LWCH7T9 compositional model: assessment and recommended simulation approach
Prepared 2026-09-21. Simulator: bxuSim. Draft for discussion.
1. Summary
Section titled “1. Summary”The LWCH7T9 deck was audited end to end, its grid and initialisation were validated, and its fluid description was analysed. Three things came out of that work.
The deck contains three data defects that will affect any simulator. One of them, an array-length error in the permeability field, is a hard stop. These are documented in section 3 with exact fixes.
The reservoir is under pure depletion. Six producers, no injection of any kind, 195 months of history. No gas cycling, no miscible displacement, no pressure maintenance.
Recommendation: run this model in a black-oil formulation. For a depletion-only history, an extended black-oil description reproduces the phase behaviour of these fluids along the pressure path the reservoir actually follows, at roughly an order of magnitude less cost per time step. Section 5 sets out the technical case; section 6 sets out what the conversion involves.
2. The model as received
Section titled “2. The model as received”Petrel 2021.5 export, FIELD units.
| Item | Value |
|---|---|
| Grid | 134 × 113 × 457 corner-point, 6,919,894 cells |
| Active cells | 1,215,707 after minimum-pore-volume screening, 17.6% |
| Fluid description | Four-component equation of state, Peng-Robinson with the modified form, three EOS regions |
| Components | C1+, C2+, C3+, C7+ |
| Equilibration | 145 regions, each with a composition-versus-depth and a temperature-versus-depth table |
| Datum | 11,527 ft at 9,127 psia |
| Temperature | 118 °F at 4,872 ft to 263 °F at 21,325 ft |
| Wells | 6, all producers |
| Well control | 388 records, all on liquid rate, with bottom-hole pressure limits |
| VFP tables | 4 |
| History | 2007-06-01 to 2023-08-01, 195 monthly steps |
| Injection | none |
What the history match consists of
Section titled “What the history match consists of”The model name carries the history-match case number, F304, and the schedule section defines the match. All 388 well control records use the WCONHIST keyword in liquid-rate mode: each month, each well is forced to produce its observed total of oil plus water. The simulator then predicts three things from the reservoir description: how that liquid splits between oil and water, how much gas comes with it, and how low the bottom-hole pressure must fall to deliver the rate. Water cut, gas-oil ratio and bottom-hole pressure are therefore the quantities on which the match is judged.
The bottom-hole pressure limit is set to 150 psia on every well, which in practice means the rate target is always honoured.
One point to note for validation: the summary section requests no observed-value vectors, so the deck carries no measured bottom-hole pressures against which the predicted ones can be compared. The observed data exist only as the rates in the control records. Any assessment of match quality on this deck is limited to the oil-water split and the gas rate. If measured bottom-hole pressures exist in the source project, exporting them adds a direct check.
Grid and initialisation are sound
Section titled “Grid and initialisation are sound”The grid loads, the 145 equilibration regions initialise, and the resulting pore volume is 6.39 × 10⁹ rb. The structural description, the fault set, the rock properties and the region definitions all pass without complaint once the data defects in section 3 are corrected.
3. Deck integrity findings
Section titled “3. Deck integrity findings”These are properties of the exported data, not of any particular simulator. All three should be corrected in the source project so that future exports are clean.
3.1 The permeability array has one value too many
Section titled “3.1 The permeability array has one value too many”LWCH7T9_HM_F304_211TRANS_1_PROP_PERMX.GRDECL expands to 6,919,895 values. The grid has 6,919,894 cells. Every other property array in the deck is exactly correct:
| Array | Values | Status |
|---|---|---|
| PERMX | 6,919,895 | one too many |
| PERMY, PERMZ, PORO, NTG, ACTNUM, MULTX, MULTY, MULTZ | 6,919,894 | correct |
The surplus value is in the final repeat count of the last data line. The last record reads ... 0.00552523 37740*0 / where 37739*0 would close the array exactly. Because the error is in the last line, every preceding value is correctly placed, so the fix is a single edit and no property data is displaced.
This is a hard stop: a simulator cannot know which value to discard.
