The Gold Standard of Petroleum Geology | Permeability Across Layers and Why Flow Direction Changes the Average

Chapter index

Start with definitions

• 1. Begin with a geological question, not a mean
• 2. Use Darcy’s law as a dimensional contract
• 3. Parallel flow: add conductance, not resistance

See the calculations

• 4. Work the same two-bed package in both directions
• 5. Cross-bed flow: add pressure drops in series
• 6. Change one thin bed and watch the asymmetry

Challenge the interpretation

• 7. Let units and boundary conditions expose mistakes
• 8. Recognise when the elementary formulas fail
• 9. Plan the evidence before trusting an upscale

Check, practise and continue

• 10. Answers to common permeability questions
• 11. Practise with a different stack and diagnose errors
• 12. Use the directional answer to ask a better next question

A permeability value describes how readily a rock transmits a fluid under a specified flow regime. A core plug measures a specimen under laboratory boundary conditions; an interval test or a well response samples a different, much larger effective system. In geological layers, those different scales can disagree without either measurement being intrinsically wrong.

Porosity measures the fraction of volume that is pore space. Permeability concerns the connected paths through which a fluid can move. Two rocks may have similar porosity and very different permeability. Lamination, sorting, cementation, clay distribution, fractures and the alignment of connected pores affect transmission.

Before averaging any values, draw the package, put a direction arrow on it and label where pressure is imposed. If the drawing does not show whether the fluid travels along beds or across them, no single average answers the stated question. This guide continues the Petroleum Geology root at the level of an inspectable flow calculation.

Contents · Next chapter · Petroleum Geology guide

For steady single-phase Darcy flow in a homogeneous specimen, the volumetric flow rate can be written Q = k A Δp /(μ L), where k is intrinsic permeability, A is the cross-sectional area normal to flow, Δp is the pressure difference, μ is dynamic viscosity and L is the flow length. The form assumes a regime where the Darcy approximation applies and ignores gravity in the stated horizontal example.

Rearranging gives k = Q μ L /(A Δp). This is why the geometry and the fluid cannot be omitted from a permeability measurement. Dividing one flow rate by another without checking the cross-section, pressure gradient and viscosity is not a permeability comparison.

The millidarcy (mD) is a practical petroleum-industry unit: 1 mD is approximately 9.869 × 10⁻¹⁶ m². Using all-SI units is necessary for a numerical Q in cubic metres per second. Our effective-permeability examples retain mD because their ratios cancel the common conversion factor.

It also matters what k means. Intrinsic permeability is a rock property in the ideal single-phase relation; phase-specific effective permeability in multiphase flow additionally reflects saturation and relative-permeability effects. The two must not be silently substituted.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Imagine several laterally continuous beds extending the same horizontal distance L. Each bed experiences the same imposed horizontal pressure difference Δp. For bed i of thickness hᵢ and width W, its flow area is W hᵢ and its Darcy flow is Qᵢ = kᵢ W hᵢ Δp /(μ L).

The total flow is the sum of individual contributions: Q = ΣQᵢ. Define one equivalent bed of total thickness H = Σhᵢ by Q = k∥ W H Δp /(μ L). Equating the two expressions gives k∥ = Σ(kᵢ hᵢ)/H. This is the thickness-weighted arithmetic average.

Its physical reason is straightforward: a more permeable bed carries more flow under the same along-bed gradient. Wider conducting pathways combine in parallel. There is no need to invent a sophisticated numerical model while the simple assumptions hold.

The derivation assumes lateral bed continuity, common length and viscosity, no important interbed exchange effects and boundary conditions that establish the shared gradient. A channel that pinches out midway is not the same geometry.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Use a hypothetical ten-metre-thick package. Bed A is 2 m thick with k = 100 mD; bed B is 8 m thick with k = 10 mD. For horizontal flow, k∥ = (2 × 100 + 8 × 10)/10 = 28.00 mD. The higher-permeability bed occupies only one fifth of the package but makes a disproportionately large contribution.

For vertical flow, the fluid must cross both beds sequentially. As derived next, k⊥ = 10 /(2/100 + 8/10) = 10/0.82 ≈ 12.20 mD. Both averages retain the same geometrical thicknesses and rock permeabilities.

Look carefully at the final column of the table. It reports a direction-dependent equivalent under a particular idealization, not a claim that permeability itself magically changes when a sample is rotated. The geometry makes different pathways available.

