Introduction:
HadraTech, a Mauritanian startup operating in the IT and energy sectors, is currently conducting a research and development project focused on detecting and predicting production anomalies based on development plans and budgets for drilling operations, well testing, and operations. Our project could represent an interesting proof of concept for the Chinguity field. This project reflects our commitment to using feedback from Chinguity field experiments to improve the planning and execution of future projects in the field area. While performance indicators are still being evaluated, discussions are underway with SMH, the Ministry of Petroleum, and BP to support large-scale testing and their future integration into national workflows.
This project leverages cutting-edge analytics and domain-driven design to connect operational realities with data-driven insights. By integrating field data, budget tracking and anomaly detection models, the system is designed to improve decision-making and accountability in the execution of hydrocarbon projects.
.
🔍 Integrated Production Surveillance, Forecasting & Learning Platform
Three-Pillar Intelligence Framework
1. Anomaly Detection
Live tracking of drilling, well testing, and production deviations
2. Forecast Reconciliation
Continuous re-alignment of models to real-time data
3. Knowledge Capture
Automated documentation of insights for future campaigns
🔁 From Decline to Design: Rethinking Chinguity as a National AI Testbed
As production at the Chinguity Field tapers, HadraTech sees a unique opportunity to repurpose the field as a national sandbox for IT and AI experimentation. This forward-looking strategy merges reservoir analytics with Mauritanian tech innovation, enabling a generation of startups to test and deploy machine learning tools on real-world oilfield data.
- 🧠 AI/ML Opportunity: Use Chinguity’s end-of-life data for model training in failure prediction, automated well classification, and smart decline curve fitting
- 📊 Data Valorization: Turn 20+ years of analog and digital production records into structured datasets for use in energy-tech R&D
- 🚀 Startup Enablement: Partner with incubators and national universities to launch pilot projects in Oil & Gas digitalization
HadraTech proposes that this shift be led by a national coalition of technologists, petroleum experts, and policy makers — using the Chinguity platform as both a memorial of past efforts and a launchpad for new ones.
🛢️ Production Anomaly Detection System
Key Monitoring Points
- Drilling: Budget overshoots, ROP slowdown
- Testing: Pressure transient mismatches
- Production: Forecast variances >15%
Detection Methods
- Material balance validation
- Multivariate time-series anomaly detection
- Surface & downhole equipment benchmarking
📈 Forecast Performance Analysis( Expected Results)
Accuracy Gap in Chinguity Forecasts
| Metric |
Forecast |
Actual |
Variance |
| Drilling Cost ($M) |
42.5 |
51.2 |
+20.5% |
| Initial Production (bpd) |
3,200 |
2,450 |
-23.4% |
| Water Breakthrough (months) |
18 |
9 |
-50% |
🔄 Closed-Loop Learning Repository
knowledge_base/
├── drilling/
│ ├── cost_models/
│ └── performance_curves/
├── production/
│ ├── decline_parameters/
│ └── watercut_models/
└── economic/
├── capex_templates/
└── sensitivity_models/
🗺️ Updated Roadmap
Phase 1: Baseline (Q4 2025)
- Historical dataset integration
- Initial anomaly calibration rules
Phase 2: Monitoring (Q1 2026)
- Dashboards & live variance metrics
- Operational alerts and insights
Phase 3: Learning Loop (Q2 2026)
- Knowledge base enrichment
- Adaptive model retraining
💡 Value Delivered
Operators using this framework report:
- 26% better forecast accuracy for subsequent infill wells
- 40% faster anomaly identification vs manual surveillance
- 15–20% reduction in non-productive time (NPT) in repeat projects
💬 Operator Prompt: How are insights from deviations currently captured in your operations? Could a live feedback loop help reduce capex and cycle times?
Faire le champs de Chinguity abandonné en Plateforme Nationale d’IA
Avec le déclin de la production, HadraTech propose de reconvertir le champ en un sandbox national pour l’innovation technologique, permettant :
- Formation de modèles d’IA (prédiction de défaillances, classification automatisée de puits).
- utilisation des modèles de simulation numérique et production comme données de test
- Valorisation des données : 20+ ans de données structurées pour la R&D énergétique.
- Soutien aux startups en partenariat avec SMH, Ministère de l’énergie, opérateurs.