TECHNIQUES FOR CALCULATING Cw AND TDS FROM LOGS

Chapter 14

Assessment of ground-water quality from well logs usually centers

around values for total dissolved solids (TDS) and specific conductance
(Cw). Well logs are used to calculate Cw and then TDS is estimated from
previously established TDS-Cw graphs. Sometimes other water properties
such as chloride content and total hardness can be estimated from similar
graphs. In some cases, TDS can be estimated directly from the log data.
Unfortunately, trace elements such as iron, fluoride, and nitrates cannot be
detected.

While it has been documented in other studies that determination of

Cw and TDS from logging data is feasible (Chapter 1). many ground-water
researchers have had little success applying the techniques. Both the cause
and effect of this is the fact that most ground-water logging literature has
only a cursory discussion. of the subject.

This chapter is a discussion of seven techniques for calculating Cw

and TDS from logging data. The first three (TDS-Ro Graphs, Roc-TDS
Graphs, and Field Formation Factor) are empirical relationships that must be
derived from the available data in a localized geographic areas (e.g. a
county). The last four (Formation Factor Equation, Ro-Porosity Graphs,
Resistivity Ratio Method, and SP) are stand-alone techniques that calculate
Rw. This chapter also contains guidelines for accurate CW calculations and
a discussion of sources of logs in Texas.

Considerable variation exists in the logging literature for two terms

important to this. discussion. When reading logging literature, one should
remember that:

1. Water that naturally saturates a formation is referred to as

formation water, connate water, or interstitial water. Theterm
formation water is used herein.

2. The term water resistivity (Rw). not water conductivity (Cw). is

used in virtually all logging literature. This chapter also uses Rw.
The relationship between the two terms is discussed in Chapter 2
and is restated here:

344
 345

10000 .
Rw (ohm-m) = I (!J.mhos/cm) (12·1)
Cw

GUIDELINES FOR Cw and TDS CALCULATIONS

The accuracy of Cw and TDS calculations will be greatly improved if:

1. The logging tools are suitable for the borehole conditions and the
petrophysical properties of the aquifer being analyzed.

2. The technique for calculating Rw is compatible with the chemical

composition of the water being analyzed and the available log data.

3. The logging data are accurate.

Suitable Tools and Techniques

Logging tools and water quality calculation techniques must be

selected according to the chemical composition of the ground water, the
petrophysical properties of the aquifer, and the borehole conditions. The
choice of logging tools, especially for resistivity tools, has considerable
bearing on the accuracy of the data. It also determines which analytical
techniques can be used. It is essential for one to be familiar with the
applications and .limitations of the various resistivity tools (Chapters 8 and 9)
and to remember that: .

1. Most logging tools and techniques are designed for clean (shale-
free), "normal" sedimentary rocks (quartz-rich sandstones,
limestones, and dolomite). Analysis of shaly formations may
require modifications to the techniques.

2. All resistivity tools are not created equal. They must be

compatible with the type of aquifer and borehole being logged. For
example, inductio.n tools are not the best choice for logging
aquifers with resistivities greater than 100 ohm-meters or zones
(beds or porous intervals) less than 5 tol 0 feet thick. Guard or
Latero tools provide better vertical resolution and more accurate
resistivity values in such circumstances (Chapters 8 and 9).
 346

3. The SP curve is good for quantitative work only in thick, clean

sands. SP values are too low if the formation (sandstone or
carbonate) is thin, shaly, highly resistive, or deeply invaded (see
SJuyod, 1966; Alger and Harrison, 1988; Sciacca, 1989; Chapter
12).

4. Most of the techniques for determining ground-water quality utilize

either an SP or a deep resistivity curve. The techniques have been
borrowed, with little or no modification, from the petroleum
logging industry. Application of the techniques to ground-water
studies has largely been unsuccessful because oilfield waters are
radically different from ground waters. Petroleum logging
techniques assume that formation waters have high salinities
(more than 50,000 ppm TDS) with predominately monovalent NaCI
ions. Ground waters, on the other hand, usually have much lower
salinities and a significant concentration of divalent calcium and
magnesium ions. The responses of SP and resistivity tools must
be interpreted differently for each of the two water types (Chapter
12).

Accurate Log Data

Log data accuracy is a major area of weakness for the ground-water

logging industry. The petroleum logging industry has devoted considerable
attention to data accuracy (see any detailed logging text and in particular
Helander, 1983; Bateman, 1985). Unfortunately, this expertise has not been
sufficiently utilized by the ground-water industry. Ground-water logging
articles that mention the subject provide little in the way of guidelines for
correcting log values.

Log data accuracy was addressed as each tool was discussed in

Chapters 8 to 13. The subject is so critical to Cw and TDS calculations that
a few points need to be reemphasized:

1. The accuracy of log data will be greatly improved by running a

logging suite which is compatible with the borehole conditions and
the petrophysical properties of the aquifer.

2. Resistivity values may require corrections for bed thickness,

borehole diameter, mud resistivity, and mud filtrate invasion
(Chapters 8 and 9).
 347
3. Tool calibration and quality control are essential for obtaining valid
data. Merkel and Snyder (19771, Hallenburg (1980), Hill (19861,
Hodges (19881, Collier (19881, and Sciacca (1989) discuss various
aspects of these two important subjects.

4. The correct assumption must be made about what part of the

borehole a particular resistivity tool is investigating. Most Cw and
TDS calculations utilize Ro, which in water wells is equivalent to
Rt. Variations in logging tool responses and borehole conditions
mean that Re is not always what is being measured. Either deep
mud filtrate invasion (not likely to occur in shallow or highly porous
clastic aquifers) or insufficient depth of investigation by the
resistivity tool (more likely to be the case) is usually the reason an
incorrect value is used for Re. The Resistivity Ratio Method
requires both Re and Rxe (the resistivity of the flushed zone). In
the case of Rxe, the resistivity tool is supposed to measure a thin
zone very close to the borehole wall. If the tool reads too deep or
too shallow, the resistivity will not be Rxe.

ACQUIRING LOGGING DATA

Fresh Water Aquifers

When studying the fresh water portion of Texas aquifers, most of the
logs will be from water wells and most of the wells will have at least one
water analysis. Virtually none of the logs will be hydrocarbon tests. Oil
companies are required to set surface casing through fresh water aquifers
and they generally do not log before setting,surface casing. Openhole logs
run through casing cannot be used to determine water quality.

Water well borehole geophysical logs are scattered throughout the files
of the ground-water industry. Unfortunately, there is no single, easily
accessible source. The following organizations are sources of logs:

1. Texas Water Commission, Central Records, Ground-Water

Technical File. This is the largest collection of water well logs in
Texas. Water well drilling contractors sometimes submit logs on a
voluntary basis, so coverage is by no means complete. Data are
\
filed by county and by complete or partial state well number.
Without a state well number, data on a particular well are not
 348
easily accessible. It can be very time consuming to track down
large numbers of logs.

2. Texas Water Commission, Watershed Management, Surface

Casing Unit. This file consists mainly of oilfield logs. However, it
does contain a small number of fresh-water logs. Some are logs
from water wells; the rest are oilfield logs that penetrated fresh
water formations.

3. Water well drilling contractors. Irrigation and domestic wells are

rarely logged. Conversely, virtually all municipal, rural water
supply, and industrial wells are logged. Drilling contracts usually
contain a clause requiring borehole geophysical logs. Drilling
companies meet the requirement by subcontracting to a logging
service company. Drilling contractors generally retain copies of
these logs. The older drilling firms have amassed extensive log
files. Access to the logs varies from company to company.

4. Ground-water consulting firms. These firms have a limited number

of logs and generally consider their data proprietary.

5. Petroleum Information Corporation (PI). PI is the main broker for

oilfield logs. They carry some water well logs, but .the logs are not
identified as such. Water well drilling contractors do not routinely
release their logs to PI. They only release the' ones that PI
requests, which are usually the wells logged by the major
commercial logging companies. Consequently, PI carries only a
small percentage of available water well logs.

6. Logging companies. Major logging companies, except for

Schlumberger, retain their logs for only a few months.
Schlumberger started archiving tapes of logging jobs as of 1987.
Presently, about 50 percent of all jobs are archived (at no fee to
the client). The client may request that a logging job be archived.
If the client releases the data, anyone can purchase a copy of the
tape. However, Schlumberger does not archive hard copies of the
logs and it takes sophisticated logging software to make a hard
copy from a tape. Smaller logging companies sometimes keep
hard copies of all their logs (e.g. Tejasl. However, they do not
make a practice of furnishing copies of the logs in their files. In
 349
fact, the logs archived by any logging company are proprietary and
cannot be released without the consent of the client.

Saline Water Aquifers

Log coverage of the saline water portions of Texas aquifers is much

more complete. In fact, too much data will be the problem in some of the
petroleum-producing areas of the state. Most of the logs will be from
hydrocarbon tests and very few will have accompanying water analyses.
When searching for logs, one should check the following sources:

1. Petroleum Information Corporation (PI). PI is the premier source of

oilfield logs. They sell copies of the logs, or one may borrow logs
from a PI log library. A membership is generally required to check
out logs from the library.

2. Petroleum log libraries. Scattered around Texas are numerous

public and private membership petroleum log libraries. Some
libraries cover the entire state, while others only cover a particular
geographic area. In public log libraries, logs can be examined in-
house for a flat per-day fee or they may be checked out by
. members.

3. Texas Water Commission, Watershed Management Division,

Surface Casing Unit. This agency has over 250,000 logs from
throughout the state. The logs are filed by county and are keyed
to county land and ownership maps. It is easy to locate logs from
a particluar geographic area. Logs may be copied from the file, or
they may be checked out to be copied and returned within 2 days.

4. Bureau of Economic Geology, Geophysical Log Facility. This

agency is the repository for oilfield logs collected by the Railroad
Commission of Texas. Since 1986, oil companies have been
required to submit a log for each new well. The regulation is
worded 50 that any wireline log is acceptable. Thus, the log which
is submitted may be of a type that is of no value for water-quality
calculations. The regulation further states that the log must be of
the producing interval. Some operators, therefore, submit a partial
log, rather than a log of the entire borehole. The end result is that
many of these logs may not be suitable for ground-water studies.
. , .)

350

The Geophysical Log Facility also has a large number of pre-1 986
logs. These logs are from the Railroad Commission and the Bureau
of Economic Geology files.

5. Ground-water firms. Water well drilling contractors and ground-

water consulting firms have few such logs.

EMPIRICAL RELATIONSHIPS FOR ESTIMATING TDS AND RW

TDS-Ro Graphs

Under ideal conditions, it is possible to accurately estimate TDS

directly from an Ro value. An Ro value is plotted on a previously established
TDS-Ro graph and the corresponding TDS value is read directly off the
graph. The procedure is as follows:

1. Construct a TDS-Ro graph using data from wells closest to the

well being analyzed. Ideally, the data should also be from the
same stratigraphic unit. In reality, well control is usually sparse
and the stratigraphy of any aquifers from which water samples
were analyzed is not always identified. For these reasons, the
data collected during this project (Volume II, Section 5) were
processed by counties. Data plotted by counties often has a high
correlation coefficient because most of the wells are from the
same aquifer (Figure 14-1).

2. Logarithmic scales are recommended. They allow a wider range of

values to be plotted on a single graph (Figures 14-1 and 14-2) and
the curve fit is a straight line.

. 3. The deepest reading resistivity curve (64" normal, deep induction,

lateral, or deep laterolog) is used for Ro.

a. An average Ro value is selected for the aquifer. Figures 8-25,

8-26, and 9-21 discuss how to select average resistivity
values. If there is more than 10 percent variation in the
resistivity of the zone, it may be worth plotting the highest
and lowest average values, Ro High (Ro H ) and Ro Low (Rod.
This was done, as necessary, for the graphs in Volume II,
SectioOn 5.
 351

1000

c;::
.9 100
.
.
0
a:
"tJ
c
~
~
'" •
.s=
0>
I 10
0
a:

100 1000 10000

sumTDS

Figure 14-1. Ro-TDS graph that has a high correlation between Ro and TDS. Data are from Harris
County. Sum TDS includes 100 percent of the bicarbonate value.

1000

c;::
.9
0
a:
"tJ
c
~

~
'." 100
0

.s=
0>
I 0

0 DB. 0
a:
l\o< •

0
'. 0

100 1000 10000

sumTDS

Figure 14-2. Ro-TDS graph that has a low correlation between Ro and TDS. Data are from Dallas
County. Sum TDS includes 100 percent of the bicarbonate value.
 352

b. If borehole conditions warrant, an environmental correction

should be made to the resistivity value on the log (Ra)
before it is used as Ro and plotted on the graph. (Chapters
8 and 9 discuss environmental corrections for each
resistivity tool.)

4. TDS values should include 100 percent of the bicarbonate value.

5. A line-fitting routine can be applied to the data, but is not·

necessary. If there is much scatter, the equation of the line does
not give an accurate TDS value. It is better to simply plot new Ro
values on the graph and to look at the range of corresponding TDS
values (Figure 14-2).

This technique has one serious limitation: it requires Ro values to be

solely a function of water salinity (TDS). This condition exists only in shale-
free sandstone aquifers that have approximately the same porosity. Such
conditions normally are approximated only in sandstones that are ,<onsistent
in lithology, unconsolidated to semi-consolidated, Tertiary or younger, and
less than about 1000 feet deep (aquifers such as the Gulf Coast and Carrizo-
Wilcox). Sandstones from the same depositional facies and a limited
geographical area give the best correlation.

TDS-Ro graphs do not work as well for consolidated sandstones and

carbonates such as the Trinity, North-Central Texas Paleozoic formations,
Edwards, and Ellenburger. Several of the graphs in Section 5 of Volume II
substantiate this fact. Figure 14-2 is an example of a poor correlation. In
these types of aquifers porosity can vary considerably within a well and from
well to well. In such cases, Ro is a function of both porosity and TDS, so
there is no consistent correlation between Ro and TDS. The water quality
calculation must account for the effect of porosity on Ro .(Roc-TDS Graphs,
Field Formation Factor, Water Saturation Equation, and Ro-Porosity Graphs)
or factor it out (Rxo-Ro Ratio).

Few Ro-TDS graphs have been published. Guo (1986) and Fogg and
Blanchard (1986) are the only examples this author found. Guo applied the
technique to Quaternary alluvial sand aquifers in the North China Plain
(Figure 14-3). The one graph that he published does have a high correlation
between TDS and Ro values.
 353

Fogg and Blanchard

constructed a TDS-Ro graph
,,----------------,
for the Carrizo-Wilcox •
aquifer system in the Sabine
Uplift area of Texas (Figure 7

14-4). The correlation is

only fair (0.80) and the (f)
graph is a good example of oI-- 5

the inaccuracy of this a:: 4

W
technique. An Ro of 30 I--
«
nhm-meters could represent ~ 3

a TDS anywhere from 300 a::

..
..
w 2
to 1400 mg/I and a TDS of LL
::>
1400 mg/I could have an Ro $1 I

ranging from 10 to 40 ohm-

meters. The scatter is o 5 15 20 25 30 35 40

probably due, in large part, MEAN AQUIFER RESISTIVITY (ohm-m)

to variations in porosity.
Grouping the data according Figure 14-3. Ro-TDS graph for North China Plain Quaternary alluvial
to smaller geographical sands (Guo. 1986). gil x 1000 = mg/l

areas would probably improve the correlation. Also, the graph would be
much easier to interpret if the data had been plotted as whole numbers.

Ro-TDS graphs were constructed from the data base assembled for
this project. Ro values are from Section 3 and TDS values are from the Sum
TDS column of Section 2 of Volume II. The data was plotted by counties.
Forty-eight counties had enough data for a graph (Volume II, Section 5,
Figures 5-1 to 5-48). Each graph contains 2 to 20 data sets.

Thirty-six graphs had sufficient data to judge the quality of the curve
fit: 33 percent (Brazos, Cherokee, Denton, Ellis, Harris, Hidalgo, Hunt,
McMullen, Milam, Rusk, Shelby, and Wood Counties) had a good fit, 47
percent (Anderson, Angelina, Collin, Dallas, Dimmit, EI Paso, Fannin,
Freestone, Gonzales, Grayson, Jefferson, Montgomery, Nacogdoches, Smith,
Tarrant, Upshur, and Van Zandt ) had a fair fit, and 20 percent (Erath,
Karnes, Robertson, Limestone, McLennan, Red River, and Walker) had a poor
fit. The graphs represent data from four major Texas aquifers: Carrizo-
Wilcox, Trinity, Gulf Coast, and Bolson Deposits. The .Trinity has the highest
percentage of poor curve fits, but it does have three counties with good
curve fits.
 354

'.0,----------------------------,

+ Correlallon.0.80
3.5 + ITDSII::l (6550) Ro -0.'1"

Standard error 0' eslimate- 0.20

(LoO unUsl

3.0

+ +

2.5
+ +
+ +
+
++
+
+ + +
+
+ +
2.0 ":-_~ __ ~ __ ~_~~_~ __ ~ __ .......o~_......::~~_"____--'

0.5 1.0 1.5 2.0 2.5 3.q

~ IRoJ (ohm-m)

Figure 14-4. Ro-TDS graph for the Carrizo-Wilcox aquifer system, Sabine Uplift area. Ro values are from 64"
normal and induction logs. Data are from water wells that are screened primarily' in channel-fill sands at
depths of 200 to 1,200 feet (Fogg and Blanchard, 19861. There is considerable scatter in the data. The graph
would be much easier to interpret if the data had been plotted as whole numbers.

Of the twelve counties with good curve fits; nine have parallel curve
fits (Figure 14-5). However, the variations in the curve fits underscore the
fact that Ro-TDS graphs are site-specific and should be used with caution.

For thirty counties the data distribution was such that the Ro values
corresponding to 1,000 mgtl TDS could be getermined from the Ro-TDS
graph. Five counties had sufficient data to determine the. Ro value
corresponding to 10,000 mg/l TDS. Table 14-1 is a compilation of the Ro
values. The data distribution for some counties was such that it was
necessary to include a range of Ro values, rather than a.single Ro value, in
the table. .
 355

1 Brazos
2 2 Cherokee
3 Denton
4 Ellis
"'" 5 Harris
6 Hidalgo

~
7 Hunt
6 8 Milam
8, '" 11

100!~~~!!gli~'~"~"~~2~11~~J9lM~C~M!UII~.nJI
10 Rusk
fE '1 Shelby
12 Wood

-r---+--+--+--H++++--+--+--1--1-'-t~~~",--

~~"r---
1010 100 1000
+--+--+-+-f--H+++---I-+--J-++-+-l++----l'<~Mt'::;"..,~++++H
~'" ~'10000
4
sumTDS

Figure 14-5. Curve fits for the Ro-TDS graphs of 12 counties. The graph of each county is in Volume II,
Section 5. The correlation for each county was good, so the curve fits were "eyeballed".

Table 14-1 shows that for a TDS of 1,000 mgtl, Ro values vary
considerably (15 to 45' ohm-meters). However, when grouped by aquifers,
the Ro values show fairly consistent patterns. In the case of the Carrizo-
Wilcox, an additional grouping by geographic area further enhanced the
consistency. Ro values for 10,000 mgtl TDS, while ranging from 1 to 10
ohm-meters, are usually less than 2 ohm-meters. Table 14-1 further
underscores the site-specific nature of this technique. It illustrates that
using an Ro cutoff for determining a particular water salinity will usually be
valid only in a limited geographic area.

When the aforementioned guidelines are followed, Ro-TDS graphs can

be used with a limited amount of success to estimate TDS. The graphs are
site-specific and should always be used cautiously.

Ro.-TDS Graphs

When an aquifer extends to depths of several thousand feet,

variations in Ro values at widely different depths will be, to some degree,
due to the fact that formation temperature and porosity vary with depth.
Alger (personal communication, 1988) proposed that the correlation
"
 356
TABLE 14-1. Ro VALUES CORRESPONDING TO TDS OF 1000 AND 10.000
MG/L ON COUNTY Ro-TDS GRAPHS

Aquifer/County Ro ohm-meters Ro ohm-meters

for 1000 mgll TDS for 10.000 mg/I TDS
Huecho Bolson Aquifer
EI Paso 15 -
Carrizo-Wilcox Aquifer (North)
Anderson 18 to 40 -
Angelina 15 1 to 2
Cherokee 20 -
Gregg 20 -
Nacogdoches 30 -
Rusk 18 . -
Shelby 25 -
Smith 25 -
Wood 25 -
Carrizo-Wilcox Aquifer (Central)
Brazos 30 -
Burleson 30 -
Freestone 40 -
Gonzales 35 -
Leon 25 -
Milam 30 10
Carrizo-Wilcox Aquifer (South)
McMullen 30 - .

Gulf Coast Aquifer

Harris 20 to 40 5
Hidalgo - 2 to 3
Jefferson 20 2
. Kleburg 25 -
Trinity Aquifer -
Collin 30 -
Dallas 40 -
Denton 25 to 45 -
Ellis 25 to 35 -
Fannin 45 -
Grayson 30 to 40 -
Hunt 28 -
McLennan 45 -
Tarrant 40 to 45 -
 357
coefficient of the corresponding Ro-TDS graph could be improved by using
Ro values normalized or corrected for variations in porosity and formation
temperature. The corrected Ro value is designated Roc'

The temperature correction normalizes Ro values, which are at

formation temperature, to equivalent Ro values at 77 0 F (designated as Roc)'
The procedure is as follows:

1. A geothermal gradient must be established. It can be calculated

for each well or an area-wide value can be used. Most counties in
Texas have geothermal gradients of 1.0 to 2.5 0 FI 100 feet.
Columns 3 and 4 of Section 3, Volume II, list geothermal gradients
for counties studied during this project. The Explanation section of
Section 3 documents the geothermal gradient calculation and
explains the problems associated with calculating a geothermal
gradient from the bottom hole data available for many water wells.

2. The geothermal gradient is used to calculate the temperature at the

depth at which Ro was measured. The equation for this
calculation is listed in the explanation for Column 7 in Section 3,
Volume II.

3. Ro is normalized to Ro at 77 0 F (Roel by substituting formation

temperature into Equation 2-4.

4. Roc is plotted versus TDS.

Temperature corrections can be made to Ro values from any type of

aquifer. Alger (personal commmunication, 1988) suggested that they would
be helpful when an aquifer extends below 1000 feet.

This study found that normalizing Ro values to a common temperature

(77 0 F) did nothing to improve the curve fit of Ro c-TDS graphs over that of
the corresponding Ro-TDS graphs. This was true even for Ro values from
, depths of up to 4,000 feet.

Temperature corrections using both site-specific and county-wide

geothermal gradients were applied to all the wells in this study (Section 3 of
Volume II). Many of these wells had Ro values from depths of 2,000 to
4,000 feet. Roc-TDS graphs utilizing county-wide geothermal gradients were
prepared for all 48 Ro-TDS graphs (Figures 14-6 and 14-7 are examples).
 358

1000

c;=
.5 100 :
0
a:
"0
c 0

'"
-:-
~

.<:
Cl 0

J: 10
0
a:

100 1000 10000

sumTDS

Figure 14-6. Roe-TDS graph for Harris County. Roc has been normalized to 77° F using a county-wide
geothermal gradient. The curve fit is no better than that of the Ro-TOS graph (Figure 14-1 I.

1000

c;=
.5
0
a:
"0
c 0

100
'"
-:-
~

.<: 0
Cl 0

J: 0
0
a: 0"" 0

0
0

0
0

100 1000 10000

sumTDS

Figure 14-7•. RO,-TOS graph for Oallas County. Ro, was normalized to 77° F using a county-wide
geothermal gradient. The curve fit is no better than that of the Ro-TOS graph (Figure 14·2).
 359

1000

~
3:
,~
.9 100
0
a:
"0
c: 0

~
'"
.c:
Ol
J: 10
0
a:

100 1000 10000

sumTDS

Figure 14-8. Ro.-TDS graph for Harris County. Ro. has been normalized to 77° F using site-specific
geothermal gradients. The curve fit is no better than that of the Ro-TDS graph (Figure 14-1).

C
3:
.9
0
a:
"0
c:
100
'"
-:--
~

.c:
!£!!
:r:
•
0 .

a:
••
•

101+0--I-+-+-+-H-ttOO±-:~,.....+-+-t--+++-+:14:1ooo:-:--+--+-+-f---t-+-:lmOOOO
1

sumTDS

Figure 14-9. Ro.-TDS graph for Dallas County. Ro. was normalized to 77° F using site-specific
geothermal gradients. The curve fit is no better than that of the Ro-TDS graph (Figure 14-2).
 360

Roc-TDS graphs utilizing site-specific geothermal gradients were prepared for

only 27 counties, since many wells lacked a bottom hole temperature·
(Figures 14-8 and 14-9 are examples). None of the 75 Roc-TDS graphs had
a curve fit that was any better than the corresponding Ro-TDS graph, as
illustrated by a comparison of Figures 14-1, 14-6, and 14-8, or 14-2, 14-,7,
and 14-9. (The rest of the Roc-TDS graphs were not included in this report.)

This lack of improvement in the curve fit, while due in part to

inaccurate geothermal gradients, demonstrates that scatter in the curve fit is
largely a function of factors other than temperature variations. Porosity
variations are the most likely explanation.

Ro can also be normalized for variations in porosity due to compaction.

The correction is valid only for unconsolidated sandstones in which porosity
decreases with depth as a function of increasing compaction. Alger
(personal communication, 1988) suggested that such Roc values would yield
better curve fits for the portions of unconsolidated aquifers from 1,000 to
3,000 feet in depth. He considered the South Texas Carrizo-Wilcox and the
Gulf Coast aquifers good candidates for the correction. .

