Whenever a fluid is produced or injected into the
reservoir, the diffusion produces changes in pressure
and temperature that may be recorded in various
places. Combine this data with the production and
/ or injection rate history and we have what we call
Not so long ago, dynamic data came only from well
tests. Analysis comprised specialized plots and
dedicated analytical models. For KAPPA, this was the
time of Saphir, our then ‘welltest analysis’ software.
Today the sources of dynamic data are multiple and
apply at various scales in time and space. Analyses
are also more complex. Different methodologies
have to be applied, sometimes using the same data,
sometimes not. However, all these techniques apply
to the same reservoir and wells, and the truth is in the
data. The name of the game today is to build a puzzle
from little pieces coming from everywhere.
There is a long list. It starts with pressure transient
analysis (PTA), the new name for welltest analysis, and
production analysis (PA). On the scale and lifespan of
the reservoir we can history match (HM) production,
pressure and temperature data.
It is also possible to obtain a vertical profile of the field
contribution with production logging (PL) and formation
tests (FT). To bring all this to datum the output of a well
performance analysis (WPA) tool is required.
For KAPPA this is the time of the integrated Ecrin
suite. Ecrin is the software environment under which
all KAPPA dynamic data analysis modules operate.
By running under a single executable Ecrin provides
complete interconnectivity between the modules and
allows data sharing and technical objects by drag-and-drop,
saving time, repetition and frustration.
Permanent downhole gauges (PDG) constantly monitor
downhole pressures and are a data-rich witness of
what is happening in the well and the reservoir. PDGs
provide long-term data to run production analysis and
short-term data, from incidental shut-ins, to perform
pressure transient analysis.
In the early 2000’s the initial bubble of excitement
about PDG data burst quickly: the data stored in the
historians were huge. If engineers could find the data
in the first place, it ground their computers to a halt,
often ending with the ‘blue screen of death’.
Work at Stanford University produced smart, wavelet
based filters that drastically reduce the number of
points without eroding the data signature.
KAPPA initially released a stand-alone application
(Diamant) to perform this task and subsequently
developed this to an asset-wide shared environment
accessing multiple data sources.
Today Diamant Master is a client-server solution
that loads, processes and shares standardized data
within a workgroup and establishes real-time links
with Intelligent Fields.
Pressures and temperatures are processed using
wavelets. Definition of rates, previously requiring
heavy manual intervention to correct inaccurate
reallocation processes, is solved with a new algorithm
that identifies shut-ins with a high and useable level
This missing link in automating the process of creating
build-up files in the intelligent field environment is
Sharing data and technical objects
One of the ways to enhance the workflow is to avoid
duplication. Whenever possible Diamant Master and
the Ecrin modules will share and/or transfer data and
technical objects. Any dynamic data acquired and
filtered by Diamant Master is immediately accessible
to the relevant Ecrin modules. One can also drag-anddrop
data loaded by an Ecrin module into Diamant
Master, making this data available to all workgroup
users with the right privilege, or to another Ecrin
module, to a different document from the same
module, or within the same document.
The same ease of manipulation applies to technical
objects such as PVT, reservoir maps, relative
permeability tables, intake curves, IPR’s, analytical
models and numerical models.
PVT objects may be tables, black oil correlations or
an EOS. In some environments (Saphir NL, Topaze
NL, Rubis) this PVT may be isothermal. In other
modules (Emeraude, Amethyste, Rubis) temperature
dependency may be required. When dragged and
dropped from an environment to another a PVT object
will adapt and request what it takes to switch from
isothermal to non-isothermal, or vice versa.
Analytical models may be exchanged between
Saphir NL and Topaze NL. It is actually possible to
convert a whole document, with information, data and
models, from PTA to PA or vice versa.
Numerical models may be exchanged between
Saphir NL, Topaze NL and Rubis. The gridding will
dynamically adapt to its new use and environment.
Numerical models are of increasing importance,
especially in the context of unconventional resources.
IPR’s may be created in Saphir NL and transferred to
Amethyste, or the opposite. Well intake curves and
well models calculated by Amethyste can be sent to
and used by Saphir NL, Topaze NL and Rubis.
