Tidally influenced meandering river deposits of the Cretaceous middle McMurray Formation are characterized by rapid vertical and lateral lithological and associated reservoir property changes. Within the reservoir, water may occur below, above, and in the middle of the bitumen column, and there may be multiple gas intervals. Although conceptual understanding about the depositional environment and its control on distribution of different fluids (bitumen, water, and gas) is documented in literature, integration of these concepts into reservoir models and history matching through flow simulation is lacking. Thus, even in areas with closely spaced wells (as much as several hundred meters apart), geostatistical modeling approaches show high degrees of randomness.

This chapter closes the gap between the conceptual mapping and numerical modeling approaches. Specifically, the workflow for creating a deterministic three-dimensional (3-D), object-based (geobody) model, which integrates data from closely spaced wells, high-quality 3-D seismic data, and sound geologic concepts is shown. The geobodies are typically large-scale depositional elements comprising meandering river deposits. Geobodies include channel lag breccia (tens to hundreds of meters wide and as much as several meters thick), lower and upper point-bar deposits (from several hundreds of meters to as much as 5 km [3 mi] wide and as much as 40 m [131 ft] thick), and mud plug deposits (as much as 500 m [1640 ft] wide and as much as 40 m [131 ft] thick). Because of the potential impact on reservoir development economics, top water, top gas, and low-bitumen, high-water saturated zones are mapped as distinct geobodies. Based on their reservoir development potential, geobodies can then be classified as reservoir flow unit types 1 and 2, reservoir flow barriers, and reservoir flow impairments.

Geobody mapping includes several steps. First, interpretation of geobodies in each well profile (one-dimensional interpretation) using well log (including dipmeter) and bitumen analysis data is conducted. Generally, clean, massive, or trough cross-bedded, bitumen-saturated sand, characterized with scattered dips and with a base marked by breccia or scour, and overlain by inclined heterolithic strata deposits are identified as lower point-bar deposits. Interbedded sand and mud with unidirectional dips are interpreted as laterally accreting, upper point-bar deposits. Intervals with neutron-density crossover are gas geobodies, and clean sand with low-bitumen and high-water saturation are interpreted as a distinct geobody. In the second step, geobodies interpreted in individual wells are then correlated on a series of closely spaced (100–300 m [330–1000 ft] apart) cross sections and horizon slice maps (every 10 m [33 ft]). In this step, a reliable correlation of certain geobodies between wells is supported by seismic amplitude changes occurring along geobody contacts. Other contacts, characterized by the lack of amplitude changes, are delineated using well data and applying process sedimentological concepts. All contacts are manually drawn using computer-drafting tools. The third step connects the geobodies three-dimensionally to create spatially defined triangulated geobody surfaces. This methodology honors and integrates a range of data, as well as sedimentological principles for tidally influenced meandering river depositional facies.

Following mapping of the 3-D geobody surfaces, a 25 25 1-m (82 82 3.3-ft) grid is generated, and properties are assigned stochastically to preserve the spatial distribution and population for each geobody. Lithofacies are populated first, followed by porosity and permeability after logs were upscaled to match the 1 m (3.3 ft) vertical resolutions of the block model. Lithofacies proportions and variograms are computed for each geobody to account for local trends and spatial orientation. Porosity and permeability histograms and variograms are computed by lithofacies for the entire model area. Water saturation is populated per geobody using an inverse correlation with porosity based on a fieldwide relationship. This approach allows for geologic interpretation to guide the distribution of lithofacies, petrophysical, and fluid properties.

Results show that in comparison with common geostatistical workflows, the stochastic simulation constrained to deterministically interpreted 3-D geobodies allows conceptual geologic interpretation to guide statistical distribution, thereby reducing uncertainties and improving the ability to visualize, simulate, and analyze production results in a geologic context. Implications for reservoir development include better placement and optimization of horizontal wells, more realistic production optimization decisions and production history matching constrained by geologic interpretation. Reduced reservoir development and production costs and maximized recoveries likely will be the outcome.

This workflow demonstrates the necessity to integrate a variety of data types from several disciplines, implying a thorough understanding of geologic heterogeneities and necessitating the teamwork of several specialists. Depending on the model complexity, this approach can be challenging in terms of workflow and resource management, but in our experience, it significantly improves the precision and the accuracy of static reservoir models built for steam-assisted gravity drainage reservoir developments.