The procedure to normalize Ro for variations in porosity due to

compaction is as follows:

1. A porosity value must be established for the depth that

corresponds to each Ro value. This can be obtained from a
porosity log, if available. Unfortunately, they seldom are available
for water wells (see Chapter 13). The oniy alternative is to .
estimate porosity values by first establishing a porosity gradient
from offsetting wells in the geographic area and depth range of the
Ro valuesin question. As mentioned above, the porosity gradient
will be valid only if porosity values are a function of compaction.

2.. A formation factor (F) is calculated for each porosity value (Le.
each sample depth) using equation 14-7. Formation factor is
discussed in the Formation Factor Equation section of this chapter.

3. A porosity value (depth) is chosen as the one to which all the rest
of the data will be normalized.

4. Ro values are normalized to the common porosity (depth) value by

the following equation:
 361

Ro e= Ro x Fe/F 14-1

Where:
Roc = Ro values corrected for variations in porosity due to compaction.
Ro = Resistivity of the uninvaded formation 100 percent saturated
with water.
Fe = Formation factor of the common porosity (depth) value.
F = Formation factor corresponding to the Ro value being normalized.

5. Roc is plotted versus TDS.

Alger (personal communication, 1988) used density and sidewall core

porosity values to calculate a porosity gradient for the Queen City and
Carrizo-Wilcox in Atascosa, Karnes, LaSalle, and McMullen Counties. His
equation is as follows:

¢ - 0.40 - .00003 x depth ± 0.01 14-2

Where:
¢ = porosity
depth is in feet

Equation 14-2 was used to normalize Karnes County Ro values

(Section 3, Volume II). Table 14-2 contains the calculations and Figure 14-
10 is a graph of the results. Only the Roc value from 4,300 feet had an
improved curve fit. All the rest of the Ro values were from a very small
depth range (700 to 1,100 feet). The scatter in the data from this interval is
most plausibly explained by porosity variations that are a function of
geological factors other than compaction.

In theory, the curve fit of Ro-TDS graphs should be improved by

normalizing Ro for the effects of temperature and porosity variations. In
practice, such was not the case for temperature normalization and porosity
normalization was only slightly successful. Ro e-TDS graphs appear to offer
•an advantage over Ro-TDS graphs only for porosity normalization and only
when porosity is a function of compaction.
 362

TABLE 14-2. KARNES COUNTY Roc VALUES NORMALIZED FOR POROSITY

Depth Porosity F F, Ro, High Ro, High

(feetl calculated from (ohm-meters) (ohm-meters)
Equation
14-2
. corrected to 77° F corrected to
using a county· 77' F and 1000
wide geothermal feet
gradient
4300 0.27 10.4 5.3 29 15
1000 0.37 5.3 5.3 , 14 14
1000 0.37 5.3 5.3 15 15
1000 0.37 5.3 5.3 10 10
1100 0.37 5.3 5.3 6 6
700 0.38 5.0 5.3 5 5.3
800 0.38 5.0 5.3 7 7.4
800 0.38 5.0 5.3 7 7.4
700 0.38 5.0 5.3 8 8.5
800 0.38 5.0 5.3 8 8.5
700 0.38 5.0 5.3 9 I 9.5

1000 _

10 _

'.')

11+o------j~+-+-l-+++-,1-!0-,-0-. -+--+++-+--I+,*'01::c=-----t-+-+-1H-+-!+.!0000
00 1

sumTDS

Figure 14-10. RO,·TOS graph for Karnes County. Ro, High normalized to 77° F with a county-wide
geothermal gradient (0 I was then normalized to 1,300 feet to compensate for porosity variations (0 ).
 363

Field Formation Factor (FFF)

Turcan (1962) introduced the Field Formation Factor (FFF) technique

as an empirical method of estimating Rw from resistivity logs. It is a slightly
modified form of Archie's formation resistivity factor equation (14-5). In
the FFF calculation Ro is first normalized to 77° F. This is done to negate
the effects of variations in formation temperature on Ro and to calculate an
Rw value at standard temperature (77 ° F).

There is one other significant difference between FFF and F.

In fresh water (water with an Rw greater than 3 ohm-meters) FFF is less
than F. In saltier water the two are equivalent. The difference between the
two is due to surface conductance and is explained below. Surface
conductance is explained in the Formation Factor Equation of this chapter.

FFF is the ratio of the Ro of a formation divided by the resistivity of

the water in the formation (Rw):

FFF = Ro(@ 77. Fl I Rw(@ 77· F) 14-3

Where:
FFF = Field formation factor
Ro(@ 77' F) = Resistivity of the uninvaded formation normalized
to 77° F
Rw(@ 77. F) = Resistivity of the formation water at 77° F

Having established the FFF of a formation, the Rw of a subsequent

well can be estimated by substituting the Ro of the interval in question into a
rearranged version of Equation 14-3.

Rw(@ 77' FI = Ro(@ 77' FI I FFF 14-4

Rw is then converted to Cw (Equation 14-1). Final!y, to determine TDS, Cw

. is plotted on the appropriate TDS-Cw graph or substituted into the equation
of the graph (Chapter 4 and Section 4 of Volume i1) •.

Equation 14-4 has one serious drawback for calculating Rw: FFF is
porosity dependent. Therefore, an FFF constant is valid only as long as
porosity remains fairly constant (Le., shallow, unconsolidated sands in a
limited geographic area). The method worked for Turcan because he was
 364
analyzing shallow sands in a limited area (Eocene Wilcox Group in north
Louisiana). His FFF values were fairly consistent, ranging from 1.7 to 3.0.

In deeper aquifers FFF increases with depth by as much as two orders

of magnitude (MacCary, 1984). MacCary calculated F's for Catahoula, Frio,
Yegua, and Wilcox sandstones of the Texas Gulf Coast. Figure 14-11, a .
graph of his Wilcox data, illustrates how F varies with depth, sometimes by
as much as a factor of 10 within a few hundred feet. The overall trend of F
increasing with depth is due to porosity decreasing due to compaction.
Other diagenetic processes may be contributing to the trend. The wide
variations in F over small vertical intervals are probably due to variations in
porosity associated with varying depositional facies. Whatever the reasons
for the variations in F, Figure 14-11 illustrates that calculating water quality
from an FFF constant is unreliable in deeper aquifers.

Alger calculated FFF's for about 400 of the wells in the data base in
Section 1 of Volume II. Values range from 0.1 to 28, with most of them
between 2 and 7. Wilson County is the best example of a consistent
county-wide FFF value. Five values range from 5.1 to 6.2, with one value of
3.9. There is considerable variation in the values for each of the other
counties, even in the
Carrizo-Wilcox and Gulf
1000 ,.---,--,---,--r--,---r---r--",
Coast aquifers. Dallas
County is a good example o
o
of widely ranging values. 0
0

FFF's range from 4.8 to 16, OJ o

o
reflecting porosity variations ... 0 0 0
o
00
within the Trinity aquifer. ",' 100 000 o
o o o
>- o
00 00
u 00
0 0
The FFF's in this data ~ 000 0 0
z Cb 00 0
base are smaller than values S! o 0 0 0
>- o {}
calculated using Archie's ~ o q, o
2
formation resistivity factor- o o 0

porosity equation (Equation ...'"o 10

o 0
o o
o 0 /
0 0
~;t:.o
0

o
o o o
14-6) and a known porosity o
value. The difference is due o
. 0

to surface conductance,
which lowers Ro in fresh 2 '--'_-l._.....L._--'-_.l.-_':---J.:---:'.
water aquifers. o 2 4 6 8 10 12 14 16
DEPTH, IN THOUSANDS OF FEET
Consequently, in fresh
water aquifers the FFF value Figure 14-11. Graph of formation factor vs. depth for the Texas Gulf
Coast Wilcox Group .(MacCary, 1984).
 365
calculated by Equation 14-3 is lower than an F value calculated with
Equation 14-6, which is independent of Ro.

Examination of the data base in Section 1 of Volume II reveals that

the FFF method of calculating water quality is not very accurate. In the
absence of porosity data, an Ro-TDS graph is preferred over an FFF
calculation. Indeed, a good linear fit to an Ro-TDS graph is really a plot of
Equation 14-4 for varying Ro's and a constant FFF. The scatter about the
linear fit demonstrates the degree to which porosity varies in the aquifer.
Examination of the scatter allows one to visually take into account the effect
of a change in porosity on Ro and consequently on the water quality
estimation. This allows one to make a more intelligent estimation of the true
water quality. An FFF calculation, on the other hand, offers no alternate
interpretations and lends a false sense of accuracy to the water quality
estimation.

STAND-ALONE TECHNIQUES FOR CALCULATING RW

Formation Factor Equation

Archie (1942) discovered that the resistivity of a water saturated rock

(Rol varies by a constant value as the resistivity of the formation water (Rw)
changes. He quantified the relationship as:
(l. "2- C- \
Ro = F x Rw 14-5

Where:

Ro = the resistivity in ohm-meters of the formation 100 percent

saturated with water
F = the formation factor, a proportionality constant
Rw = the resistivity in ohm-meters of the water saturating the
formation

The proportionality constant in Equation 14-5 is called the formation

resistivity factor (F R ) or formation factor (Fl. F ranges from 5 to several
hundred in sandstones and from 10 to several thousands in carbonates
(Helander, 1983).
 366

Archie derived Equation 14-5 by saturating core samples of different

porosities (10 to 40 percent) with waters of various salinities (20,000 to
100,00 mg/I of NaCI) and then measuring Ro. He found Equation 14-5 to be
valid for his entire range of porosities and salinities.

Archie also observed that Ro, and consequently F, decrease as

porosity increases. He inferred that F is a function of porosity and was able
to derive an empirical relationship between the two:

14-6

Subsequent investigation by Winsauer, et. al (1952) led to the addition

of a variable a in the numerator of Equation 14-6:

14-7

Where:

F = formation factor
a = tortuosity factor
rp = porosity in decimal form
m = cementation exponent

In the ideal case of pore spaces that are parallel cylindrical channels, F
would be inversely proportional to porosity, a and m would both equal 1, and
F would equal 1/rp. The pore system in almost all rocks, however, departs
from the ideal case. Depositional and diagenetic processes result in pore
diameters of varying cross-sectional areas and pore paths of varying lengths
or tortuosities. a and m quantify the degree to which the pore system
departs from the" ideal case. The names tortuosity factor and cementation
exponent are really misnomers for a and m, because the variables are the
product of several factors. For instance, Helander (1983) lists eight factors
that influence m: I

1. Degree of cementation·
2. Shape, sorting, and packing of the particulate system
3~ Type of pore system (intergranular, intercrystalline, vuggy, etc.)
4. Tortuosity of the pore system •
 367

5. Constrictions existing in porous system

6. Presence of conductive solids (clays, pyrite, etc.)
7. Compaction due to overburden
8. Thermal expansion

The pore system of natural rocks is too complicated for a and m to be

measured. Instead, they are empirically derived from the best line fit of a
logarithmic plot of Ro/Rw (F) and porosity in decimal form (Figures 14-12,
14-13, and 14-14). The Ro and porosity vCllues are obtained either from
core (Figure 14-12) or log (Figures 14-13 and 14-14) measurements. Rw
must be known independently.

The data is plotted on three-cycle log-log paper: F on the y axis with a

scale of 1.0 to 1000 and porosity on the x axis with a scale of .001 to 1.0.
m is the slope of the line that fits the data (Figure 14-13) and a is the value
ofF when ¢ = 1.0. a can also be expressed in terms of a rearranged
version of Equation 14-7 :
a = F x ¢m 14-8

Theoretically, a
should always equal 1, 1000.,1Jp:mrrommmommllIlll

since a porosity of 1.0

(1 00 percent) has an
Ro/Rw or F of 1 and
these values substituted ::.
into Equation 14-8 100
calculate an a of 1. In £
~
reality, a sometimes
:f
varies from 1.0. Such .~
cases may be an artifact •
§

~"
of the curve fitting
routine or compensation 10

for consistent variations

in pore geometry
(Hilchie, 1987).

Over the years a 1.0

0.1
and m have been 0.001 0.01 1.0

calculated for thousands

Poro,lly Fraclloll
of rock samples. a's of
0.6 to 4 and m's of Figure 14-12. Example of a formation factor-porosity graph constructed
from core data (Collier, 1988). For an a of 1, m is 1.95..
 1000.00

.. -
M=1.1 a_-
I\.• .. _
100.00
.! f\,. M=1.4

F
'2. 5
•I
.,
\
. ~=

I: ""'f-\ _-, A_I

~ :!i' :~
A \
10.00
,*. ., . I
Et=H=t-m~+ttttttl=1J:~ . :.,
t=tW+mt=-+-Ktt1ttt-tt11
..
;1E'~ ,rAJ· '1
1.00 _
.00 •
.. .01 ,. .10
I
.
X ' /~ .
DO
fin
.

Figure 14-13. Formation Factor-Porosity graph' constructed from log data. The porosity values are density-neutron crossplot porosity. The data clusters
into three groups, with the two main clusters (A and BI correlating to the portion of the borehole above or. below 500 feet. The column to the right of
the graph is a diagram of the borehole with the intervals plotted on the graph noted by blocks. These intervals were chosen because water samples
were available. The data clusters into distinct groups because of differences in the pore systems. m is 1.4 for group A and 1.1 for group B. Other
differences such as water chemistry and SP behavior (refer to Plate 2) between the two intervals reinforce the conclusion of a significant difference in
the petrophysical properties of the two intervals. The well was drilled in the Gulf Coast aquifer. The well is the TWOB PUB Test Well Site F, Cameron
County. Texas (state well number 88-59-411). Plates 1 to 4 and Figures 13-12 and 14-20 are also from this well. .
~
.0'1
OC
 '·'"1 mil ~. mil
\ ",. ><
100.00 .' :,,}"?x
v.·.·
: :••J<" ;.:.::0;. ...,.:. :.'~ •
.",,: "'..Y.":'- ." .-,:'.:
,1.'-;:;';,.' "
F :' : ';;~0: ~-i~·:~~~~:~,..,2_~t',I-++++-l-I
~-'

x ;.:: ...-- .
10.00 vXX
•• :4'-~<',
'·~.:~~t;;i
A ~_

1.00 I I I I I I I I II I I I I I I III I I I I I I I N
.nOl .01,J. .10 \,00

Figure, 14-14. Formation Factor-Porosity graph constructed from log data. The porosity values are density-neutron crossplot porosity. The data shows a
high correlation. Forcing the line fit through 1.0 on the x axis (a = 1.91 gives an m of 1.8. An m of 2 was used in the Rwa calculation for this well and
yielded accurate Rw values (Poteet, Collier. and Maclay, 19921. The well is in the Edwards aquifer. San Marcos, Texas. The well is the Edwards
Underground Water District.D, Hays County, Texas (state well number 67-01-8141. ' ~
0'1
\0
 370

0.57 to 5.4 have been documented (Porter and Carothers, 1970 and Focke
and Munn, 1987). Although specific a and m values have been calculated
for some formations, most of the time log analysts use one of three versions
of Equation 14-7:

F = .62 I (/)2,15 14-9

was developed by researchers at Humble Oil and Refining Company

(Winsauer, et. ai, 1952) and is known as the Humble equation. It is used for'
granular formations such as sandstones with porosities of 13 to 35 percent.

F = .81 I (/)2 14-10

is known as the Tixier equation. It also applies to granular formations. It is

often substituted for the Humble equation when porosity is approximately 20
to 10 percent because it is easier to calculate and the two equations are
equivalent for this porosity range (Figure 14-15).

14·11

is called the Archie equation. It is used for carbonates and low porosity (Le.
cemented or compacted) sandstones.

Figure 14-15 is a graphical solution of the three equations. Plate 1

demonstrates that the Humble equation is more accurate than the other two
in high porosity, uncompacted sandstones (the environment of most shallow
clastic aquifers, as well as the Gulf Coast and Carrizo-Wilcox aquifers).
I

The most accurate approach, however, is to calculate formation-

specific a and m values. Many log analysts set a equal to 1 and just deal
with m (Figures 14-13 and 14-14). For an a of 1, certain patterns emerge
regarding m: .

1. For uncemented grains m increases as sphericity decreases. m

ranges from 1.3 for uncemented spheres to 1.8 for uncemented
plates (Hilchie, 1982).

2. m increases as the cementation of the rock increases. m ranges

from 1.3 for completely unconsolidated sands to 2.2 for well
cemented sandstones (Helander, 1983).
 100

60
50
40

30
25
'"
..
cf.
>-
20

'iii 15
0
~
~

o 10
lL 9
8
7
6
5
4
3 :t,
1
2 +
j
I
1
2 3 4 5. 30 40 60 80 100 200 300 400 600 1000 2000 4000 6000 .10000
Formation Factor

Figure 14-15. Formation Factor versus Porosity and m (modified from Hilchie, 1982). The graph is used to calculate F, given m and porosity. ~
...
 372

3. m is 2 for carbonates with interparticle and intercrystalline porosity

greater than 5 percent (Focke and Munn, 1987). Below 5 percent,
m decreases as porosity decreases, down to 1.5 at 1 percent
porosity.

4. For carbonates with vuggy and moldic porosity, m increases as the

ratio of total porosity to vuggy/moldic porosity decreases (Towle,
1962). This is true when total porosity is greater than 5 percent.
At less than 5 percent total porosity, m decreases as porosity
decreases, down to about 1.3 at less than 1 percent porosity
(Focke and Munn, 1987).

5. Fractures decrease m. Wide open fractures can have an m as low

as 1.1 (Hilchie, 1982). This is because the pore geometry is
approaching the ideal case of parallel cylindrical. channels.

. Rw calculated from Equation 14-5 using an F obtained from the Ro/Rw

ratio in an offsetting well is the previously discussed Field Formation Factor
technique.· The previous discussion of F clarifies why the technique is so
site specific. F will be the same in two wells only when porosity, m, anda
are the same.

A much more accurate method, herein called the Formation Factor

Method, is to substitute Equation 14-7 into 14-5 and rearrange the equation
in terms of Rw:

Rw = Ro x ¢m I a 14-12

This equation is called an Rwa calculation in petroleum logging

literature. The term Rwa (apparent Rw) is used because when hydrocarbons
are present in a formation Ro becomes Rt ( the resistivity of the uninvaded
formation with its naturally occurring fluidS) and Equation 14-12 calculates
an incorrect Rw value (Rwa) that is greater than Rw. The Rwa curve is an
optional curve, so it is not found on many logs. It is usually placed in Track
1. The values are at formation temperature (Figure 14-16).

When the Formation Factor Equation is used to calculate water quality,

Rw must be converted to 77° F. This is done by first establishing the
geothermal gradient of the well (refer to the explanation for column 3 in
Section 3, Volume II) and then calculating the temperature at the depth of
interest (refer to the explanation for colume 6 in Section 3, Volume II). The
 373
~,

--r- ~...,

-- _ _
.,
l--l

,---t=
~-.~ -~ f=t}j
~ I - r - --+-1-

~ ~~

L, ,

2200

, r I' ,
'=1~
1- .-t-- 1-+-1
-!=,-:11=:
-\-r-:
11-
.1-
F.R.
=!i: ._
____ .... _'='~f. ... _
1.0000 11.000
___
' ~llJllll,L.L _~-=-
21000. 1000.0
npHI
.60000 0.0
____ J!Rli.llU;L~1. _ ~, ~ D~L-'.lN_) _
-.9000 .10000 -20.00 . . 20.000
, - - _ ...BIliLilllllll1L _~ _ ____________ ~~~Q~~L ~-

'i);o 5.0000 .£0000 • £000,0

~ ~~ ~ ~ ~ ~ ~.!<L ~(GR~I)~ ~ ~ ~ ~ ~ ~ ~ ~ _. . _. .1)..1:1..(I;ItlJ1t1? _. _ _.
0.0 150.00 .20000 2000.0
op (MV ) Of' ,,(nHMM'
o 20.0 0 .20000 000.0

Figure 14-16, Example of a petroleum type Rwa curve. Track 1 contains an Rwa curve computed by the
Formation Factor method using 1/(/J2 and density porosity (track 3). Rwa values are at formation temperature.
The arrows point to Rw values obtained from water samples and adjusted to formation temperature. The Rwa
value of zone A is 55 percent less than the measured Rw, while the Rwa value of zone B is 40 percent less
than the measured Rw. The difference between Rw measured and Rwa is greater than the difference
observed in laboratory tests by Evers and Iyer (Figure 14-18) for this Rw range. However, the differences may
be due to factors other than surface conductance. The gamma ray curve shows shale in the sands, which
would lower Rwa. The interval is part of the Carrizo~Wilcox. The well is the McKinley Drilling Company,
George Strait #1, Webb County, Texas. Rm is 5.23 ohm~meters at 90° F and Rmf is 5.87 ohm~meters at 75°
F. T.D. is 2,280 feet. bottom hole temperature is 105° F, and bit size is 97A, inches. Density porosity was
computed using a matrix density of 2,65 g/cm'.
 374

10‫סס‬oo.,....-----------------------,

,‫סס‬oo

~
..'"
c

'000
TOS .. 0.982 Cw o -
• r" 0.994
•

,oo+-.<:------~------~-------l
'00 1000 ,‫סס‬oo 100000

CBIcula1ed Conductivity
ij.Imhoslcm)

Figure 14-17. TOS-Cw graph for the Gulf Coast aquifer in Cameron, Hidalgo, Starr, and Willacy
Counties. The equation is used to calculate TOS in Plate 3. TOS includes 100 percent bicarbonate.

calculated Rw value is then converted to Rw at 77 0 F by substituting the

formation temperature into Equation 2-4. Rw at 77 0 F is converted to Cw
using Equation 14-1 . Finally, Cw is converted to TDS using the appropriate
TDS-Cw graph (Section 5 of Volume II). Figure 14-17 is the TDS-Cw graph
for Plate 3 . · .

Equation 14-12 will calculate accurate Rw values when the following

conditions are met:

1. The formation is shale-free. Shale lowers Ro, which means that

the calculated Rw will be too low.

2. The porosity value is accurate. If a single porosity log is used, the

lithology must be constant. If the lithology varies, a crossplot
porosity should be used (Chapter 12). The porosity log(s) must be
on depth with the resistivity curve. Porosity logs are run in very
few water wells. However, they are often available for post-1960
hydrocarbon tests.

3. The proper a and m values are used. Ideally, they should be

calculated for the formation being analyzed. However, as
discussed above, they can only be calculated if Rw is known I The
 375

only alternative is to estimate a and m based on available

geological information about the formation (sample description,
porosity log, regional geology, etc.). Most of the time a should be
set at 1.0 and m considered a variable. It may be advantageous to
establish a range for Rw, using high and low values of m (Plate 1).

4. Rw is less than 2 to 3 ohm-meters at 77° F. For higher Rw's

surface conductance causes the logging tool to record an Ra value
that is less than Ro. There is no way to accurately correct Ra to
Ro using just log data, so Ra is used for Ro in Equation 14-12.
This results in a calculated Rw that is too low and Cw and TDS
values that are too high. (This is the problem in Figure 14-16).

The surface conductance effect is a major drawback to using the

Formation Factor Equation to calculate the water quality of aquifers with
fresh to slightly saline water. Surface conductance has such a profound
. .
effect on calculated Rw values that it deserves further explanation.

Surface conductance increases the conductivity (or reduce~ ,he

resistivity) of an aquifer. Chemists (McBain, et aI., 1929; Urban, .et aI.,
1935) have shown that solid surfaces in contact with aqueous solutions
attract a layer of ions which, in turn, attract a layer of oppositely charged
ions which, in turn, attract... The result is double layer conductivity -- a more
concentrated solution of predominantly positively charged ions near the
surface of the solid, which has a higher conductivity than the rest of the
solution. Electrical current preferentially travels through the more conductive
double layer, thus making an aquifer appear to have a lower resistivity than
is actually the case.

Surface conductance occurs in all aquifers, fresh or saline. Its

magnitude is related to the ion concentration of the solution. In fact, surface
conductance increases with increasing salinity. However, the increase in
surface conductance always remains within the same order of magnitude,
while the increase in water conductivity may be several orders of magnitude
(McBain, et aI., 1929). This means that only in fresh water aquifers will
surface conductivity be a large enough percentage of the log-measured Ra
value to noticeably affect Ro. For aquifers in Texas, this study found that Ro
is significantly affected when Rw is greater than 2 ohm-meters (conductivity
less than 5000 pmhos/cm). This agrees with the laboratory work of Evers
and Iyer (1975b) which found a cutoff of 3 ohm-meters (3333 pmhos/cm).
 376
Evers and Iyer
(1975b) documented 100,..-'------------_-----,
~ tlaCl
that the magnitude of o Na,~04
a l:afiCC3
surface conductance T Cael
2
is also a function of
grain size. The
smaller the grain size
the greater the
surface conductance.
Figure 14-1 8 is an
example of two of 10

their data sets.

They also quantified Figure 14-18. Relationship of F and Rw for two experimental cores(Evers and
the extent to which Iyer, 1975). Grain size is in parentheses. Note: divide the Rw values by 100
to get ohm-meters.
Ro is reduced as Rw
increases: from a few percent at 3 ohm-meters to as much as 60 percent at
50 ohm-meters (Figure 14-18).

The amount of surface conductance is difficult to calculate from log

data. There are so many factors that affect Ra (shale, borehole effects, etc.)
that from log data it is very difficult to isolate and quantify the effect of
surface conductance on Ra. This makes it is very difficult to establish any
type of guidelines for correcting calculated Rw values. Calculations on a
well in the Carrizo-Wilcox (Figure 14-16) s'how a much larger difference
between Rwa and Rw measured than would be extrapolated from the lab
data. However, other factors (e.g. shale) may be contributing to the
difference.

Plates 1 to 5 are two examples of water quality calculations using the

Formation Factor Equation. Plates 1 to 4 are from a well in an
unconsolidated, high porosity sandstone (Gulf Coast aquifer). Plate 5 is from·
a well in a high porosity carbonate (Edwards aquifer).

