Dedicated Sessions

This 80‑minute session convenes operator, vendor, and academic perspectives to demonstrate how agentic and multimodal AI integrate geoscience and reservoir engineering across the value chain (G&G → reservoir → wells → production) with a clear business case (cycle time, recovery, cost, risk). It frames why integration has been hard—divergent problem frames, data/scale mismatches, modeling cultures, and organizational silos—and what is new: multimodal agents that read seismic, logs, reports, schedules, and SCADA; surrogate models that de‑bottleneck simulation; and standardized data products/knowledge graphs that make handovers machine‑executable. Short talks deliver concrete use cases on both sides: seismic interpretation, generative
geomodels, and structural uncertainty mapping; AI‑guided history matching, surrogate‑accelerated forecasting, and closed‑loop reservoir management. A moderated panel distills guardrails and value realization—calibration and physics audits, uncertainty and economics, governance/MLOps and data lineage, human‑in‑the‑loop and accountability—and addresses pitfalls (domain shift, false confidence, process/incentive misfit). Attendees leave with actionable integration patterns (hybrid physics–ML, agentic plan–act–check orchestration, shared subsurface data products), a minimum quality bar, and a skills roadmap for near‑term adoption.

Considerations:

• Map the end‑to‑end workflow (G&G interpretation → static/dynamic modeling → wells/production → economics) and show how artifacts (faults, horizons, wells, schedules) are exchanged with traceability.
• Present operator, vendor, and academic case studies demonstrating surrogate‑enabled scenario screening, hybrid physics‑ML calibration, and closed‑loop optimization in real projects.
• Identify common pitfalls and guardrails—validation and physics audits, governance/model cards, human‑in‑the‑loop roles—and outline the skills and team structures needed for near‑term adoption.
  Session format (80 minutes total):
• 0–5 min — Opening context from the committee side (Roderick Perez).
• 5–20 min — Academic perspective (TBC Antorweep Chakravorty, University of Stavanger).
• 20–35 min — Operator perspective (Kristoffer Jack Mann, Aker BP).
• 35–50 min — Vendor perspective (Atilla Melillo, Halliburton).
• 50–80 min — Moderated panel led by Roderick Perez with audience Q&A (precirculated seed questions + open floor).