In Part 1 of this series, we introduced the principles of rule-based reservoir modeling and explained how geological-rule–driven geomodels differ from conventional stochastic approaches. By embedding depositional rules, stacking patterns, and stratigraphic architecture directly into model generation, rule-based modeling captures the subsurface heterogeneity that truly matters to fluid flow. In this second part, we explore how Shell researchers have applied such rule-based models—built using ReservoirStudioTM—in a series of pioneering studies that reveal the dynamic consequences of geological realism.
Understanding the Challenge
One of the central challenges in reservoir modeling is translating geological complexity into models that remain both realistic and computationally efficient. Traditional stochastic or object-based approaches often oversimplify the geometry of channels, barriers, and drapes, leaving out the fine-scale structures that control permeability pathways.
Between 2010 and 2015, researchers from Shell published seven landmark papers that systematically examined these issues using ReservoirStudioTM, a geologic-rule–based modeling software capable of constructing realistic turbidite, estuarine, and fluvial architectures. These publications—while purely scientific and noncommercial—collectively demonstrate how rule-based geomodels improve flow predictions, guide upscaling strategies, and identify which heterogeneities truly matter in production forecasting.
Simplified Modeling of Turbidite Channel Reservoirs (Alpak, Barton & van der Vlugt, 2010)
The 2010 SPE Journal paper introduced an innovative workflow for representing fine-scale geologic features in simplified reservoir simulations. Instead of explicitly modeling every meander or shale drape, the authors derived effective properties—notably geologically based pseudorelative permeability functions—that preserved the influence of small-scale architecture on flow.
Using ReservoirStudio, high-resolution sector models were constructed to quantify how parameters such as meander-belt width, degree of amalgamation, and shale-drape coverage influence dynamic connectivity. These fine-scale simulations were then upscaled into coarse, full-field dynamic models through Monte Carlo analysis.
The simplified models achieved recovery factors consistent with detailed simulations, but with computational times reduced from days to hours. The study concluded that dynamic connectivity in turbidite channel systems is governed primarily by large-scale architectural patterns and shale distribution, rather than every micro-scale detail. This laid the foundation for practical, geologically consistent upscaling in real-world field development planning.
Retaining Geological Realism in Dynamic Modeling: A Channelized Turbidite Reservoir Example from West Africa (Alpak, Barton & Castineira, 2011)
Building on their 2010 findings, the Shell team demonstrated how rule-based geological realism can be retained in dynamic modeling for a real turbidite reservoir offshore West Africa. The challenge: how to integrate detailed channel architectures into flow simulation without overwhelming computational resources.
The authors employed a two-stage, Bayesian-assisted history-matching workflow. ReservoirStudio models captured realistic channel storeys and shale drapes, while pseudo-relative permeability functions incorporated the effects of subseismic barriers. Using this approach, multiple geologically consistent realizations were generated and history-matched to production data within hours rather than weeks.
The result was a set of models that matched field performance while preserving geological credibility. This study proved that rule-based models—when integrated with Bayesian data assimilation—enable fast, realistic, and reliable forecasting, bridging the gap between geology and engineering in a practical, workflow-ready manner.
A Multiscale Adaptive Local–Global Method for Modeling Flow in Stratigraphically Complex Reservoirs (Alpak, Pal & Lie, 2012)
By 2012, Shell’s research expanded into numerical methods designed specifically for stratigraphically complex, rule-based models. The authors developed a multiscale adaptive local–global (ALG) solver, a method that dramatically accelerates flow computation while maintaining fine-scale accuracy.
In the ALG method, the flow field is solved globally on a coarse grid while local basis functions represent fine-scale stratigraphic variability. This allows local geological heterogeneities—like shale drapes or channel lenses—to influence global flow without full fine-scale resolution.
Applied to ReservoirStudio-derived turbidite models, the method achieved substantial computational speed-ups with minimal accuracy loss. Importantly, it captured multiphase flow behavior and mobility-ratio effects better than conventional upscaling. The work established a numerical foundation for integrating rule-based geological realism into efficient reservoir simulators, a critical step for modern AI-assisted flow modeling.
The Impact of Fine-Scale Turbidite Channel Architecture on Deep-Water Reservoir Performance (Alpak, Barton & Naruk, 2013)
In 2013, a comprehensive sensitivity study explored how variations in channel geometry and stratigraphic detail influence reservoir performance. Using over 1,700 realizations generated in ReservoirStudio, the team systematically varied channel width, degree of amalgamation, and shale-drape frequency.
Results showed that shale drapes and base-channel shales exert the strongest control on recovery efficiency, followed by net-to-gross ratio and channel stacking style. Where shale drapes were continuous, recovery dropped sharply due to compartmentalization and early water breakthrough.
The study’s major conclusion—that “not all heterogeneities are equal”—remains a guiding principle in reservoir modeling. Fine-scale details matter only when they affect connectivity, and rule-based modeling provides the framework to identify those critical features. This insight has since influenced numerous workflows focused on prioritizing geologic realism where it matters most to fluid flow.
