Page 160 - KAPPA - Ecrin v4.30 tutorials

Ecrin v4.20 - Doc v4.20.01 - © KAPPA 1988-2011 Saphir Guided Session #2 • SapGS02 - 6/8
B02 • Kappa intelligent well test interpretation (KIWI)
KIWI is an Artificial Intelligence (AI) module, an interpretation model advisor, aimed at helping
you to make quick interpretations by suggesting possible analytical models and estimating the
relevant parameters. It can also be used as a training tool or simply to 'ring a bell' from time
to time and remind you of possible alternative interpretation models.
KIWI does NOT have ANY knowledge of the geology, the geophysics, the history of the field or
any other information on the reservoir and fluids. KIWI bases its analysis purely on the shape
of the derivative and gives you a list of models (with parameters) that may reproduce the
shape of the derivative, it is your decision as to what makes sense physically.
KIWI is NOT a substitute for user judgment and common sense. The only time that it does not
give you a CHOICE of solutions is in the case of homogeneous infinite behavior.
KIWI is only enabled when there is a single dP set displayed. It is disabled if several gauges or
several groups are displayed.
The KIWI option is not enabled in case of Multilayer, NonLinear or Formation Test analysis.
When KIWI is invoked the first time, a loglog plot is created on which the trend identification is
done. The data is plotted together with a sketch (in white) and a spline (in yellow). The cubic
spline fit is made on the sketch and non linear regression minimizes the curvature. After this
pre-processing, the spline provides a (hopefully) smooth description of the response that can
be used for model identification. The response is represented as 'sections' characterized as Up,
Down, Flat, etc.
Immediately after, the symbolic description and the spline are passed to the knowledge base
module. The description of analytical model components is embedded in this module together
with rules defining how they can be combined. In addition, each model is associated with
quantitative constraints expressing slopes of characteristic regimes, etc. Upon completion of
the search, a dialog pops up listing the possible interpretation models found by KIWI.
When a KIWI plot already exists, pressing the 'KIWI' icon of the control panel calls the dialog
with the list of solutions. KIWI specific options can be accessed directly in the plot pop-up
menu (right-click on the plot).
Choose a new analysis by clicking on the tab New , choose to start 'from nothing' and
re-extract the loglog plot, hitting the Extract dP button .
Open the panel (page) and click on the KIWI icon. The loglog plot with its
sketch and spline is made automatically and transparent to you but immediately you will see
the suggested list of models. In the list illustrated in Figure B02.1 it can be observed that the
first choice made by KIWI is indeed a double porosity model, however the PSS model is only
the third choice.