Ro-Porosity Graphs

The Ro-Porosity graph, commonly known as the Pickett plot, is a

graphical solution of Equation 14-12. It is a slightly modified version of a
Formation Factor-Porosity graph (Figure 14-13). However, it is discussed
separately in logging literature and is therefore discussed separately in this
chapter.
 377
The data is plotted on log-log paper. Ro is plotted on the x axis and
porosity on the y axis. The data plots as a straight line as long as m and Rw
remain constant and there is e~ough variation in porosity to establish a
straight line. m is the negative slope of the line, the measured distance on
Ro divided by the measured distance on ¢ (Figure 14-19). The point at
which the line intersects the 100 percent porosity line is a x Rw at
formation temperature. If a is 1 (as it usually is), the point of intersection is
Rw at formation temperature.

The Ro-Porositygraph is subject to the same limitations as Equation

14-12. In many aquifers it is difficult to use the method to calculate Rw. In
homogenous, high porosity aquifers there will not be enough variation in
porosity to establish the slope of the line. In fresh to slightly saline aquifers
the curve fit will estimate an Rw that is too low and an m that is too high
due to surface conductance. An additional complication is the accuracy of
the curve fit (refer to Appendix II). However, an Ro-TDS graph should
always be constructed in conjunction with the Formation Fac~or Equation
method. Under the right circumstances (refer to the Formation Factor
Equation section) it can be used for several purposes: .

1. Rw can be estimated if m is' constant, Rw is constant, and a is

known (Figure 14-19).

2. m can be estimated if Rw is known and is constant (Figure 14-19).

3. Variations in m can be discerned if Rw is constant.

4. Variations in Rw can be discerned if m is relatively constant

(Figures 14-20 and 14-21).

Resistivity Ratio Method

Petroleum log analysts have been using the Resistivity Ratio method to
determine Rw for a number of years. Alger and Harrison (1988) proposed
that the technique should be utilized more often in ground-water logging.

The Resistivity Ratio method calculates Rw by comparing the

resistivity of the flushed zone (Rxo) with the resistivity of the uninvaded
zone (Ro). The technique is based on four assumptions, all of which are
normally valid:
 1

"7::.........,
"'x ~l>:

--n~' xx r·
x ':,,:., '<I x 1

I I I I I I I I I' "~~~::k{f,;;:8~
-.'. - .~:;2-50-,".;;:-~;,
I I II
{:\'~.;,.=::@<>~'.. ~,g;,>1 ~
It ':.:l I I 1 1
I I
1.6 ~ ~~1t~~Y' .~~ ,
j .. ~'-::'i1:~:'jo:" I-. I I I I I I I I I
f.!r=2.5 X I, < ~
I J
'" ~ I, 4 . .........

..
.011 I I I I I I I I I 1'1 I I I I I I I 1 1 1 I I I I I I
.1 1 RO 10 100

Figure 14-19. Using an Ao-Porosity plot to estimate Aw and m. 'Ao values are from the deep induction curve and porosity values are density-neutron
crossplot porosity. The well is in the Edwards aquifer and the data is from part of the Kirschberg Evaporite and Dolomitic Members of the Kainer
Formation. The data is from an interval that has a constant Aw (1.8 ohm-meters at formation temperaturel. Aw is from water samples obtained during
pump tests. Although there is scatter in the data, a fairly consistent pattern is present. The data is consistent enough to estimate Aw and/or m from the
graph. Assuming an a of 1, a line with an Aw of 1.8 ohm-meters was fitted to the data. This line yields an m of 2.5, not an unreasonable value in a .~

vuggy-moldic carbonate. However, an m of 2 was used in the Formation Factor Equation method and it gave accurate Aw's (Plate 5). Not knowing Aw, '-l
00
a best fit line through the data would yield an Aw close to the actual value. The well is the Edwards Underground Water District, A-l. Figure 13-32 and
Plate 5 contain additional information on this well.
 1.00

1-+--;f-I-l-+-U-I-!-.....1'~~
. . =~:>f"" ~
--.
" '

="'~,. .:,.,'~';;r~~~~
." " .... '" '•.
'"
'''',.A
~ """
a:!'-
1---t-+-+++1-tj-l-Jc~'.=~~-~'-~'~"~
I '" ~;~. . 7"'8: ," ."+'
B ~1-'''-b'-:'=-4';@~-1-UjJ~'- ,.-!, ---J.-l--ll
•.• .
lJ
yO'
.1:'
--s
" "A ;f!
~
.10
""", 3
5-
J

d=Dttttttt-=Rffi+m=+±±ffffE i' } I
I Ic"
•n
I ~,.. """"
.01
.10
4

1.Q0
4ft
... "..
10.00 ~=
" •••:='----L-_
100.00
_...J
RO

Figure 14-20. Using an Ro-Porosity graph to distinguish waters of different salinities. Ro values are from the deep induction curve and porosity values
are from density-neutron crossplot porosity. The data is from the entire length of the borehole. The data cluster into three groups lA, B, and C),
suggesting three different water salinities. Rw values from water samples taken during-pump tests confirm three Rw ranges. The scatter in groups A
and C is such that it is impossible to estimate either m or Rw. A good line fit can be drawn through group B, yielding an Rw that is too high (about 1.05
ohm-meters versus an actual Rw of about 0.54 ohm-meters at formation temperature) and an m of 1.3. This m is consistent with m's of 1.1 and 1.4
estimated from the Formation Factor-Porosity graph (Figure 14-13). The column to the right of the graph contains the deep induction curve and a few
100 foot depth markers. It also identifies the intervals in the well that correspond to the three clusters of data. The well is the TWDB PUB Test Site F,
Cameron County, Texas. Refer to Figure 13-12 and Plates 1 to 4 for additional information on this weill.
~
-a
 1.00, '~I s ~
v
s.
_
>'~
mm c=
~-

-
f'---......'- ,.

A
ojJ

.10 I _ 1_ J J I II III I I
_. x
.,

v I
-.

-,
I I I
,''-:1i:.IA,J I
'}J Ill:x:,.;'
",""."',+---l-=lc-+c44
l·-~.· '",,-,,=:;,,~.,.J'"

c
';"""""'-','0.-
,",

-~---
c,
H
\,--_.;':/
~_'-<.l:

I
I I I II I I 11 ~P-J--'-li I Ii" i I I I I I I II
B

.01
1.00 10.00
" 10nN 1000.no.
RC

Figure 14·21. Using an Ro-Porosity graph to distinguish waters of different salinities. Ro values are from an averaged spherically focused curve and
porosity values are from density-neutron crossplot porosity. The data is from the entire length of the borehole. The data cluster into three groups lA, B,
and Cl. suggesting three different water salinities. Group C, however, is the Georgetown Formation and the Regional Dense Member of the Person
Formation. Both are low porosity and very low permeability limestones. Therefore, in the case of group C the cluster represents a different pore
structure, rather than a distinct water salinity. Rw values from water samples faken during pump tests confirm two Rw ranges: 16 to 17 ohm·meters for
group A and 2 ohm-meters for group B. These Rw's agree closely with Rw's estimated from the graph. Group A has an m of 2.0 and B has an m of 2.2
12 was used in the Formation Factor Equation}. The column to the right of the graph identifies the intervals in the well that correspond to the groups.
\,j>
The well is in the Edwards aquifer. It is the- Edwards Underground Water District, C-l. New Braunfels, Texas (state well number 68-23-619). Poteet,
Collier. and Maclay (19921 contains detailed information on the well. ~
 381

1. The pore geometry and lithology of the formation near the borehole
is the same as it is laterally in the formation at the depth of
ir:lVestigation of the deep reading resistivity tool (normally a few
feet). '

2. The amount of surface conductance is the same in both the

flushed and the uninvaded zones.

3. The difference between Ro and Rxo is solely a function of the

difference in Rmf and Rw. .

4. There is mud filtrate invasion. However, near the bottom, of the

hole there may not be enough invasion for a resistivity contrast
between'the two zones.

When the above assumptions are valid, the Formation Factor Equations
for the flushed zone and the uninvaded zones (Equations 14-13
, and 14-14)
are equivalent (Equation 14-15). The Formation Factor Equation for the
flushed zone is

F = Rxo / Rmf 14-13

The Formation Factor Equation for the uninvaded zone is

F = Ro / Rw 14-14

Equations 14-13 and 14-14 are equivalent, so they can be Set equal to
each other:

Rxo / Rmf = Ro / Rw 14-15

The equation can then be rearranged to solve for Rw:

,
Rw = Rmf / Rxo / Ro 14-16
 382

Since F has been factored out of Equation 14-1 5, the Rw calculation

(Equation 14-16) is independent of porosity, m, and a., Surface conductance
effects are also factored out. If Rmf is adjusted to 77 0 F, Rw will be
calculated at 77 0 F, thus eliminating any need to know formation
temperature. The only data required for the calculation are accurate Ro,
Rxo, and Rmf values.

Accurate Ro values are usually not a problem if the formation is thick

and shale-free. iChapters 8 and 9 explain how to correct resistivity curves
for borehole effects.) If the formation is shaly, both Ro and Rxo must be
corrected (Alger and Harrison, 1988):

(Rxo I Ro) ' 1(' - Vel) = (R xo I R0 )ele.n 14-17

Where:

Rxo I Ro = log values uncorrected for shale

Vcl = clay volume
(Rxo I RO)ele.n = log value corrected for shale

The clay volume can be calculated from either the gamma ray (Figure 10-7)
or the SP curves (Equation 12-1).

Accurate Rxo values are essential. In wells with fresh-water muds and
fresh-water aquifers, Rxo curves read much too high. Any subsequent Rw
calculations will be too low. The appropriate borehole correction is not
difficult to make, but it has usually not been applied to the'Rxo curve on the
log. It must be made before using the Resistivity Ratio method (refer to the
FOCUSED PAD MICROELECTRODE TOOLS section in Chapter 9).

A greater problem with Rxo data is its availability. Rxo tools are
virtually never run in water wells and they are normally utilized by the
petroleum industry only in certain sections of the state, the Permian Basin
being the chief area. In the absence of an Rxo curve, an Rxo value can be
estimated from an Ri value obtained from a shallow reading resistivity curve
(LL8, SFL, Short Guard, or Short Normal). As shown by Figure 8-3, the
Short Normal reads deeper than the other three tools. Consequently, it will
be the least accurate. Figure 14-22 is a nomograph correcting Ri I Ro to
Rxol Ro for various curve combinations. The chart assumes an invasion
 383

--- ----
- _ - _ : : : 9'L-T"'ntl'
~()---~ ~: - :::::: :=. ---.--- - ------ ---
- :: -= :: - ~, - : : - : : ::
----
=~ =-: -= :
1-
- : - -= - :- ,
----6·

2.6
:!A

2.0

I.'

. - - - - -
--:':::-:-·:=~:L~
1.2

i+I-HH~1
1.1
• :1

g:~ ~~~;:' ~~- ~:::~: jii;;;;-: ;~::~; ~~;:~:;~"':: ~~;~ ~ ;II;":!' j~:! I; ~;
0.6 __ 0.9

Figure 14-22. Nomograph for converting Ri I Ro to Rxo I Ro. The chart converts the most common log
combinations: 16" normal 164" normal, 16" normal I deep induction, LL8 I deep induction, and SFL I deep
induction. An Ri I Ro value is entered on the appropriate line on the right side of the chart and a horizontal line
is drawn to corresponding Rxo I Ro value on the left side. The chart was calculated for an invasion diameter
of 20 inches, which is fairly typical for shallow, high porosity sandstones. Therefore, it contains generalized
conversion factors. Bob Alger constructed the chart.
 ·384

diameter of 20 inches, which is fairly typical for shallow, high porosity

sandstones (Alger, personal communication). The chart, therefore, contains
generalized conversion factors and the Rxo / Ro value obtained from the
nomograph will usually be an estimat~ that has an unknown degree of
accuracy. An Rw value calculated from data derived from Figure 14-22 is
subject to a great deal of error.

An accurate Rmf measurement mayor may not be a problem~ Rm and

Rmf are often measured on a sample of the drilling mud at the time of
logging (refer to the Drilling
I
Fluid Invasion section of Chapter 6). The values
are usually included on the log header. If only Rm is measured, Rmf can be
calculated from Rm. In fact, Alger (personal communication) believed that
an Rmf value calculated from Rm is more accurate than a measured Rmf.
This is because such a small volume of mud filtrate is available for the Rmf
measurement. However, most log analysts prefer measured values
(Schlumberger, 1988).

When calculating Rmf from Rm, some log analysts simply multiply Rm
by 0.75. The most commonly used conversion factor is the one developed
by Overton and Lipson (1958):

14-18

The value of the constant Km is a function of the mud weight (Table 14-3).
The equation is for drilling muds with Rm in the range of 0.1 to 10 ohm-
meters at 75° F. Most chart books contain a nomograph of Equation 14-18.

TABLE 14-3. Km VALUES FOR VARIOUS MUD WEIGHTS

Mud Weight K",

Ib/gal
10 0.847
11 0.708
12 0.584
13 0.488
14 0.412
16 0.380
18 0.350
 385
Lowe and Dunlap (19861. after reworking Overton and Lipson's data
and making Rmf measurements on additional samples, developed their own
conversion equation. The test data from both studies (Figure 14-23) shows
that the Rmf I Rm value of muds lighter than 11 pounds per gallon is much
more variable than it is for denser muds. In fact, for these lighter muds Rmf
can be greater than Rm. In the wells analyzed for this study, a wide range
was also observed in the ratio of Rmf to Rm, and in 14 percent of the wells
Rmf was greater than Rm (Section 1, Volume II).

The best policy is to accurately measure the Rm and Rmf of a

circulated sample of the drilling fluid. If possible. the measurements should
be taken daily during drilling because mud properties can change during the
course of drilling a well. In the absence of a measured value, Rmf can be
estimated from Equation 14-18. However, the calculated Rmf may be too
low for muds lighter than 11 pounds per gallon.

The Resistivity Ratio method will calculate an accurate Rw, if accurate

Rmf and Rxo values are available and if there is some invasion. Accurate
Rw values were calculated for approximately a dozen wells analyzed during

10.0

t,
Rmf/Rm
1.0
..... •• ! :. "
• •
••
Off"
.. 4,_'
• ~ff
....• • ... +)

4 • "
•
• ~.. $
~
\' i
."
• • T• ...t,
~.:',
.' ··.1·•r. .J ... ~;.( •
hi t;. •
-~
• •
•
4 +. ii:::--
:.
•
-- •

0.1
~
8 9 10 II 12 13 14 15 16 17 18' 19
Mud Density, PP9

Figure 14-23. Rml I Rm versus Mud Weight (Lowe and Dunlap, 1986). The graph includes data lrom
Overton and Lipson (1958), The data lor muds less than 11 pounds per gallon has considerable scatter.
 386

this study (Figures 14-24 and 14-25 and Plates 2 and 3 are examples). The
technique should be utilized more widely in ground-water studies, but Rxo
tools must first start being run.

SP

Rw calculations from an SP curve date back to the earliest days of

quantitative log analysis. Wyllie (1949) and Gondouin, et al. (1957)
authored the technique. Alger (1966) elaborated on the difficulties of
determining water quality with the SP curve. He also attempted to use the
curve to estimate hardness (Alger and Harrison, 1988). McConnell (1983,
1985, 1988, and 1989) has been the latest to pUblish on the use of the SP
curve in water quality calculations.

The technique requires a shale-free formation that is thick enough to

have a static SP (SSP). It also assumes that static SP is solely a product of
an electrochemical potential and that the shale adjacent to the formation of
interest is an ideal ionic permeable membrane. (The various aspects of the
. SP curve are discussed in Chapter 12.) Under these conditions, SSP is
related to the chemical activities of the cations in the formation water (a w )
and the mud filtrate (ami) by the formula:

14-19

For sodium chloride solutions and shales that are ideal ionic permeable
membranes, K is solely a function of temperature:

K - 61 + 0.133 T' F 14-20

For pure sodium chloride solutions that have an Rw or an Rmf of

greater than 0.12 ohm-meters (83,333 pmhos!cm), resistivity is inversely
proportional to the activity of the sodium ions (Figure 14-26). This means
that Equation 14-19 can be rewritten as:

SSP = -K log Rmf!Rw 14-21

 387

.~~ ~. . :... ~: If" ~:~

,,/ :~.
; . ii"
'1++ "'I'~'
I::C~I=
}+~
=-: ::::I:C:~
, '-
._~

_.f..
,
- I· _
_
1':..: :_

~_" __. ·1··

,
"
:1 ~,~ . ~ :.,
-:: I:'" ::: - ~
~... ~ ..
:I:'::'~:~ :
, "·"H!!' :~~ :.. ~ :::: p. ;~~
/.:
I·"" if,..· :. ......~~ y: ~ ~ ,~~:
"

~: :('t~
, cc"II+-:I:::::1:
,, ,:j:: ~::;; I: :~ ••~:.
800
.--,. _·t
.. ,.. 1
.:' c':: I:: ..: ..:
.--,- , ( :::.I·~ I:~'

-,"\
,)
,,' ..
",."

I~':
~~)+ .
=:e ::_ ;:;~ I::: :::
.,,
1_ ",_
•. 1. __
~~ :'~?J~: CC

_ ... \. -.-
'.

!~"ffi-~~';02:~
~80. 20. I. 1000_\
RSfL OHlIn
$1' ttV

.1·:·~irr1···ii~~?t!:·~ " . l'HIE .,

DEC

I' '-Gil"C ·'M>Y'-·-··" -';..

O. 150.10. I.
i.---oic-··~r -_. I.
, r'--Rii.L;
\000.110000.
'·Ottiiii:-',-":=ii".,
10001.5·
-'' ' ' ' "'=:'' ' ' "' ' ' ' ' ' ' ' ':':';1
O.

Figure 14-24. Water quality curve calculated by the Resistivity Ratio method using an Rxo curve. Track 4
contains a TDS curve calculated from Rw values obtained from the Resistivity Ratio method~ Rw was
converted to Cw and then TDS was calculated from Cw. The water analysis from a test hole drilled to 81 3
feet has a TDS of 590 mgll. The calculated TDS of the two sands from 730 to 813 feet varies from 500 to
700 mgll, very close to the TDS values that would give a composite TDS of 590 mgtl for the two sands. The
resistivity curves nearly overlay, which indicates that Rmf and Rw are about the same. Rmf is 8.2 ohm-meters
at formation temperature (82 0 Fl. Track 1 contains SP and borehole corrected gamma ray curves (GR C).
Track 2 is a lithology column calculated from the gamma ray and porosity logs. The volume of clay (VCL) and
the volume of quartz (dot pattern) are in decimal form (DECI. Track 3 contains microspherically focused
(MSFL) and dual induction curves. Track 5 contains a density-neutron crossplot porosity curve (PHIEI in
decimal form (DECI. The well is the Alsay, Cypress Creek U.D., #3, Harris County, Texas. The interval is part
of the Gulf Coast aquifer. Figure 9-14 contains additional information on this well.
 388

~~3~~ [[!llEl
"-- '<1''-:\;::' ::::\\ . . . . <ii
i
1850

.....\ '> .•. :ff; . . 6:i

;>'}y]'~
- -- -. -1= I·'.
1_ ;! L_:__ I~~I .• ~
'- ?I.. : t D
Ii ~~ \950

, i[!.!~
1- " I::t (.:1::' ,

"~ ~~. ·!~t;~~ .

-,i-
". e.
·ao. to. II.
COflL Oft
,.
" "' ". '" '"' ,.
von Of'
o. 160.11. ,.
--Oit-C-ipl--- veAL Of'

Figure 14-25. Water quality curve calculated by the Resistivity Ratio method using an Rxo curve. Track 4
contains a TDS curve calculated from Rw values obtained from the Resistivity Ratio method. Rw was
converted to Cw and then TDS was calculated from Cwo The water analysis of the screened interval (1850 to
1866 and 1910 to 1974 feet) has a TDS of 2330 mgtl. The calculated TDS of the main sandstone (1910 to
1970 feet) varies from 1900 to 2100 mg/l, an error of about 10 percent. Some of the error may be due to the
wrong exponent being used to convert Cw to TDS. The resistivity curveS nearly overlay. which indicates that
Rmf and Rw are about the same. Rmf is 3.38 ohm-meters at 81 0 F. Track 1 contains SP and borehole
corrected gamma ray curves JGR C). Track 2 is a lithology column calculated from the gamma ray and
porosity logs. The volume of clay (VCll, quartz (VQT21. and calcite (VCAll are in decimal form (DECI. Track
3 contains microspherically focused (MSFl) and dual induction curves. Track 5 contains a density-neutron
crossplot porosity curve (PHIE) in decimal form (DECI. The well is the J.L. Myers, Bristol Water Supply #2,
Ellis County, Texas. The aquifer is the Woodbine. Figure 9-11 contains additional information on this well.
 389
By rearranging Equation 14-

lom~~
21, Rw can be calculated:
1 P'.o.
5 Resistivity 01 Noel vs Na+
3 Acti,ity (Temp. 75°F) \
Rw = Rmf/10,-SSP/K) 21-+--i"l:1+fttt-H--H+tttl---+--+-HitHl
14-22 '"
Q; 7
0;1.0 _
E 5
,
E 3 --
--
--
.c ,
If Rw or Rmf is less than o 2 1-+-H-++I+H--H>..t-IH+t+t--+--+-t+t+Hl
l, rue Resistivity NaGI
0.12 ohm-meters (Figure 14-261.
if the formation water is a type
other than sodium chloride, or if
polyvalent cations are present in
either the formation water or the
mud filtrate, Rw and Rmf are no
longer inversely proportional to aw
and amI' Under any of these
circumstances, Rw and Rmf in Figure 14-26. Chart of aN. vs. NaCI resistivity IGondouin. et
al. 19571.
Equations 14-21 and 14-22
become equivalent resistivities (Rmfe and Rwe). An equivalent resistivity is \

the resistivity value of the sodium chloride solution that will genera~e the
same SSP as that generated by a non-sodium chloride solution .. Any basic
petroleum log analysis text has a discussion of equivalent resistivities for low
resistivity waters. This research was limited to non-sodium chloride waters,
. including those with polyvalent cations.

Polyvalent cations were found to be common in fresh to moderately

saline ground 'waters in Texas (Section I, Volume 2). Most commonly the
polyvalent cations are divalent calcium and magnesium ions. These two ions
have a chemical activity that is about an order of magnitude greater than
that of sodium ions (Figure 14-27). Consequently, divalent ions generate a
negative SP that is much larger than the SP generated by an equivalent
concentration of sodium ions. Therefore, Rwe is much lower than Rw. Rwe
must be converted to Rw by means of an empirically derived correction
factor.

The relationship between Rw and Rwe for a particular water type can
be established if a sample of the water in question is available. The
conversion factor is the ratio of the Rw of the water sample to the Rwe
calculated using Equation 14-22. Figure 14-28 is a graph of Rw-Rwe
 390
10 10

t..o
1.0 1.0

I'

0.1 ~ ~ ~
0.1
Cl
E
III

+
=
u
0.Q1
~ ~
1-- l,,; - 0.Q1

-
III

y
0.001 0.001

,
0.0001 1,;'.
0.0001

0.00001 t..o 0.00001

~ 0.1 1.0 10 100 1,000 10,000 100,000

PPM c • + PPM Mg or PPM N•

Figure 14-27. Cation concentrations vs. activities for sodium and calcium plus magnesium ions at 77° F (modified from Alger, 1966).
 391

relationships for several pure end-

member water types. Natural
waters will fall somewhere in
between the end members.

If a water sample is not

available, the only option is to use
an Rw-Rwe relationship from an
offsetting well that has the same
concentration of divalent ions.
Unfortunately, the only way to be
sure that the concentrations are
the same is to have an analysis of
both waters! An additional
difficulty is that the water sample •• 1.0 2 • 10 20 50 100
may not accurately reflect the
chemical composition of the
formation water. Mud filtrate Figure 14-28. Rw-Rwe relationships for various types of
waters ISchlumberger, 19721.
invasion may alter the
sodium/calcium/magnesium ratio of formation water passing through the
invaded zone until several pore volumes of water have been produced
(McConnell, 1985).

Calcium and magnesium ions can occur in mud filtrate, but the filtrate
is usually a sodium chloride solution. This is because the divalent ions are
preferentially adsorbed by clay platelets in the mud, while sodium ions are
released into the drilling fluid (Gondouin, et aI., 1957). If aquagel has been
added to the drilling fluid or if long shale sections have been drilled, there is
usually enough clay in the mud to adsorb any polyvalent ions (Alger and
Harrison, 1988). At least, this is the assumption that has to be made, since
a chemical analysis of the mud is seldom available (Alger, 1966).

There are several other conditions under which the SP curve will
calculate an erroneous Rw value:

1. The presence of clay in an aquifer, or a thin-bedded aquifer. Both

reduce the amount of negative SP deflection, so a calculated Rw
value will be too high (Plate 2).

2. An SSP with an electrokinetic potential component. SSP will be

larger than expected, so a calculated Rw value will be too low.
 392
3. A formation water or a mud filtrate that is predominately sodium
bicarbonate. K will be less than that calculated by Equation 14-20
(Schlumberger, 19891. This will result in a calculated Rw that is
too high.

4. Shales that are non-ideal membranes. Since K is a function of the

cation exchange capacity of the shale (Silva and Bassiouni, 1981 I,
it will be different than the K calculated with Equation 14-20.
Unfortunately, cation exchange capacities are very seldom
measured. Silva .and Bassiouni (1981 I recommend determining K
from a previously established empirical relationship between K and
shale resistivity. Shales that behave as non-ideal membranes are
usually Tertiary or younger and K is less than what Equation 14-20
calculates (Alger and Harrison, 1988).

A water analysis can also be used to predict SSP. A known Rmf is

converted to amI (Figure 14-26). aw is calculated from the sum of aN. and
-Vac. + aMg (Figure 14-27). Equation 14-19 is then used to calculate SSP.
However, such an exercise is mainly of academic interest and has little
practical application in ground-water studies.

