Tomographic Pore-Network Imaging¶

Structural characterization test — instantiates Functional Porosity Design

Reconstructs the real three-dimensional void network from X-ray slices — actual connectivity, constrictions, and dead ends — instead of trusting a bulk average.

Tomographic Pore-Network Imaging measures the void architecture spatially. From many X-ray projections it reconstructs the pore space in three dimensions, voxel by voxel, then extracts the network as a graph of pore bodies joined by throats. The defining idea is that it sees where the porosity is, not just how much — the actual connectivity, the constrictions that throttle flow, the dead-end pockets that hold volume but carry nothing, and localized defects — all of which a bulk-average metric collapses into a single number and hides. It is how you learn why two bodies with identical porosity behave nothing alike.

Example¶

Two sandstone cores from the same reservoir report the same bulk porosity, yet one flows oil freely and the other barely at all. Micro-CT explains it. Each ≈2 cm core is scanned at ≈3 µm voxels, the volume is segmented into pore and grain, and a pore-network model is extracted — pores as nodes, connecting throats as edges. The free-flowing core turns out to have a well-connected network with wide throats; the poor one has the same void volume locked in dead-end pores and pinched throats that the average could never reveal. From the reconstructed geometry the tool computes connectivity, tortuosity, and an effective permeability directly, turning "equal porosity" into a mechanistic account of the difference.

How it works¶

The distinguishing pipeline is reconstruct → segment → skeletonize. Many projections taken at rotated angles are reconstructed into a 3D grayscale volume; a threshold segments it into pore and solid; then the pore space is reduced to a network graph — pore bodies and the throats between them — or fed to direct simulation. From that geometry the tool reads coordination number, throat-size constrictions, dead ends, tortuosity, and a representative volume. The differentiator against bulk methods is completeness in space: nothing is averaged away, so anisotropy and local defects survive into the result.

Tuning parameters¶

Voxel resolution vs field of view — finer voxels resolve smaller throats but shrink the volume imaged; the central trade, since throats below the voxel size simply vanish.
Segmentation threshold — where pore is cut from solid moves the measured porosity and connectivity, and is the single largest source of bias.
Imaged volume vs representative volume — the scan must span at least a representative volume or its numbers will not generalize.
Contrast method — added contrast agents or dual-energy scanning to separate phases of similar X-ray absorption.
Static vs in-situ (4D) — a single snapshot, or time-lapse imaging under load or flow to watch the network evolve.

When it helps, and when it misleads¶

Its strength is being the only method that exposes the actual network topology — dead ends, constrictions, anisotropy, and localized defects — so it explains behavior that bulk numbers cannot and grounds every geometry-based transport model. It misleads through its resolution limit: throats near or below the voxel size fall victim to the partial-volume effect and are mis-segmented or lost, systematically making the network look more disconnected than it is.^[1] The segmentation threshold silently sets the porosity, a small imaged volume may not be representative, and because scans are slow and costly people over-generalize from one. The discipline is to report resolution and segmentation choices, confirm the volume is representative, and cross-check the geometry against a functional measurement rather than trusting the picture alone.

How it implements the components¶

Tomography fills the components read off the reconstructed 3D geometry:

connectivity_and_throat_topology — its core output: the actual pore-and-throat graph, with coordination, constrictions, and dead ends made explicit.
tortuosity_and_effective_path_model — it computes true path length and tortuosity directly from the reconstructed network.
multiscale_representative_volume — it extracts a representative volume and its effective properties empirically, from the real image rather than an assumed cell.

It does not create any voids — that is the fabrication methods — nor measure the bulk pore-size distribution and surface area by intrusion or adsorption, which is Multi-Method Porometry; and it does not measure live flow, storage, or breakthrough performance, which is Transport, Storage, and Breakthrough Testing.

Instantiates: Functional Porosity Design — supplies the as-built 3D network geometry the design must be checked against.
Consumes: a fabricated specimen from any of the making methods, whose network it then reconstructs.
Sibling mechanisms: Transport, Storage, and Breakthrough Testing · Multi-Method Porometry · Mechanical Coupon and Fatigue Testing · Perforation, Microchanneling, or Drilling · Phase Separation and Selective Extraction

Notes¶

Imaging reveals the geometry, not the function. A network that looks well connected on the reconstruction can still conduct poorly because of sub-resolution constrictions or wettability effects the image cannot show, so a tomographic result should be paired with Transport, Storage, and Breakthrough Testing before its connectivity is trusted as performance.

References¶

[1] Partial-volume effect — a voxel straddling pore and solid records an averaged intensity, so features near or below the voxel size are mis-segmented or lost; a tomographic network is trustworthy only for features spanning several voxels, which is why sub-resolution throats must be confirmed by a functional flow measurement. ↩