Deepwater turbidite QI: Girassol-style fluid prediction

The basin and the field

Block 17 sits in 1300-1500 m water depth offshore Angola. The Congo deep-sea fan deposited a stack of sand-rich turbidite intervals through the Late Miocene and Pliocene. Salt mobilization from the underlying Aptian Loeme salt has created complex structural traps over much of the area. Girassol is a STRATIGRAPHIC TRAP: the reservoir sands are channel-lobe complexes whose shapes themselves are the trap geometry. Specifically:

• Reservoir interval: stacked turbidite sand bodies (channels + lobes) at ∼2000-2400 m TVDSS. Net reservoir thicknesses of 30-100 m per channel-lobe complex.
• Source rocks: deeper Cenomanian-Turonian organic-rich shales, oil-mature in the syn-rift section.
• Migration: vertical from source through fault networks, charging the Pliocene reservoirs.
• Seal: regional Pliocene mudstones above each turbidite package.
• Fluids: about 30 to 32 API medium crude with a small dissolved-gas content; expected reservoir pressure modestly under-pressured.

The QI question

In the late 1990s, the operator faced a familiar deepwater problem: a 3D seismic survey covered the entire area, showing dozens of bright amplitude anomalies. Which of these are oil-bearing turbidite sands worth drilling? Which are wet sands or non-reservoir bright spots that look promising but won’t pay back? At deepwater drilling costs of $50-100M per well, every dry hole hurts.

The QI program was designed to answer one question per anomaly: is this a commercial-scale oil-bearing reservoir? Answering it requires three pieces of evidence working together (the THREE-LEG TEST):

• SAND: Is reservoir-quality sand present in this area? Net-sand cubes from inversion answer this.
• FLUID: Is the sand hydrocarbon-charged? AVO Class III intensity answers this.
• CLOSURE: Is the trap geometry capable of holding a column? Top-reservoir structure maps answer this.

Only a prospect that PASSES ALL THREE is a high-confidence drill target. Failing any one leg drops the prospect substantially.

Exercise, cycle the QI products and synthesize

• Open the widget in RMS amplitude mode. You see the channel-lobe architecture: a main NW-SE feeder splits into two distributaries that empty into a broad lobe complex in the SE corner. This is the GEOLOGY, inferred from seismic attribute analysis (Part 6) without any fluid-prediction information yet.
• Switch to AVO Class III intensity. Now you see only the FLUID-BEARING regions: the SE lobe (Lobe A), a mid-fan anomaly, and a small upstream pocket. Most of the channel-lobe system DOES NOT show AVO anomaly, those are wet sands or partially-saturated sands. The AVO map TELLS YOU WHICH SANDS HAVE OIL.
• Switch to Net-sand thickness. This is the QI inversion product (§7.4). High-net-sand (yellow) traces the channel axes and lobe interiors quantitatively. This is what feeds NET PAY computations (§7.6).
• Switch to Top-reservoir depth (structure). Regional dip is NW-shallow to SE-deep, with two LOCAL CLOSURES (red highs): one over Lobe A, one at the mid-fan. Without closure, hydrocarbons leak updip; closure is the structural prerequisite (Part 3).
• Switch to Composite prospect ranking. The brightest gold spots are the BEST DRILL TARGETS, they have all three legs passing simultaneously. The Lobe A target stands out as the highest-ranked; mid-fan is the secondary target. Other parts of the channel system drop out because they fail one or more legs.
• Note how the ranking map COMPRESSES the integration: a single visualization that the asset team can use to decide which 2-3 prospects to drill first. This is the SYNTHESIS DELIVERABLE that QI exists to produce.

