The Zavanna Workstation
The major differences between our workstation
and others have to do with:
The way data is loaded
The way data is related within the database
The ability to create entirely new data sets
The ability to view and analyze data sets on
an interactive display window
The ability to create unique reports
The ability to statistically analyze the data
and to perform Exploratory Data Analysis and to map the results
The ability to relate important aspects of wells
to the wells that are nearby or surrounding them
In most commercial workstations the emphasis
is on:
Storing data
Making maps
Displaying maps
Many workstations, such as GeoGraphix, perform
these functions very well. The major differences listed above
are important because they have to do with the ability to explore,
finding new ways to look at old data and relating it to the presence
of hydrocarbons. The ability to make maps is important. The Zavanna
ExlorationStation was designed to help answer the question which
map to make. Commonly the map is one which has never been created
before - ever!
The following is a brief summary of the major
differences itemized above with specific examples:
The Way Data Is Loaded
Zavanna has developed a program called WorldLoader
which analyzes the data that is put into the workstation. WorldLoader
allows data from many different sources, such as Petroleum Information
digital well history data and production data, other companies'
digital land files, and a variety of different companies' correlated
log tops files such as GDS, GCS, Digitech, RPI, WBA and others.
These tops files are merged into a single file -- not a series
of separate files.
WorldLoader makes use of artificial intelligence
to analyze the data, report on missing data and warn you if data
from the same vendor or from different vendors is inconsistent.
The objective is always to avoid "garbage in." Examples
of this include specific reports that indicate
Missing elevations or tops
Log tops that are "out of order"
Perforation results that are not consistent with
the results of the well
Large differences between reported scout and
electric log tops
Big differences between isopach and elevations
in nearby wells
Instances where total depth is reported as shallower
than reported mapped horizons
Such reports help us to understand the data and
to make corrections before we start to analyze the data further.
The Way Data Is Related within the
Database
Drill stem tests, cores, samples, and perforations
are directly related to the correlated log tops. Other systems
relate such important data to nonstandardized and subjectively
defined formation names. For example, PI commonly reports the
producing horizon as the Mission Canyon in the Williston Basin.
However, our database contains correlated horizons within the
Mission Canyon consisting of the Rival, Top State A Zone, Bluell,
Sherwood Argillaceous Zone, Sherwood, TK1 Argillaceous Zone,
Mohall, TK2 Argillaceous Zone, and Glenburn. Each of these horizons
have very different producing trends that are completely masked
when the production or test data is lumped into a single category.
Another difference is that drill stem test recoveries
are analyzed by whether their best results were of mud, water,
oil or gas cut fluids, free oil or gas to surface. The results
are then "coded" and related to each correlated log
horizon. This allows us to know more about the "test history"
of any individual interval or group of intervals.
The Ability to Create Entire New Datasets
The Zavanna ExplorationStation has been designed
around a "rich tool environment." The idea is that
given the right tools you can work with the data in any way you
want. Other workstations commonly contain more digital data than
we do. It is stored mainly as a "library" of data at
your fingertips. It is very handy and very useful, for instance,
to have the entire digital well history of a given well in your
workstation to review.
Our ExplorationStation is designed to identify
which "book" to read in the library (which wells we
ought to be most interested in knowing more about). Although
I admit that the analogy sounds a little corny, the fact that
you have the data doesn't mean that you can explore easily with
it.
An example of a new data set might be --
All
wells that drill stem tested free oil or gas to the surface from
the Belly River with a shut-in pressure gradient of more than
0.4 psi per foot, but were only perforated in some deeper horizon
and are still producing but at a rate less than 50 BOPD with
a decline rate for the last five (5) years of less than 10% per
year. Such wells might be acquisition targets with possible behind
pipe pay with additional long term production potential from
their current producing interval.
Another example is --
Create
a map set that consists of the subsea value of the "Second
White Specks" using only the control that went to the Nisku
prior to 1986. Depending on supporting statistical analyses,
certain shallow structural datums can be shown to be highly correlating
with deeper datums and predictive of subsequent new field discoveries.
A final example comes from the Permian Basin
of the United States --
A trimodel
distribution of producing perforations was found to exist with
respect to the depth below the top of the thick Queen Formation.
Because many wells were missing the Queen tops, we constructed
a new data set using wells that did have Queen tops and for the
wells with no Queen tops, we created "estimated Queen tops"
based on interpolated values from the subsea structure grid.
The new data set allowed for the correct
placement of the producing interval with respect to the top of
the Queen which was the basis for identifying wells that did
not reach deep enough to test the entire prospective Queen interval.
