Sampling Error: Core Plug Heterogeneity, Representative Elementary Volume, and WCSB Reservoir Characterization
Sampling error is the uncertainty introduced into any formation measurement because the analysis is performed on only a small, limited portion of the rock rather than on the entire reservoir. When a petrophysicist measures porosity, permeability, grain density, or fluid saturation on a one inch diameter core plug drilled from a full diameter core, that single number is treated as if it represents cubic kilometres of buried rock, yet the plug samples only a few cubic centimetres. The gap between the value measured on the sample and the true value of the population it is meant to describe is the sampling error, and it grows directly with the heterogeneity of the material being sampled. In a clean, well sorted Viking sandstone the pore system is fairly uniform, so a handful of plugs can characterize a whole zone with modest error. In a fractured, vuggy, or laminated rock such as the Montney siltstone or a Nisku reef margin, most of the porosity and permeability lives in features that a small plug either captures entirely or misses entirely, so the same plug spacing produces error bars several times wider. The concept is closely tied to the representative elementary volume, the minimum sample size above which a measured property stabilizes and stops swinging with each new sample; below that volume, results scatter badly and the mean is unreliable. Sampling error is not the same as measurement error. Measurement error is the imprecision of the instrument or laboratory procedure and can be reduced by better calibration, while sampling error stems from the selection and volume of material examined and can only be reduced by taking more samples, larger samples, or samples positioned to capture the true variability of the formation. Whole core analysis, which tests the entire length of full diameter core, deliberately maximizes sample volume to shrink this error in heterogeneous rock, while routine plug analysis trades volume for speed and cost. Sampling error propagates forward into every downstream calculation: it feeds directly into porosity and permeability estimates, into the net pay cutoff, and ultimately into booked reserves, so a poorly designed sampling program can bias an entire field development economics case. In Western Canadian Sedimentary Basin work, operators such as Tourmaline and ARC Resources manage this risk through statistically designed plug spacing, duplicate sampling, and quality control programs that quantify the error rather than pretend it does not exist.
Key Takeaways
- Heterogeneity Drives the Error: The single largest control on sampling error is the variability of the rock itself. A clean, well sorted Cardium sandstone yields low error from a few plugs, while a fractured Duvernay or vuggy Leduc carbonate can produce permeability values that differ by three or more orders of magnitude between adjacent plugs, so a small sample set badly misrepresents the true reservoir behaviour.
- Sampling Versus Measurement Error: Sampling error comes from examining only a limited portion of the formation and is reduced by taking more or larger samples. Measurement error comes from instrument imprecision and is reduced by calibration. Confusing the two leads teams to over invest in laboratory precision while ignoring the far larger uncertainty created by too few, too small, or badly located samples.
- Representative Elementary Volume: Every property has a minimum sample volume above which the measured value stabilizes. Below this REV, results scatter wildly with each new sample. Whole core analysis on full diameter core, roughly 100 mm across, is used in fractured or vuggy WCSB reservoirs precisely because plug scale volumes fall below the REV and generate unacceptable error.
- It Propagates Into Reserves: Sampling error does not stay in the laboratory. Biased porosity and permeability numbers flow into net pay cutoffs, saturation models, and volumetric reserve calculations reported under AER Directive 059 and evaluated per COGE Handbook standards, so a flawed sampling design can shift booked reserves and the economic case for an entire development.
- Quantify Rather Than Eliminate: Sampling error can never be reduced to zero because no operator cores the whole reservoir. Best practice is to quantify it with duplicate samples, statistically designed spacing, and confidence intervals, so decision makers work with honest error bars rather than a false sense of a single precise answer.
Plug Spacing and Statistical Design in a Montney Core Program
A typical Montney vertical core in northeast British Columbia might recover 30 m of siltstone at a cost of roughly CAD 250,000 to CAD 400,000 for cutting, retrieval, and analysis. If a petrophysicist pulls plugs on a fixed one metre spacing, only 30 samples describe a laminated interval whose permeability varies over centimetres. Statistical design instead concentrates plugs where log response changes fastest and adds duplicate plugs to measure short range variance directly. This lets the analyst report porosity as a mean with a confidence interval rather than a bare number, exposing the true sampling error before it contaminates the volumetric model and the reserve booking.
Whole Core Analysis Where Plugs Fail
In fractured Duvernay shale or a vuggy Slave Point carbonate, plug scale samples routinely miss the fractures and solution vugs that carry nearly all the flow capacity. Here operators switch to whole core analysis, measuring porosity and permeability on the entire full diameter core segment so the sample volume finally exceeds the representative elementary volume. The cost per metre rises, but the sampling error on effective permeability drops from a factor of ten or more down to a manageable band, which is the difference between a defensible reserve estimate and one that regulators or auditors will challenge.
Fast Facts
The French mining engineer Pierre Gy formalized sampling theory in the 1950s and showed mathematically that sampling variance scales inversely with sample mass and directly with the size of the largest particles or heterogeneities present. His work, developed for ore grade control, is why a fractured carbonate demands a full diameter whole core while a uniform sandstone tolerates a thumb sized plug. The same equations that govern gold assay quality in a mine also govern whether a WCSB porosity number can be trusted.
Related Terms
Sampling error sits at the centre of reservoir characterization and connects to several core concepts. It directly shapes core analysis, which is the laboratory workflow that generates the measurements in the first place, and it governs the reliability of both porosity and permeability, the two properties most sensitive to sample volume. Because biased samples flow into volumetric math, sampling error also influences net pay determination and the confidence attached to booked reserves.
Real-World WCSB Scenario: A Duvernay Permeability Dispute
An operator evaluating a Duvernay land block near Fox Creek, Alberta cut a 45 m core at a cost of roughly CAD 500,000 including analysis. The first pass used routine one inch plugs on a fixed 0.75 m spacing, returning an arithmetic mean permeability of 340 nanodarcies with wild scatter. When a reservoir engineer flagged that the fractured intervals were being missed, the team re sampled with whole core segments across the natural fracture zones and ran duplicate plugs to measure variance. The revised effective permeability came in materially higher with a tight confidence band.
The corrected sampling design changed the type well forecast enough to move the block from marginal to commercial under prevailing AECO and condensate pricing. The extra CAD 60,000 spent on whole core and duplicates was trivial against a multi well development decision, illustrating that the cheapest way to reduce sampling error is to design for it before the core ever reaches the plug saw.