RESERVOIR
Pressure-transient analysis from gauge data to a regressed analytical model. Read the Bourdet derivative, draw the straight lines, deconvolve the full rate history and fit a model with confidence intervals on every parameter.
WHAT IT DOES
A well test changes the rate and records how pressure responds. The shape of that response, read on a log-log plot of pressure change and its Bourdet derivative, tells you about wellbore storage, skin, permeability, boundaries, natural fractures and hydraulic fractures.
Converge takes the derivative against superposition time, so multi-rate histories plot correctly, flags end effects and names each flow regime in time order with a confidence. Quick estimates of permeability, storage and skin come straight from the regimes, and Horner and MDH straight lines confirm them.
You then build an analytical model from a well type, a reservoir type and a boundary, computed in Laplace space and inverted by Stehfest, and regress it to the data by Levenberg-Marquardt. Every fitted parameter comes with a 95% confidence interval, and deconvolution recovers the constant-rate response hidden in the whole rate history.
KEY CAPABILITIES
The derivative and the regimes, before any model is chosen.
The semilog methods engineers check by hand, done for you.
The near-well geometry, with constant or changing storage.
Heterogeneity and limits, combined freely with the well model.
Fit the model to the data and say how sure the fit is.
Gas tests on the right basis, and a report that matches the screen.
WORKFLOW
Paste the gauge record and rates. Shut-ins are found from the pressure, and you pick the period to analyse.
Check the derivative, the regimes and the quick estimates of permeability, storage and skin.
Confirm with Horner or MDH, and read p*.
Recover the constant-rate response from the full rate history, to see further in radius.
Combine well, reservoir and boundary models, then regress them with confidence intervals.
Write the report with the fitted values, intervals and plots.
VALIDATION
A 48-hour drawdown at 800 STB/d followed by 72 hours shut in, generated from the line-source solution with superposition. The known answer: k = 180 md, skin +4.5, initial pressure 3,650 psia.
| Check | Reference | Converge | Tolerance |
|---|---|---|---|
| Horner slope (V19) | 26.43 psi/cycle | 26.46 psi/cycle | 3% |
| Horner permeability (V19) | 180 md | 179.8 md | 3% |
| Extrapolated pressure p* (V20) | 3,650 psia | 3,650.0 psia | 2 psi |
| Skin, Horner (V21) | +4.5 | +4.47 | ±0.3 |
| Permeability and skin, log-log derivative (V22) | 180 md, +4.5 | Within tolerance | 3%, ±0.3 |
| Deconvolution of a multi-rate test | Known unit-rate response | Recovered | 5%, or 10% with 0.05 psi noise |
| Earlougher (1977) Example 5.1 | k = 12.8 md, skin 8.6 | Both matched | 5% |
The model library is also checked against the published analytical limits of Theis, van Everdingen and Hurst, Gringarten, Ramey and Raghavan, Cinco-Ley and Samaniego, and Warren and Root.
INPUTS AND OUTPUTS
CONNECTED MODULES
Permeability and skin from a test calibrate the well and reservoir models.
FAQ
Yes. The derivative is taken against superposition time, so multi-rate histories plot correctly, and deconvolution recovers the constant-rate response from the whole history.
Each comes with a 95% confidence interval. The intervals are linearised, so they can understate the spread when parameters are strongly correlated, for example storage and skin.
Yes, on normalised pseudo-pressure. Pseudo-time and rate-dependent skin are not built yet.
Yes. Shut-in periods are detected from the pressure alone; you type the rates for each period.
Homogeneous, Warren-Root and transient-slab dual porosity, and two-zone radial composite, with infinite, fault, channel, rectangle and circle boundaries. Multilayer models are not built yet.