Workflow walkthrough, Parts 0-8 in action

• Acquisition + Processing (Parts 1-2): wide-azimuth 3D survey acquired over Block 17. Anisotropic Kirchhoff pre-stack depth migration produces a depth-converted volume suitable for inversion. Anisotropy modeling (§8.3) corrects for VTI overburden of stacked Pliocene shales.
• Structural framework (Part 3): top-of-reservoir horizons picked semi-automatically (§2.3); fault network mapped (§2.5); structural closure analysis (§3.6) identifies the Lobe A and mid-fan closures.
• Stratigraphic interpretation (Part 4): channel-lobe architecture mapped (§4.4-4.5); the channel system’s downflow direction (NW→SE) constrains where lobe deposition is concentrated.
• Rock-physics calibration (Parts 5 + 7.2): well-log data from any nearby wells used to build a basin RPT. Brine sands plot at Ip~7400, Vp/Vs~1.80; oil sands at Ip~6800, Vp/Vs~1.72. Clear separation, a QI-favourable basin.
• Pre-stack inversion (§7.3): simultaneous Ip + Is + density inversion delivers 3D elastic cubes. Wavelet estimation per angle stack; LF model from regional structural framework + analog wells.
• Rock-property transforms (§7.4): Ip + Vp/Vs cubes → net sand, Vsh, Sw cubes via per-facies linear regression calibrated against well logs.
• AVO attribute extraction (§5.4 + §6): Class III intensity computed per voxel from near-/far-stack difference. Cross-validated against the inversion-derived Sw cube.
• Probabilistic facies + uncertainty (§7.5-7.6): each voxel assigned per-class probabilities (oil sand, brine sand, shale). Monte-Carlo prospect ranking integrates all uncertainties.
• Composite ranking + drill program: highest-ranked prospects drilled first. Girassol’s actual drill program achieved >60% commercial-success rate, well above the deepwater industry average of 30-40%.

Outcome and lessons

Girassol came online in December 2001 with first oil from the Pliocene reservoirs above expectations. The full Block 17 hub (Girassol + Dalia + Pazflor + CLOV) eventually produced more than 2 billion barrels by 2024 with declining contribution as the field matured. Multiple Lobe A and mid-fan-equivalent prospects were drilled successfully on the back of the integrated QI workflow. The lessons that transfer to other deepwater plays:

• QI works best where all three legs cooperate. Bright AVO + clean sand + structural closure is the magic combination. Fields where one leg is marginal (e.g., wet sand cluster, no closure) have much lower hit rates even with the best QI.
• Composite ranking is the deliverable, not individual attribute maps. The asset team needs ONE map that tells them WHERE TO DRILL, not five maps they have to integrate themselves.
• Ranked drill programs build portfolio momentum. Drilling the best prospects first keeps the success rate high through the early development. Investor confidence builds, capital becomes available for the next phase.
• Iteration matters: each well drilled adds CALIBRATION to the QI model. Girassol’s QI improved over time as more wells confirmed (or refined) the rock-physics template. Today’s QI in Block 17 benefits from 20+ years of well-tied calibration.
• Failures are informative: even a few unsuccessful wells in the original program identified RPT calibration issues in specific stratigraphic intervals. Today those intervals are flagged with extra uncertainty.

Where this workflow generalizes, and where it doesn't

The Girassol workflow is the GOLD STANDARD for QI-favourable deepwater clastic plays:

• Generalizes to: West African deepwater (Niger Delta deepwater, Equatorial Guinea, Ghana), Brazil Campos Basin Tertiary, Gulf of Mexico Miocene-Pliocene, Norwegian Sea Tertiary turbidites. Anywhere with bright AVO Class III on Pliocene-Miocene clastic reservoirs.
• Does NOT generalize to: tight gas sands (Class II AVO, ambiguous fluid signal); carbonate reservoirs (different rock physics, often non-Class III); over-pressured Paleogene sands with complex AVO; reservoirs in heavily faulted compartments where compartment-by-compartment uncertainty dominates over QI signal.
• The Girassol-style workflow is also the template for CO2 storage screening. Replace "is there oil here?" with "is the storage reservoir present + sealed + structurally closed?", same three-leg test, slightly different attributes.