6

Summary for complete analysis. He thought a program could be written that

Engineering programs often are a collaborative effort between en- would, in some way, include technical information to help analysts
gineers, managers, and software designers. This paper chronicles take advantage of the many interpretation techniques available.
how a microcomputer program for analyzing weUlogs from shaly An early concern in program development was which platform
sand formations evolved when an "expert" was used not only to to choose. IBM PC compatible computers were readily available
provide technical information but also to aid in developing the over- for analysts in small or large companies, so the program was de-
all program structure. veloped for IBM PC's with 640-KB RAM and EGA (or higher)
graphics. These lowly specifications were the standard at the be-
Introduction ginning of this project for laptop and desktop computers. Because
Windows ™ was in its infancy and did not yet have many
Analyzing well logs from shaly sand formations can be confusing programming tools, this program was developed for DOS and does
and frustrating. Three years ago, George Asquith, a noted expert
not require Windows.
in well-log interpretation, joined John Walsh and Susan Brown in
In its earliest conception, SSA was a traditional program in which
a development project that would aid companies trying to analyze
a user would input data, select methods, and choose calculations
logs in shaly sands. The original idea was to develop a program
and graphs to be displayed. After several meetings where we had
affordable to small operators that would run on easily available hard-
difficulty finding a suitable program structure, we began document-
ware. This paper presents the development method used to create
ing not how the program should be, but how Dr. Asquith analyzed
a program containing the" advice" of an expert but still allowing
logs. Over the next several months, Dr. Asquith analyzed dozens
user flexibility in performing calculations and interpreting results.
of examples of shaly sand formations where the steps were carefully
This flexibility is necessary in well-log analysis where available
documented. Looking at these analyses and at how Dr. Asquith
data vary from well to well.
decided which methods to use, we formulated a general procedure
Because log interpretation is done both at the well site and in the
for analyzing any set oflogs from shaly sand formations. This proce-
office, Shaly Sand Advisor (SSA) was written for IBM PC com-
dure was transported easily to a program structure. Once the struc-
patible computers. The project was begun with Dr. Asquith creating
ture was programmed, modifications to aid in the analysis (such
a flow chart for four classifications of shaly sand problems (low
as specific graphs, calculations, or notes) could be added easily.
salinity, low volume of clay, laminar shale, and dispersed clay).
This paper presents the steps used in organizing how an expert
On the outline, methods to use in calculations were listed, along
solves problems to achieve a systematic approach so that users of
with the restraints on when the methods would be inappropriate.
various skill levels can analyze logs with confidence. This procedure
We then expanded the outline into a general program structure by
can be used with other topics to translate knowledge into a program
identifying model types and checking Dr. Asquith's interpretations
that will teach and aid in interpretations.
through calculations, crossplots, and comparisons. The program
structure included how to solve problems with a variety oflog suites
or additional data. The user is led through decision points in the Beginning Flow Chart
program. Several elements were identified from the beginning as important
The program design was enhanced from Dr. Asquith's experience to a successful log interpretation program.
in analyzing many types oflogs and distinguishing between the ap- • Analyze old electric logs.
propriateness of different methods. The resulting program allows • Provide environmental corrections.
both novice and expert analysts to perform interpretations quickly • Calculate clay volume in different ways.
and with more confidence because of the many checks provided • Display reference material with on-line technical notes.
in the program. • Compute clay volume from transformations.
This paper outlines how the program structure changed from a • Establish producibility from a Q-plot.
simple straight-branched flow chart to a more complicated system • Compute Rz when Rxo is not available. 2
that presents a unique way to analyze shaly sands. The paper pro- • Include different water-saturation models.
vides specific examples on how, through the development proce- From an original flow chart by Dr. Asquith, a working flow chart
dure, several new analysis techniques emerged, including different was created (Figs. 1 and 2). The flow chart enters log data, performs
crossplots to look at log data statistically. The approach used to environmental corrections, enters laboratory data, computes clay
create SSA can be used to solve a variety of problems where soft- volume by different methods, compares log- and laboratory-derived
ware developers want to create a program not only to perform anal- clay volume, and can transform clay volume by simple regressions.
yses but also to help users analyze available data fully with existing Users then must select the clay volume to use in the program. Next,
methods and the experience of persons who have solved many of effective porosity is computed, depending on the available logs.
these problems. A Q-plot 3 is displayed with a decision point that an analysis should
be continued only if depths fall in the producible region of the
Background Q-plot.
Over the last 5 years, Dr. Asquith has been doing research and Once producibility is established, the program would look for
teaching classes in shaly sand and carbonate well-log analysis. I special cases where traditional shaly sand water-saturation models
An avid microcomputer user, he envisioned a program for well- could not be applied. The first is whether the formation water is
log interpretation that would include calculations and graphs needed brackish or cation-exchange-capacity (CEC) data are available. If
CEC data are available, Waxman-Smits 4 would be used to com-
Copyright 1993 Society of Petroleum Engineers pute water saturation. If CEC data are not available and the forma-
SPE Computer Applications, February 1993
 7
tion water is brackish, water saturation would be computed by the I. Compute clay volume from the gamma ray log (or by other
Poupon-Levlaux 5 method. methods when the gamma ray curve is unreliable).
The next decision point was to check for the low-clay-volume II. Compute effective porosity from the density-neutron log.
problem. If clay volume is low and the Archie water saturation high, III. Compute Rz if an Rxo device does not exist.
the WMethod 6 is the preferred Sw method when the dielectric log IV. Plot the neutron-density crossplot to look for gas effect.
is available; Waxman-Smits is preferred when CEC data are avail- V. Plot the Q-plot to look for producibility.
able. Nonbrackish formations and those without low clay volumes VI. Check for hydrocarbon presence by
would be analyzed by traditional shaly sand water-saturation models. A. Deep-resistivity-derived porosity vs. porosity-log-porosity
Last, a net pay would be computed from user-defined cutoffs. crossplot.
B. Pickett plot.
Taking Advantage of an Expert's Knowledge C. Archie water-saturation vs. ratio water-saturation plot.
D. R/R w vs. RsIRz Dew* plot.
Even from this beginning flow chart, a program would be more VII. Check that the reservoir is at irreducible water saturation
useful ifit helped analysts at decision points. Some decisions were with the bulk-volume-water plot.
straightforward, such as selecting an appropriate clay-volume VIII. Compute water saturation by the Archie, Fertl,7 and dual-
method when laboratory- and log-derived clay volumes are trans- water 8 models.
formed and the desired method is selected by the highest correlation IX. Compare the Fertl and dual-water results by performing the
coefficient. Others were not even conceived in the chart, such as Fertl and Hammack 9 correction if necessary.
how to analyze a low-clay-volume reservoir when CEC data or a X. Compute net pay.
dielectric log was not available. Other questions that the program Similar outlines were prepared for various logging suites: old
could aid in answering were outlined: electric logs, sonic log as the only porosity device, density log as
• Do these log data fit the profile for a shaly sand? the only porosity device, both neutron and density logs, and all three
• How does a user select the best water-saturation model? porosity tools available. These outlines attempted to take into ac-
• How do you determine in shaly sands which zones are potential count all possibilities when a decision point was reached. From these
hydrocarbon-bearing zones? outlines, a structure was formed to create a two-pass analysis
• What graphs are generated and when are they examined? program.
• What is done to check results?
As these basic questions arose, we began to characterize analyses Two-Pass Structure
by available logging suite and type of shale formation, brackish When Dr. Asquith analyzes logs, he first establishes potential
water, low-clay-volume/high-CEC-content clay, laminar shale, dis- hydrocarbon productivity before trying to select a shaly sand water-
persed clay, and unknown clay. Each possible log combination was saturation model. Thus, Pass I was designed to establish whether
outlined with how Dr. Asquith would analyze it. For example, if a zone appears to be hydrocarbon-bearing to merit computing water-
the clay type is not known and a neutron-density dual-induction log
suite is available, the outline was as follows: • Personal communication with E. Dew (t986).

GR D-S
SP Res
N-D bulk den
N-S Lab

Effective Porosities
No
N·S Neutron
N-D Sonic
D-S RES
Density .M:L
Poupon No Porosity Log
S., Dual Water - Swe
if Brackish No Porosity Log
RatioS we Auto Compensation - Swe
If No! AUlO Compensation - Swe
Dispersed Clay - Swe
Simandoux - Swe
Fertl.S w
Yes Dual Water - Swe

No 4( Next Zone)

Detennine Net Pay Using Cutoffs

Fig. 1-Working flow chart for SSA (V2). Fig. 2-Working flow chart of SSA (%).
8
saturation values. In Pass 1, clay volume, effective porosity, and suite to compare various methods, including methods added by the
R" are computed. Then, a series of crossplots is provided to es- user.
tablish whether a zone appears hydrocarbon-bearing or wet. Cross-
plots also are included to test for the low-clay-volume/high-CEC Cross plot Example
problem. Pass 1 would establish whether a shaly sand analysis Each detail of the program, from the graphs displayed to how cal-
should be done and could be used at the wellsite to aid in the decision culations are performed, was examined and outlined to fit each of
to run pipe. the defined paths that an analyst might take based on clay type and
In Pass 2, water saturations and net pay would be computed. The logging suite. During the detailing of the project, the power of an
program selects at least two water-saturation models, one that uses interactive tool was obvious; users often did not need to enter in-
an Rsh value and one that does not. The selection is based on avail- formation because the program could choose appropriate parameters
able logging suite. If the models compare, the shale is laminar and from data available.
compatible with adjacerit shales and either method can be used. If Each of the crossplots in SSA was enhanced to take full advantage
the Sw values from a model using Rsh are greater than Sw values of a computer display and automatic selection of data to use in the
from a model that does not require R sh , then the Fertl and Ham- crossplots. As an example, one of the graphs of hydrocarbon indi-
mack correction can be applied for dispersed clays and the models cation plots a resistivity-derived porosity vs. porosity-log-derived
should give similar results. If CEC data are available, Sw from porosity. The resistivity-derived porosity uses the deep-resistivity-
Waxman-Smits always is computed. log reading to compute porosity. The porosity-log-derived porosity
is computed by the neutron-density log, density logs, or sonic log,
Final Flow Chart depending on which logs are available. Fig. 6 gives an example
Once a two-pass strategy was established, the mechanics and "bells of this crossplot. Note that the crossplot is provided on the left side
and whistles" could be added to the program easily. On each screen, of the screen with a depth graph on the right. Data on the graph
the graphical data are presented with explanations. Technical notes are plotted in yellow and dark blue, with the dark blue being shale
about choosing between methods or discussions on what the cross- determined by the depth's clay-volume value and a clay cutoff de-
plots mean are included on every screen. fined by the user. The shaded region (blue) indicates the depths
Figs. 3 through 5 display a revised flow chart. The program takes that would be wet. Any depths that fall in this region are highlighted
any logging suite and first establishes whether the formation has a in the graph on the right by a blue column.
potential for hydrocarbons through a series of erossplots. Results For each crossplot, the user can highlight data from specific depths
from erossplots are summarized in a single screen so that users can and view the notes that discuss the crossplot. The cross plot above
scan depths to see which meet the criteria of being hydrocarbon- has four on-line notes.
bearing and permeable enough to produce. The user can override I. On first pass, use no hydrocarbon correction, letting
automatic selection of shaly sand water-saturation models by logging Sxo=lOO%.

Hydrocarbon Versus Water

Low VcI and CEC Crossplots

<PN - <l>D
<l>Rs - <1>. (CEq

OR D-S <l>Rt-<I>t(HC)'
SP Res Highlight On-Line
N-D Pb Intervals Pickett' Technical Notes
N-S Lab in a Zone To Explain
LowVcI Check Crossplots

Swa - Swr
, And Aid
Interpretation

BVW'
DEW'
VclSP - VclGR

N-D Neutron
N-S Sonic
D-S RES
* Denotes Hydrocarbon
Indicators
Density ML

NO~

Fig. 3-Final flow chart for SSA (V3). Fig. 4-Final flow chart for SSA (213).
 9
Thus, an analyst can quickly view the crossplot and know depths
with a potential to be productive.

Generalized Procedure
In SSA, the knowledge of an expert was successfully implemented
into a computer program by writing a program that leads the user
through an analysis based on available data. To be flexible to all
users, the program allows users to override selections to perform
other calculations. Users can include their own calculations, and
because many published shaly-sand-analysis models are included,
the program can be used by analysts who use a different methodol-
ogy. Technical notes throughout the program help the user know
when such a deviation is justified. The following procedure was
used to create the program:
1. Produce a general outline.
2. Account for all types of analysis.
3. Establish how an expert selects among available data and anal-
ysis methods.
4. Categorize methods on the basis of paths at decision points.
5. Automate as many decision points as possible from the
categorization scheme.
Rsh SweMethod
Dual Water Method
6. Organize the analysis and interface to be straightforward from
start to finish of each analysis.
Rsh Swe Method Non.Rsb Swe Method
Dual Water Ratio Sw 7. Provide documentation at decision points so that users can devi-
Non.Rsh Swe Method
ate from paths if justified by theories or experience.
FertI 8. Test the program continually with a wide range of data to en-
sure that all analysis paths are consistent and accurate.
SSA takes advantage of an expert's experience to provide calcu-
lations and graphs appropriate to an analysis. The program is not
a learning machine that can code its own experience on the basis
of each set of data analyzed. Whether it is considered an expert
Detennine Net Pay and Hydrocarbon system depends somewhat on semantics. However, by using the
hq> Using Cut-offs experience and knowledge for both the structure and details of the
program, we created a tool to help log analysts evaluate and learn
Fig. 5-Final flow chart for SSA (%). about evaluating well logs in shaly sands. This tool would not have
evolved into a comprehensive program without the input of the ex-
pert at every phase of development.
2. Data clusters around the 45° line (i.e., PhiR=PhiD) indicate
low-CEC clays with no excess conductivity. Continue using standard Conclusions
shaly sand analysis.
3. If data fall above the 45° line (i.e., PhiR~ PhiND), then high- SSA is written so that users with a wide level of experience can
CEC clays are indicated with excess conductivity. Use the Waxman- analyze logs accurately. By studying how Dr. Asquith analyzed logs,
Smits method (with CEC data) or the WTechnique (with dielectric we developed a generalized interpretation methodology that was
log) to determine effective water saturation. If CEC data or a di- implemented into a computer program. Besides establishing a sys-
electric log are not available, use ratio water saturation. tematic approach to interpret logs, several new technical ideas were
4. If the data fall below the 45° line, redo the plot after reducing developed as the program was written. These include creating a
Sxo' but not below 70%. low-clay-volume check plot, making a summary crossplot histo-

.0

.0

J: .0

'0

10 2. o.
lIND (x)

nu: (FS) ~t $c~n (F7) UclMaw (Fe) HI (I'HI> Return

Fig. 6-Example of a hydrocarbon-indicating crossplot tPRIO VS. tPO-N' Points lying in the
shaded area represent depths that are wet. These depths are also shaded between the V cl
and tP, graphs.
10------------------------------------------------------------
gram that compares results from several crossplots, and using differ- References
ent x-y axes for traditional crossplots for limited logging suites. I. Asquith. G.B.: Log Evaluation ofShaly Sandstones: A Practical Guide,
Examples of modifying traditional crossplots include plotting AAPG Publishing, Tulsa (1990).
deep-resistivity-derived vs. resistivity-derived porosity in a Pickett 2. Tixier. M.P.: "Electric Log Analysis in the Rocky Mountains," Oil &
plot when only an old electric log is available, or computing a Q-plot Gas 1. (June 23, 1949).
with the total porosity and effective porosity when a sonic log is 3. Log Interpretation Charts. Dresser Atlas. Dresser Industries Publishing,
not available. Houston (1979).
An unexpected outcome of this project was that, by organizing 4. Hilchie, D.W.: Advanced Well Log Interpretation, D.W. Hilchie Inc.
the analysis of many types of formations, the general method de- Publishing, Golden, CO (1982).
rived for shaly sands has changed the way each developer inter- 5. Poupon, A. and Levlaux, J.: "Evaluation of Water Saturation in Shaly
prets logs. Pass I, which looks at the potential of a formation to Formations," Proc., SPWLA 12th Annual Logging Symposium (1971)
be hydrocarbon-bearing, provides an automatic and quick means paper O.
to determine whether further analysis is merited. 6. Guillotte. J.G., Schrank, J., and Hunt, E.: "Smackover Reservoir: Inter-
Working with an expert provides software developers with a pretation Case Study of Water Saturation vs. Production," Trans.,
GCAGS (1979) 29, 121-26.
unique opportunity to program the steps of an analysis and to in-
7. Fertl, W.H.: "Shaly Sand Analysis in Development Wells," Proc.,
clude "smarts" in a program to help users at decision points and
SPWLA, 16th Annual Logging Symposium (1975) paper A.
provide analysis paths that will anticipate different cases, including 8. Clavier, C., Coates, G., and Dumanoir, J.: "Theoretical and Experimen-
unusual or difficult-to-analyze situations. tal Bases for the Dual-Water Model for Interpretation of Shaly Sands,"
SPEJ (April 1984) 153-68.
Nomenclature 9. Fertl, W.H. and Hammack, G.W.: "A Comparative Look at Water Satu-
Rs shallow resistivity
= ration Computations in Shaly Pay Sands," Proc., SPWLA 12th Annual
Rsh =
shale resistivity Logging Symposium (1971) paper R.
R, =
true formation resistivity
Windows is a registered trademark of Microsoft Corp. SPECA
Rw =
water resistivity
R xo =
resistivity of the flushed zone
Original SPE manuscript received for review July 19, 1992. Revised manuscript
Rz = resistivity of the mixed zone
received Jan. 8, 1993. Paper accepted for publication Jan. 8. 1993. Paper (SPE
Sw = water saturation 24453) first presented at the 1992 SPE Petroleum Computer Conference held
Vel = clay volume in Houston, July 9-22.