Oil, Gas and Energy
Where a wrong action has physical consequences
Energy is the sector where the gap between an analytical recommendation and a control action matters most. Advising a human is a different proposition from actuating a valve, and the assessment treats them as different decisions with different gates, even when the underlying model is identical.
Where AI is most often proposed
These are the use-case families the pack recognises. Appearing on this list is not an endorsement — several of them resolve to a no-AI outcome more often than their sponsors expect.
- Predictive maintenance on rotating and static assets
- Production and generation optimisation
- Grid balancing, load and demand forecasting
- Environmental monitoring and emissions reporting
- Inspection support from imagery and sensor data
- Trading analytics and hedging support
What actually constrains deployment here
Safety-critical control
Anything capable of influencing a safety instrumented system, a protection function or a physical control loop meets a blocking gate unless meaningful human control and independent safety validation are demonstrated.
The operational technology boundary
Moving data or decisions across the boundary between operational and corporate technology is a security question with its own governance, not an integration detail to be resolved during delivery.
Fail-safe behaviour
What the system does when the model is unavailable, degraded or wrong is assessed explicitly. A capability with no defined safe state is not ready.
Environmental and reporting obligations
Where output feeds statutory environmental or emissions reporting, accuracy, auditability and traceability requirements apply to the model itself.
Outcomes that come up more often than expected
- Hybrid human and AI, where advisory output supports an operator who retains control
- Process redesign first, where maintenance planning cannot act on a prediction anyway
- Data foundation first, where sensor coverage or historian quality will not support the claim
These reflect what the framework tends to conclude given the constraints above. They are not statistics drawn from completed engagements, and no such figures are published because none have been gathered.
Content status
| Property | Value |
|---|---|
| Version | 1.0.0 |
| Maturity | Foundation |
| Validation status | Unvalidated |
| Validated by | Not validated by a named specialist |
| Functions | 17 |
| Processes | 40 |
| Use Case Families | 32 |
| Questions | 20 |
| Hard Gates | 8 |
| Risk Patterns | 12 |
| Controls | 0 |
| Alternatives | 0 |
| Kpis | 34 |
| Financial Templates | 0 |
| Implementation Patterns | 0 |
| Assessment Templates | 5 |
Internally authored by DUN-AI and not externally validated. No named specialist has reviewed this content.
A functional safety engineer and an operational technology security specialist should review this pack before its safety content informs a decision.