Nnenna Henchard
Industrial AI for O&G Operations

Unplanned equipment failure on offshore production facilities is one of the costliest operational risks in Oil & Gas — driving emergency maintenance, lost production, and regulatory exposure. Maintenance teams typically lack early warning systems that account for the full operational context of each asset type: vibration thresholds, maintenance history, and degradation trajectory together.

ORPMI — a predictive maintenance platform aligned to ISO 14224, using a Random Forest model trained on 80 engineered features across vibration, temperature, pressure, efficiency, and maintenance history. 10-page executive dashboard with asset health monitoring, failure probability gauges, and maintenance recommendations with cost framing.

Temporal validation — train Jan–Sep, test Oct–Dec. Random k-fold leaks future data into training and inflates metrics that will not hold in production.

ISO 14224 schema — mirrors production CMMS architecture so the platform integrates with real systems without translation layers.

Threshold at 0.4 — a business decision. Missing a failure (false negative) costs far more than a false alarm (false positive) in a PdM context.

HSE compliance reporting in Nigerian upstream operations is largely manual — incident data collected on paper, regulatory submissions to NUPRC and NOSDRA compiled at month-end from multiple Excel files, and corrective actions tracked informally with no visibility into close-out rates between audits.

HSEI — an 8-page executive HSE analytics platform with API RP 754 Tier 1–4 classification, ISO 45001 leading indicator framework, and NUPRC and NOSDRA reporting fields embedded as structural database design decisions. Compliance reporting becomes a query, not a manual exercise.

NUPRC and NOSDRA at schema level — not added as afterthoughts but designed into the data model from the start. Every incident is classified on intake.

API RP 754 Tier 1–4 — Tier 1 and 2 events automatically flagged as regulatory reportable, matching the exact workflow used by Nigerian IOC operations teams.

Leading vs lagging separation — the dashboard surfaces observation rates, near-miss frequency, and CAPA close-out rates alongside TRIR and LTIR.

Lost Time Injuries are tracked after they happen. The leading indicators that predict where the next LTI will occur — observation rates, training compliance, CAPA close-out rates — are either not collected systematically or not connected to safety outcome prediction.

HSIP — a safety culture intelligence platform that combines five leading indicators into a composite Safety Culture Index (SCI) and uses a calibrated Gradient Boosting model to predict 30-day LTI probability per department. Validated on 2,288 weekly records across 11 operational departments.

Isotonic calibration — the model's output is presented as a true probability, not just a relative score. A prediction of 0.68 means approximately 68% of departments with this profile had an LTI within 30 days — a number a safety manager can act on.

Evidence-based SCI weights — near-miss reporting rate (25%) and training compliance (25%) weighted highest based on their empirical correlation with LTI rates in the training data.

Department-level granularity — predictions are per department, not per facility, enabling targeted intervention rather than blanket campaigns.