If a data set can be imagined, it can be made
and often in more ways than one.
The Ability to View Such New Data Sets
on an Interactive Display Window
The Zavanna ExplorationStation includes an interactive
"Display Window" in which data is displayed. This window
allows one to view all well control and/or just wells of interest.
It is particularly helpful when used to display mapping values,
showcoded by formation, and production data. It graphically identifies
trends and outliers, and provides statistical summary data about
the extremes and averages of the data being displayed.
A few examples are:
To view the variability of mapping values and
locate anomalous points of interest
To compare the range and average of production
data from a specific formation over different parts of the study
area
To "animate" the data by sorting the
data in time and create a "movie" showing the drilling
history of a given formation through time.
The interative nature of the window allows you
to instantly access all of the data for a given well by clicking
on its symbol in the window.
The Ability to Create Unique Reports
about the Data
Five specific types of reports have been designed
to be used repeatedly throughout the exploration process. These
reports were written by us to provide commonly needed information
about the database with respect to the entire stratigraphic interval,
specific formations, and operators.
The first report, for instance, identifies all
of the correlated tops and the frequency of different kinds of
"shows" in each horizon as well as the number of producers
of oil and gas, the number of wells which penetrated that horizon,
the number of wells that have a correlated log top, the amount
of oil and gas produced and an average cumulative production
per well all in a one-page format. This data is constantly changing
as different portions of the database are analyzed.
Another report is horizon-specific and displays
the complete test histories of all wells within each zone. It
provides needed information about what kinds of tests are important
in any given horizon, and where they need to be in the section
with respect to the correlated tops.
The Ability to Statistically Analyze
the Data and to Perform Exploratory Data Analysis and to Map
the Results
The mathematical field of statistics has been
largely ignored with respect to the analysis of geologic data.
However, with so much data available in digital format, it is
hard to imagine not applying the use of statistics. The ExplorationStation's
ability to utilize statistics in a visual, interactive and graphical
sense is matched by no other commercial workstation. It is our
most important exploration tool to analyze data.
The ExplorationStation allows the Expert User
to identify patterns and relationships and to test hypotheses
and find new ways to think about, display, model and map data.
Built into and linked to the ExplorationStation
are statistical formulas for measuring the relationships between
individual data sets. The most common application is to measure
the variance between two structural horizons (Multivariate Analysis).
Another use is to differentiate the significance of different
variables between producers and dry holes (Discriminate Function
Analysis).
An
example of such a statistical approach is to identify quantitatively
which shallow horizons or which combinations of multiple shallow
horizons, isopachs and trend surfaces are "predictive"
of deeper stratigraphic and structural events. New derived variables
created from such analyses are often mapped as being closely
associated with deep production. It's a new approach to exploration
that you have to see to really understand.
The Ability to Relate Important Aspects
of Wells to Other Nearby or Surrounding Wells
The newest tool we have incorporated into the
ExplorationStation is called Z-tool. It allows us to locate
specific wells of interest because of the wells that are nearby
(or that are not nearby). It was designed initially as a program
to identify wells that may have exceptional additional production
potential, and to therefore target them as production acquisition
targets.
For example, one type of target may be to find
every well that drilled through the Cardium Formation without
testing it and that was completed in a deeper formation which
is still producing today, and yet is surrounded (by at least
two opposing quadrants, for instance) by wells that were subsequently
completed in the Cardium, and each of those wells had produced
at least 200,000 BO and no well has been drilled within 1/4 mile
of the target well, and the surrounding producers are no further
than 3/4 mile from the target well. The idea in this case is
to find wells that probably have oil pay behind pipe in the Cardium
because of their location with respect to other Cardium producers,
but those producers are far enough away that they have not totally
drained the target well. Target wells can be hi-graded by such
factors as the per well reserves of the surrounding wells, distances
to those wells, and operators of those wells.
This tool may be applied to all types of data.
The algorithms to set up a specific search of this type can be
very complicated. The analytical time, however, is short. A database
of tens of thousands of wells can be evaluated very quickly and
the results have proven to be extremely valuable!
Working a new database is exciting. In most cases,
we have little first-hand experience with the geology. We have
very few pre-conceived ideas of what we will find. Some would
consider this a handicap, but I consider it an advantage.
In mature basins today, there are fewer and fewer
new big fields being found using old ideas and pre-conceived
notions. We let the patterns in the data teach us about the geology.
Data analysis is like detective work. Analyzing data can be like
playing a computer game except that the goal is to puzzle out
the world rather than to match wits with a game designer -- the
beauty of it all is that there is always another good puzzle
after you solve the current one.