Calculating Rw from the SP curve was very unsuccessful for the

wells studied in this project. This was especially true in cemented
sandstones with varying porosities, such as the Trinity aquifer. Plate 2
contains an Rw curve calculated from the SP of a well in Cameron County,
Texas. The calculated Rw values are accurate down to 500 feet, but below
this depth they are much too high. Possible explanations for the.
discrepancies are discussed in the explanation for Plate 2.

Calculating Rw from an SP curve works well for sodium chloride

waters as long as the aforementioned conditions are met. If, however, the
water ha's appreciable divalent ions, as is the case of many fresh to
moderately saline ground waters, the technique is very difficult to use. It
will work only if an Rw-Rwe relationship has been established.
Unforturiately, the only way to establish the relationship is to first measure
the Rw 6f a sample of the water; if a water sample is available, logging
techniqubs are not needed! An Rw-Rwe relationship can be used from an
adjacent l well, as long as the calcium and magnesium concentrations are the
same in both wells. But again, the only way to be sure of this is to have a
water analysis from both wells. Thus Rw calculations from an SP curve in
ground waters with divalent ions are only reliable after the fact.
 SUMMARY AND CONCLUSIONS

Borehole geophysical techniques are a viable method of determining

the water quality and hydrogeological parameters of aquifers. However,
accurate answers require utilization of the correct logging suite, the
appropriate interpretation techniques, and appropriately constructed TDS-Cw
graphs.

This study found a wide variety in the accuracy with which labs in
Texas measure specific conductanCe. Some labs need to improve their
calibration procedures. It was found that the Texas Department of Health's
method of calculating specific conductance from diluted conductance does
not give accurate values. Diluted conductance should not be substituted for
measured conductance. It was documented that in the absence of a
measured conductivity value, specific conductance can be accurately
calculated from ionic concentrations.

Temperature-conductivity relationships for slightly to moderately saline

non-NaCI type waters have not been published in either petroleum or ground-
water literature. One of three equations is normally used to adjust
conductivity for temperature fluctuations: Arp's, simplified Arp's, and 2
percent per· C. Analysis of published data and laboratory experiments
conducted during this project revealed that no one equation consistently
yielded more accurate values. All three equations give acceptable
temperature-corrected conductivity measurements, but the 2 percent per· C
equation is less likely to yield extreme values.

This study found a difference in the way labs calculate TDS. Some
labs include 100 percent of the bicarbonate value, while others use 49.2
percent. Since ionic concentration governs Cw, 100 percent of the
bicarbonate value should be included. Also, some labs do not document and
perform a check on the accuracy of their TDS measurements.

TDS can be accurately estimated if Cw is known by using a TDS-Cw

graph. These graphs have not preViously been constructed for Texas ground
waters. TDS-Cw graphs were constructed in this study for the major
aquifers in the state. Also, a data base consisting of water analyses and log
values from over 700 wei lis in the major Texas aquifers was compiled and
analyzed.

393
 394
, This study researched and established the proper procedure for
constructing TDS-Cw graphs: guidelines for where to obtain water analyses,
how to correct Cw to standard temperature (25" Cl. how to compute the
proper TDS value, how to critique the accuracy of water analyses, and how
to construct the graphs. Important conclusions include:

1.· TDS should include 100 percent of the bicarbonate value.

2. The best graph is a log-log scale plot with the equation of the line
fitted by reduced major axis and the equation of the line
transformed to a power law.

3. Because of interionic interference, the graphs need to be

established for a local area in order to conform to local water
chemistry.

4. When the data is extrapolated to a TDS greater than the range of

the data, TDS will be too low due to the effects of interionic
interference on specific conductance.

There are two important fundamental differences between petroleum

and ground-water/environmental logging:

1. Surface conductance has a significant effect on resistivity curves

in ground-water logging. It was discovered that for ground waters
in T~xas, surface conductance becomes a factor when Rw is
. greater than 2 ohm-meters.

2. SP interpretation in petroleum logging is based on sodium chloride

formation water. Many of the ground waters in Texas have
calcium and magnesium, which complicates SP interpretation.

There are also important practical differences between the two types'
of logging. In ground-water/environmental logging, it is usually true that:

1. Bit sizes vary more, from a few inches to a few feet.

2. Poor quality drilling mud is common.

3. Little attention is paid to selecting the proper logging suite and to

quality control.
 395

4. There is a dearth of knowledge about logging techniques.

5. Slimhole logging tools are often used.

All of these factors complicate ground-water log interpretation.

Borehole geophysical tools have been run in ground-water wells since

the early days of the science. However, little attention was ever paid to the
subject by petroleum logging companies. Consequently, ground-water
logging technology is decades behind petroleum logging. The applicability of
many logging tools to ground-water studies was' not previously documented,
and guidelines for utilizing these tools in groul)d-water environments were
not established. '

Existing and new openhole logging tools were run in this study in the
main types of aquifers in the state (unconsolidated clastics, consolidated
clastics, and carbonates). Each tool was evaluated for its applicability to
ground-water/environmental studies. Guidelines were also written for using
the tools in ground-water/environmental studies. The most important
conclusions of this phase of the study are:

,. The single-point resistance tool, popular in environmental and

mineral logging, should never be the primary resistivity tool. The
tool has few strengths and several weaknesses.

2. Apart from the problem with bed definition, normal logging tools
are ideally suited to openhole ground-water logging. The tools
have been abandoned by the petroleum logging industry in favor of
inducti,on tools. However, in ground-water environments it is
easier to get accurate resistivity values with normal curves than
with induction curves.

3. Focused electrode (guard) tools are designed for the exact borehole
environment of most water wells. The tools give accurate
resistivity values and excellent vertical resolution. They should be
utilized much more widely in ground-water/environmental studies.

4; The gamma ray and caliper logs are infrequently run in water wells,
but should be a standard part of every logging suite. The gamma
ray provides lithology information and the caliper provides a picture
 396

of the hole diameter. Both curves are useful in interpreting other

curves.

5. Porosity logs are very seldom run in water wells. They should be
run routinely in carbonate aquifers such as the Edwards and in
consolidated clastics such as the Trinity.

6. The Litho or Spectral Density tool is the porosity tool of choice. It

is an excellent stand-alone porosity tool, with the ability to .both
identify lithology and calculate porosity.

7. Contrary to popular opinion, the sonic log will give accurate

porosity values in shallow, unconsolidated formations if the correct
transform is used. The Raymer-Hunt transform gives accurate
porosity values in such circumstances.

8. Considerable improvement needs to be made in the quality of some

slimhole logging tools. Specifically, gamma ray tools must be
calibrated in API units; density and neutron tools must be
calibrated in such a manner that porosity can be calculated with
the tools.

Seven techniques for calculating Cw and TDS from logs were

evaluated, three empirical relationships (TDS-Ro Graphs, Roc-TDS Graphs,
and Field Formation Factor) and four stand-alone techniques (Formation
Factor Equation, Ro-Porosity Graphs, Resistivity Ratio Method, and SP). The
Field Formation Factor method produced the least accurate results ..
Theoretically, Roc-TDS Graphs should be more accurate than TDS-Ro Graphs,
but in reality they were not. Both types of graphs are site-specific. Ro-TDS
graphs were constructed for 48 counties.

The Formation Factor Equation calculates accurate Rw values, as long

as surface conductance is negligable (Rw less than 2 ohm-meters). The
drawback is tha.t a porosity curve is required and the cementation exponent
must be accurately estimated. The Ro-Porosity Graph is simply a graphical
solution of the Formation Factor Equation. The advantage is that m can be
estimated and trends in water quality and m can sometimes be discerned.
The Resistivity Ratio Method has the advantage of canceling the effect of
surface conductance. It can, therefore, accurately calculate Rw regardless
of the salinity of the water. The drawbacks are that it requires an accurate
Rmf measurement and an Rxo tool. Rmf measurements are sometimes not
 397

recorded and sometimes not representative of the zone of interest. Rxo logs
are scarce, except in the Permian Basin. Rxo tools were run in a number of
wells to demonstrate the viability of the technique. The SP methOd works
only in sodium chloride water, which ground waters in Texas usually are not.
In addition, there are a number of other conditions that must be met in order
for Rw calculations from an SP to be accurate. The SP method is the least
reliable of the four stand-alone methods.

When calculating water quality from geophysical logs, it is best to use

all of the techniques for which data is available. It is also essentfal to make
sure that any necessary environmental corrections have been applied to the
logging curves and that one is making the correct interpretation as to the
relationship of a resistivity curve to the invasion profile.

This study has shown the need for research in several areas:
.
1. TDS-Cw graphs need to be constructed for all aquifers and
counties in T e x a s . )

2. Several areas of slimhole logging need to be researched. Accurate

density and neutron tools, along with additional guard and.
induction tools, need to be developed.

3. Temperature-conductivity relationships for slightly to. moderately

saline non-NaCI type waters need to be documented.

4. The applicability of logging technology to ground-water and

environmental studies needs to be better documented in the
literature. Quality control guidelines also need to be better
documented.

5. The applicability of cased hole logging technology to ground-water

and envi~onmental studies needs to be studied. Subjects of .
particular importance include:

a. Techniques to evaluate the integrity of bentonite grout in PVC

cased holes.
b. A cased hole resistivity tool.
c. Cased hole porosity tools.
 398

BIBLIOGRAPHY

Alger, R.P., 1966, Interpretation of electric logs in fresh water wells in

unconsolidated formations, paper CC, in 7th an"nual symposium
transactions: Society of Professional Well Log Analysts, 25 p.

Alger, A.P., and Harrison, C. W., 1988, Improved fresh water assessment in
sand aquifers utilizing geophysical well logs, paper I, in 2nd
international symposium on borehole geophysics for minerals,
geotechnical, and groundwater applications [Golden, Colorado, Oct. 6-
8, 1987], proceedings: Society of Professional Well Log Analysts,
.Minerals and ,Geotechnical Logging Society Chapter-at-Large, p. 99-
118. Also published in 1988, in FOCUS conference on southwestern
ground water issues [Albuquerque, New Mexico, March 23-25]
proceedings: National Water Well Association, Dublin, Ohio, p. 281-
309. Also published in 1988, in 2nd national outdoor action
conference on aquifer restoration, ground water monitoring, and
geophysical methods [Las Vegas, May 23-26] proceedings, volume 2:
Association of Ground Water Scientists and Engineers (AAASI. p. 939-
968. Later published in 1989, The Log Analyst, v. 30 (11. January-
February, p. 31-44.

Allaud, L.A., and Martin, M.H., 1977, Schlumberger, the history of a

technique: John Wiley & Sons, New York City, 333 p.
.
American Public Health Association, American Water Works Association,
Water Pollution Control Federation, 1985, Standard methods for the
examination of water and wastewater, (16th ed.): American Public
Health Association, Washington, D.C., 1268 p.

Archie, G.E., The electrical resistivity log as an aid in determinin'g some

reservoir characteristics: Transactions AIME, v. 146, p. 54-62.

Arps, J.J., 1953, The effect of temperature on the density and electrical
resistivity of sodium chloride solutjons: Transactions AIME, v. 198, p.
327-330. .

Baroid, 1991, Drilling fluid products for water well, mineral exploration, and
geotechnical drilling: NL Baroid Corp., Houston, variously paginated.
 399

Bateman, R.M., 1985, Openhole log analysis and formation evaluation:

IHRDC Press, Boston, 647 p.

Beeson, C.M., and Wright, C.C., 1952, Loss of mud solids to formation
pores: Petroleum Engineering, p. 40-52.

Bevington, P.R., 1969, Data reduction and erro'r analysis for the physical
sciences: McGraw-Hili Book Company, New York, p. 93-103.

Brown, C.E., 1988, Determination of rock properties by borehole-

geophysical and physical-testing techniques and ground-water quality
and movement in the Durham Triassic Basin, North Carolina: U.S.
Geological Survey Professional Paper 1432, 29 p.

Brown, D.L., 1971, Techniques for quality-of-water interpretations from

calibrated geophysical logs, Atlantic Coastal area: Groundwater, v. 9
(4), p. 25-38 .

. Clenchy, D.R., 1985, Effect of borehole diameter on log quality and'

interpretation--Grand Banks area, east coast Canada, paper V, in 10th
formation evaluation symposium transactions: Canadian Well Logging
Society, Calgary, 26 p.

Coates, G., Collier, H., Milligan, B., Vasil ache, M., and Carter, J., in press,
MRIL * an environmentally safe measure of porosity and permeability:
in 4th international symposium on borehole geophysics for minerals,
geotechnical, and groundwater applications [Toronto, Ontario, August
18-22), proceedings: Society of Professional Well Log Analysts,
Minerals and Geotechnical Logging Society Chapter-at-Large.

Cohen, S.C., 1981, Relationships among the slopes of lines derived from
various data analysis techniques and the associated correlation
coefficient: Geophysics, v. 46 (11), p. 1606..

Collier, H.A., 1988, The effect of isolated biomoldic porosity on the log
analysis of a Pennsylvanian carbonate reservoir in north Texas, paper
X, in 29th annual symposium transactions: Society of Professional
Well Log Analysts, 16 p.

_ _-::---;-,-" 1989, Assessment of the dielectric tool as a porosity device, in

3rd national outdoor action conference on aquifer restoration, ground
 400

water monitoring and geophysical methods [Orlando, Florida),

proceedings: National Well Water Association, Dublin, Ohio, p.
1 51-165. Also published in 1989 as paper l, in 3rd international
symposium on borehole geophysics for minerals, geotechnical, and
groundwater applicatons, proceedings: Society of Professional Well
log Analysts, Minerals and Geotechnical logging Society,
Chapter-at-large, p. 183-197.

Davis, J.C., and Doveton, J.H., 1990, Overview of statistical methods in

log analysis: Society of Professional Well log Analysts computer
applications workshop [lafayette, louisiana, June 281 course notes:
Society of Professional Well log Analysts, Houston, 19 p.

Davis, S.N., 1988, Where are the analyses?: Groundwater, v. 26 (1),

p.2-5.

Desai, K.P., and Moore, E.J., 1969, Equivalent NaCI determination from
ionic concentrations: The log Analyst v. 10 (3), p. 12-21.

Dewan, J.T., 1983, Essentials of modern open-hole log interpretation:

PennWell Books, Tulsa, 361 p.

Doll, H.G., 1949, SP log: theoretical analysis and principles of

interpretation: Transactions AIME, v. 179, p. 146-185.

Dresser Atlas, 1982, Well logging and interpretation techniques: Dresser

Industries, Inc., Houston, variously paginated.

_ _....,...,.._' 1985, log interpretation charts: Dresser Industries, Inc.,

Houston, 157 p.

Driscoll, F.G., 1986, Groundwater and wells, (2nd ed.): Johnson Division,
St. Paul, Minnesota, 1089 p.

Dunlap, H.F., and Hawthorne, R.R., 1951, The calculation of water

resistivities from chemical analyses: Transactions AIME, v. 192,p.
373-375.

Dyck, J.H., Keys, W.S., and Meneley, W.A., 1972, Application of

geophysical logging to groundwater studies in southeastern
Saskatchewan: Canadian Journal of Earth Sciences, v. 9, p. 78-94.
 401

Ellis, A.J., 1976, The I.A.G.C. interlaboratory water analysis comparison

programme, Geochimica et Cosmochimica Acta, v: 40: Pergamon
Press, Great Britain, p. 1359-1374. .

Emerson, D. W., and Haines, B.M., 1974, The interpretation of geophysical

well logs in water bores in unconsolidated sediments: Bulletin of
Australian Society of Exploration Geophysics, v. 5 (31. p. 89-118.

Etnyre, L.M., 1984a, Practical application of weighted least squares

methods to formation evaluation, part I: The logarithmic transformation
of non-linear data and selection of dependent variable: The Log
Analyst, v. 25 (1 I. p. 11-21.

_ _ _---::_' 1984b, .Practical application of weighted least squares

methods to formation evaluation, part II: Evaluating the uncertainty in
least squares results: The Log Analyst, v. 25 (3). p. 11-20.

_ _-=--;-;-:' 1989,. Finding oil and gas from well logs: Van Nostrand
Reinhold, New York, 306 p.

'Evers, J.F., and Iyer, B.G., 1975a, A statistical study of the SP log in fresh
water formations of northern Wyoming, paper K, in 16th annual
symposium transactions: Society of Professional Well Log Analysts
logging symposium, 8 p.

_ _ _---,:--' 1975b, Quantification of surface conductivity in clean

sandstones, paper L, in 16th annual symposium transactions: Society
of Professional Well Log Analysts, Houston, 11 p.

Focke, J. W., and Munn, D., 1987, Cementation exponents in Middle Eastern
carbonate reservoirs: SPE Formation Evaluation, v. 2 (21. p. 155-167.

Fogg, G.E., and Blanchard, P.E., 1986, Empirical relations between Wilcox
ground-water quality and electric log resistivity, Sabine uplift area, in
Geology and ground-water hydrology of deep-basin lignite in the
Wilcox Group of east Texas, W.R. Kaiser, Principal Investigator, Bureau
of Economic Geology, University of Texas at Austin .
.
Frank, R. W., 1986, Prospecting with old E-Iogs: Schlumberger Educational
Services, Houston, 161 p.
 402
Freeze, A.A., and Cherry, J.A., 1979, Groundwater: Prentice-Hall, Inc.,
Englewood Cliffs, N.J., 604 p.

Gardner, J.S., and Dumanoir, J.L., 1980, Litho-density log interpretation,

. paper N, in 21 st annual symposium transactions: Society of
Professional Well Log Analysts, 23 p.

Gearhart, 1981, Basic open hole seminar: Gearhart Industries, Inc., Fort
Worth, variously paginated.

Glenn; E.E., and Slusser, M.L., 1957, . Factors affecting well productivity - II.
Drilling fluid particle invasion into porous media: Transactions AIME,
v. 210, also published in Journal of Petroleum Technology (May). p.
132-139.

Glenn, E.E., Slusser, M.L., and Huitt, J.L., 1957, Factors affecting well
productivity- I. Drilling fluid filtration: Transactions AIME, v. 210, also
published in Journal of Petroleum Technology (May), p. 126-131.

Goetz, J.F., Dupal, L., and Bowler, J., 1979, An investigation into
discrepancies between sonic log and seismic check shot velocities:
unpublished.

Gondouin, M., Tixier, M.P., and Simard, G.L., 1957, An experimental study
on the influence of the chemical composition of electrolytes on the
S.P. curve: Transactions AIME, v. 210, also published in Journal of
Petroleum Technology (February 1958), p. 58-70.

Guo, Y.A., 1986, Estimation of TDS in sand aquifer water through

resistivity log: Ground Water, v. 24 (5). p. 598-600.

Guyod, H., 1944, Electric well logging, part 3; The single-point resistance
method: The Oil Weekly, August 21, p. 44-52.

~_ _....,....,,:-:-' 1957, Resistivity determination from electric logs: privately

published, variously paginated.

_ _ _-::-_' 1966, Interpretation of electric and gamma ray logs in water

wells: The Log Analyst, v. 6, (5). p. 29-44.
 403
Guyod, H., and Pranglin, J.A., 1959, Analysis charts for the determination
of true resistivity from electric logs: Houston, 202 p. .

Hallenburg, J.K., 1980, A comparison of nonpetroleum and petroleum

logging, chapter E, in 21 st annual symposium transactions: Society of
Professional Well. Log Analysts, 7 p.

_~_----,,:--' 1984, Geophysical logging for mineral and engineering

applications: PennWell Books, Tulsa, 254 p.

Hansen, H.J., and Wilson, J.M., 1984, Summary of hydrogeologic data

from a deep (2,678 ft.) well at Lexington Park, St. Mary's County,
Maryland: Maryland Department of Natural Resources, Maryland
Geological Survey, open-file report 84-02-01, 61 p. '

Hartline, no date, Logging course notes: privately published, variously

paginated.

Hartmann, D.J., 1975, Effect of bed thickness and pore geometry on log
response, paper Y, in 16th annual symposium transactions: Society of
Professional Well Log Analysts, 14 p.

Hassan, M., Hossin, A. and Combaz, A., 1976, Fundamentals of the

differential gamma ray log interpretation technique, paper H, in 17th
annual symposium transactions: Society of Professional Well Log
Analysts, 18 p. .

Helander, D.P., 1983, Fundamentals of formation evaluation: Oil and Gas

Consultants International, Tulsa, 332 p.

Hem, J.D., 1982, Conductance; a collective measure of dissolved ions, in

R.A. Minear and L.H. Keith, editors, Water analysis, v.l,lnorganic
species, pt. 1: Academic Press, New York, p. 137-161.

_ _---::----,., 1985, Study and interpretation of the chemical characteristics

of natural water: U.S. Geological Survey Water-Supply Paper 2254; p.
1-249.

: Hertzog, R., Colsen, L., Seeman, B., O'Brien, M., Scott, H., McKeon, D.,
Grau, J., Ellis, D., Schweitzer, J., and Herron, M., 1987, Geochemical
logging with spectrometry tools, SPE-16792, in SPE annual technical
 404

conference and exhibition, proceedings, volume omega, Formation

evaluation and reservoir geology: Society of Petroleum Engineers, p.
447-460. Later published in 1989, SPE Formation Evaluation, v. 4(21;
p.153-162.

Hilchie, D.W., 1968, Caliper logging--Theory and practice: The Log Analyst,
v. 9, (1), p. 3-12.

_ _--;-_' 1979, Old (pre-1958) electrical log interpretation: D. W. Hilchie,

Inc., Golden, Co., 161 p.

_ _-;-:-:-;-:-' 1982, revised, Applied openhole log interpretation: D.W.

Hilchie, Inc., Golden; Co., variously paginated.

1987, The geologic well log interpreter notes: D. W. Hilchie, Inc.,

_ _-=:-----,-'
Boulder, Co., variously paginated.

_ _--;--:-' 1990, Wireline: a history of the well logging and perforating

business in the oil fields: D.W. Hilchie, Inc., Boulder, Co., 200 p.

Hill, D.G., 1986, Geophysical well log calibration and quality control, in P.G.
Killeen, editor, Borehole geophysics for mining and geotechnical
applications [international symposium and workshop, (Toronto,
Ontario, August 29-31, 1983), proceedings]: Geological Survey of
Canada Paper 85-27, p. 379-392.

Hodges, R.E., 1988,. Calibration and· standardization of geophysical well-

logging equipment for hydrologic applications: U.S. Geological Survey
Water Resource Investigations Report 88-4058, 25 p.

Hounslow, A. W., 1987, Practical interpretation of water quality data,

course notes: Oklahoma State University, Stillwater, Oklahoma,
variously paginated.

Hurst, A., Lovell, M.A., and Morton, A.C., editors, 1990, Geological
applications of wireline logs: Geological Society of London Special
Publication, No. 48, 357 p.

International Critical Tables, 1928, International critical tables of numerical

data, physics, chemistry and technology, v. III, prepared under the
auspices of the International Research Council and the National
 405
Acadamy of Sciences by the National Research Council of the, United
States of America: E. W. Washburn, editor-in-chief, New York'
Publishers, McGraw-Hili Book Company, Inc., p. 444. '

_ _--;_--,-' 1929, International critical tables of numerical data, physics,

chemistry and technology, v. III, prepared under the auspices of the
International Research Council an~ the National Acadamy of Sciences
by the National Research Council of the United States of America:
E.W. Washburn, editor-in-chief, New York Publishers, McGraw-Hili
Book Company, Inc., p. 481.

Itoh, T., Miyairi, M., and Kimura, K., 1980, The high temperature well
logging system for geothermal well, paper G, in 21 st annual logging
symposium: Society of Professional Well Log Analysts, 21 p.
,

Jacobson, L.A., and Fu, C.C., 1990, Computer simulation of cased-hole

density logging: SPE Formation Evaluation, v. 5 (4), p.465-468.

Jansson, M., 1985, A comparison of detransformed logarithmic regressions

and power function regressions: Geografiska Annaler, v. 67 A(1-2), p.
61-70. '

Johnson, H.M., 1962, A history of well logging: Geophysics, v. 27 (4), p.

507-527.

Jones, P.H., and Buford, T.B., 1951, Electric logging applied to ground
water exploration: Geophysics, v. 16, (1), p. 115-139.

Jones, T.A., 1979, Fitting straight lines when both variables are subject to
error, part I; Maximum likelihood and least-squares estimation:
Mathematical Geology, v. 2 (1), 25 p.

Jorden, J.R., and Campbell, F.L., 1984, Well logging I--Rock properties,
, borehole environment, mud and temperature logging: Society of
Petroleum Engineers, Dallas, Monograph series No.9, 167 p.

_ _---:;-::---,-" 1986, Well logging II--Electric and acoustic logging: Society

of Petroleum Engineers, Dallas, Monograph series No. 10, 192 p.
 406
Jorgensen, D.G., 1988, Using geophysical logs to estimate porosity, water
resistivity, and intrinsic permeability: U.S. Geological Survey, Denver,
Water Supply Paper 2321, 24 p.

Keys, W.S., 1988, Borehole geophysics applied to ground-water

investigations: U.S. Geological Survey open-file report 87-539, 305 p.
Also published in Techniques of Water-Resources Investigations, 1990,
02-E2, 1 50 p.

Keys, W.S., and MacCary, L.M., 1971, Application of borehole geophysics

to water-resources investigations: U.S. Geological Survey, Techniques
of Water-Resources Investigations, book 2, chapter El, 126 p.

Kienitz, C., Flaum, C., Olesen, J-R., and Barber, T., 1986, Accurate logging
in large boreholes, in 27th annual logging symposium: Society of
Professional Well Log Analysts, 21 p.

Kwader, T., 1982, Interpretation of borehole geophysical logs in shallow

carbonate environments and their application to ground-water
resources investigations: Florida State University, unpublished
dissertation, 335 p.

