Wellbore Genius
Performance modeling
SURF crystalline — DFIT closure vs strain
South Dakota (SURF, Lead) · DOE SURF / EGS Collab · Guglielmi 2023 · IJRMMS · 10.1016/j.ijrmms.2023.105521
First author: Guglielmi, Y.
DOI: 10.1016/j.ijrmms.2023.105521
No digitized data cached yet. “Auto-fetch” asks the AI to read the open-access figure(s) and return (x, y) samples for each measured series. Review before saving as ground truth — model output may be noisy and should be cross-checked against the paper.
| Series | n | RMSE | MAE | Bias | MAPE % |
|---|---|---|---|---|---|
| BHP falloff | 20 | 328.87 | 328.35 | 328.35 | 5.10 |
Composite 83/100 — passes blind match.
- 01Gate identified — DFIT closure
Dispatcher routed entry "guglielmi-2023-dfit-closure-strain" to the dfit-closure forward evaluator.
- 02Reference inputs check
All required `referenceInputs` (and `extra` knobs, if any) resolved.
- 03Measured series check
1 series, 20 samples evaluated.
- 04Fit — BHP falloff
RMSE within 10 % of series mean — tracks measured curve.
n: 20RMSE: 328.868MAE: 328.345bias: 328.345MAPE: 5.1 %maxAbs: 354.169
No saved runs yet. Click Save run after a rerun to snapshot the score, per-series metrics, and current rerun inputs (unit overrides, preview values, preset). Snapshots are scoped to the active project.
| Factor | Spearman ρ | Score swing | Sign flip? |
|---|
±25% half-range, 7 samples per factor. Score = exp(-3.6 · rel-RMSE) × 100.
Forward-run only — no post-hoc tuning. Digitized series carry the listed reading error. DOI links to publisher; data reuse subject to publisher license.