Bed	Thickness	Intrinsic k	Parallel numerator h × k	Cross-bed resistance h/k
A	2 m	100 mD	200 m·mD	0.020 m/mD
B	8 m	10 mD	80 m·mD	0.800 m/mD
Total	10 m	—	280 m·mD	0.820 m/mD
Effective answer	—	—	28.00 mD	12.20 mD

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Now move fluid vertically through the same 2 m and 8 m layers, with equal flow area A across each layer. At steady state, conservation requires the same flow Q through both. By Darcy’s law, the drop across layer i is Δpᵢ = Q μ hᵢ/(kᵢ A).

Pressure drops add: Δp = ΣΔpᵢ = (Q μ/A) Σ(hᵢ/kᵢ). For an equivalent thickness H, Δp = Q μ H/(k⊥ A). Cancel the common factors and obtain H/k⊥ = Σ(hᵢ/kᵢ), hence k⊥ = H/Σ(hᵢ/kᵢ).

The harmonic average gives more influence to low-k beds because they consume more of the pressure drop for the same flux. This is resistance in series. Treating cross-bed pathways as independent parallel pipes would violate the chosen geometry.

Check two limiting cases: if all kᵢ are equal, both averages equal that common value; if one finite-thickness layer becomes nearly impermeable, vertical equivalent permeability tends toward zero, while parallel transmission may remain appreciable.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Add a hypothetical 0.10-metre lamina with k = 0.10 mD to the ten-metre package. Total thickness becomes 10.10 m. Parallel effective permeability is (2×100 + 8×10 + 0.1×0.1)/10.1 ≈ 27.72 mD, only slightly lower than 28.00 mD.

Perpendicular permeability instead becomes 10.1 /(2/100 + 8/10 + 0.1/0.1) = 10.1/1.82 ≈ 5.55 mD. A lamina representing less than one percent of the thickness has cut the ideal vertical equivalent by more than half.

This is a model sensitivity test, not proof that a real reservoir contains an unbroken barrier. An actual thin shale may be discontinuous, fractured, offset by faults or bypassed by connected sand bodies. Conversely, a thin high-resistance layer may matter greatly even when coarse-resolution logging does not resolve it.

Ask two questions rather than concluding from one calculation: is the bed present across the scale of interest, and do its boundaries permit a bypass? A good geological interpretation combines the arithmetic with the geometry and the observations.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Both Σ(kᵢhᵢ)/H and H/Σ(hᵢ/kᵢ) return units of permeability. The parallel numerator has mD·m and is divided by metres. The cross-bed denominator has m/mD and is divided into metres. Any remaining unit of length or inverse permeability indicates a lost thickness factor.

Never compare a total horizontal transmissivity Σ(kᵢhᵢ), whose unit is permeability times length, with vertical intrinsic permeability in mD. Transmissivity and equivalent permeability are related but not interchangeable. A model may require transmissibility including geometrical and fluid factors.

More layers do not justify an unweighted numerical average unless the chosen weights are genuinely equal. If a 2-m bed and an 8-m bed are counted once each, the simple average (100+10)/2 = 55 mD loses the thickness information and answers neither flow problem.

State whether the pressure gradient is imposed parallel or normal to bedding, whether cross-section stays constant, whether anisotropy exists inside each bed, and whether the fluids and saturation are comparable. Assumptions are part of a calculation rather than decorative fine print.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Real reservoirs may have cross-bedding, lenticular channels, faults, fractures, variable fluid saturation and permeabilities that differ within a single bed. The arithmetic and harmonic limits apply cleanly to a particular ideal stratified arrangement. They do not provide a universal upscale for every three-dimensional geological distribution.

Flow direction may curve. Permeability may be tensor-valued rather than a single scalar. Near a well, radial geometry replaces the uniform rectangular slab, and perforation intervals constrain which beds contribute. A model calibrated on one streamline may not represent another.

For gas flow, pressure-dependent gas properties, slippage at very low pressures and changing conditions can matter. For multiphase displacement, saturation-dependent relative permeabilities and capillary forces can dominate connectivity. None of those complications disappear because a weighted average is easy to compute.

The right response is to use these simple cases as bounding diagnostics. They test whether a simulator, interpretation or claim has the correct direction of effect before moving into flow-based upscaling, tensor treatment or a reservoir-specific model.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

A core plug can characterise small-scale rock under controlled flow. Image logs, sedimentary descriptions and core photographs can establish bed orientation and continuity at their respective resolutions. Pressure-transient tests sample a larger connected volume but are themselves affected by well completion and interpretation assumptions.

