Volve field data

Contains data from the Volve field data set, (c) Equinor ASA and the former Volve licence partners
ExxonMobil Exploration & Production Norway AS and Bayerngas Norge AS (or their successors), used
under Equinor's "Terms and conditions for licence to data - Volve":
https://equinoropendata.blob.core.windows.net/disclaimers/HRS%20and%20Terms%20and%20conditions%20for%20license%20to%20data%20-%20Volve.pdf
A dated copy of the terms is in this folder: TERMS-Volve-2026-10-07.txt.

This folder is Adapted Material under those terms. It is shared free of charge and is not for sale.
It is not connected with, sponsored or endorsed by Equinor or the former Volve licence partners.
Every recipient may use it under the same terms; the template's MIT licence covers its code only,
never this folder.

Where the files come from
Equinor now distributes Volve through a Databricks Marketplace listing. These files were built from
public copies of Equinor's own files, each pinned by commit or URL and by sha256
(volve/pipeline/sources.json):
  deck and includes   github.com/tpp-grupo-166/opm-datasets at 1c323c2 (folder volve/), the files as
                      first mirrored, before that repository's own edits for OPM
  Equinor's run       github.com/f0nzie/rEcl at 4b187da: VOLVE_2016 SMSPEC, UNSMRY, INIT, RSSPEC
                      github.com/f0nzie/volve_eclipse_reservoir at 0d34eaf: VOLVE_2016.PRT (Eclipse 2015.1)
  wells, production   github.com/awgeo/Volve_field_data at 7cdecf1: 21 'ACTUAL' deviation surveys and
                      'Volve production data.xlsx'
                      github.com/andymcdgeo/spwla_volve at c293edf: Well_picks_Volve_v1.dat
  seismic, horizons   OSDU's public test-data bucket (osdu-seismic-test-data, volve/): the ST0202R08 PS
                      PSDM full-offset depth stack (read in part, by range), Hugin Fm top and base
The deck is checked against Equinor's own run before use: every line Eclipse echoed into Equinor's
VOLVE_2016.PRT is found, in order, in the fetched files, and the model OPM builds from them matches
the depth of all 183,545 active cells in Equinor's VOLVE_2016.INIT to 0.01 m, and its porosity,
permeability, pore volume and regions.

What we changed (clause 3.2: every adaptation, stated)
1. The simulation deck (VOLVE_2016.DATA and includes), before our run:
   - Report settings only: restart files every month (the deck wrote every 4th, then every 3rd);
     RPTSCHED's RESTART mnemonic removed; WGPT (gas produced per well) added to the summary output.
   - Lines Eclipse itself discarded when Equinor ran the deck, which OPM rejects instead (each one
     logged in Equinor's PRT as "SPURIOUS DATA"): the 13th ROCK row and the 13th RSVD table (the deck
     declares 12 PVT regions), and five producer records at 1 Jan 2015 that have no keyword above
     them.
   - What OPM Flow 2026.04 cannot run: the 100 ADDZCORN records (fault throws) are applied to the
     grid's corner depths in advance, exactly as Eclipse applied them (proved against Equinor's INIT),
     and the keyword commented out; PRIORITY is removed (the deck has no group production target it
     could act on); EQLOPTS keeps THPRES and loses QUIESC and MOBILE, which OPM does not have. Without
     MOBILE, 1,200 cells keep the deck's SOWCR and SOGCR instead of Eclipse's corrected values, and
     the initial state can differ slightly from Equinor's.
2. Saturations and pressures (in the model files, once the workflow has made them) come from our
   own OPM Flow 2026.04 run of that deck (31 Dec 2007 to 1 Oct 2016), not from Equinor's run. It is
   compared with Equinor's Eclipse 2015.1 summary vectors in validation.json; the model is only
   extracted when every check there passes.
3. The model files repack the run: corner coordinates in ED50 / UTM zone 31N (the grid's MAPAXES
   applied) as float32 relative to a centre point; quarterly frames (the first restart in each
   calendar quarter, plus the first and last); saturations quantized to 8 bits and pressure to 16.
4. The seismic (survey ST0202, the 2002 baseline; the PS PSDM full-offset stack in depth, 2008
   processing) is cropped to the model (reservoir top - 500 m to base + 500 m, footprint + 250 m),
   low-pass filtered across inlines and crosslines and decimated from 12.5 m to 25 m traces, and
   quantized to 8 bits with zero at code 128 and a symmetric clip at the 99.9th percentile of
   |amplitude| (0.104697; 0.1 % of samples reach it). Depth samples are kept at their original
   5 m interval and original depths: nothing is shifted, stretched or tied to the model.
5. The Hugin Fm top and base horizons (interpreted on survey ST10010, 2011 processing, and adjusted
   to the wells) are sampled at the 25 m trace positions.
6. Well paths are converted from depth below the rotary Kelly bushing to depth below mean sea level,
   using the KB elevation in Equinor's well picks (54.9 m); production is the monthly sheet,
   re-encoded; values are rounded to 0.01 m and 0.1 Sm3.

Files
  model.json       dimensions, map frame (crs, mapaxes), frame dates and steps, wells (path,
                   completions, role and history targets per frame), rates (Sm3/day, averages between
                   frames; simulated, and history), the comparison with Equinor's run
  geometry.bin     float32, 8 corners x (x, y, depth) per active cell, metres, relative to model.center
  neighbours.bin   int32, 6 face neighbours per active cell (-I, +I, -J, +J, up, down), -1 = open face
  ijk.bin          uint8, 0-based I, J, K per active cell
  static.bin       float32 blocks of PORO, PERMX, PERMZ, NTG (1: the deck has none), FIPNUM, DEPTH
  dynamic.bin      per frame: SWAT uint8 (/255), SGAS uint8 (/255), PRESSURE uint16 scaled to
                   dynamic.pressureRange
  validation.json  our run against Equinor's: field and well totals, field pressure, active cells
  seismic.bin      uint8, inline-major (inline, crossline, depth); amplitude = (code - 128) / 127 * clip
  seismic.json     the seismic's grid (origin, inline and crossline vectors, corners, depth axis),
                   clip, polarity, filter, crop and sources
  horizons.bin     float32, Hugin Fm top then base, depth in metres on the seismic's grid, NaN = no pick
  wellpaths.json   21 deviation surveys (MD; x, y; depth below sea level) with Equinor's well picks
  production.json  monthly oil, gas, water, gas and water injection (Sm3) and hours per wellbore
  TERMS-Volve-2026-10-07.txt   Equinor's terms, as retrieved on 7 Oct 2026
