The default well test derivative may be misleading !
In some programs such as Kappa Saphir, there is a default smoothing on the well test derivative that could mask a particular problem with the data. Worse, it could create some erroneous features in the default derivative that could be misinterpreted as flow regimes or a wrong well behaviour.
Saphir, for example, uses a default smoothing coefficient of 0.1. So the data are not the raw ΔP plot and derivative but smoothed responses. As a result, the analyst may not spot some data quality issues, tidal effects, wellbore phase redistribution that could corrupt the entire PBU, etc…
Tidal effects may not be observed with the default derivative:
And below is the derivative that we should see instead. This derivative -with no smoothing- shows some problem with the data, due to the tidal effects.
These effects mask the pressure transient response and need to be removed from the data in a proper way.
The plot below shows another example.
While the change in the derivative shape is a bit abrupt, this could be mis-interpreted as a decrease in reservoir properties, a boundary or a baffle.
If we set the smoothing coefficient to 0, the derivative becomes:
The data after 48 hours are affected by a valve handling issue and is therefore not representative of the well and reservoir. Similarly wellbore phase redistribution may not be detected.
On some cases, the default smoothing could create some artificial features that could be misleading. In this next example, we see a fall in the derivative instead of the correct stabilization.
We need to be cautious with the derivative plot when a default smoothing is applied.
Poor quality data, tides, noise, wellbore phase redistribution may then be difficult to spot. Worse, the derivative may be misleading and not representative of the well and reservoir.
Conclusion: always set the smoothing coefficient to 0. Better to check an “unprocessed” derivative first.