Build an evidence ladder: identify bed types and thickness; measure or estimate permeability distributions; map lateral continuity; compare directional response; and test the final model against independent observations. In an uncertain system, show plausible low, central and high cases rather than one falsely precise number.

Do not let a close match to one production observation erase structural uncertainty. Different connected networks can yield similar aggregate flow while forecasting different outcomes when pressure fields change. The more a decision depends on cross-bed communication, the more important it is to test the continuity of tight intervals.

Evidence	What it can constrain	What it cannot prove alone
Core plug permeability	Small-specimen flow response	Field-scale connected pathways
Sedimentology and image logs	Bedding orientation and local facies	Uninterrupted barrier continuity far from the well
Pressure and interference tests	Larger-scale communication	One unique bed-level permeability distribution
Flow simulation sensitivity	Consequences of alternative architectures	Truth of the architecture used

Contents · Previous chapter · Next chapter · Petroleum Geology guide

These questions are useful whenever a chart or report offers a single permeability for a strongly layered interval. A number without a direction, flow model and spatial scale is a starting clue, not a finished description.

Is the harmonic mean always the correct vertical permeability?

No. It is exact for the ideal series arrangement with laterally continuous layers, consistent cross-section and Darcy conditions. Faults, bypass pathways or heterogeneous three-dimensional flow can invalidate that geometry.

Does the arithmetic average mean the reservoir is naturally isotropic?

No. It describes an equivalent property for one chosen direction. A different direction can have a different effective value even when each individual bed is isotropic.

Can a very thin low-permeability bed matter?

Yes for cross-bed flow if it is continuous. Its pressure-drop contribution is thickness divided by permeability; a small numerator may still be outweighed by an extremely small permeability.

Can permeability be inferred from porosity alone?

A porosity-permeability trend can be useful when locally calibrated, but connectivity, fabric, clay and diagenesis can break a simple relationship. A direct test and geological context remain important.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

Transfer the reasoning to a new fictional package: 4 m at 60 mD, then 6 m at 5 mD, with equal fluid properties. For parallel bedding flow, the answer is (4×60 + 6×5)/10 = 27 mD. For perpendicular flow, it is 10/(4/60 + 6/5) = 10/1.2667 ≈ 7.89 mD.

Now ask whether 27 and 7.89 are two alternative ways of describing one undefined direction. They are not: each belongs to a different boundary-value problem. A learner who can say that before writing an equation has understood the central idea.

Challenge the new calculation by halving the low-permeability bed’s k from 5 to 2.5 mD. The parallel value changes to (240+15)/10 = 25.5 mD. The perpendicular value falls to 10/(4/60+6/2.5) ≈ 4.05 mD. This asymmetric response is a useful check on the physical model.

As a final test, write a two-sentence memo to a fictional reservoir team explaining why a net reservoir thickness and an average core permeability do not uniquely determine vertical communication. Name one additional measurement that would help discriminate competing models.

Contents · Previous chapter · Next chapter · Petroleum Geology guide

The gold standard is not memorising that arithmetic means go with one arrow and harmonic means with another. It is deriving which quantities combine under shared gradient versus shared flux, checking units and identifying the geological arrangement that makes those operations credible.

Return to the Petroleum Geology subject guide to connect this transport question with source rocks, depositional environments, trap timing and subsurface evidence. The related gas formation volume factor guide addresses a different distinction: how reservoir-state volume and standard reporting volume are related.

One article concerns how easily fluid moves; the other concerns what volume is being reported. Combining them prematurely can obscure both. Good reservoir reasoning keeps flow capacity, fluid phase and volumetric reporting as separate steps until the problem calls for their connection.

Sources and further reading

The references below support the physical principles and the scope of the methods. All numerical cases in this article are constructed educational examples; they are not field measurements, product specifications or a substitute for specialist analysis.

Reference	What to verify
USGS: Heterogeneous aquifer effective conductivity and layering	Derivations and conditions for arithmetic and harmonic directional averaging
Society of Petroleum Engineers: Upscaling grid properties	Bounds and flow-based considerations beyond simple averages
USGS MODFLOW Layer Property Flow package	Boundary and averaging choices in actual groundwater models
Society of Petroleum Engineers: Reservoir resource definitions	Distinguish physical subsurface estimates from formal reserve classifications

Continue learning

Return to the Gold Standard of Petroleum Geology · How X Works hub · Gold Standard of Education

Contents · Previous chapter · Petroleum Geology guide

References checked 21 October 2026. Fictional worked examples state their simplifying assumptions so their calculations can be checked or challenged.