Crude Oil Quality Prediction From Reservoir to Trading
Challenges
- A crude oil blend assay can drift over time due to variations in contributing oil fields, production methods or reservoir depletion.
- Assay tests are costly and time-intensive, so producers often rely on periodic assessments to characterize their blends.
- Investment decisions based on inaccurate predictions of crude quality can result in increased project costs.
Approach
A unified data framework consolidates well and field level production, PVT and density tests, condensate spiking and reservoir conditions.
A model trained with hyperparameter optimization predicts shipment-level API weekly, and the projected quality is converted into a downstream value estimate across market sentiment scenarios.
Key features
- Unified data framework across production, lab tests and reservoir conditions
- Weekly shipment-level API prediction instead of periodic assays
- Downstream value projection under different market scenarios
- Quality tracked from reservoir through to trading and the downstream facility
Impact
up to 1%
Value increase under an average market scenario
Weekly
Shipment-level API predicted, instead of periodic assay tests
Quality tracked from reservoir to trading and downstream
As stated in the company deck