A PRACTICAL DCA–RTA–AI/ML WORKFLOW FOR OIL AND ASSOCIATED GAS PRODUCTION FORECASTING IN A SILTSTONE RESERVOIR: A CASE STUDY

Abstract

This paper is framed as a field case study and applied forecasting workflow for hydraulically fractured vertical wells producing oil and associated gas from a siltstone interval in the Cambay Basin. The reservoir setting is characterized by low matrix permeability, solution-gas/depletion-drive behavior, rapid pressure loss, and dependence on hydraulic fracturing for sustained commercial deliverability. Daily production histories from Wells 101, 112, and 201 are used to integrate empirical decline-curve analysis (DCA), rate-transient-analysis (RTA) screening, and AI/ML-assisted short-term forecasting. The forecast is extended to 2030 and is conditioned by reservoir attributes including porosity, water saturation, net pay, production/drainage area, depth, bottomhole-pressure references, and oil gravity.

The case study demonstrates how a transparent, auditable workflow can be constructed when complete bottomhole-pressure, PVT, and fracture-treatment datasets are not available. DCA provides the deterministic production-forecast baseline, RTA screening provides diagnostic evidence for flow-regime and data-quality assessment, and AI/ML supplies a repeatable feature-based layer for holdout testing and interpretation. The contribution is therefore not a new machine-learning algorithm, but a practical field workflow that combines engineering diagnostics, reservoir conditioning, and data-driven validation for low-permeability fractured siltstone wells.