3.2 Two conflicting minimum-pore-volume cut-offs
Section titled “3.2 Two conflicting minimum-pore-volume cut-offs”The deck sets a minimum pore volume of 50 rb in the main file and 0.629 rb in the grid include file. The two settings apply to the same screening step and differ by a factor of eighty. Whichever is honoured determines how many cells survive, so the ambiguity propagates into every volumetric result.
Only one should be kept. Given that 50 rb is the value carried in the main data file and 0.629 rb appears to be an automatically written default, 50 rb is the more likely intent, but that should be confirmed against the Petrel project.
3.3 Grid file reference lacks its extension
Section titled “3.3 Grid file reference lacks its extension”The grid is imported by file name without the .EGRID suffix. Simulators differ in whether they append the extension automatically. Writing the full file name removes the ambiguity.
3.4 Two well dimension entries are left at their defaults
Section titled “3.4 Two well dimension entries are left at their defaults”The maximum well count and the maximum group count are both defaulted. The model has 6 wells and 3 non-field groups. Where a simulator interprets the default as zero rather than as unlimited, the run stops at the first well. Writing the actual counts avoids this.
3.5 Summary of corrections
Section titled “3.5 Summary of corrections”| Location | Change |
|---|---|
| PERMX file, last data line | 37740*0 to 37739*0 |
| Main file or grid include | keep one minimum-pore-volume setting |
| Grid file reference | append .EGRID |
| Well dimensions | write well count 26 and group count 8, or values of your choice above 6 and 3 |
4. Fluid system
Section titled “4. Fluid system”The 145 equilibration regions carry two distinct compositions.
| Composition | Regions | C1+ | C2+ | C3+ | C7+ | Character |
|---|---|---|---|---|---|---|
| A | 106 | 0.8615 | 0.0373 | 0.0427 | 0.0585 | Gas condensate |
| B | 39 | 0.6696 | 0.0276 | 0.0650 | 0.2378 | Volatile oil |
At the datum, 9,127 psia and roughly 186 °F, composition A is a rich gas and composition B is an undersaturated volatile oil. Three EOS regions are defined; regions 1 and 3 share the same critical properties, so there are two distinct fluid characterisations in the model, matching the two compositions.
This matters for the recommendation in section 5: both fluids sit close enough to their critical points that their phase behaviour is genuinely pressure sensitive, which is why the depletion path, rather than the fluid type alone, decides which formulation is appropriate.
5. Recommended approach: black oil
Section titled “5. Recommended approach: black oil”5.1 The case for it
Section titled “5.1 The case for it”The reservoir is depletion-only. Six producers, 195 months of history, zero injection records of any kind. Pressure declines monotonically from its initial state. Every cell therefore traverses a single, one-directional pressure path.
Compositional simulation earns its cost when composition changes in ways a pressure-indexed table cannot capture: gas cycling that revaporises condensate, miscible or near-miscible displacement, CO₂ or enriched-gas injection, or compositional grading that is actively disturbed. None of these occur here.
For depletion of a gas condensate or a volatile oil, an extended black-oil description carries both directions of mass transfer, dissolved gas in the oil and vaporised oil in the gas, as functions of pressure. Tables generated from the deck’s own Peng-Robinson description along the depletion path reproduce the phase behaviour that this reservoir actually experiences. This is long-established practice for depletion studies of near-critical fluids.
The cost difference is large. A compositional formulation solves a phase-equilibrium problem in every cell at every non-linear iteration. With 1.2 million active cells and four components, that is the dominant cost. Black oil replaces it with table interpolation.
The companion black-oil model on this field, LM_HM116, gives a measured reference point: 668,529 active cells, 545 report steps spanning 45 years, completed in 50 minutes 44 seconds on 16 parallel processes. LWCH7T9 is roughly twice the active-cell count over one third of the report steps, so the same class of run time is a reasonable expectation once converted.