_____, 1984, The use of geophysical logs for determining formation

water quality, in D.M. Neilsen and M. Curl, editors, 1984 NWWA/EPA
conference on surface and borehole geophysical methods in ground
water investigations, proceedings: National Water Well Association,
Worthington, Ohio, p. 833-841. Later published in 1986, Ground
Water, v. 24 (1), p. 11-15.

_--;---' 1985, Resistivity-porosity cross plots for determining in situ

formation water quality--Case examples, in 1985 NWWA conference
on surface and borehole geophysical methods in ground water
investigations [Fort Worth, Texas, February 12-14], proceedings:
National Water Well Association, Worthington, Ohio, p. 415-424.

_____, 1986, The use of geophysical logs for determining formation

water quality: Ground Water, v. 24, (1), p. 11-15.

Labo, J., 1987; A practical introduction to borehole geophysics--An

overview of wireline well logging principles for geophysicists: Society
 407
of Exploration Geophysicists, Tulsa, Geophysical References Series, v.
2,330 p.

Lancaster, J.R., and Atkinson, A., 1987, Computer processing of selected

drilling, lithologic, and geophysical log data produces accurate
permeability values and comprehensive aquifer analysis, including
specific yield: COGS [Computer Oriented Geological Society)
Computer Contributions, v. 3, (1), February, p. 13-36.

Logan, J., 1961, Estimation of electrical conductivity from chemical

analyses of natural waters: Journal of Geophysical Research, v: 66
(8), p. 2479-2483.

Lowe, T.A., and Dunlap, H.F., 1986, Estimation of mud filtrate resistivity in
fresh water drilling muds: The Log Analyst, v. 27 (2), p. 77-84.

MacCary, L.M., 1978, Interpretation of well logs in a carbonate aquifer:

. U.S. Geological Survey Water-Resources Investigations No. 78-88,
30 p. Later published in 1983 as, Geophysical logging in carbonate
aquifers: Ground Water, v. 21 (3), p. 324-342.

_ _ _ _--:' 1980, Use of geophysical logs to estimate· water-quality trends

in carbonate aquifers: U.S. Geological Survey Water-Resources
Investigations, No. 80-57, 23 p.

_ _---,-__' 1984, Relation of formation factor to depth of burial in aquifers

along the Texas Gulf coast, in D.M. Nielsen and M. Curl, editors, 1984
NWWA/EPA conference on surface and borehole geophysics methods
in ground water investigations, proceedings: National Water Well
Association, Worthington, Ohio, p. 722-.742.

Mann, C.J., 1987, Misuses of linear regression in earth sciences, in W.B.

Size, editor, Use and abuse of statistical methods in the earth
sciences: Oxford University Press, New York, p. 74-106.

Mark, D.M., and Church, M., 1977, On the misuse of regression in earth
science: Mathematical Geology, v. 9 (1), p. 63-75.

McBain, J.W., Peaker, C.R., and King, A.M., 1929, Absolute measurements
of the surface conductivity near the boundary of optically polished
 408
glass and solutions of potassium chloride: Journal of American
Chemical Society, v. 51, p. 3294-3312.

McConnell, C.L., 1983, Spontaneous potential corrections for ground-water

salinity calculations; Carter County, Oklahoma, U.S.A.: Journal of
Hydrology, v. 65, p. 363-372.

_~---..,... ' 1985, Time dependence of the equivalent water resistivity in

fresh water wells: The Log Analyst, v. 26 (3), p. 12-17.

_ _--;-_.....,...-:' 1988, A general correction for spontaneous potential well

logs in fresh water: Journal of Hydrology, v. 101 (1-4), p. 1-13.

______, 1989, Characterization of wellhead protection areas using·

spontaneous potential well log analysis, paper II, in 3rd international
symposium on borehole geophysics for minerals, geotechnical, and
groundwater applications, proceedings: Society of Professional Well
Log Analysts, Minerals and Geotechnical Logging Society, Chapter-at-
Large, p. 671-697.

McCoy, R.L., Kumar, R.M., and Pease, R.W., 1980, Identifying fractures
with conventional well logs: World Oil, v. 191 (7), Dec., p. 91-98.

Merkel, R.H., and Snyder, D.O., 1977 Application of calibrated slim hole
logging tools to quantitative formation evaluation, chapter X, in 18th
annual symposium transactions: Society of Profes!;;ional Well Log
Analysts, 21 p.

Miller, R.L., Bradford, W.L., and Peters, N.E., 1988, Specific conductance:
theoretical considerations and application to analytical quality control:
U.S. Geological Survey Water-Supply Paper 2311, p. 1-16.

Moelwyn-Hughes, E.A;, 1961, Physical chemistry, (2nd ed.): Pergamon

Press, New York, 1333 p.

Moore, C.V., and Kaufman, R.L, 1981, Your unsuspected problems--Fluid

resistivity and water analysis, paper C, in 22nd annual symposium
transactions: Society of Professional Well Log Analysts, Houston,
15 p..
 409

_ _ _----:-.,---' 1983, Resistivity techniques can cause unsuspected

problems: World Oil, p. 49-53, 62.

Moore, E.J., 1966, A graphical description of new methods for determining

equivalent NaCI concentration of chemical analysis, paper M, in 7th
annual symposium transactions: Society of Professional Well Log
Analysts, Houston, 34 p.

Moses, P.L., 1961, Geothermal gradients, drilling, and production practice:

American Petroleum Institute.

National Research Council, 1928, International critical tables .of numerical

data, physics, chemistry and technology: McGraw-Hili Book Company,
Inc., New York, v. 3, 444 p.

National Research Council, 1928, International critical tables of numerical

data, physics, chemistry and technology: McGraw-Hili Book Company,
Inc., New York, v. 4, 480 p.

Overton, H.L., and Lipson, L.B., 1958, Correlation of electrical properties of

drilling fluids with solids content: Transactions AIME, v. 213, p. 333-
336.

Peterson, B.R., 1'991, Borehole geophysical logging methods for dissolved

solids; a case study, in Ground water management: no. 5, [Las Vegas,
Nevada, May 13-16). proceedings of the 5th national outdoor action
conference on aquifer restoration, ground water monitoring, and
geophysical methods,.p. 1047-1072.

Pirson, S.J., 1963, Handbook of well log analysis for oil and gas formation
evaluation: Prentice-Hall, Enlglewood Cliffs, N.J., 325 p.

Poteet, D., Collier, H., and Maclay, R., 1992, Investigation of the
fresh/saline water interface in the Edwards Aquifer in New Braunfels
and San Marcos, Texas: Edwards Underground Water District Report
92-02, San Antonio, Texas, 171 p.

Prensky, S.E., 1987, Geological applications of well logs--An introductory

bibliography and survey of the well logging literature through
September, 1986, arranged by subject and first author: The Log
Analyst, parts A and B, v. 28, (1), January-February, p. 71-107, part
 410

C, v. 28, (2), March-April, p. 219-248, (previous editions published as

U.S. Geological Survey Open-File Reports 85-441, 86-170, 87-16).

Raymer, L.L., Hunt E.R., and Gardner J.S., 1980, An improved sonic transit
time-to-porosity transform, paper P, in 21 st annual"symposium
transactions: Society of Professional Well Log Analysts, Houston,
13 p.

Repsold, H., 1989, Well logging in groundwater development: International

Association of Hydrogeologists, International Contributions to
Hydrogeology, v. 9, 136 p.

Rider, M.H., 1986, The geological interpretation of well logs: John Wiley &
Sons, Inc., New York, 175 p.

Rink, M., and Schopper, J.R., 1974, Interface conductivity and its
implications to electric logging, paper J, in 15th annual symposium
transactions: Society of Professional Well Log Analysts, Houston,
15 p.

Sanyal, S.K., and Jusbasche, J.M., 1979, Calculation of geothermal water

salinity from well logs--A statistical approach: Geothermal Resources
Council Transactions, v. 3, September, p. 613-616 .
•
Sarma, V.V.J., and Rao, V.B., 1962, Variation of electrical resistivity of
river sands, calcite, and quartz powders with water content:
Geophysics, v. 27 (4), p. 470-479.

Schlumberger, 1958, Introduction to Schlumberger well logging:

Schlumberger Well Surveying Corp., Houston, 176 p.

_ _--:-:---:-_:,1972, Log interpretation: Volume I - principles: Schlumberger

Limited, New York, New York, 112 p.

__--=-__' 1979, Log interpretation charts: Schlumberger Well

Surveying Corp., Ridgefield, Conn., 97 p.

__--=-...,...,_"
1987, Log interpretation principles/applications:
Schlumberger Educational Services, Houston, 198 p.
 411

_ _--:::-_-;--' 1988, Log interpretation charts: Schlumberger Educational

Services, Houston, Document No. SMP-7006, 150p,.·

__-=---;-:-_' 1989, Log interpretation principles/applications:

Schlumberger Educational Services, Houston, Document No. SMP-
7017, 223 p.

Sciacca, J., 1989, Operational and quality assurance considerations for

borehole geophysical logging in hydrogeologic investigations, in 3rd
national outdoor action conference on aquifer restoration, ground
water monitoring and geophysical methods [Orlando, Florida),
proceedings: National Water Well Association, Dublin, Ohio, p. 891-
907.

Scott, J.H., 1978, A FORTRAN algorithm for correcting normal resistivity

logs for borehole diameter and mud resistivity: U.S. Geological
Survey, Open-File Report 78-779, 12 p.

Segesman, F., and Tixier, M.P., 1959, Some effects of invasion on the SP
curve: Transactions AIME 216. Also published in 1959 in Journal
of Petroleum Technology, p. 138-146.

Serra, 0., 1984, Fundamentals of well-log interpretation, volume 1--The

acquisition of logging data: Elsevier, New York, Developments in
. Petroleum Science, No. 15A, 423 p.

_ _--:::---;-;_' 1985, Sedimentary environments from wire line logs:

Schlumberger Technical Services, Paris, Document No. M-
081030/SMP-7008, 211 p.

Sherman, H., and Locke, S., 1975, Depth of investigation of neutron and
density sondes for 35 percent-porosity sand, paper Q, in 16th annual
symposium transactions: Society of Professional Well Log Analysts,
14 p.

Shuter, E., and Teasdale, W.E., 1989, Application of drilling, coring, and.
sampling techniques to test holes and wells, chapter Fl in Collection
of environmental data: U.S. Geological Survey Techniques of Water-
Resources Investigations, Book 2, 97 p.
 412

Silva, P., and Bassiouni, Z., 1981, A new approach to the determination of
formation water resistivity from the SP log, paper G, in 22nd annual
symposium transactions: Society of Professional Well Log Analysts,
14 p.

Snyder, D.D., and Fleming D.B., 1985, Well logging -- a 25-year

perspective: Geophysics, v. 50 (12), p. 2504-2529.

Society of Professional Well Log Analysts, Houston Chapter, 1979, The art
of ancient log analysis: Society of Professional Well Log Analysts,
Houston, 131 p. plus reprints of 22 classic papers.

Society of Professional Well Log Analysts, 1982, Geothermal log

interpretation handbook: Society of Professional Well Log Analysts,
Houston, Reprint Volume, variously paginated.

Summers, W.K., 1972, Factors affecting the validity of chemical analyses

of natural water: Groundwater, v. 10 (2), p. 12-17.

Taylor, K.C., Hess, J. W., and Mazzela, A., 1989, Field evaluation of a slim-
hole borehole induction tool: Ground Water Monitoring Review, v. 9
(1) p. 100-104.

Taylor, K., Molz, F., and Hayworth, J., 1988, A single well electrical tracer
test for the determination of hydraulic conductivity and porosity as a
function of depth, in 2nd national outdoor action conference on aquifer
restoration, ground water monitoring and geophysical methods [Las
Vegas, May 23-26], proceedings, volume 2: . Association of Ground
Water Scientists and Engineers (AAAS), p. 925-938.

Taylor, T.A., and Dey, J.A., 1985, Bibliography of borehole geophysics as

applied to ground-water hydrology: U.S. Geological Survey Circular
No. 926, 62 p.

Tittman, J., and Wahl, J.S., 1965, The physical foundations of formation
density logging (gamma-gamma): Geophysics, vol. 30 (2) p. 284-294.

Towle, G., 1962, An analysis of the formation resistivity factor-porosity

relationship of some assumed pore geometries, paper 3, in 3rd annual
symposium transactions: Society of Professional Well Log Analysts.
 413

Troutman, B.M., and Williams, G.P., 1987, Fitting straight lines .in the earth
sciences, in W.B. Size, editor, Use and abuse of statistical methods in
the earth sciences: Oxford University Press, New York, p. 107-128.

Truman, R.B., Alger, R.P., Connell, J.G. and Smith, R.L., 1972, Progress
report on interpretation of the dual-spacing neutron log (CNL) in the
United States, paper U, in 13th annual symposium transactions:
Society of Professional Well Log Analysts, 34 p.

Turcan, A.N. Jr., 1962, Estimating water quality from electrical logs, in
Geological Survey Research 1962: U.S. Geological Survey
Professional Paper 450-C, p. C135-C136.

_ _ _....,-_' 1966, Calculation of water quality from electrical logs--Theory

and practice: Louisiana Geological Survey, and Louisiana Department
of Public Works Resources Pamphlet 19, 23 p.

Urban, F., White, H.L., and Strassner, E.A., 1935, Contribution to the
theory of surface conductivity at solid-liquid interfaces: The Journal qf
Physical Chemistry, v. 39 (3), p. 311-330.

Veneruso, A.F., and Coquat, J.A., 1979, Technology development for high
temperature logging tools, paper KK, in 20th annual symposium
transactions: Society of Professional Well Log Analysts, 13 p.

Vonhof, J.A., 1966, Water quality determination from spontaneous-

potential electric log curves: Journal of Hydrology, v. 4, p. 34; -347.

Wahl, J.S., Tittman, J., Johnstone, C.W. and Alger, R.P., 1964, The dual
spacing formation density log: Journal of Petroleum Technology, p.
1411-1416.

Weiss, J.S., 1987, Determining dissolved solids concentrations in

mineralized ground water of the gulf coast aquifer systems using
electric logs, in Aquifers of the Atlantic and Gulf Coastal Plain,
American Water Resources Association, Monograph No.9, J. Vecchioli
and A.1. Johnson, editors, p. 139-150.

Welex, 1985, Log interpretation charts: Welex, Houston, variously

paginated.
 414

Williams, G.P., and Troutman, B.M., 1987, Algebraic manipulation of

equations of best-fit straight lines, in W.B. Size, editor, Use and abuse
of statistical methods in the earth sciences: Oxford University Press,
NewYork,p.129-141. !

Winsauer, W.O., Shearin, H.M. Jr., Masson, P.H., and Williams, M, 1952,
Resistivity of brine-saturated sands in relation to pore geometry:
Bulletin of the American Association of Petroleum Geologists, v. 36
(2), p. 253-277.

Winsauer, W.O., and McCardell, W.M., 1953, Ionic double-layer

conductivity in reservoir rock: Transactions AIME, v. 198, p. 129-
134.

Worthington, A.E., Hedges, J.H., and Pallatt, N., 1990, SCA guidelines for
sample preparation and porosity measurement of electrical resistivity
samples, part I, Guidelines for preparation of brine and determination
of brine resistivity for use in electrical resistivity measurements: The
Log Analyst, v. 31, (1) p. 20-28.

Wyllie, M.R.J., 1949, A quantitative analysis of the electrochemical

component of the SP curve: Transactions AIME, v. 186, p. 17-26.

Wyllie, M.R.J., Gregory A.R. and Gardener, L.W., 1956, Elastic wave
velocities in heterogeneous and porous media: Geophysics, v. 21, (1) .
p.41-70.

Yearsley, E.N., Crowder, R.E., and Irons, L.A., 1991, Monitoring well
completion evaluation with borehole geophysical density logging:
Ground Water Monitoring Review, v. 11 (1), p. 103-111.

I
GUIDELINES FOR VERIFYING THE ACCURACY OF WATER ANALYSES

Appendix I

Examination of hundreds of water analyses from several laboratories in

Texas revealed significant differences in analytical techniques and in the
accuracy of TDS and specific conductance measurements (Chapters 2 and
3). Therefore, it was deemed necessary to include a section on guidelines
for determining the accuracy of water analyses. Hem (1985) is the source
of most of the guidelines. HounsloiN (1987) is also a useful reference.

Virtually all of the analyses examined during this study were from six
laboratories: the Texas Department of Health (TDH). Pope Testing, Edna
Wood (formerly Microbiology Service Laboratories). Curtis (out of business).
Texas Testing (out of business). and United States Geological Survey
(USGS). Other laboratories exist in Texas, but these six laboratories have
analyzed the vast majority of ground-water samples'taken in the state. The
petroleum industry has an entirely different group of laboratories.
, ,

The procedures listed below apply to water analyses from all of the
laboratories named above and are for establishing the accuracy of TDS and
specific conductance values. If a water analysis is run for another purpose
(e.g. to measure the concentrations of minor ions). these guidelines, while of
some help, will not be sufficient to verify the accuracy of the analysis.

The following guidelines apply to complete, routine analyses of normal

waters. A complete, routine water analysis measures the following normal
major constituents: silica (Si0 2 ). calcium (Ca). magnesium (Mg). sodium
(Na). chloride (CII, bicarbonate (HC0 3 ), sulfate (S04). and carbonate (C0 3 ).
Normal waters are those which do not have excessive concentrations of
organics, nitrate, sulfate, or suspended matter. TDS for such water analyses
should be within ±5 percent of the actual TDS and conductivity within ±2 to
±5 percent of the actual value (Hem, 1985, pp. 163 and 69).

Al
 A2

METHODS FOR ASSESSING THE ACCURACY OF TOTAL DISSOLVED

SOLIDS MEASUREMENTS

Three methods exist for determining the accuracy of TDS

measurements: anion-cation balance, comparison of. TDSca'culated versus
residue on evaporation, and the TDS-Cw relationship. Either of the first two
methods is an excellent check of TDS. All water analyses should consist of
the constituents mentioned above to allow one of the two checks to be
done. The TDS-Cw relationship should be used only as a last resort, unless
the relationship has been firmly established from other water analyses in the
vicinity of the sample in question.

Anion-Cation Balance

An anion-cation balance is an efficient method of checking TDS

measurements. The Texas Department of Health, the United States
Geological Survey, and some petroleum laboratories routinely run this
calculation. A laboratory may run the calculation, but not display it on its '
water analysis report (e.g. United States Geological Survey). PC programs
are available to perform the calculation (contact Arthur Hounslow at
Oklahoma State University). The procedure is as follows:

1: The concentration of each ion in the analysis is converted from

mg/l, ppm or kg/m 3 to meq/l.

meqII = mgII x-
r
v.a,=:le.::.:n.:.:c.:.,e. :. :n""um:.:=be:c.. (1-1 )
formula weight

This equation can be simplified to a series ,of conversion factors

(see Table 1-1):

meqll = mgll x F (1-2)

Where:
F is a conversion factor.
'" a pa n'ICU la'
£Oor r IOn F'IS valence number
h'
formula weig t
 A3

2. The anions are summed and the cations are summed. Silica, which
is electrically neutral, is excluded.

3. The two sums are compared. The closer the two values are, the
more accurate the TDS calculation is. In theory, since all waters

TABLE 1-1. MULTIPLIERS FOR CONVERTINGMG/L TO MEQ/L

FOR THE MOST COMMON IONS

Ion Symbol V slenee Number F

Bicarbonate HC03 -1 0.01639
Calcium Ca +2 0.04990
Carbonate CO, -2 0.03333
Chloride CI -1 0.02821
Fluoride F -1 0.05264
Iron F. +2 0.03581
Iron F. +3 0.05372
Magnesium Mg +2 0.08229
Manganese Mn +2 0.03640
Nitrate NO, -1 0.01613
Potassium K +1 0.02558
Sodium Na +1 0.04350
Sulfate SO, . -2 0.02082
Modified from Hem (1985).

are electrochemically neutral, the sums should be equal. In

practice, however, the sums will not be exactly equal. Usually the
cation sum is less than the anion sum. This is because most
waters have far more cation than anion species and a number of
the minor cation species are not analyzed in a routine water
analysis. The difference is usually expres'sed as a percentage:

Anion-Cation balance ~J sum of cations - sum of anions l x 100 (1-3)

sum of cations + sum of anions

Having expressed the anion-cation balance as a percentage, the value

should be interpreted according to the following guidelines:

1. If the sodium concentration is reported as either "Na by

difference", "Na (diff.)", or "Na and K (diff.)", an anion-cation
balance should not be performed. It is worthless as'a quality
control check. The ions will balance perfectly because of the
procedure used to determine sodium.
 A4

Prior to about 1960 sodium and potassium were difficult to analyze

accurately (Hem, 1985, p. 164). Consequently, all the common
ions except sodium and potassium were measured, converted to
meq/I, and then the difference between the anions and cations
(calcium and magnesium) was attributed to the presence of sodium
and potassium. This value was converted to mgll and placed on
water analyses as Na (diff.). Potassium concentrations are
normally so small in relation to sodium that potassium was lumped
under sodium. This means that analytical errors and/or .
undetermined ions are automatically included in the sodium value.
The anion-cation balance is always perfect, yet worthless as a
quality control check.

Two laboratories in Texas, Edna Wood and Pope Testing, still

determine sodium by difference. If requested, they will measure
sodium and potassium for a fee of $12 to $15 per ionic species.
For internal quality control they measure the two ions when the
results of a water analysis do not look right (Edna Wood
Laboratories and Pope Testing Laboratories, personal·
communication, 1990). Measured sodium and potassium values
are more accurate than calculated values but, for the purpose of
this study, calculated values are fine as long as another quality
control check of TDS is available.

2. If sodium is not listed as Na (diff.) and the ionic concentrations are

in perfect agreement, the anion-cation balance still should not be
used to judge the validity of the TDS value. A perfect balance
more than likely indicates that sodium was determined by
difference, whether or not it is identified as such on the water
analysis. (Pope Testing is an example of this practice.) As
discussed in previous paragraphs,· determining sodium by difference
invalidates an anion-cation balance.

3. If sodium was measured rather than calculated and the anion-

cation balance is less than 5 percent, the analysis is probably
accurate and the TDS value is reasonably correct (Hounslow,
1987).

4. If the balance is much greater than 5 percent to 10 percent, the

TDS value is incorrect. The analysis may have one of several
problems:
 AS
a. The analysis is complete but inaccurate due to analytical
errors.
b. The analysis is incomplete.

i. Other ions (s\Jch as heavy metals) are present but were not
analyzed. .
ii. The water is very acidic (pH is less than 4) and the
hydrogen (H +) ions were not analyzed.
iii. Organic ions are present in significant quantities (often .
indicated by colored water), but were not analyzed
(Hounslow, 1987). '

TDSCalcu,otod vs. Residue on Evaporation

Two methods are used to determine TDS: TDSco,cu'atod' also known as

TDScomputod' and residue on evaporation, commonly labeled dissolved residue
at a specified temperature. Comparing the two answers provides a good
check of TDS.

For many years TDS was determined by evaporating a known amount

of water and then weighing the residue. Normally this gives an accurate
TDS. Errors can be introduced if the drying time or temperature is incorrect.
Water will be retained by the residue or solutes other than bicarbonate
(HC0 3 ') will be volatized. One unavoidable result of the procedure is that
during evaporation HC0 3' is converted to C0 3 ~, CO 2 , and H2 0 with 50.8
percent of the HC0 3 ' driven off as CO 2 and H2 0 vapor while 49.2 percent
remains as C0 3 ~.

Laboratories today report TDSco,cu,atod' The value is obtained by

summing the measured dissolved constituents. The sodium value may be
measured or calculated, depending on the laboratory (see the previous
section). The bicarbonate value is either 100 percent (Edna Woo'd, Pope
Testing, and Curtis) or 49.2 percent (Texas Department of Health, United
States Geological Survey, and Texas Testing) of the measured value (see the
Measurement Techniques section in Chapter 3).

Edna Wood is the only currently operating laboratory to still report

dissolved residue. (Many Curtis water analyses reported it.) This is done as
a quality control check on TDSca'cu'atad' TDSca,cu'atod is considered accurate if it
is within ±3 percent of TDSactua' (Alice Perry, personal communication,
 A6

1990). TDSactua' is determined by summing the dissolved residue and the

amount of bicarbonate ions volatized during the drying process:

TDS act1Ml1 = Dissolved Residue + (0.508 x bicarbonate) (1-4)

Either an anion-cation balance or a TDSca,cu,ated-residue on evaporation

check should be performed on every water sample. Both methods provide
stand-alone verification of the accuracy of a water analysis. Of the existing
laboratories surveyed in this study, Pope Testing is the only one that did not
use one of these two methods. Therefore, they cannot use anion-cation
balance, because sodium is determined by difference. They can measure
residue by evaporation, but it must be requested by the customer and an
additional fee of about $5 is charged.

TDS-Specific Conductance Relationship

The third method for checking a water analysis is to examine the

relationship between specific conductance and TDS in mgt!. This is the least
reliable of the three methods. The relationship is too variable and
conductivity values are sometimes not measured accurately, Hem (1985, p.
168) provides several guidelines for using this relationship as an approximate
accuracy check:

1. For normal waters up to a few thousand mgtl TDS:

TDS = 0.55 to 0.75 . (1-5)

Specific Conductance

a. These values are for TDS's calculated by using 49.2 percent of

the bicarbonate ions.
b. Silica has been excluded from the TDS value.
c. Bicarbonate and chloride waters will be at the low end of the
value.
d. Waters high in both divalent cations ICa + + and Mg + +) and
divalent anions (S04"") may reach a value as high as J .0. The
ions form uncharged pairs and thus do not contribute to
conductivity. This allows TDS to increase at a faster rate than
specific conductance (Hem, personal communication, 1990).
These waters are often saturated with gypsum (Ca 2 S0 4).
 A7

e. Waters containing large concentrations of silica may reach a

value as high as 1.0.
f. Saline waters with more than 100,000 ppm TDS will have
much higher values than nonsaline waters.
g. Acidic (low pH) waters may have much lower values than
neutral waters.
h. This ratio is so variable that it is only good for determining gross
errors in TDS. Confidence can be placed in the value only when
it has been firmly established by utilizing water analyses in the
vicinity of the sample in question.

