ORPMI Platform

Operational Reliability & Predictive Maintenance Intelligence

Predicts equipment failure 30 days ahead — ISO 14224 aligned, deployed on Docker and Streamlit Cloud.

🔗 Live Demo: orpmi-platform-ninahenchy.streamlit.app

Production-grade industrial analytics platform for Oil & Gas production facility operations, aligned to ISO 14224 reliability data standards.

---

Model Performance

Metric	Value	Notes
ROC-AUC	0.9381	Temporally validated — train Jan–Sep, test Oct–Dec
Recall	0.294	Threshold set at 0.4 — optimised for PdM cost structure
F1 Score	0.454	Balanced against false alarm rate
Features	80 engineered	Vibration, temperature, pressure, efficiency, maintenance history
Validation	Temporal split	No data leakage — mirrors production deployment reality
Tests	76 / 76 passing	Full automated test suite

Why Recall 0.294? In predictive maintenance, the cost of a missed failure (unplanned downtime) far exceeds the cost of a false alarm (unnecessary inspection). The decision threshold is set at 0.4 — a business decision, not a statistical one.

---

Dashboard Preview

---

Platform Overview

The ORPMI Platform transforms fragmented operational data into decision-ready intelligence across four capability layers:

Layer	Capability	Audience
Data Foundation	ISO 14224 asset database, ETL pipeline, 29-check validation	Data Engineer
Reliability Analytics	MTBF, MTTR, availability, downtime cost, maintenance compliance	Reliability Engineer
Predictive Intelligence	ML failure probability (ROC-AUC 0.9381), risk scoring, maintenance recommendations	Maintenance Superintendent
Executive Reporting	Fleet KPIs, AI narratives, financial impact, operational risk register	Operations Director

---

Architecture

Data Entry (Streamlit Forms)
        │
        ▼
  SQLite Database
  (ISO 14224 schema · 7 tables · 29-point validation · 4,700+ records)
        │
        ▼
   ETL Pipeline
   (automated transforms · data quality checks · incremental loading)
        │
        ▼
 Feature Engineering
 (80 features · rolling stats 7/14/30d · MTBF trajectory · vibration slope · degradation rate)
        │
        ▼
  Random Forest Model
  (class_weight=balanced · threshold=0.4 · temporally validated)
        │
        ▼
  10-Page Streamlit Dashboard
  (asset health · failure probability · maintenance recommendations · KPIs · data entry)

---

Platform Pages

Page	Content
Executive Overview	Fleet-level KPIs — OEE, MTBF, MTTR, availability, downtime cost
Asset Health Monitor	Real-time health scores and risk classification per asset
Predictive Maintenance	30-day failure probability gauges and trend charts
Failure Analysis	Historical failure patterns by asset type and failure mode
Maintenance Intelligence	Work order analytics and PM effectiveness tracking
Reliability KPIs	MTBF, MTTR, OEE trends over time by asset and department
Cost Analytics	Downtime cost tracking and maintenance cost breakdown
Work Order Management	Open, in-progress, and completed work order pipeline
ISO 14224 Explorer	Equipment taxonomy browser aligned to standard
Data Entry	Live operational data input — writes directly to database

---

Standards Alignment

Standard	Application
ISO 14224	Equipment taxonomy, failure mode classification, 7-table relational data model
ISO 10816	Vibration severity zone classification (A/B/C/D) embedded in feature engineering
ISO 45001	HSE management system context — see also HSEI and HSIP platforms

---

Assets Monitored

6 critical asset types across the OPC-Alpha simulated offshore facility:

Pumps · Compressors · Heat Exchangers · Tanks · Vessels · Separators

---

Tech Stack

Python 3.11    Scikit-Learn    Pandas    NumPy
SQLite         SQLAlchemy      Plotly    Streamlit
Docker         Git             pytest

---

Run Locally

# Clone the repository
git clone https://github.com/NinaHenchy/orpmi-platform
cd orpmi-platform

# Run with Docker (recommended)
docker-compose up --build
# Open http://localhost:8502

# Or run directly
pip install -r requirements.txt
streamlit run dashboards/app.py

The platform initialises its own database and runs ETL automatically on first launch. No manual setup required.

---

Project Structure

orpmi-platform/
├── dashboards/
│   ├── app.py                    # Main Streamlit application
│   ├── components/               # Reusable UI components and theme
│   └── pages/                    # Individual dashboard pages (p1–p10)
├── database/
│   ├── schemas/                  # ISO 14224-aligned SQL schema
│   └── db_connection.py          # Database connection management
├── etl/
│   ├── extractors/               # Synthetic data generation
│   └── loaders/                  # Database loading pipeline
├── models/
│   ├── predictor.py              # Random Forest training and inference
│   └── artifacts/                # Serialised model files (.pkl)
├── tests/                        # 76 automated tests
├── scripts/                      # Setup and training scripts
├── requirements.txt
└── docker-compose.yml

---

Live Demo

🔗 orpmi-platform-ninahenchy.streamlit.app

The platform initialises its database and runs ETL automatically on first load. Allow up to 60 seconds on first visit for the database to populate.

---

Related Platforms — OPC-Alpha Analytics Suite

All three platforms run on the same simulated offshore production facility (OPC-Alpha) and share a consistent data architecture.

Platform	Focus	Standards	Tests	Live Demo
HSEI	HSE Incident Analytics & Process Safety Intelligence	API RP 754 · ISO 45001 · NUPRC · NOSDRA	29 ✅	Launch
HSIP	Safety Culture & LTI Prediction	ISO 45001 · Safety Culture Index (SCI)	19 ✅	Launch

---

Author

Nnenna Henchard — Reliability Data Scientist · 15 years O&G Operations & HSE