5.2 What is preserved and what changes
Section titled “5.2 What is preserved and what changes”| Preserved exactly | Changed |
|---|---|
| Grid geometry, faults, non-neighbour connections, pinch-outs | Fluid description: EOS replaced by pressure-indexed PVT tables |
| Porosity, permeability, net-to-gross, transmissibility multipliers | |
| Relative permeability and capillary pressure tables, end-point scaling | |
| All 145 equilibration regions and their contacts | Composition-versus-depth replaced by dissolved-gas and vaporised-oil versus depth |
| Region definitions: saturation, rock, fluid-in-place | |
| All 6 wells, their completions, and all 388 history records | |
| Temperature-versus-depth | Retained where the PVT tables are generated at the corresponding temperatures |
5.3 Where the approximation is weakest
Section titled “5.3 Where the approximation is weakest”Stated plainly so that it can be checked rather than assumed.
The black-oil tables are generated along one depletion path. Cells that deviate substantially from that path, for example a cell that repressurises because a neighbouring compartment is still at initial pressure, will be described less accurately than cells that follow it. Given the compartmentalised nature of this field, that situation is worth checking once the model runs, by comparing the pressure history of cells in well-drained compartments against those in undrained ones.
The two compositions should be given their own PVT region sets rather than being averaged into one. With 106 regions on composition A and 39 on composition B, the split is clean and the existing region arrays already carry it.
6. Conversion procedure
Section titled “6. Conversion procedure”- Correct the deck defects in section 3 in the source project and re-export.
- Generate PVT tables from the existing EOS. Using the deck’s own Peng-Robinson characterisation, including the volume-shift and binary-interaction parameters already present, run a constant-volume depletion for composition A and a differential liberation for composition B, over the pressure range the reservoir traverses, at the temperature of each region set. Fit dissolved-gas and vaporised-oil tables plus water properties.
- Assemble the black-oil deck. Replace the compositional section with the tables. Declare the three phases and both directions of mass transfer. Keep every other section unchanged.
- Convert the initialisation. The 145 equilibration regions are retained. The composition-versus-depth tables are replaced by dissolved-gas and vaporised-oil versus depth, derived from the same EOS at the same depths.
- Run and validate. Compare against the ECLIPSE reference described in section 7.
Steps 2 and 4 are standard PVT-package work and use tooling already available on the project.
7. Validation plan
Section titled “7. Validation plan”The conversion should be validated at two levels.
Fluid level. Compare the generated tables against the EOS they came from: formation volume factors, dissolved-gas ratio and vaporised-oil ratio at a set of pressures spanning the depletion range, for both compositions. This confirms the tables reproduce the EOS before any flow simulation is involved.
Field level. Run the converted deck and compare cumulative oil, gas and water, field average pressure, and the bottom-hole pressure of each of the 6 wells against an ECLIPSE run of the same converted deck. This separates any difference in the flow solution from the difference introduced by the fluid description.
A reference run is available: the earlier ECLIPSE attempt on the companion model failed only on the parallel-options licence feature, with the data check itself passing. Removing the parallel keyword and running serially produces the reference.
8. Recommended next steps
Section titled “8. Recommended next steps”- Correct the three deck defects and re-export from the source project. The permeability array error blocks any simulator and should be fixed first.
- Confirm which minimum-pore-volume value is intended. This changes the active cell count and therefore every volumetric result.
- Generate the black-oil tables for the two compositions and assemble the converted deck.
- Obtain the ECLIPSE reference for the converted deck.
- Run and compare, then proceed to history matching on the black-oil model.
Appendix: what was verified
Section titled “Appendix: what was verified”| Item | Method | Result |
|---|---|---|
| Grid loads and initialises | Full initialisation of all 145 equilibration regions | Pass, 1,215,707 active cells, 6.39 × 10⁹ rb pore volume |
| Property array lengths | Token expansion of all nine grid property files against the cell count | One defect found, PERMX |
| Well inventory | Well specification and history records parsed | 6 producers, 388 records, all on liquid rate, no injection |
| History span | Date records | 2007-06-01 to 2023-08-01, 195 steps |
| Fluid characterisation | Composition-versus-depth tables and EOS region parameters | Two distinct compositions, gas condensate and volatile oil |
| Keyword inventory | Every keyword in the deck classified | Listed in section 3 where action is needed |