Sharing data and technical objects goes beyond
KAPPA applications. Diamant Master already has
plug-ins to connect to Intelligent Field models and
third party databases. In its next generation Diamant
Master will also offer OpenServer APIs for third party
applications to retrieve data and Ecrin analyses.
Looking for a numerical reservoir proxy
With Ecrin we seek, step by step, to build an
understanding of the reservoir and its wells from the
various dynamic data available. First we use the data
for analysis, then we use the result to history match all
available data. Finally we want to forecast the future.
One way or another we need to feed a unique model,
which we will use as a proxy of the reservoir.
One such a proxy could be the geological model.
But this will seldom work and, arguably, even when
it does, there is generally no time to use and update
it in a practical sense. At the other end of the scale
an alternative might be to use simple proxies such
as decline curves or analytical models, but these are
an over simplification and not up to the increasing
complexity of drive mechanisms.
KAPPA’s position is that such a proxy has to be an
intermediate numerical model between a single tank
material balance and the upscaled geological model.
Such a proxy should be complex enough to reproduce
the main reservoir drives, but simple enough to allow
shorter, practical work cycles and above all that it must
honor the dynamic data.
This numerical model also needs to ‘talk’ to each
analysis module, to allow corroboration between the
various methods and data sources. This is where the
specific nature of the numerical tools shared by Saphir
NL, Topaze NL and Rubis can be a game changer.
This original specification of Rubis was and remains
diametrically opposite to the development of the next
generation of zillion cell simulators with massive
parallel processing. The solution is as simple as is
necessary to solve the problem... but not simpler.
KAPPA numerical models
When the first numerical model was released in
Saphir the relatively simple objective was to simulate
linear diffusion problems for geometries too complex
for analytical tools. It used the unstructured Voronoi
grid and could generate models on a logarithmic time
scale with a speed and accuracy comparable to an
analytical model. This allowed interpretations to be
achieved within the time frame usually allocated to
making a transient or production analysis.
Gridding was of limited concern to the engineer who
could therefore focus on the real issue of the physical
problem. The sequence starting on the right shows
the typical steps in building a 2D numerical model.
The starting point of an interactive build will be a
scaled bitmap representing the reservoir, its main
inner boundaries and its wells (1).
From this bitmap the user interactively builds a
vector representation of the problem, including the
reservoir outer contour, the inner faults and the wells
(2). Composite zones (3) and thickness / porosity /
permeability fields (4) may also be defined.
Ecrin then builds the grid automatically (5)!
After simulation, the static and dynamic properties can
be visualized and animated in 2D using a color lookup
table (6), in pseudo-3D where the Z axis represents
one property and the color coding may represent
another property (7), and in standard 3D (8).
Around each well, the 2D unstructured grid is replaced
by a 3D unstructured grid whenever needed, as in the
case of a limited entry well (9) or a horizontal well (14).
These models also account for vertical and horizontal
In Rubis models different horizons and gravity can
be considered. In later releases, horizons, volumes
and static properties can be directly imported from
Geomodelers such as Petrel™ using GRDECL or
CMG formats (11).
It was natural progression to start solving nonlinear
(NL) PTA and PA problems that had been hitherto
overlooked. Saphir NL, Topaze NL and Rubis can
be used to model real gas diffusion, multiphase flow,
water and gas injectors, water drives, nonDarcy
flow (Forscheimer), unconsolidated formations and
desorption models for unconventional resources.
All Ecrin numerical modules use the same technical
kernel and the same basic reservoir grids (12), but with
a flexible upscaling adapted to their very different
requirements around the well.
Considering the case of a horizontal well, Saphir NL
needs very significant refinement, with around two
thousand cells to perfectly simulate the different 3D
flow regimes on a loglog scale (13). Topaze NL will
only require a detailed 2D or limited 3D representation
with around 300 cells around the well (14). Rubis time
steps being days or weeks a very coarse grid with only
six cells is suitable (15).
Comparing these three different refinements on a
loglog scale shows that all responses merge on a
reference analytical model after a time adapted to the
required minimum time steps (16).
We have a trick: rather than using an analytical well
index, before any run coarse grids are calibrated with
a refined PTA grid with a small single phase simulation
around each well. The value of the coarse grid well
index is adjusted to match the long term productivity
of the refined grid. Numerical problems can therefore
be transferred between modules, upscaled or
downscaled, and they will remain consistent.