RESERVOIR
Tank models on the project fluid, from diagnostic plots to an aquifer history match and a forecast. When the data cannot support an answer, Converge says so instead of printing a number.
WHAT IT DOES
Material balance treats the reservoir as a tank: what came out, plus the expansion of what remains, must equal what was there to begin with, plus any water that flowed in. Paste the pressure and production history, and Converge computes the underground withdrawal F and the expansion terms from the project fluid, then plots them the ways that reveal the drive mechanism.
The Havlena-Odeh straight line gives the oil in place when the aquifer is right. Campbell and Cole plots show when it is not: a rising trend means influx not yet accounted for. You choose an aquifer model, and a global search fits its parameters while solving for N at every trial, so the match does not stall in a local minimum.
Every fit is checked for misfit, slope drift across the history and drive-index closure. A fit that fails is labelled not reliable, with the reasons, rather than shown as an answer. Randomised refits give a P90, P50 and P10 on the in-place volume, and the accepted tank predicts forward with relative permeability and a rate target.
KEY CAPABILITIES
One tank formulation for each fluid type, all reading the project fluid.
The classic straight-line forms, each clickable to exclude a point.
Five influx models, from the simplest to the full unsteady-state solution.
A global search finds the aquifer parameters; the in-place volume is solved exactly at every trial.
The matched tank forecasts forward under a rate target and a pressure floor.
Connect compartments that communicate through faults or a shared aquifer.
WORKFLOW
Paste pressure and cumulative production, with unit-aware headers, or send it from the Data module.
Check the Havlena-Odeh line and the Campbell and Cole trends, and click out any doubtful point.
Pick an aquifer model and run the history match. Review the reliability checks and the P90/P50/P10 band.
Accept a fit that passes; the in-place volume and aquifer go forward to prediction.
Set a rate target, pressure floor and relative permeability, and run the forecast.
Send the profile to Economics, Reserves or the integrated forecast.
VALIDATION
The synthetic tanks were built from a known oil in place, so the truth is exact. Two textbook examples check the aquifer influx and the withdrawal arithmetic.
| Check | Reference | Converge | Tolerance |
|---|---|---|---|
| Volumetric tank built from N = 45 MMSTB, no aquifer (V10) | 45.00 MMSTB | 45.03 MMSTB | 1% |
| The same tank with an aquifer it does not have (V11) | No wrong answer shown | Fit blocked as not reliable | — |
| History match on the sample tank (V12) | 150 MMSTB | Within tolerance | 3% |
| P90–P10 band contains the truth (V13) | 150 MMSTB inside the band | Band contains the truth | At least 80% coverage over 10 histories |
| Infinite radial aquifer influx, Ahmed (2006) Example 10-4 | Constant and influx at 6 and 12 months | Reproduced | 0.5% |
| Underground withdrawal F, every row, Ahmed (2006) Example 11-1 | Published table | Reproduced | 0.01 MMbbl |
Multi-tank transmissibility links are entered in rb/d/psi in this release, and links carry no gravity or capillary terms.
INPUTS AND OUTPUTS
CONNECTED MODULES
History arrives from Data, fluid properties from PVT, and the matched tank feeds the forecast and the money.
FAQ
No. Every fit is checked for misfit, slope drift across the history and drive-index closure. A fit that fails is labelled not reliable and the reasons are shown, instead of an answer.
Pot, Schilthuis, Fetkovich, van Everdingen-Hurst and Carter-Tracy. Fetkovich and van Everdingen-Hurst can be radial or linear, and van Everdingen-Hurst is computed by Stehfest inversion rather than read from tables.
By randomised maximum likelihood: 300 refits that perturb the pressures, initial pressure, PVT and rock compressibility. The spread gives the P90, P50 and P10.
Yes. Use Corey curves, or paste SWOF and SGOF tables; three-phase oil uses the Baker model.
Yes. Gas tanks have Cole and p/Z plots. Condensate tanks are handled on a wet-gas equivalent basis.