a site-wide ILI database containing vendor-reported bending strain features, their associated strain values, and geospatial coordinates. This study leverages data from both mentioned databases to explore a probabilistic data-driven approach for estimating future strain at vendor-reported features. The approach quantitatively accounts for two major sources of uncertainty: variability in reported IMU strain measurements and uncertainty in the strain projection methodology. The first step involved extracting a dataset of reported bending strain features from multiple inspections at flagged geohazard sites by mapping between the two databases. Then future bending strain was predicted using a Monte Carlo simulation framework incorporating simple linear and weighted linear regression. The weighted regression approach used an age-weighted function to increase inspection weight based on recency. Results showed that linear approximations provided reasonably accurate predictions, with the age-weighted approach providing improved accuracy. The results indicate that the strain estimation methodology and the associated data mapping procedure can be included in the operator's geohazard management workflow as an additional tool to help screen vendor-reported strain features and flagged geohazard sites.

Read/Download PaperGradeAvailable after the event