Optimizing the Prioritization of First-Time ILIs Using Quantitative Risk and Machine Learning
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract Inline inspections (ILIs) are one of the most effective methods for managing the integrity of pipelines. However, many older pipelines were not designed to accommodate ILI tools. Pipeline operators often prioritize which pipelines to make inspectable on a risk-basis. While this risk-based approach has many merits, it does not necessarily result in the maximum risk reduction for a given budget as the risk-reduction from completing the inspection is not considered. An optimized prioritization strategy should consider both the uninspected risk and amount of risk reduction. Since post-ILI risk are calculated based on the detected imperfections, it is not possible to directly calculate the risk-reduction from performing a first-time ILI. To overcome this, TC Energy (TCE) completed an exploratory analysis of numerous first-time ILI results to identify key parameters and built machine learning models which predicts the risk impact of performing first-time ILIs. Several machine learning algorithms (neural network, decision tree, etc.) were trained on data from pre and post-ILI risk results from TCE’s quantitative risk assessment. The models were trained at a dynamic segment level and aggregated to an ILI assessment path evaluation. The best-performing machine learning model was selected that accurately predicts the risk reduction achieved from a first-time ILI. These results demonstrate the risk-reduction of a first-time ILI can be accurately predicted before the inspection is performed. Combining the traditional risk-based prioritization approach with the predictive abilities to estimate risk-reduction will allow TCE to optimize the selection of first-time inspections by maximizing the amount of risk reduction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it