2. The TDS values used in equation 1-5 must be consistent in the

treatment of the HC0 3 - value, either 100 percent or 49.2 percent
of the measured bicarbonate value (see Chapter 3).

3. The relationship becomes more poorly defined for waters exceeding

50,000 mg/I TDS and for very fresh waters (such as rainwater). if
the. nature of the principal solutes is unknown.

Assorted Other Checks

If concentrations are converted to meq/I, a number of other, checks can

be used to determine whether the reported values of certain ions are
reasona~le (Hounslow, 1987). As long as one remembers that many
accurate analyses have exceptions, the following general checks can be
used:

1. Sodium in meq/I will be much greater than potassium in meq/I,

,unless both are less than about 5 mg/1.

2. Calcium in meq/I will be greater or equal to magnesium in meq/I

unless the calcium has been removed by precipitation.

3. Calcium in meq/I will equal sulfate in meq/I unless calcium has been
. removed by precipitation. Note: in some oilf.ield waters the sulfate
concentration may be decreased by 'sulfate reduction and can be
unrelated to calcium concentration (Hem, personal communication,
1990).

4. Sodium in meq/I will equal chloride in meq/I unless sodium has

been added by ion exchange.
 AS
5. If carbonate is absent, pH should be less than 8.

METHODS FOR ASSESSING SPECIFIC CONDUCTANCE ACCURACY

Specific conductance measurements should always be scrutinized

closely. Some laboratories do not measure conductivity accurately.
Unfortunately, verifying the accuracy of specific conductance measurements
is not as straightforward as it is for TDS.

Three methods are available to determine the accuracy of specific

conductance measurements: TDS-Cw relationship, ionic concentrations, and
anion concentration in meq/1. All the available methods utilize TDS.
Because TDS values are usually more !lccurate than specific conductance
measurements, it is best to use the relationships to calculate specific
. conductance from TDS and not vice versa.

General Guidelines

1. The measurement must be recorded in pmhos/cm at 25° C (77°

F). Field conductivity measurements and oilfield water resistivities
(Rw) may require conversion to this temperature and/or this unit of
measurement. Note: many conductivity meters have a built-in
conversion factor to convert the conductivity value from the
measured temperature to 25° C. Temperature-conductivity
conversion factors vary according to the chemical composition
(Hem, 1982). Little research has been done on how much this
relationship varies, so this is a possible source of error. . (Refer to
the discussion of temperature under Factors Controlling Water
Conductivity in Chapter 2.)

a
2. In order to assess the accuracy of specific conductance measure-
ment, determine whether a laboratory uses standard methods to
routinely calibrate its conductivity meter and if the meter is
calibrated for the conductivity range of a particular water sample.
Such is not the case with Pope Testing and Texas Department of
Health (Pope Testing Laboratories and Texas Department of Health,
personal communication, 1990).

3. Diluted conductance is not valid as a substitute for specific

conductance. The Texas Department of Health uses dilluted
conductance, and most of the conductivities in the Texas Water
 A9

Development Board Ground-Water Data Base are diluted.

conductances. See item 4. in the section Preparing the Data in
Chapter 4 for further details on the types of conductivities in the
Ground-Water Data Base.

TDS-Specific Conductance Relationship

Normally only gross errors in specific conductance can be detected

with the TDS-specific conductance relationship (see the previous TDS-
Specific Conductance Relationship section). For a particular geographic area
or water bearing unit where a number of water analyses exist, a precise
relationship can be established between TDS and specific conductance. In
such cases the relationship is a valid check' of the accuracy of a specific
conductance value. All the TDS values must be calculated the same way,
using either 100 percent or 49.2 percent of the measured bicarbonate value.

Specific Conductance from Ionic Concentrations

It is possible to calculate water conductivity (Cw) from a chemical

analysis. The ionic concentration of a water is converted to the equivalent
NaCI concentration (Figure AI- 1 l. which is then converted to Cw (Figure AI-
2). Calculated Cw is accurate to within ±5 percent of measured Cw (Desai
and Moore, 1969).

Figure AI- 1. lists the ionic charge of common ionic species relative to
Na and CI: The chart shows the relative contribution of each .ion to water
conductivity and how the contribution varies with TDS. A NaCI type water
was chosen as the standard because most oilfield waters are NaCI type
waters. The chart, compiled by Schlumberger, is constructed from data in
Desai and Moore (1969) for TDS greater than 1000 ppm and from the
Variable Dunlap Method for TDS less than 1000 ppm. Some curve
smoothing was done (Schlumberger, 1972).

Figure AI-2 is a modified, quick and simple version of a Resistivity-

TDS-Temperature chart found,in logging chart books. It converts equivalent
NaCI concentration directly to Cw in jJmhos/cm at 25° C. The graph is
constructed from data that Arps (1953) obtained from the International
Critical Tables (1928, 1929). The equation for the graph, which was
modified from the Dresser Atlas log interpretation chart book (1985), is as
follows:
 AID

I-:.~-+':--_I-~'+--H-.HH..----/-'---+--·j- 'j-f'
'-t-' t+"I::--:';--~:~'~.-
o
.r- .
-.--- ·_·c
I.O-No'a 1(.
-1- .

- 1 - - -. ----1-- ----... ---- _\ I-

..... - -- - - .-. - I-
_.~._.

--- _. -- --- . --I--.f-. - c --- - -l--H+I-H-"" r-

-LOIO 20 50 100 200 500 Ik 2k 5k 10k 20k ·5Ok lOOk :lOOk

Totol Solids Concentration, ppm .or mg/kg

*MultipUor. which do not 'lor)" approciably for low cone-lIt'Clllo,,1 (I... thon
about 10,000 PPIII) '1'1 .hown I;Illh.l.ft ",argin of ,h. chari.

Figure AI-'. Chart for converting the ionic concentrations of a water sample to the equivalent
NaCI concentration (Modified from Gen-a Chart, Schlumberger, 1979).

Rw@25°C=0.012+ 3_5_52_.7 _ 11-6)

, (equivalent NaCrl 955

The procedure for using Figures AI.-' and AI-2 to calculate Cw from
ionic concentrations is:

,. The appropriate TDS value is entered into Figure AI-'. The TDS
value should include , 00 percent of the bicarbonate value and
exclude the silica value. The TDS value can be in ppm or mgtl.
 All

Equivalent NoCI Concentration (ppm)

~
u>
~ 1,000
ro
n
'"
ii-
n
0
~
~
c 2,000
n
0'
~
n
ro 3,000
'3 4,000
3
~
0
5,000
-.....
~

n
3 6,000
@ 7,000
N 8,000
~ 9,000
- 10,000

30,000

40,000
50,000
60,000

80,000
90,000
70,000
100,000 ~illliililliilillllillillii 11111111111111;

Figure AI-2, Graph for converting NaCI concentration to specific conductance, The data used to
construct the graph is from Aarps (19531. The graph agrees with Chart Gen-9 (Schlumberger, 19881,
 A12

2. The appropriate multiplier for each ionic species in the water

analysis is established by moving vertically to the intersection of
the TDS value and the particular ion. For most water only Ca. Mg.
Na, Cl i HC0 3 , C0 3 , and S04 occur in concentrations which are
significant enough to be included in the calculation.

3. the concentration of each ionic species is multiplied by the

appropriate multiplier. For the bicarbonate concentration' 00
percent of the value is used. '

4. The products calculated in step 3 are summed to yield the

equivalent NaCI concentration.

5. The equivalent NaCI concentration is entered into Figure AI-2 and

the corresponding Cw value is read. This is Cw pmhos/cm at 25°
C for the water sample in question.

Specific Conductance from meq/I

, The total of eitner anions or cations in meq/I multiplied by' 00 usually

approximates specific conductance. While not exact, this relationship is less'
variable than a general TDS-Cw relationship in mgt!. It is not as accurate as
specific conductance calculated from ionic concentrations.

Two guidelines must be used with this method:

,. The anion-cation balance should be within acceptable limits (less

than 5 to '0 percent).

2. If an acceptable difference exists, the anion value is used. Usually

, it will be more accurate than the cation value. This is because
most waters have far more cation than anion species and a number
of the minor cation species are not analyzed in a routine water
analysis.

SUMMARY

-,. Laboratories in Texa's, have significant differences in analytical

,- techniques and in the accuracy of TDS and specific conductance
measurements.
 AI3

2. A water analysis should not be assumed to be correct until a

quality control check is performed.

3. The TDS values should be verified by either an anion-cation balance

(within less than 5 to 10 percent), or a TDSco,cu,oted-residue on
evaporation comparison (iNithin less than 3 percent).

4. When comparing different TDS values, the method used to

calculate TDS must be consistent, using either 100 percent or 49.2
percent of the measured bicarbonate value.

5. The TDS-Cw relationship should be used to verify TDS or specific

conductance only when the relationship has been firmly established
from other water analyses in the vicinity of the sample in question.

6. Specific conductances to within ± 5 percent of the actual value can

be calculated from ionic concentrations.

7. Diluted conductance is an acceptable measurement only of fresh

waters.

8. A laboratory should use standard methods to routinely calibrate its

conductivity meters.
 GUIDELINES FOR SELECTING AND UTILIZING LINE-FITTING ROUTINES

Appendix II

x-v plots are the basis of many log analysis techniques. Plots such as
TDS-Cw, Ro-TDS, RocPorosity, and Ro-Field Formation Factor are used to
determine water quality. X-V plots also are used to calibrate logging
equipment and to calculate various rock properties.

X-V plots are constructed in order to analyze the relationship between

two variables. Scatter usually exists in the data, which makes the
relationship less than perfect and difficult to quantify. In such cases, curve-
fitting routines are utilized to establish the relationship that best fits the
distribution of the data.

This section provides guidelines for selecting the proper straight-line

fitting routine and correctly utilizing it. The discussion is limited to
procedures for fitting straight lines to either untransformed data or data that
have been transformed to their natural logarithms. CurVilinear-fitting
routines are not discussed because most wireline log data sets fit a straight
line.

Straight-line fitting routines warrant a review in this study because:

1. They are critical to many log analysis techniques.

2. There are many fitting routines from which to choose and the most
appropriate one is frequently not used.

, 3. They are: often misinterpreted.

4. Such a discussion provides the fundamentals for understanding the

rationale behind the treatment of TDS-Cw graphs in this study.

Two excellent articles on straight-line fitting routines are Mann (1987)

and Troutman and Williams (1987). The following discussion is a distillation
of these two articles and verbal communications with Lee Etnyre, ARCO Oil
and Gas; John Davis, University of Kansas; and Joe Booth, Abilene Christian
University. .

A14
 A15

Step 1. Choose Between a Linear and a Curvilinear Fit.

The first step is to decide if a straight-line fit is appropriate for the data
set:

1. The data should be plotted on a linear (arithmetic) scale. If the

data appear to have a linear trend, a line-fitting routine can be
applied.

2. If the trend is curvilinear, the data should be plotted on a

logarithmic scale. If the trend is now linear, a straight-line fitting
routine can be used.

3. If the trend is curvilinear on a logarithmic scale, a nonlinear fitting

routine has to be used.

4. The data may have so much scatter that it is difficult to tell

whether or not a linear relationship exists. In such cases one
should apply the line-fitting routines, use a technique to analyze"
the quality of the lines, and then decide. Residual analysis is the
best technique to determine linearity, but several other methods
also can be used (e.g. Mann, 1987, pp. 94-100).

Step 2. Fit the Best Line to the Data.

The second step is to fit a straight line to the data set. No single, all-
purpose, correct fitting technique exists. Each method usually gives a
different line. The best line to use is dictated by the quality of the data, the
amount of scatter in the data, and the intended use of the line. Personal
biases also enter into the selection process. Straight-line fitting is not an
exact science. Researchers sometimes differ as to which technique is best
and what is the proper way to handle a data set.

This section reviews eight straight-line fitting routines common in .

scientific studies: eyeballing, averages, ordinary least squares, inverse least
squares, weighted least squares, robust methods, least normal squares, and
reduced major axis. Recommendations on when to use each method are
also included. This section is, for the most part, a summary of Troutman
and Williams (1987).
,
 A16

Quick-look Methods

The first two methods, eyeballing and averages; are simple, somewhat
subjective, quick-look methods.

Eyeballing. Simply draw a straight line through what looks to be the

best fit and calculate the equation of the line. Note: if the graph has
logarithmic scales, the logs of X and V must be used to calculate the
equation. '

While the method is quick and easy, it is subjective and there is no

way to quantify the precision of the fit. At times, however, eyeballing may
'be all that is needed.

Averages. This method is also called semiaverages or groups averaged

by residue summation. First, divide the data into two or more groups,
u~ually according,to increasing values of X. Calculate the average X
_ _ J . __

(identified as X) and average V IV) for each group. Plot each IX, V) on the
graph, connect the points, and calculate the equation.

This technique is more precise than eyeballing, yet it is still relatively

uncomplicated and easy to use. Unfortunately, the line may vary according
to how the data are grouped. The equation cannot be reversed and the
method is not as statistically sound as the following six mathematical
methods.

Mathematical Methods

The following six procedures employ rigorous mathematical methods

to fit the line. Each method varies as to how to minimize the deviation of
points from the line (Figure AII-1). Consequently, each method usually
draws a different line.

Ordinary least squares IV on X). This method is the most commonly

used straight-line fitting method. It is the method normally used for
prediction. Ordinary least squares draws the line that has the minimum sum
of the squared deviations (distances) of V from the fitted line. It minimizes
line AS in Figure AII-1 A. This means; as shown in Figure AII-2, that the
fitted line minimizes high values and maximizes low values.
 A17

(A) Y 011 X A
(B) XonY

c
• Y Y

y=a+bx x = a + bY
x x

(C) Least Normal Squares (D) Reduced Major Axis

A
c

x x

Figure AII-1 (A-D). Graphs showing the differences in line-fitting. criteria for the four most
common straight-line fitting routines (Modified from Jones, 19791.

An advantage of ordinary least

squares is that the equation does not X Y on
• ~.

vary if one or both of the scales are ~,

"'V'E~AGES
Y X Dn

changed (invariant to scale changes).

One disadvantage is that unless the
scatter of the data points about the line
is very low (Le. a very high correlation •
coefficient) the equation can only be
used to predict Y; it cannot be reversed. N eo 010 'N a:I HXI
.....TEI\AI,. DISTANCE 101. ~ I;ILOMETtI'lS
In other words, the only valid form of
the equation is Y = a + bX. It cannot Figur. AII-2. Hypothetical data set showing the line
fits determined by five straight-line fitting methods
be rearranged to X = - alb + Y lb. (Modified from Troutman and Williams. 19871.
Another disadvantage is that outliers,
points having a large deviation from the line, have a disproportionate
influence on the location of the fitted line.

Ordinary least squares gives the best fit when errors in the data points
are normally distributed with equal variances (Etnyre, 1984a). When this
 Al8

condition is satisfied, the line has the minimum variance from the data points
in the Y direction and the method gives the best possible prediction of Y. If
the errors are not normally distributed, the method still can be used but the
line may not be any better than lines obtained with other line-fitting routines.
When the errors are not normally distributed, robust weighted least squares
gives a better fit (Etnyre, 1984a). When the errors are normally distributed
with unequal variances, weighted least squares provides the best fit when
the weights are assigned as the reciprocal of the error variance for each data
point (Etnyre, personal communication, 1990). Troutman and Williams
(1987, p. 123) and Mann (1987, pp. 76-79) give details on other
assumptions that underlie ordinary least squares.

Inverse least squares (X on V). This method is a variation of ordinary

least squares and is usually discussed under ordinary least squares. Inverse
least squares draws the line that has the minimum sum of the squared
deviations of X from the line. This is line AC in Figure AII-1 B. This means,
as shown in Figure AII-2, that the fitted line maximizes high values and
minimizes low values. The resulting equation is X = a + bY. The equation
cannot be rearranged. Inverse least squares has the same advantages and
limitations as ordinary least squares.

Weighted least squares (WLS). This method applies a weighting factor

to the squared deviation of each data point before fitting the line. It can be
applied to both ordinary and inverse least squares. The weighting factors
are used to decrease the effect of less reliable points on the line fit, thereby
equalizing the errors in the data. It is also used to remove the weighting
effect of a logarithmic transformation. The key to the method is assigning
the proper weighting factors. Aside from being more complicated, weighted
least squares has about the same advantages and disadvantages as ordinary
and inverse least squares.

Robust methods. Robust methods are similar"to weighted least

squares. These methods are used to minimize the effect of outliers (points
having large deviations from the line). which can adversely affect ordinary
and inverse least squares lines. Robust methods should be used only after it
has been determined that the outliers are erroneous.

These methods are more complicated than most other straight-line

fitting routines. Please refer to Troutman and Williams (1987. pp. 116, 124,
and 125) for further details.
 A19

Least normal squares (LNS). This method, also known as major axis,
principal axis, and Pearson's 1901, draws the line that minimizes the sum of
the squared perpendicular deviations of the data points from the line (line AD
in Figure AII-1 C). Line AD is therefore the shortest distance that can be
obtained between a straight line and each point in the data set (see Figure
AII-2).

The equation is reversible. Scale changes, however, require the slope

to be recalculated. Even though this line gives the closest fit to the data, it
is the best line to use only if the magnitudes of the errors in both variables
are nearly the same. '

Reduced major axis IRMA). This method minimizes the sum of the
areas of right triangles formed between the data points and the fitted line.
This is area ABC in Figure AII-1 D. RMA assumes that on a percentage basis
the measurement accuracy between the two variables is nearly equal.

The equation is reversible and scale changes do not alter it.. However,
unless the correlation coefficient is very high, the equation is not statistically
as good as a least squares equation for predictive purposes.

Selecting the Proper Line-Fitting Routine'

Remember that the selection of a line-fitting routine is dictated by the

intended use of the line, the quality of the data, the amount of scatter in the
data, and personal preference. Selection of the fitting routine also involves a
decision as to which axis each variable is to be plotted on.

Linear regression. One objective of fitting a straight line to a data set

is to provide an equation for predicting values of one variable, given values
of the other variable (linear regression). Use the following guidelines to select
the best line-fitting routine for linear regression:

1. If a data set has perfect correlation, it makes no difference which

straight-line fitting routine is used and which variable is Y. Each of
the eight methods will draw the same line and calculate the same
rearrangeable equation.

2. If a data set has very high correlation, it makes little difference for
the main body of the data set which fitting routine is used and
which variable is Y. In the tails of the data set and beyond,
 A20

however, small variations from perfect correlation can mean

considerable' differences in the position of the various lines. .If the
.. tails are the important part of the data set, use guideline 3. below
to choose the proper line.

3. If scatter exists in the data, each routine will draw a different line.
Ordinary least squares and inverse least squares are the extremes
of the fits, with all other lines falling between these two (Figure
AII-2). All lines agree at the centroid (X, Y) and disagreement
increases away from this point. As the scatter in the da~a
decreases, the lines converge on the RMA line. When one variable
has no error and the other variable has all the error (as is the case
in some laboratory experiments), or when one variable has most of
the error, the best equations for predictive purposes are ordinary
least squares, inverse least squares, weighted .Ieast squares, and
robust methods. The variable with the least amount of error is
designated the independent variable. The variable with the
greatest amount of error is the dependent variable and is usually
assigned to the Y-axis.

a. Ordinary least squares (Yon X) is used to predict values of Y,

given v.alues of X, when Y is the dependent variable.. The
equation solves for the mean value of Y given a particular value
of X. The same answer will be obtained by assigning the
variable with the most error to X and using inverse least
squares (X on Y) to solve for X. Y on X, however, is the
conventional arrangement.
b. When the variable to be predicted (Y) has less error than X,
there are two ways to regress the data. Some workers use (X
on V). X is treated as the dependent variable, although in truth
it is not. Most statisticians, however, advocate calibration:
regress Y on X and then invert the equation to solve for X. For
values near the centroid, both methods give similar answers,
while calibration gives better answers at and beyond the tails
(Davis and Doveton, 1990). This is the only time that Y on X
should be inverted. .
c. In order to minimize the effect of less reliable data points on
either ordinary or inverse least squares, weighted least squares
should be used. It can also be used to remove the weighting
effect of a logarithmic transformation of the data.
d. For data sets with spurious outliers, one can either use robust
methods or discard any known or suspected outliers from the
 A21 '

data set. However, one should avoid rejecting any data

without investigating the reasons for the outliers (Etnyre,
personal communication, 1990).

4. Least normal squares and reduced major axis are the best routines
for linear regression when the magnitudes of the errors in the two
variables are nearly the same. Since both methods have reversible
equations, it makes no difference which variable is Y.

Structural and functional analysis. Another reason for fitting a straight

line to a data set is to estimate the true underlying linear relationship
between two variables (structural and functional analysis). 1 Least normal
squares and reduced major axis are the best methods for determining the
true underlying linear relationship between the two variables. Working on
the assumption that errors exist in X and Y, these methods use statistics to
estimate error-free counterparts of X and Y and establish the relationship
between these unobservable variables.

Step 3. Calculate the Equation of the Line.·

In addition to plotting a straight line, each of the aforementioned fitting

routines calculates the equation of the line. The equation for a straight line
is as follows:
Y = a + bX (11-' )

Where:
X and Yare the measured values of the two variables under
consideration.
tJ is the Y-axis intercept of the line when X =' o.
b is the slope of the line - the number of units that Y changes for
each one unit change in X.

A line-fitting routine applied to logarithmically transformed data uses

the logarithms of the data points, not the arithmetic values. The equation of
the line must be written accordingly:

See Troutman and Williams (1987. p. 118) for an explanation of the differences between structural and
functional analysis.
 A22

log Y = loga + blogX (11-2)

Where:
a is now the Y-axis .intercept·of the line when X = 1. Some
authors denote loga as simply a.
By a they mean that the term is a constant - the log of a.
This equation can be rewritten in a simpler form by taking the antilogs of Y,
X, and a. The new equation is in the form of a power law as follows:
(11-3)

Where:
a is now a proportionality constant. It is the log of a in (II-1).
b is an exponent in the nonlinear relationship.

Step 4. Assess the Degree to Which the Line Fits the Data.

A standard part of every line-fitting routine should be to

mathematically, rather than just visually, determine how well the line fits the
data set. This can be done by calculating the correlation coefficient and
examining the residuals.

Correlation coefficient. The correlation coefficient (r) is a measure of

the linear association between X and Y or how well X predicts Y.
Coefficient correlations will be -1.0 :s r :s 0 Or 0 :s r :s + 1.0. When r =
0, there is no linear association between X and Y. An r of -1.0 or + 1.0
means perfect correlation. A direct relationship between X and Y or an
equation with a positive slope yields a positive r. Negative values occur
when an equation has a negative slope or the two variables have an inverse
relationship.

Correlation coefficients should be interpr~ted according to the .

following guidelines:

1. When various line-fitting routines are applied to a data set, the line
with the highest correlation coefficient is not necessarily the best.
fit (see Step 3. Cah;ulating the Equation of the Line).
 A23

2. When comparing different pairs of variables from the same data

set, remember that coefficients have relative, not absolute, value.
An r of 0.9 for variables X and Y is not twice as good as an r of
0.45 for Y and Z.

3. Technically, correlation coefficients from different data sets should

not be compared.

4. A high correlation coefficient implies nothing about the statistical

or geological significance of the linear association. All it signifies is
that the two variables have a high degree of linear association
(Mann, 1987).

A more useful comparison may be made using the coefficient of

determination (r 2 ). It gives the fraction of data spread which is accounted
for by the fitted line (Etnyre, 1984a).

Residual analysis. Residual analysis is another excellent method of

checkinOg the quality of a fitted line. A residual (d) is the difference between
an observed value and the predicted value. The equation is as follows:

d=Y-Y (11-4)

Where:
d is the residual.
Y is the observed value.
Y is the predicted value as determined by the equation of
the fitted line.

There are several types of residuals and various ways to display them
(Mann, 1987, p. 99). One technique is to plot raw residuals against X, Y, or
'Y and examine the magnitude and pattern of the residuals (Figure AII-3).
The closer the line fits the data, the closer the residuals cluster randomly on
a narrow band along d = O. The nature of the distribution of the residuals
about d = 0 also helps to identify outliers, reveal trends, and verify the two
,most. often violated linear regression assumptions: linearity and constant
vanance of errors (Mann, 1987, p. 79).
 A24

..
• ..,,
'3

;~'!::oo::-,:::"'=00:-'::-:""'00""'' ' ' ' 00''-;'-;;;;'' ' 00-;''"' 00"'' -C";;;;OO;;;',7.""'00"-'-

x

• •

•
y ,
..
•
y
:" .
.
'.':-,------:.,:----:-,-.,--
• •
Figure AII-3. Example of three ways to plot raw residuals: against predictor
variable X (top). regressor variable Y (lower left). and predicted values Y (lower right!
(Mann. 1987). . . '

Step 5. Properly Use the Data.

Extrapolating linear relationships beyond the range of the data set is

very tenuous. When there is no alternative, one should at least be aware of
two potential errors:

1. The trend and/or the relationship between the two variables may
change in the region beyond the data base. For example, the
relationship between TDS and Cw is linear for fresh water and
. curvilinear for saline waters.

2. Confidence in predictions decreases away from the center of the

data and drops off even faster beyond the data set. Confidence
intervals on a graph readily show this (Figure AII-4). This explains
one reason why predicting Rw from Ro-Porosity plots is often
unsuccessful (Etnyre, 1984a). Confidence intervals should be kept
in mind when predicting the TDS of waters with either very high or
very low conductivities (those at the extreme ends of the data
set).
 A25

_ 100
'"
E
g" 30% Confidence Interval
""
:;
ro ,v
~
.~ 10
o
ll-
05 '-------~~-----_;_;~~
.1 10.0
Resistivity (Ohm-Meters)

Figure AII-4. Confidence interval about the regression line (Etnyre,

1984a).