Shale-Drape Modeling for the Geologically Consistent Simulation of Clastic Reservoirs (Alpak & van der Vlugt, 2014)
Recognizing the outsized influence of shale drapes, Alpak and van der Vlugt (2014) introduced the Shale-Drape Function (SDF)—a deterministic algorithm for inserting shale barriers consistent with depositional geometry.
Rather than random shales, the SDF generated drapes that conform to channel and lobe surfaces, with optional “holes” to represent erosion or partial deposition. This made it possible to capture realistic barrier patterns in both sector and full-field models.
Simulation results revealed that once shale coverage exceeds about 50–70%, it becomes a dominant control on recovery and pressure communication. The authors recommended explicit shale-drape modeling when lateral continuity is high, and transmissibility multipliers when discontinuous. This work was pivotal in operationalizing the concept of geologically conditioned barriers, now a staple in modern rule-based geomodeling.
Dynamic Impact and Flow-Based Upscaling of the Estuarine Point-Bar Stratigraphic Architecture (Alpak & van der Vlugt, 2014)
A parallel 2014 study applied similar principles to estuarine point-bar deposits, where lateral accretion, shale drapes, and abandoned channels create complex internal architectures. Using fine-scale models generated in ReservoirStudio, the authors explored how these features affect flow performance and recovery.
They discovered that shale drape continuity and channel abandonment geometry dominate dynamic behavior, while well spacing and mobility ratio have secondary effects. When barriers align with flow direction, sweep efficiency decreases; when oblique, they act as partial baffles that delay breakthrough.
The study also validated a multiphase, flow-based upscaling method, which better captured saturation-dependent effects than traditional single-phase approaches. Notably, deviated wells were shown to outperform long horizontals in draining compartmentalized systems. The paper demonstrated how combining rule-based modeling with flow-based upscaling can bridge the gap between geological understanding and engineering optimization.
Quasiglobal Multiphase Upscaling of Reservoir Models with Nonlocal Stratigraphic Heterogeneities (Alpak, 2015)
The final study in this series, published in SPE Journal (2015), introduced Regional-scale Multiphase Upscaling (RMU)—a quasiglobal approach that calibrates pseudofunctions from intermediate-scale sector models rather than relying solely on local upscaling.
RMU captures nonlocal heterogeneities—such as laterally extensive shale drapes and interconnected channel systems—without full-field fine-scale simulation. When applied to ReservoirStudio geomodels, the approach reproduced the key dynamic signatures of fine-scale models with 2–3 orders of magnitude computational efficiency.
This paper completed a conceptual evolution: from building realistic geologic rules, to simplifying flow representation, to achieving scalable, accurate, and predictive simulation. It highlighted how geological consistency—when carried from modeling to upscaling—enables both accuracy and efficiency in reservoir forecasting.
Concluding Reflections
Together, these seven Shell publications demonstrate the enduring value of geologic-rule–based reservoir modeling. They collectively show that:
- Geological realism, when based on depositional rules, directly improves flow predictions.
- Shale drapes and connectivity patterns are the most influential heterogeneities in many clastic reservoirs.
- Multiphase, flow-based upscaling provides a reliable bridge from high-resolution models to full-field simulation.
Although these studies are independent of any commercial endorsement, they remain a benchmark for the industry—showing how rule-based reservoir modeling captures the heterogeneity that truly matters most to fluid flow.
References
- Alpak, F. O., M.D. Barton and F. v. d. Vlugt, 2010, Simplified modeling of turbidite channel reservoirs. SPE Journal June 2010, pages 480-494.
- Alpak, F. O., M.D. Barton and D. Castineira, 2011, Retaining geological realism in dynamic modelling: a channelized turbidite reservoir example from West Africa. Petroleum Geoscience. Vol 17, No 1, February 2011 pp. 35 – 52. DOI: 1144/1354-079309-033.
- Alpak, F. O., Mayur Pal, Knut-Andreas Lie, 2012, A Multiscale Adaptive Local-Global Method for Modeling Flow in Stratigraphically Complex Reservoirs, SPE Paper #, 140403, SPE Journal, December 2012
- Alpak, F. O., Mark D. Barton, Stephen Naruk, 2013, The impact of fine-scale turbidite channel architecture on deep-water reservoir performance, AAPG Bulletin, V. 97, No. 2 (February 2013), P. 251–284
- Alpak, F. O. and F. v. d. Vlugt, 2014, Shale-drape modeling for the geologically consistent simulation of clastic reservoirs, SPE Journal, October 2014, pages 832-844.
- Alpak, F. O. and F. v. d. Vlugt, 2014, Dynamic impact and flow-based upscaling of the estuarine point-bar stratigraphic architecture. Journal of Petroleum Science and Engineering 120, pages 18-38.
- Alpak, F. O., 2015, Quasiglobal multiphase upscaling of reservoir models with nonlocal stratigraphic SPE Journal, April 2015, pages 277-293.
