Skip to content

DRP-317 Sandstone Validation Overview

This report summarizes the current voids validation workflow for all eleven DRP-317 sandstone notebooks and all currently configured image-to-network extraction backends:

  • PoreSpy snow2, the historical baseline for these notebooks
  • PREGO, the native PREGO-style seeded region-growing backend
  • Native maximal-ball, the dependency-light maximal-ball backend in voids

The comparison is a workflow-level comparison: every backend starts from the same segmented DRP-317 image, ROI selection, boundary-pressure convention, fluid model, and conductance model, but each backend produces a different reduced pore network. Agreement or disagreement with experiment should therefore be interpreted as extraction-plus-solve behavior, not as isolated solver parity.

A focused set of same-ROI studies is also available for the newer map-based single-phase solvers:

Sources

  • Dataset: Neumann, R., ANDREETA, M., Lucas-Oliveira, E. (2020, October 7). 11 Sandstones: raw, filtered and segmented data [Dataset]. Digital Porous Media Portal. https://www.doi.org/10.17612/f4h1-w124
  • Experimental reference paper: Neumann, R. F., Barsi-Andreeta, M., Lucas-Oliveira, E., Barbalho, H., Trevizan, W. A., Bonagamba, T. J., & Steiner, M. B. (2021). High accuracy capillary network representation in digital rock reveals permeability scaling functions. Scientific Reports, 11, 11370. https://doi.org/10.1038/s41598-021-90090-0

Current Notebook Setup

Rendered results need refresh

Most DRP-317 notebook sources use the published-reference generic_poiseuille conductance model for all extraction backends. The Bentheimer report has additionally been rerun with the new PoreSpy/PREGO external-reservoir transport geometry and a same-network conductance-model audit. Re-run the full notebook batch before treating cross-sample trends as final.

All DRP-317 notebooks share the same current modeling choices:

  • a porosity-matched coarse ROI scan rather than a fixed centered ROI
  • extraction with backend="porespy", backend="prego", and backend="native_maximal_ball"
  • conductance model: generic_poiseuille for all extraction backends
  • Bentheimer additionally reports a same-network conductance audit for generic_poiseuille, hagen_poiseuille, valvatne_blunt, and auto
  • pressure-dependent water viscosity from the thermo backend
  • absolute outlet pressure 5.0 MPa
  • imposed pressure gradient 10 kPa/m
  • reported aggregate permeability as the quadratic mean across Kx, Ky, and Kz

Summary Table

The underlying summary CSV is committed as docs/assets/validation/drp317_summary.csv.

Sample Backend Exp. phi [%] Network phi [%] Exp. K [mD] Agg. K [mD] Rel. K error [%]
Bandera Gray PoreSpy snow2 18.10 20.70 9.0 49.99 455.40
Bandera Gray PREGO 18.10 20.40 9.0 95.13 957.01
Bandera Gray Native maximal-ball 18.10 20.40 9.0 22.74 152.68
Parker PoreSpy snow2 14.77 12.90 10.0 20.43 104.32
Parker PREGO 14.77 12.57 10.0 45.62 356.24
Parker Native maximal-ball 14.77 12.57 10.0 8.79 -12.14
Kirby PoreSpy snow2 19.95 22.00 62.0 127.01 104.85
Kirby PREGO 19.95 21.30 62.0 217.91 251.48
Kirby Native maximal-ball 19.95 21.30 62.0 63.73 2.79
Bandera Brown PoreSpy snow2 24.11 21.25 63.0 63.12 0.18
Bandera Brown PREGO 24.11 20.70 63.0 121.44 92.76
Bandera Brown Native maximal-ball 24.11 20.70 63.0 30.33 -51.85
Berea Sister Gray PoreSpy snow2 19.07 20.20 80.0 127.01 58.77
Berea Sister Gray PREGO 19.07 19.55 80.0 207.94 159.92
Berea Sister Gray Native maximal-ball 19.07 19.55 80.0 64.75 -19.06
Berea Upper Gray PoreSpy snow2 18.56 19.91 86.0 113.97 32.52
Berea Upper Gray PREGO 18.56 19.26 86.0 194.80 126.52
Berea Upper Gray Native maximal-ball 18.56 19.26 86.0 59.79 -30.48
Berea PoreSpy snow2 18.96 21.82 121.0 139.98 15.69
Berea PREGO 18.96 20.96 121.0 244.93 102.42
Berea Native maximal-ball 18.96 20.96 121.0 80.05 -33.84
Castlegate PoreSpy snow2 26.54 25.16 269.0 296.91 10.37
Castlegate PREGO 26.54 24.29 269.0 515.24 91.54
Castlegate Native maximal-ball 26.54 24.29 269.0 179.69 -33.20
Buff Berea PoreSpy snow2 24.02 23.40 275.0 313.23 13.90
Buff Berea PREGO 24.02 22.56 275.0 594.86 116.31
Buff Berea Native maximal-ball 24.02 22.56 275.0 217.36 -20.96
Leopard PoreSpy snow2 20.22 20.01 327.0 321.66 -1.63
Leopard PREGO 20.22 19.30 327.0 580.20 77.43
Leopard Native maximal-ball 20.22 19.30 327.0 220.38 -32.61
Bentheimer PoreSpy snow2 22.64 27.57 386.0 465.11 20.49
Bentheimer PREGO 22.64 26.53 386.0 877.84 127.42
Bentheimer Native maximal-ball 22.64 26.53 386.0 292.38 -24.25

Figures

DRP-317 permeability scatter

DRP-317 permeability comparison

DRP-317 porosity comparison

Interpretation

Across all eleven samples, backend choice is now a first-order part of the validation result. The PoreSpy snow2, PREGO, and native maximal-ball networks can have similar image porosity context but substantially different pore counts, throat counts, coordination statistics, and permeability estimates.

These pages should therefore be read as a backend-sensitivity study for the current voids image-to-network workflow. The comparison does not establish that one backend is physically correct in isolation; it identifies how much the final single-phase permeability depends on the extraction/reduction stage for each sample.

Sample Reports