Step 6. Properly Handle Logarithmic Transformations.

Many data sets that have a curvilinear trend on an arithmetic scale can
be converted to a linear trend by mathematically transforming the data. A
transformation common in log analysis is to convert the values of the
variables to their natural logarithms. This is the transformation discussed in
this section.

Logarithmic transformations are usually performed by simply plotting

the data on log-log paper, which is the method used in this study. An
alternate technique is to calculate the logarithms for each data point and plot
the logarithmic values on an arithmetic scale (see Fogg and Blanchard, 1986,
for examples). However, this method is time consuming and cumbersome.
Computer programs sometimes make the transformation, calculate the line,
convert the data and the equation to their antilogs (arithmetic values), and
then graph them without ever displaying a logarithmic plot. This procedure
has a disadvantage in that no one ever visually inspects the logarithmic plot.

The advantage of logarithmically transforming data to a straight line is

that all of the above mentioned line-fitting routines and quality checks are
then applicable. After the appropriate line-fitting routine has been applied,
then the data can be plotted on arithmetic or logarithmic scales.

As discussed in Step 3., the equation of the line can be written either
of two ways:
 A26

logY=loga+blogX (II-51

or
(11-61

Equation 11-6, called a power law, is a little faster to solve and is more
commonly used.

Logarithmic transformation involves an often overlooked pitfall -- the

addition of an unavoidable weighting factor to the data. A weighting factor
is introduced because the line-fitting routine deals with the logarithms of the
data points, not the arithmetic values. Because of the logarithmic
transformation, the relative weight or influence of small values on the fitted
line is greatly increased and the relative weight given to large values is
decreased.

To demonstrate why a logarithmic transformation introduces a

weighting factor, consider two numbers: 100 and 10,000, and their
logarithms, 2 and 4. The relative weight of 4 to 2 is considerably less than
the relative weight of 10,000 to 100 and vice versa.' Consequently, during
the line-fitting of logarithmic data, the line will be skewed toward the small
logarithms. When a line is fitted to the arithmetic values, the large numbers
will have more influence and the line will move in their direction. Thus, the
best-fit line for a logarithmically-derived straight line, when recalculated into
the untransformed equivalent (antilog) and plotted on an arithmetic scale,
will plot lower than the line calculated from the arithmetic scale (Figure 4-3).
Jansson (19851 has a good discussion of this subject. ,

Adding this weighting factor to the data mayor may not be

advantageous. When it is not advantageous, as is often the case, the
remedy is to use weighted least squares. Etnyre (1984a,bl has an excellent
treatment of how to apply weighted least squares to logarithmically
transformed data.

Etnyre 11984a, p. 14) mathematically derives the weighing factor for ordinary least squares that a
logarithmic transformation creates.
 A27

Sometimes only one variable is converted to its natural log (i.e., a core
porosity versus core permeablility graph). In this case the following
procedures apply:

1. Regressing Y on the natural log of X does not affect the deviations

of Y from the fitted line (Figure All-51. The equation of the line is
as follows:

Y=u+blnX (11-7)

10

6
y
4

0.1 1
x x
Figure AII-5. Regressing Y on the natural log of X (Davis and Doveton, 1990).

2. Regressing the natural log of Y on X will mean that the deviations

of InY from the line will be symmetrical, but the deviations of Y
will not be (Figure AII-5). The equation of the line is as follows:

lnY = a + bX (11-8)
 A28

10

6
Y
4

Figure AII-5. Regressing 1he natural log of X on X (Davis and Doveton, 1990).

3. All of the guidelines mentioned in this chapter for straight-line

fitting routines also apply to semi-Iogarith.mic plots.
 TECHNIQUES TO EVALUATE THE QUALITY OF DRILLING MUD

Appendix III

Chapter 8 demonstrated that the chemical composition of the drilling

fluid has an important bearing on the quality of the log responses. Drilling
fluids also will effect drilling efficiency. Even more. important, well
productivity is directly related to the quality of the drilling fluid. Because
drilling fluids are so critical to all aspects of the assessment and development
of a water well, the addition of this appendix was deemed necessary. This
material is from Baroid Drilling Fluids, Inc.'s 1991 seminar on Drilling Fluid
Technology As Applied To Minerals Exploration, Water Well, and
Geotechnical Drilling. Bob Pace of Baroid Drilling Fluids provided additional
comments.

Drilling fluids perform a number of important functions that effect the

. drilling operation. They:

1. Cool and lubricate the bit and drill string.

2. Clean the cuttings from the bit and the bottom of the hole.

3. Suspend solids.

4. Transport cuttings and sloughings to the surface ..

5. Minimize fluid loss into permeable formations.

6. Stabilize the well bore.

7. Prevent the flow of formation fluids into the well bore.

8. Minimize wear and corrosion of the drilling equipment.

9. Assist in gathering subsurface geological data and formation

evaluation.

Good quality drilling fluid more efficiently performs the above

functions. The performance of drilling fluid depends on four physical
properties: mud weight (density), viscosity, filtration (filter cake and filtrate

A29
 A30

volume), and sand content. It also depends upon the chemical properties of
pH and ion concentrations. Ion concentrations determine which additives
can be used to design specific properties into the drilling fluid.

Many people tend to forget that the primary purpose of a drilling fluid
is to convey drilled solids from the hole to make room for the production
casing and down hole pumps. A 10 inch hole drilled to 200 feet moves
approximately nine tons of drilled solids. The drilling fluid must deal with
these solids chemically and physically. If the solids are not physically
removed from the fluid then they must be treated chemically to minimize
their effects on the fluid.

A drilling fluid may be as simple as clean, fresh water, or it may be

designed to densities of greater than 18 pounds per gallon in order to
tolerate temperatures greater than 500 degrees Fahrenheit. Most non-oilfield
applications are not so complex, except for geothermal applications. The
most common designs are simple fresh water, extended bentonite
dispersions. Even so, it is a dynamic fluid in a constant state of change as
active ingredients, mainly commercial bentonite, are constantly being diluted
or replaced with inactive ingredients (solids from the formations). The only
way to maintain a good mud quality is to routinely measure the mud
properties and adjust the mud accordingly.

This chapter defines the most important physical and chemical

properties of drilling mud and discusses their effect on drilling and logging
operations. Recommendations on how to maintain the proper balance of
each property are included.

PHYSICAL PROPERTIES

Mud Weight (Density)

Mud weight, or density, is the most important mud property. Density.

is measured in pounds per gallon (Ib/gal) by weighing a known volume of
mud with a mud balance. Density must be measured. Visual estimates will
not be accurate:

For simple water-base muds, it is a measure of the percentage of

suspended solids (Table 9-1). It also indicates the hydrostatic pressure of
the mud column as follows:
 A31

HydrostDtic pressure (pst) = 0.052 (mud weight, lb/gal) x depth (ft) (111-11

TABLE 111-1. RELATIONSHIP BETWEEN

SOLIDS CONTENT AND MUD WEIGHT

Percent Solids· Mud Weight

Ibs/gcl

0 8.33
1 8.47
2 8.60
3 8.74
4 8.88
5 9.02
6 9.15
7 9.29
8 9.43
9 9.57
10 9.70
11 9.84
12 9.98
13 10.12
14 10.25 .

15 10.39
16 10.53
17 10.67
18 10.80
19 10.94
20 11.08

(Baroid course notes, 1991). • Assumed Solids Specific Gravity = 2.65 gm/cm'

Mud weight is very important to both the drilling and logging process
because excessive solids create a number of problems:

1. A pressure differential between the mud column and the

formation water which is too high increases impregnation and
can either reduce formation permeability or cause loss of
circulation by fracturing the formation.
 A32

2. Excessively thick mudcake, which impairs water production,

affects· pad-type logging tools, and can stick pipe.

3. Excessive wear on the mud pump, drill string! and bit.

4. The settling of cuttings at the surface is retarded, so the mud

pump does unnecessary work recirculating solids and the drilling
rate is reduced.

For most water well drilling and coring, density should be kept below
about 9 lb/gal. (Water is 8.33 Ib/gal.llf the solids control equipment or
settling pits are inadequate to control solids density, it may rapidly increase
to 9.5 Ib/gal or greater. In the case of artesian water pressure, mud weight
will need to be increased (preferable with barite) to control the overpressured·
lone.
However, a low mud weight does not guarantee a good quality mud.
The type, as well as the quantity, of solids in the mud determines mud
· quality. A 9.0 lb/gal natural mud will be composed of clays and other solids
from the formations. It will not be nearly as effective in fUlfilling the
previously mentioned drilling mud functions as will a 9.0 Ib/gal mud
· containing commercial bentonite and other additives.

With a poorly designed drilling fluid handling system, the density will
increase as the hole is deepened. This means that at the most critical time,
when the bit reaches the target aquifer at the bottom of the hole, the drilling
fluid. is at its highest density and poorest quality.

Several techniques can be used to control density. The· most

important ones are: have a good mud pit design, use a shale shaker with fine
screens, use desander cones, keep the pump suction off the bottom of the
· mud pit, and add water to the mud.

Filtration (Filter cake and Filtrate)

A filtration test measures the filter cake thickness to the nearest '/32 of
an inch and the mud filtrate in cubic centimeters (cm 3 ) produced by a filter
press in 30 minutes atl 00 psi under static conditions. The results are used
to estimate the quality of the mud cake forming in the borehole and the
amount of filtrate invading the formations.
 A33

A good mudcake is essentially impermeable (low filtration rate). thin,

slick, pliable, and easily removed during well development.. The quality of
the mudqake is a function of the type and the concentration of solids in the
mud. A good mudcake requires commercial bentonite rather than native clay
and a low percentage of natural solids. Otherwise, problems may develop:'

1. A high water-loss mud with excessive mud filtrate invasion will

affect the resistivity response and increase the chance of
differentially sticking of the drill pipe.

2. An abnormally thick mudcake that will affect pad-type logging

tools, reduce well productivity, and create tight spots.'

3. Less stable borehole walls.

For water well drilling and coring the filter cake should be less than 3/32
of an inch thick, slick, and soft. The filtrate loss during the test should be
15 cm 3 or less. Mud cake quality and filtrate loss are controlled by keeping
the ratio of commercial bentonite to natural solids high.

Viscosity

Viscosity is the resistance offered by a fluid to flow. The term is

loosely applied to drilling mud as a measurement of carrying capacity and gel
strength. It is measured with a Marsh funnel in secqnds/quart (which is
commonly abbreviated to seconds). As a rule of thumb,

viscosity ... 4(mud weight) (9-2)

Although the Marsh funnel is not a true viscosity measurement, its use is
considered very useful for simple drilling fluids.

The viscosity and velocity of the drilling fluid determine how

effectively cuttings are removed from the face of the bit and are carried to
the surface. Low viscosity muds are best at these tasks if adequate
pumping is available. However, higher viscosity muds may be needed to
remove coarse sand from the borehole and to stabilize some unconsolidated
sand and gravel formations "that cave into the well bore.

A natural mud with a viscosity of 45 seconds and a density of 10.3 Ib/gal will form a volume of filter cake
36 times greater than a bentonite additive mud with a weight of 8.6 Ib/gal (Baroid course notes. 1991 I.
 A34

When drilling fluid circulation ceases, the mud thickens or "gels·. The
force needed to break the gel when circl:llation resumes is" the gel strength.
For normal water-base muds viscosity is a good indicator of gel strength.
Gel strength from commercial bentonite suspends cuttings in the borehole
when circulation stops, thus preventing the drill pipe from getting stuck.

High viscosity and gel strength create the following problems:

1. Drilling rate is reduced because cuttings are not removed

efficiently from the bit face and from the mud when it reaches
the surface.

2. High pump pressure is required to break the gel, which can

cause impregnation and loss of circulation.

3 .. Mud weight increases. , "

Viscosity should be as thin as practical and still clean cuttings from the
hole. The recommended range is 32 to 38 sec/qt. (Water is 26 sec/qt.)
However, for hole stability in very unconsolidated formations, viscosity may
need to be much higher.

Viscosity is increased by adding more bentonite or a polymer. It is

decreased by adding water or, in some cases, a phosphate deflocculant.

Sand Content

Sand content is defined as percent by volume of mud solids retained

on a 200-mesh sieve. The recommended limit is less than 2 percent by
volume.

Excessive sand content creates a number of problems such as:

1. ", Thickens the mud cake.

2. Increases the mud weight.

3. Abrades the pump, drill string, and bit.

4. Slows the drilling rate.

 A35
Sand content is controlled by utilizing a shale shaker, desanders, and a
good mud pit design that maximizes settling time .. Adding water to lower
the viscosity will also decrease the sand content. .. .
:'

CHEMICAL PROPERTIES

pH

The term pH is defined as -log of the hydrogen.ion concentration (H').

In its simplest terms, 1 to 7 is, acidic
, and 7 to 14 is alkaline. In fresh water
,

it is easily measured with pHydrion paper or ColorpHast sticks. Most drilling

fluids should have a pH of 8.3 to 9.5 (or moderately alkaline) because:

1. The effectiveness of bentonite is greatly reduced in acidic

conditions (pH less . than 7) and . viscosity, gel .strength, and
. filtratiol'l are adversely affected ..

2. Drilling equipment exposed to acidic conditions may corrode

rapidly. . ,

3. A pH greater than 9.5 can accelerate clay dispersion and may

cause thick drilling fluid and borehole deterioration, which can
hamper both drilling and logging.

Drilling fluids should not be overtreated with alkaline materials (caustic

soda, bicarbonate of soda, or soda ash). Cement contamination is the main
cause of excessively high pH. The pH is lowered by adding sodium
bicarbonate. When using fresh water add caustic soda to raise the pH.
Soda ash should be added if the available water is considered hard (soluble
calcium).

Alkalinity

Most non-oilfield drillers do not test their drilling fluids for any chemical
property except pH. However, many drilling, problems could be eliminated if
they tested the mud alkalinity.
. .
3
Pm" Pm is defined asthe volume, in cm , of standard (N/50) sUlfuric'
acid to reduce the pH of 1 cm~ of whole mud ·from actual pHto pH 8.3 using
phenolphthalein as the pH indicator. It quantifies the amount of soluble and
 A36

insoluble alkaline material in the mud (lime or cement, caustic soda,

bicarbonate of soda, or soda ash.)

PI - MI' PI is defined as the volume, in cm 3 , of standard (N/50)

sulfuric acid to reduce the pH of 1 cm 3 of mud filtrate from actual pH to pH
8.3 using phenolphthalein as the pH indicator.

M f is defined as the volume, in cm 3 , of standard (N/50) sulfuric acid to

reduce the pH of 1 cm 3 of mud filtrate from the actual pH to pH 4.3 using
methyl orange as the pH indicator. The relationship between PI and M I
quantifies the amount of hydroxyl, bicarbonate or carbonate ions in the
filtrate.

Total Hardness

There is a standard test for total hardness (calcium and magnesium

ions.) The amount of total hardness is important because small amounts of
either or both elements will effect the yield of commercial bentonite. The
amount will dictate whether caustic soda, bicarbonate of soda, or soda ash
should be used.

Chloride

There is a standard test for determining the amount of chloride. A

knowledge of the amount of chloride is important for two reasons. First, the
yield of commercial bentonite is affected by chloride in the mud, and
second, the quality of the electric log responses is affected.

Field testing procedures for drilling fluids (API RP 138) are available
from the AMERICAN PETROLEUM INSTITUTE, Production Department, 211
North Ervay, Suite 1700, Dallas TX 75201.

MAKE-UP WATER

Make-up water should be tested and treated before bentonite is added.

Most city drinking water is adequate. Water that is drinkable is adequate for
drilling 'fluid. In some areas of the United States, the water in the local
streams and creeks will have a total hardness that needs to be treated with
soda ash to reduce it to an acceptable level before the commercial bentonite
will yield properly.
 A37

Testing Program

All mud property measurements and any changes made to the mud
system should be documented and included in the permanent well file. For
water wells the testing normally will be done by the drilling crew. For
petroleum tests, a representative of the mud company (mud engineer)
periodically does a detailed analysis, in addition to those performed by the
drilling crew. The mud sample should be taken from the flowline before the
mud has traveled through any surface eOquipment (shale shaker, pits, etc.).

The more important mud properties, for characterizing the borehole

environment for log analysis, are mud type, density, filtration, chloride ion,
and any major changes to the mud system. The standard log heading has
spaces for type fluid in hole, density, viscosity, pH, and fluid loss (same as
filtrate loss). The logging company does not measure these properties, but
only records what is provided to them. The information is usually present on
petroleum logs. Water well logs usually contain the type fluid in the hole.
Density and viscosity are sometimes noted but pH and fluid loss are seldom
recorded. Log headings do not contain information regarding changes to the
° mud system. The mud properties on the heading should be those at the time
of logging.
 A38

ABBREVIATIONS AND SYMBOLS

Appendix IV

a- Variable used in several equations in Chapter 4; meaning

varies according to the equation
a - Tortuosity factor
A, - Guard electrode
A- Electrode
AC - Alternating current
aCa - Activity of the calcium ions
AI- Aluminum
AM - Electrode spacing of normal tools
AM 0 0 - Antiquated term for AM
ami - Activity of the mud filtrate
aMg - Activity of the magnesium ions
aNa - Activity of the sodium ions
AO - Electrode spacing of'lateral tools
APHA - American Public Health Association
API - American Petroleum Institute
aw - Activity of the formation water

b- Variable used in several equations in Chapter 4: meaning

varies according to the equation
.Bop - Compadion correction factor (same as Cpl
BGT - Borehole Geometry Tool
B.H.V. - Borehole volume
BPB - British Plaster Board
Br - Bromide
'oBRe- Balcones Research Center

c- Conductivity
DC· Degrees Celsius
Ca· Calcium
Cali - Caliper
Ca 2 S0 4 • Calcium SulFate
CDL· Compensated Density Log
CEC· Cation exchange capacity
CGR· Computed gamma ray curve minus the uranium count
 A39

CILD - Deep induction conductivity curve

CI - Chloride
CNT-G - Epithermal-Thermal Compensated Neutron Log
CO a - Carbonate
Corr - Corrected·
Cp - Compaction correction factor (same as Bop)
CPS Counts per second
CSNG - Compensated Spectral Natural Gamma Ray
Cw - Specific conductance or specific conductivity
Cw Anion Sum .. Specific conductance calculated from the· anion sum
Cw Calculated .. Calculated specific conductance
CW10n Cone... Specific conductance calculated from the ion concentration·
Cw Moasured .. Measured specific conductance

DCAL- Differential caliper

DEC - Decimal
At - Travel time
A~. Fluid travel time
Atm•. Matrix travel time
At.h - Shale travel time
D.F. - Drill Floor
dh - Borehole Diameter (also d h )
diff - Difference·
DIL - Dual Induction Log
DIL-SFL - Dual Induction Log - Spherically Focused Log
DIS-EA - One version of Schlumberger's Dual Induction Tool
DLL - Dual Laterolog
DLS/DE ~ One version of Schlumberger's Dual Laterolog
DN - Density neutron
DNL - Dual Porosity Compensated Neutron Log
DNPOROS - Density-neutron crossplot
DPHI - Density porosity
DPT - Deep Propagation Tool
DRHO - Bulk density correction (L~p)

e- Bed thickness (in older literature)

E- Emitter
E• - Electrochemical potential
Ek - Electrokinetic potential
El- Electric log
Elj - Liquid-junction potential
 A40

E log - Electric log

Em - Shale membrane potential
emp - Equivalents per million
ENPH - Epithermal porosity
EPHI - Electromagnetic propagation porosity
EPT· Electromagnetic Propagation Tool
ES - Electrical Survey

F- . A conversion factor
F- Flouride
F- Formation Factor
o F_ Degrees Fahrenheit
Fc - Formation factor of the common porosity (depth) value
FOC - .Compensated Formation Density
FFF - Field formation factor
FLUIORES - Fluid resistivity
FoRxo Log - . Flushed zone microelectrode tool
FR - First reading
FR - Fluid resistivity
FR - Formation resistivity factor

g- Gram
G· Geometric factor
GAM - Gamma
g/I - Grams per liter
G.L. - Ground level
GLT - Geochemical Logging Tool
GR- Gamma ray
GRC· Gamma ray corrected
GR cl - Gamma ray response in a clean zone
Gr-ion/I - .
GRSh - Gamma ray response in 100 percent shale
grain/gal - Grains per gallon

Bed thickness
Hydrogen ion
Bicarbonate

I- Survey current
IOPH - Phasor Deep Induction
IEL - Induction Electric Log
 A41

IES - Induction Electric Survey

IFG - Instruments for Geophysics
lOR - Gamma ray shale index
I.H.V. - Integrated hole volume
ILO - Deep Induction Log
ILM- Medium Induction Log
IMPH - Phasor Medium Induction
ISF - Induction Spherically Focused Log
in • Inches, also (n)

J - Pseudogeometric factor

K" Constant dependent on the electrode configuration

K- Constant that is a function of temperature
K- Potassium
K.B. - Kelly bushing
KCI _. Potassium chloride
KGS - Kansas Geological Society
kg 1m - Kilograms per meter
Km • Constant that is a function of the mud weight
I- length
Lat. - Lateral
Ib/gal - Pounds per gallon
LOT - LithoDensity Tool
LL - Laterolog
LL3 - Laterolog 3
LL8 - Laterolog 8
LLO - Deep laterolog
LLS - Shallow laterolog
LN· Long normal
LNS - Least Normal Squares
LONG NOR· Long normal

m - Cementation exponent
M" M 2 , M", M/ - Monitoring electrodes
md - millidarcy
MEAS- Measured
meq/l - Milliequivalents per liter
MeV - Million electron volts
Mg - Magnesium
mg/kg - Milligrams per kilogram
 A42

mgtl - Milligrams per liter

mm- Millimeters
mmhostm· Millimhos per meter
pmhostem - Micromhos per centimeter
pseetfoot - Microseconds per foot
ps - microsiemens
MINV - Microinverse
ML- Microlog
MLL - Microlaterolog
MLS - Mineral Logging Systems
Mn- Manganese
MNOR- . Micronormal
MRIL - Magnetic Resonance Imaging Log
mStm - Millisiemens per meter
MSFL- MicroSpherically Focused Log
mv- millivolt.

N- Electrode
NA - Not Available
Na - Sodium
NaCI- Sodium Chloride
NaCIEqUiV. - Sodium Chloride equivalent
Na (diff.) - Sodium by difference
NaHC0 3 - SodiulT' bicarbonate
NAT - Natural
NGS· Natural Gamma Ray Spectometry Log (Schlumberger)
NGWA- National Ground Water Association
N0 3 - Nitrate
NPHI - Neutron porosity

0- Oxygen
0.0'.00' - Electrodes
n- Ohm
ohm-em - Ohm-centimeter
ohm-m - Ohm-meter

P- Permeable bed
PC - Personal computer
Pe - Photoelectric factor
PEF· Photoelectric factor
(/J- porosity (phi)
 A43

PHIE - Effective porosity

PI - Phasor Induction
PI - Petroleum Information Corporation
pi -
PL - "Proximity Log
ppm - Parts per million
psi - Pounds per square inch
PSp· Pseudostatic SP
pu - Porosity units
PUB - Public
PVC - Polyvinylchloride

R- Resistivity
R- Receiver
Ra - Apparent resistivity
redox- Oxidation-reduction
Ri - Resistivity of the invaded zone
RID - Deep induction resistivity
IRldcor - Corrected induction resistivity
RllO - Deep induction resistivity
RIM - Medium induction resistivity
Rm - Mud resistivity
Rmc- Mudcake resistivity
Rmf- Mud filtrate resistivity
Rmfe - Equivalent mud filtrate resistivity
RMA - Reduced Major Axis
Ro - Resistivity of the uninvaded formation when it is 100%
saturated with water
Ro c - Corrected resistivity of the uninvaded formation when it is
100% saturated with water
ROH - High Ro value
ROl - Low Ro value
Rs - Resistivity of shoulder beds
RSFl - Resistivity of the spherically focused log
Rt - Resitivity of the uninvaded formation
Rw - Water resistivity
Rwa- Apparent water resistivity
Rwe- Equivalent water resistivity
Rxo - Resistivity of the flushed zone

S - Siemens
 A44

SBR - Shoulder bed restivity

SOL - Spectral Density Log
SFL - Spherically Focused Log
SFLA - Averaged Spherically Focused Log
SFLU - Unaveraged Spherically Focused Log
SGR - Spectral Gamma Ray
SGR - Total gamma ray count, including uranium
SH - Shale
SH NORM - Short Normal
SI - International System
SiO z _ Silica
SMIN - Synthetic microinverse curve
SMNO - Synthetic micronormal curve
SN - Short normal
S04- Sulfate
SP - Spontaneous potential or self potential
SPR - Single-point resistance
SPWLA - Society of Professional Well Log Analysts
SRG - Natural Gamma Ray Spectral Log (Gearheart)
SSP - Static SP

TO - Total depth
TOH - Texas Department of Health
TOS - Total dissolved solids
TOSactual - Actual total dissolved solids
TOScolculated - Calculated total dissolved solids
TOS-Cw. - Total dissolved solids - specific conductance
TOSMEEQU - TDS calculated using the equation of the line fitting the
TDS-Cw graph and CW MEAS
Temp - Temperature
TENS - . Tension
Th - . Thorium
TNPH - Thermal neutron porosity .
TNRIS - Texas Natural Resources Information System
TWC- Texas Water Commission
TWOB - Texas Water Development Board

U- Uranium
USGS - United States Geological Survey

var. - variation
 A45

VCAl- Volume of calcite

VCl - Volume of clay
Vmea - Voltage measured
VSh - Volume of shale
VQTZ - Volume of quartz

W- Water
WC & 10- Water Control & Improvement District
WIOCO - Well Investment Development Co.
WlS - Weighted Least Squares
WSC - Water Supply Corporation

x- Variables in formulas

y- Variables in formulas

Z- Atomic number
ZOl - . Compensated Z-Densilog

2Z - Halliburton designation, electrode spacing for normal

3iZ - Halliburton designation, electrode spacing for lateral
6FF28 - Schlumberger slim hole induction tool
6FF40 - Schlumberger deep induction curve
8FF34 - Schlumberger medium induction curve
 LOGGING BOOKS

Appendix V

Modern Logging Books

Aguilera, R.: Naturally Fractured Reservoirs, Penn Well (1980), 703 pp.
(800) 627-3212

Allaud, L. A. and Martin, M. H.: Schlumberger, The History of a Technique,

John Wiley & Sons, Inc., New York City (1977), 333 pp.

