Answers are grounded in this paper and cite their sources.
Static reservoir characterization integrates seismic interpretation, well-log analysis, structural modeling, facies simulation, petrophysical property modeling, and uncertainty assessment, but reproducibility is often limited by incomplete recording of software settings, stochastic controls, intermediate products, and quality-control decisions. This study presents the Reproducible Integrated Reservoir Modeling Workflow (RIRMW), a Petrel-oriented framework that separates workflow provenance from numerical validation and makes modeling decisions auditable. The framework is evaluated with a location-independent 120 × 90 × 48-cell synthetic clastic benchmark. Twenty matched random seeds are evaluated under five conditions—a stochastic baseline and structural, facies-proportion, spatial-correlation, and petrophysical perturbations—yielding 100 realizations. Baseline mean effective porosity is 17.61%, geometric-mean horizontal permeability is 5.94 mD, net-to-gross is 0.739, and connected-sandstone fraction is 0.988. Matched perturbations show response-dependent sensitivity: facies-proportion and coupled petrophysical perturbations have the largest effects on storage and permeability metrics, whereas the tested fault-displacement perturbation has comparatively small effects on the selected bulk responses. Ensemble stabilization is assessed at n = 5, 10, 15, and 20, and five prespecified realizations are reconstructed exactly within the deterministic Python benchmark. The provenance boundary is explicitly defined: Python generates the open synthetic benchmark and quantitative validation outputs, whereas Petrel provides the integrated reservoir-modeling environment and documented implementation workflow. Python reconstruction tests are therefore not presented as Petrel reruns. RIRMW therefore provides a practical framework for traceable Petrel-based modeling while defining measurable reproducibility claims that can be tested independently for the non-proprietary benchmark components.
No takes yet. Share an insight, caveat, or question.
Innocent J. Aluka (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: