Boolean Object Simulation: Modeling Channelized Sand Bodies in WCSB Fluvial and Deep Basin Reservoirs
Boolean simulation (also called object-based simulation or marked point process simulation) is a stochastic geostatistical technique for building 3D reservoir models in which the spatial distribution of reservoir facies (sand channels, carbonate lobes, reef mounds, or any discrete geological object type) is represented by the random placement of geometric objects with prescribed dimensions, orientations, and spatial arrangements — as opposed to pixel-based methods that assign facies probabilities to each individual grid cell independently. The term "Boolean" refers to the binary nature of the method: each simulation cell either belongs to one of the defined object types (typically reservoir facies: sand, limestone, carbonate) or belongs to the background facies (non-reservoir: shale, anhydrite, mudstone), with cell assignment determined by whether the cell falls inside any of the randomly placed objects. The geometric parameters of each object type — channel width, thickness, length, sinuosity, orientation azimuth, and thickness variability — are derived from analogue outcrop measurements, core data, and depositional environment interpretations, then used to constrain the object generator in the simulation algorithm. The spatial density of objects (the number of channel objects per unit volume, or the probability of a point in 3D space being inside a channel) is calibrated to honor the net-to-gross ratio measured from the available wells. Boolean simulation is the preferred method for modeling two classes of WCSB reservoirs: (1) fluvial channelized systems — the Viking Formation in central Alberta, whose productive sand bodies are meandering to braided fluvial channel fills 50-400 m wide and 5-15 m thick, separated by extensive floodplain shale that produces no oil — where the reservoir connectivity (and therefore the number of wells needed to drain the pool) depends on whether the sand channels are amalgamated into a connected network or isolated as discrete sand pods that each require a separate well to produce; and (2) turbidite and mass-transport systems in deep-water analogue plays — where lobes and channels of specific dimensions define the reservoir architecture — though most WCSB Cretaceous reservoirs are continental shelf and fluvial rather than deep-water. The primary limitation of Boolean simulation is its difficulty honoring hard well conditioning data (the modeled objects must pass through the well at the depth and thickness observed in core and logs), which becomes computationally expensive when many wells constrain the model and the objects are large relative to the inter-well spacing. Most commercial reservoir modeling software packages (Petrel, RMS, Roxar) include both Boolean and pixel-based simulation options, with the geologist selecting the appropriate method based on the depositional environment and the available conditioning data density.
Key Takeaways
- Viking Formation channel modeling in Alberta: The Viking Formation in central Alberta (Pembina, Caroline, Crossfield, Drumheller areas) consists of a succession of stacked fluvial and estuarine sand channels deposited in the Late Cretaceous interior seaway shoreline environment. Individual Viking sand channels are 50-300 m wide, 3-12 m thick, and 1-20 km long — dimensions well-suited to Boolean simulation. The key modeling uncertainty is channel connectivity: in amalgamated (high net-to-gross) Viking sections, channels stack and merge into continuous sand sheets, and conventional waterflood drainage works efficiently. In isolated (low net-to-gross) sections, individual channel pods may be only 0.5-2 km long, meaning a well 1.5 km from the producing well may be in a different channel with no hydraulic connection — causing waterflood pattern inefficiency that can only be diagnosed by Boolean simulation of the channel geometry and subsequent reservoir simulation of the flooding sweep pattern.
- Object parameterization from outcrop analogues: The accuracy of a Boolean model depends entirely on the accuracy of the input object parameters. For Viking channel modeling in Alberta, the primary data sources for channel width, thickness, and length distributions are: (1) modern meandering river channels in the same climate zone as the Viking depositional environment (Canadian Great Plains rivers such as the Red Deer and Saskatchewan rivers, whose widths scale with the preserved channel dimensions); (2) outcrop exposures of Viking equivalent strata (Medicine Hat Brick Formation analogues in Saskatchewan); and (3) dense well data in mature Viking pools (Pembina and Crossfield areas have hundreds of wells on 200-400 m spacing, allowing inter-well correlations to directly measure channel width and connectivity statistics). Misidentifying a low-net-to-gross isolated channel reservoir as a high-net-to-gross amalgamated reservoir leads to waterflooding programs that miss entire portions of the field, leaving oil in uncontacted channel pods.
- Conditioning Boolean models to well data: A Boolean model must honor the well observations: where a well penetrates a sand channel, the model must place a channel object that intersects the well at that depth and thickness. In multi-channel stacked reservoirs, this conditioning is achieved iteratively: the simulation algorithm places objects randomly, checks whether they honor all well data, and rejects placements that conflict with observations. For widely spaced exploration wells (4-8 km between wells), this conditioning is achievable because the channel objects (1-5 km long) need to pass through only a few widely separated points. For densely drilled development areas (Pembina Viking at 200 m well spacing), the conditioning problem becomes combinatorially complex as the algorithm must simultaneously satisfy hundreds of constraints — requiring either a more sophisticated multi-point statistics approach or a simplification of the channel geometry to enable tractable conditioning.
- Boolean simulation for carbonate build-up modeling: In WCSB Devonian carbonate plays (Leduc reef mounds, Swan Hills carbonate platforms, Nisku sub-aqueous shoals), Boolean simulation models reef mounds as ellipsoidal objects with prescribed long-axis/short-axis/height ratios derived from seismic amplitude data and well penetrations. The simulated reef mound distribution honors the regional trend of reef development (controlled by basement structure and paleowind direction for carbonate accumulation) while allowing stochastic variation within the uncertainty range defined by available well and seismic data. A key advantage over pixel-based simulation for reef modeling: Boolean reefs have realistic 3D geometry (convex tops, flanking debris apron geometry) that controls where porosity is highest (reef core versus reef flank) — geometry that pixel-based methods fail to capture without special treatment of the flank/core proportion curves.
- Boolean versus sequential indicator simulation (SIS) for facies selection: Sequential indicator simulation (SIS) is the main pixel-based alternative to Boolean simulation for categorical (facies) modeling. SIS assigns facies to each cell sequentially using a variogram model to ensure spatial correlation of facies proportions — it honors all well data exactly at each well and produces spatially continuous facies trends, but generates facies bodies with geometrically unrealistic shapes (blocky, irregular patches rather than the elongated channel forms seen in reality). For WCSB Viking channel reservoirs where channel geometry controls sweep efficiency and well placement, Boolean simulation produces more geologically defensible connectivity predictions than SIS despite the more complex conditioning workflow. For massive homogeneous reservoirs (thick Cardium conglomerate sheets, massive Devonian carbonates without significant internal facies variation), SIS or pixel-based indicator simulation may be adequate because the channel-scale geometry is not the controlling variable for drainage.
Viking Channel Connectivity Study: Development Planning at Redwater
A Viking producer at Redwater (Alberta, 85 producing wells on 400-m spacing, 22% net-to-gross, 3 stacked Viking sand intervals) uses Boolean simulation to evaluate whether expanding the waterflood injection pattern to tighter 200-m spacing will improve recovery or simply add redundant wells in already-swept channel segments. Input parameters from core and well log analysis: mean channel width 150 m (log-normal distribution σ = 45 m), mean channel thickness 6 m (σ = 1.5 m), mean channel length 2.5 km (σ = 0.8 km), channel azimuth 045° ± 25° (NE-SW paleocurrent direction from cross-bedding measurements in core). Twenty stochastic Boolean realizations are conditioned to all 85 wells and run through full reservoir simulation with the current waterflood injection pattern. Results: the realizations show bimodal behavior — 8 of 20 realizations show a connected sand framework where existing well spacing adequately drains most channel volume (20 years to reach 90% sweep efficiency), while 12 of 20 show isolated channel pods covering 30-45% of the reservoir volume that are completely uncontacted by the existing injectors. The 12 poorly-connected realizations are the risk scenario: if the Viking channels are as isolated as the simulation suggests, tightening to 200-m well spacing on the 300 ha central area of the pool could access CAD 38 million in additional oil (4.5 million barrels at CAD 65/bbl net). The study is presented to the AER as part of the pool scheme amendment application for the expanded waterflood pattern.
Fast Facts
Boolean simulation in petroleum reservoir modeling was pioneered by the Norwegian oil industry research institute (IFE/NR) in the late 1980s and early 1990s, in response to the practical problem of modeling the fluvial and turbidite channel systems that dominate North Sea Brent and Statfjord oil fields — reservoirs where inter-well connectivity and sweep efficiency were the primary controls on recovery factor, and where pixel-based methods produced facies distributions that bore no resemblance to the geologically recognized channel architecture. The key publication, by Viseur, Brakel, and Dubrule at Elf Aquitaine in 1998, established the conditional simulation algorithm for placing channel objects while honoring well data — the same algorithm implemented in the Petrel Boolean simulation module that WCSB reservoir geologists use today to model Viking and Cardium channelized reservoirs more than 25 years after the North Sea methodology was first published.
Related Terms
The geological interpretation that provides the channel geometry parameters for a Boolean simulation — particularly the net-to-gross ratio, channel orientation, and stacking pattern — is derived from the same bivariate crossplot formation evaluation tools described under bivariate analysis, where acoustic impedance versus gamma ray or acoustic impedance versus Vp/Vs crossplots discriminate Viking sand channels from non-reservoir shale intercalations on 3D seismic amplitude extractions. The connectivity model produced by Boolean simulation directly controls the number and spacing of production wells needed to drain the Viking or Devonian carbonate reservoir, which in turn determines the bonus consideration economics: an isolated, poorly connected Viking section requires more wells per section to drain the same OOIP compared to a high-net-to-gross amalgamated section, making the isolated case worth a lower bonus per hectare at Crown land sale despite apparently similar well log indicators at the available offset wells.