Asquith, G. B.: Log Analysis by Microcomputer, Penn Well (1980), 104 pp.

Asquith, G. B. and Gibson, C.R.: Basic Log Analysis for Geologists, Methods
in Exploration Series, AAPG, Tulsa (1982), 215 pp., AAPG, Box 979,
Tulsa, Oklahoma 74101, (918) 584-2555.

Asquith, G. B.: Handbook of Log Evaluation Techniques for Carbonate

Reservoirs, Methods in Exploration Series No.5, AAPG, Tulsa (1985),
47 pp.

Atlas Wireline Services: Fundamentals of Diplog Analysis, Gulf Publishing

Company, Houston (1988), 216 pp., Book Division, Dept MA, Box
2608, Houston, Texas 77252-2608, (713) 520-4444.

Bateman, R. M.: Cased-Hole Log Analysis and Reservoir Performance

Monitoring, IHRDC Press, Boston (1984), 380 pp., IHRDC, 137
Newbury Street, Boston, Massachusetts 02116, (617) 536-0202.

Bateman, R. M.: Log Quality Control, IHRDC Press, Boston (1984), 398 pp.

Bateman, R. M.: Open-Hole Log Analysis and Formation Evaluation, IHRDC

Press, Boston (1985), 647 pp., .

Brock, J.: Analyzing Your Logs; Vol. I: Fundamentals of Open Hole Log
Interpretation, Petro-Media, Inc., Tyler (1984), 270 pp., -1729 Rose
Road, Tyler, Texas 75701 (214) 592-8348, (also has video tapes).

A46
 A47

Brock; J.: Analyzing Your Logs, Vol. II: Advanced Open Hole Log
Interpretation, Petro-Media, Inc., Tyler (19841, 186 pp.

Brock, J.: Applied Open-Hole Log Analysis, Contributions in Petroleum

Geology and Engineering, vol. 2, Gulf Publishing Company, Houston
(19861, 292 pp.

Cased-Hole Logging, Oil and Gas Production Series, No.5, The University of
Texas at Austin, Austin (1981).

Crain, E. R.: The Log Analysis Handbook, Vol. I--Quantitative Log Analysis
Methods, PennWell, Tulsa (19861, 684 pp.

Desbrandes, R.: Encyclopedia of Well Logging, Gulf Publishing Company,

Houston (19851, 585 pp.

Dewan, J. T.: Essentials of Modern Open-Hole Log Interpretation, Penn

Well, Tulsa (19831, 360 pp.

Doveton, J. H.: Log Analysis of Subsurface Geology--Concepts and·

Computer Methods, John Wiley & Sons, .Inc., New York City (1986)
273 pp.

Ellis, D. V.: Well Logging for Earth Scientists, Elsevier Science Publishing
Company, Inc., New York City (1987) 532 pp.

Etnyre, Lee M.: Finding Oil and Gas from Well Logs, Van Nostrand Reinhold
(1989), 305 pp. .

Foster, Norman H. and Beaumont, Edward A. (editors): Formation Evaluation

I: Log Evaluation, AAPG, Tulsa, Treatise Reprint Series. 742 pp.

Foster, Norman H. and Beaumont, Edward A. (editors): Formation Evaluation

II: Log Interpretation, AAPG, Tulsa, Treatise Reprint Series. 600 pp.

Galperin, E. I.: Vertical Seismic Profiling and Its Exploration Potential, P.

Kennett, editor, E. Reidel Publishing Company, Dordrecht, Holland
(1985).

Glossary of Terms & Expressions Used in Well Logging, second edition,

Society of Professional Well Log Analysts, Houston (1984), 74 pp.
 A48

Gore, N.: Wireline Operations; Oil and Gas Production Series, J. Paxson,
editor, The University of Texas at Austin, Austin (1984), 80 pp.

Gruner, Schlumberger, A.: The Schlumberger Adventure, Arco Publishing,

Inc., New York City (1982).

Hallenburg, J. K.: Geophysical Logging for Mineral and Engineering

Applications, Penn Well (1984), 264 pp ...

Hearst, J. R. and Nelson, P. H.: Well Logging for Physical Properties,

, . McGraw-Hili Book Company, New York City (1985), 576 pp.

Helander, D. P.: Fundamentals of Formation Evaluation, Oil and Gas

Consultants International (OGCI) Publications, Tulsa (1983),344 pp.,
OGCI, 4554 South Harvard, Tulsa, Oklahoma, 74135, (918) 742-
7057.

Hilchie, D. W.: Advanced Well Log Interpretation, Douglas W. Hilchie Inc,

Golden Co. (1989), variously paginated. 37 Perkins St., Prescott,
Arizona -86301.

Hilchie, D. W.: Applied Openhole Log Interpretation for Geologists and

. Engineers, Doyglas W. Hilchie Inc., Golden, Co. (1978), variously
paginated.

Hilchie, D. W.: The Geologic Well Log Interpreter, Douglas W. Hilchie, Inc.,
. Golden, Co., (1987), variously paginated.

Hilchie, D. W.: Wireline: A History of the Well Logging and Perforating

Business in the Oil Fields, Boulder, Co. (1990), 200 pp.

Hill, A. Daniel: Production Logging--Theoretical and Interpretive Elements:

-Society of Petroleum Engineers, Monograph Volume 14, Dallas (1990),
154 pp., SPE Box 833836, Richardson, Tx. 75083 (800) 527-6863
(Texas) (214) 669-3377 .

.Hurst, A:, Lovell, M. A. and Morton, A. C., editors: Geological Applications

of Wireline Logs, Geological Society of London Special Publication
(1990), No. 48, 357 pp. Order from AAPG.
 A49

Johnson, David E. and Pile, Kathryne E.: Well Logging for the Nontechnical
Person, Penn Well (1988). 200 pp.

Jorden, J. R. and Campbell, F. L.: Well Logging I--Rock Properties, Borehole

Environment, Mud and Temperature Logging, Society of Petroleum
Engineers, Monograph Volume 9, Dallas (1984). 167 pp.

Jorden, J. R. and Cambell, F.L.: Well Logging II--Electric and Acoustic

Logging, Society of Petroleum Engineers, Monograph Volume 10,
Dallas (1986). 182 pp. .

Kerzner, Mark: Image Processing in Well Log Analysis, IHRDC Press, Boston
(1986). 140 pp.

Keys, W. Scott: Borehole Geophysics Applied to Ground-Water

Investigations, USGS Open-File Report 87-539 (1988), 305 pp.
NGWA Bookstore, Box 182039, Dept. 017, Columbus, Ohio 42018.
TWI02-E2

Kobranova, V.N. , English Translation by V.V. Kuznetsov: Petrophysics,

Springer Verlag, New York (1990), 375 pp.

Labo, J.: A Practical Introduction to Borehole Geophysics--An Overview of

Wireline Well Logging Principles for Geophysicists, Society of
Exploration Geophysicists (1987), 330 pp., Box 702740,
Tulsa, OK 74170-2740 (918) 493-3516.

Merkel, R. H.: Well Log Formation Evaluation, Continuing Education Course

Note Series No. 14, AAPG, Tulsa (1983), 82 pp.

Open-Hole Logging, Oil and Gas Production Series No.4, The University of
Texas at Austin, Austin (1984), 87 pp.

Paillet, FrederickL. and Cheng, C.H.: Acoustic Waves in Boreholes, CRC

Press, 000 Corporate Blvd. Boca Raton, Florida, 33431 407/994-
0555 176 pp. $89.95

Pirson, S. J.: Geologic Well Log Analysis, third edition, Gulf Publishing
Company, Houston (1983), 476 pp.
 A50

Repsold, H.: Well Logging in Groundwater Development, International

Association of Hydrogeologist, International Contributions to
Hydrogeology, v. 9, 136 pp. Verlag Heinz Heise GmbH & Co. KG
. (1989) 136 pp. Hannover, West Germany. 80x 610407, D-3000
Hannover 61, West Germany.

Rider, M. H.: The Geological Interpretation of Well Logs, John Wiley &
Sons, New York (1986) 175 pp.

Schlumberger: Log Interpretation Principles/Applications, Schlumberger

Educational Services, Houston (1989), Document No. SMP-7017,
variously paginated.

Schlumberger: Cased Hole Log interpretation principles/Applications,

Schlumberger Educational Services Houston (1990), Document No.
SMP-7025, variously paginated.

Sengel, E. W.: Well Logging Handbook, Institute for Energy Development

(lED), Oklahoma City (1984), 168 pp.

Serra, 0.: Fundamentals of Well-Log Interpretation, Vol. I: The Acquisition

of Logging Data, Elsevier Science Publishing Company Inc.,
Developments in Petroleum Science, No. 15A, New York City (1984),
423 pp.

Serra, 0.: Fundamentals of Well-Log Interpretation, Vol. II: The

Interpretation of Logging Data, Elsevier Science Publishing Company
Inc, Developments in Petroleum Science, No. 158, Amsterdam (1986),
684 pp.

Serra, 0.: Sedimentary Environments From Wireline Logs, Schlumberger

Technical Services, Paris, Document No. M-081030/SMP-7008
(1985), 211 pp.

Theys, Phillips P.: Log Data Acquisition and Quality Control, Gulf Publishing,
Houston (1991), 380 pp.

Tittman, J.: Geophysical Well Logging, excerpted from Methods in

Experimental Physics Vol. 24: Geophysics, Academic Press, Inc., New
York (1986), 175 pp.
 AS!
Ward, Stanley H., editor: Geotechnical and Environmental Geophysics, v. 1,
Review and Tutorial, Society of Exploration Geophysicists.

Ward, Stanley H., editor: Geotechnical and Environmental Geophysics, v. 2,

Environmental and Groundwater, Society of Exploration Geophysicists.

Ward, Stanley H., editor: Geotechnical and Environmental Geophysics, v. 3,

Geotechnical, Society of Exploration Geophysicists.

Well Log Response Chart, Penn Well (1983).

Wireline Logging Tool Catalog, second edition, Gulf Publishing Co. (1984)
450 pp.

Wireline Operations and Procedures, second edition, Vocational Training

Series, American Petroleum Institute, Dallas (1983) No.5.

Wyllie, M. R. J.: The Fundamentals of Well Log Interpretation, third edition,

Academic Press, Inc., New York City (1963).

Old Electric Log Books

The Art of Ancient Log Analysis, Houston SPWLA (1979). 131 pp., plus
reprints of 22 classic papers.
Society of Professional Well Log Analysts
6001 Gulf Freeway, Suite 129
Houston, Texas 77023 (713) 928-8925

. Handbook of Well Log Analysis for Oil and Gas Formation Evaluation, Sylvain
J. Pirson, Prentice-Hall, Inc.: Englewood Cliffs, New Jersey (1963).
Chapters 7-12.

Old (pre-1958) Electrical Log Interpretation, Douglas Hilchie (1979), 163 pp.
Douglas Hilchie, Inc.
37 Perkins St.
Prescott, Az. 86301
 A52

Practical Log Analysis, a series of reprints from the Oil and Gas Journal. pp.
45-56 are on old logs.
Penn Well .
Box 21288 .
Tulsa, Oklahoma 74121 800/627-3212

Prospecting with Old E-Logs, R. W. Frank (1986), 161 pp.

Schlumberger Educational Services
Box 2175
Houston, Texas 77252 (713) 928-4920

)
 Prinud on rtcyckdpapu

EQUALOPPORTUNl1YEMPLOYER
Th~ To:as Wat~r Developmmt Board do~s not discriminate on the basis ofraa, color, national origin, sex.
r~/igion, ag~. or disability in mtploymmt or th~ provision ofsffVic~s, prof?ams, or activiti~.

1-800-RELAY TX ({or the hearing impaired)

Well Nane: TWOS SITE F 88-59-411

__ ..A. JI,O ~O~O.l _. DEPTH --'•..".-.I.. j~ ...0;9___ 10 00

____..1 _ J.g. • FEET BlAIR 15 10,00 ", L______ ..L 1Q.JlP _
,10 SFLU 100,00 __ ~~ ~~ ~~~ ~~ DU~ ~~

2A _JI'FL 200 00

100

150
~
-.
200
-.
250

300

350

400

450

500

550

600

650

700

750

800

850

900

950

1000

1050

PLATE 1. Rw calculated by the Formation Factor method using different a, m, and porosity values, Track 1 contains deep induction (IlD). medium induction (ILMI.
unaveraged spherically focused (SFlUI, and microspherically focused logs (MSFl). The MSFl has been borehole corrected, Tracks 2 and 3 contain Rwa curves calculated
by the Formation Factor Equation using five different sets of variables and Rw values calculated from the ionic concentrations of water samples taken during pump tests
(RWA MEASI, There were discrepancies in the measured conductivity values, so calculated values were used instead, All curves in tracks 2 and 3 have been normalized
to 77 0 F, The Rwa curves in track 2 were calculated from density porosity (2.65 gfcm 3 matrix density) using either 1/~·81 (RWA M.81 I, ,62/q>2.IS (RWAM2.15). or 1/tY
(RWA M1). The RWAM2.15 curve is the most accurate of the three. This is because the aquifers are high porosity unconsolidated sandstones, which is the appropriate
geological environment for .62/~2.'s. As predicted by Figure 14-15. the RWA M.B1 curve has Rw values that are close but slightly lower than the RWAM2.15 curve and
the RWA M1 values are even lower. All three Rwa curves in track 3 were calculated with .62/q>2.15 and different porosity curves: density porosity (RWAM2.15 from track
2), density-neutron crossplot porosity (RWADN2.1) and density-neutron crossplot porosity corrected for shale (RWADNSHI. Overall, the RWADN2.1 curve has the most
accurate Rw values. The curve utilizing only density porosity has Rw values nearly as accurate as the two lithology corrected curves. This is because the density curve is
affected very little by shale. Thin section analysis of core samples and the PEF curve (Plate 4) revealed that the mineralogy includes calcite and shale, The spiky nature of
the curves, due to thin beds and washouts. illustrates the advantage of calculating Rw over entire formations rather than at a few specific depths. It is much easier to
determine the best Rw value for a formation by scanning the curve for the entire interval. Track 1 shows that the invasion profile changes down the borehole from Rw >
Rmf to Rw = Rmf to Rw < Rmf. Bottomhole temperature is 110 0 Fat 1060 feet. The well is the TWOB PUB Test Well Site F. Cameron County. Texas (state well
number 88-59-411). Plates 2 to 4 and Figures 13-12. 14·13, and 14-20 contain additional data on this well.
Well Nam:
We1 l Nana: TWOS
TWDB SITE F88-59-411
F 88-59411
ohm· mete.. ohm·meterl ohm·metell
__ J.Q.... .u.D .10Q...lll_ J DEPTH 10 RIlDN2 1 to.OO
____ .IIJl J.DlUD FEET
~2.

r-~LII.... ............"""- .......

!II_ --.IQ.--------.1E'----------.lll..QD---.
. --=----
.::~ :~~ :::b .. "'----~

100

150

200

250

300

350

400

450

500

550
(
•
';.":..-=..~
C,.. ___ •

c--
600 ~==~-- -,,
~

i;---
650 C;,. ....
;I,)

700 ,
I
•
.--'"
l

~'
........ _------
,..--... ---"
r-
750 "--~
~._-i,f""
~

~!;::---.....,..
.... _-- ...--
-.-:_----'V........
800
,- -- ..
"

...... _- .. _-----~
~------

850
,~::::-="J
'"i:____ ~
....
.".-.::::_-::::

900

...,.--=
950

1000
..-:::::
1050

PLATE 2. Rw calculated by the Resistivity Ratio Ratio and SP methods. Track 1 contains Induction and
contains Dual Induction Microspherically Focused
and Microspherically (MSFLI logs, Track 2 contains
Focused resistivity (MSFLllogs. contains
Rw curves calculated
calculated by the Resistivity
Resistivity Ratio using the MSFl
method using
Ratio method MSFL curve Rxo. The drilling
curve for Rxo. report and
drilling report reported different Rmf values, so Rw
log header reported
and the log Aw curves
curves
calculated for each
were calculated each Rmf value: 1.77 ohm-meters 7 7 O F from the log header (RWA
ohm-meters @ 77"' MSFL) and 3.3
(RWA MSFl) 3.3 ohm-meters @ 77 77'0 F from the drilling (RWAMSFL31.
drilling report (RWAMSFL3j.
Neither curve is accurate for the bottom 100 feet of the borehole.
bottom 100 borehole. This is to be expected
expected since there is little invasion
invasion over this interval (refer to the invasion in
profile in
invasion profile
1). Overall, the RWA
track 1). MSFL curve is very accurate. Track 3 contains Rw
AWA MSFL Aw curves calculated
calculated by the Formation Equation (RWADN2.
Formation Factor Equation Plate 1 for
(RWADNZ. "1, refer to Plate
details) and
detailsl methods. The RWADN2.1
(RWA SP) methods.
and SP (RWA RWADNZ.l curve is very accurate for most of the borehole. borehok. The RWA SP is very accurate down to 500 feet, but it reads much much
high below 500 feet and
too high and is very spiky. The sandstones are very thinly beddedbedded below 700 feet. feet, which means that the SP curve is reading
which means much less than SSP. This
reading much
would result in in calculated much too high, just as are present
calculated Rw values that are much present on the FflNA Correcting the SP curve for thin
W A SP curve. Correcting bed effects would significantly
thin bed
improve the calculated Rw values. The error may also be due,
improve in part, to significant differences
due, in differences in between the two intervals
in the water chemistry between (confirmed by the water
intervals (confirmed
analyses) and/or differences in
analyses) andlor membrane potential of the shales.
in the membrane shales. The radioactive interval at 480 feet, which
radioactive interval which may be an unconformity, lends support to the idea that
the shales on each side of 480 feet are diffferent. Rw values (RWA MEASJ MEAS) are calculated from the ionic concentrations
concentrations of water samples
samples taken during tests.. All
pump tests
during pump
curves in tracks 2 and 3 have been
in tracks normalized to 77 O0 F. The well is the TWDB
been normalized TWoS PUBPUB Test Well Site F, Cameron County, Texas (state well number 88-59-41
F, Cameron 88-59-411). Plates 1,
11. Plates 1.
3, and
3, and 4 and t 3-12. 14-13,
Figures 13-12,
and Figures 1 4-13, and 14-20 contain
and 14-20 additional data on this well.
contain additional
Well Nan: TWOS SITE FF 88-5941
TWDB 88-59-4111

--A ---- a--. *a,# t

- - _JllL4I._ 1~§4~ ~_~f\ Jg~qp~ r-~~4~l---------I~JDEl ~~~ .
, L a r ....._-------- m. 01 DEPTH
o 0

m.m FEET

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

.,
- *'f +
.-
,'
850 .) .-
4 -, ,
,$ .-': -
I . . '
..._ '

900

950

1000
~
--..
_ 0....- ....
.. ,

--~

1050 ---- --_1

-...
PLATE3.
PlATE 3. Comparison
Comparisonof of water
water conductivity (Cwland
conductivity (ew) tDScurves
andTOS cwvescalculated
calculatedby ResistivityRatio,
byResistivity Ratio, Formation FactorEquation,
FormationFactor Equation, andandSP methods. Track
SPmethods. Track 11contains
containsDual·'
Dual '
Inductionand
Induction andMictospherlcally
Microspherlcally Focused
Focused(MSFl)
(MSFL)logs. Track22contains
logs. Track containsfour fourCw cwvescalculated
Cwcurves calculatedfrom
fromRw cwveson
Rwcurves Rates 11and
onPlates and 2.2. CW MEASwas
CWMEAS was calculated fromthe
calculatedfrom the
RWAMEAS
RWA MEASdata.
data. Track containsros
Track33contains f O Scurves
curvescalculated
calculatedfrom
fromthetheCwCwcurves track22using
cwvesinintrack usingthe
theequation
equationofof the
theline fittingaaTDS·Cw
linefitting TDS-Cwgraphgraphthat
that includes Cameron
includesCameron
C o w (FigUl'8
County 14-17 I.1. The
(Figure14·17 T hTOS curvesinintrack
TDScurves demonstratethat
track 33demoosuate accurateTOS
thataccurate valuescan
TDSvalues canbebecalculated
calculatedfrom logsand
fromlogs andthat Increaser ififseveral
accuracy Increases
theaccuracy
thatthe severalmethods
methods
areused
are andthe
usedand m i r ewell
theentire wellisisevaluated.
evaluated. The TDSMEAS
TheTOS MEASvalues
valuesare arefrom
fromthe wateranatyses,
thewater analyw, andandthe TDSMEEOUvalues
ffieTOSMEEQU werecalculated
v a l w were calculated using
usingthetheequation
equationofof the lim
theline
fitting the TDS-Cw graph and CW MEAS values. Comparison of the M S MEAS and TDSMEEQU values confirms that accurate
fitting the TD5-Cw graph and CW MEAS values. Comparison of the TDS MEAS and TDSMEEOU values confirms that accurate TDS values can be calculated from TOS- TDS values can be calculated from TDS-
ew
C Wgraphs. TDSMEEOUisisgreater
graphs. TDSMEEQU greaterthan
thanTOSTDSMEAS below480
MEASbelow 480feet, whichisisthe
feet,which thedepth
depthat whichthe
atwhich thewater
waterchemistry
chemistrychanges.
changes. AtlAllcurves tracks22and
curvesinintracks and33have
havebeen
been
normalizedto
normalized
and
to77
14-20contain
and14-20
77 0*F.F. The
Thewell
containadditional
EheTWOB
wellisisthe
additionaldatadataon
W BPUB
onthis
thiswell.
W8Test
well.
Site F,F, Cameron
WellSite
TestWell County,Texas
CameronCounty, (state~!.Ot.lftlber
Texas(state
... ,
88-59-41t I. Plates
w l l nymber88·59-4111. I , 2,and
Plates1,2, and44andandFigures 14-13.
13-12, 14-13,
Figuns13·12, C1
,
;.
,-;J
Well Norm: TWOS SITE F 88-59-411
SP In mv, GR In APt unit., CALI In inc:h". ohm·meters
t
DEPTH
FEET

100

150

200

250

300

350

400

650

700

750

800

850

900

950

1000
.'13
t

1050

PLATE 4. logging suite used to calculate water quality. Track 1 contains SP, gamma ray corrected for borehole effects (GRl, and caliper (CALI). Track 2 contains Dual
Induction and Microspherically Focused resistivity (MSFlI curves. The MSFL has been corrected for borehole effects. Track 3 contains density pOrosity (DPHll, neutron
porosity (NPHII, density-neutron crossplot porosity (D-NPOROS), and photoelectric factor (PEFI curves. The porosity curves are calculated on a sandstone matrix. The
well is the TWDB PUB Test Well Site F, Cameron County, Texas (state well number 88·59·4111. Plates 1 to 3 and Figures 13-12, 14-13, and 14·20 contain additional
data on this well.
YYeI I NO~IB:NEW BRAUNFELS A-1

PLATE 5. Rw calculated by the Fomatlon Factor method. Tmck 1 eorrtalns fluid resislivlty {FLUIW#SI and mperah#e m e s , along with an RWA curve cakwted bv
the Formath Factor method using It& , unaveragd spherically f d mWvhy vakws (not shown), and lithology corrected porosity values (track 31. llw blue
vertical lim a n measwed Rw values from water samples taken duing pump m. Them is exceHmt agreement hatween the RWA curve and measured Rw values.
Spikes in the RWA c w e are due to thin intervals that cannot be accurately measured by t h raristivw and pamih tools, resulting in inam;urata RWA calcuhtiau for
thew Intervals. However, averaging the RWA m e for a large interval gives an a c m e pi- d Rw. RWA is less than RW abbvs 650 feet. This may be due to
$&ace conductance. This interpretation is supported by the fact that as wetar salinity dacraam below 650 feet the RWA Md Rw weday. The RWA curve may be
accurately r e f k t h g an increase in water salinrty below 660 feet, which the commits warn sample taken over tht entire interval cannot show. The RWA curve clearb
shows transitions in water quality t h r q h w t the entire borehole. Tnck 2 contains M mrmal (SH NORMI and bna normal (LONG NOR) am-. The t w o cunres agree
so wstl becape t b drilling fluid was formatiom water and there is no invasion. Werenuts in the normal curves are due to t b d i e m vertical mwtvtion of each tool.
Track 3 contains dmity-neutron crossplot porosity 40-NWROS) and photoelectric factor (mm e s . The lithology is abrmting IimaPtont and dolomite, So only a
crassplot ponrslty cuve will be accurate. At 635 to 652 feet on t h normal logs th4 cuwes show a false flat top because ths curvas want otf scale on the hard eopy of
the log from which the cutves were digitized. This is a well in the Edwards aquifer, Mew Braunfels, Texas. The well is th4 Edwards Underground Water MsUkt. A-7
(state well number 68-23-616). The bit site is 7 'A inches. Figures 13-32 and 14-19 contain additional information on this well. Po w ,Cdlier, and Maclay (1 992)
contains a detailed petrophysical analysis of this well.