You are on page 1of 5

Evaluation of Gas-Extraction

Efficiency During Mud-Logging


Operations
Suzanne S. Elder, * SPE, Texaco U.S.A.

Summary. Field and laboratory tests described in this paper were designed to develop a better understanding of conditions that
affect gas analysis during mud logging. The test objectives were to quantify gas losses, if any, from the mud at the bell nipple or
in the flowline during drilling and to study the effects of changing mud properties on gas-trap extraction efficiency.
Introduction
The monitoring of gas in mud and of lithological information during drilling operations is done both for safety reasons and to identify potential productive zones. However, the standard method of
extracting gas from the mud with a mechanical agitator and of transporting this gas through a vacuum line to be measured has been
criticized for inconsistency and inefficiency. The possibility that
gas may break out of the mud in the open bell nipple or flowline
also raises the question of how much gas is lost before the mud
reaches the gas trap. These criticisms and questions prompted the
study described in this paper.
Laboratory tests using a loo-bbl [15.9-m 3 ] flowloop were performed to study the efficiency of the gas trap and the behavior of
the gases in the mud as a function of mud properties. Seven tests
were run with two different mud weights and six yield-point/plasticviscosity combinations. A known amount of gas was metered into
the mud system to measure the efficiency of the gas trap in removal
and transportation of this gas. Two gas chromatographs analyzed
the gas samples on alternating I-minute intervals.
Field testing provided the opportunity to analyze and compare
the amount of gas in the mud at three sampling points: below the
bell nipple, in the flowline, and in the shaker box. These values
revealed the amount of gas lost from the mud before it reached the
gas trap for this specific test well. The efficiency of the gas-removal
system was also monitored during the field test by comparing the
gas removed by a steam still to the gas detected in the mud-logging
unit.
A research project concerned with various aspects of mud logging has been under way at Texaco for several years. The purpose
of the project is to investigate current mud-logging techniques to
identify areas where improvement would result in better-quality information from mud logs. A major part of the investigation is directed toward gas analysis.
In current mud-logging operations, a mechanical agitator is placed
in the shaker box to remove the lighter hydrocarbons from the mud.
Once removed, they are transported through a vacuum line to a
gas-detector system within the mud-logging trailer. There, the gas
is quantified by a total hydrocarbon analyzer (THA) and analyzed
for individual components by a gas chromatograph. Our study did
not compare different types of detectors available for THA's or
gas chromatographs. In field and laboratory tests, flame ionization
detectors (FID's) were used for THA and chromatographic gas analysis. The FID is accepted by the industry as the most accurate and
sensitive commercial detector available for commercial mud-logging
purposes.
Test Objectives
Very little material has been published giving results from field or
laboratory testing on the mud-logging gas-analysis system. However, references are made in the literature to gas losses from the
'Now at Aramco.
Copyright 1988 Society of Petroleum Engineers

686

return mud at the bell nipple and in the flowline. References are
also made, without specific test information, to the effects of mud
viscosity on the gas-trap extraction efficiency. Therefore, the objectives of our study were to quantify the gas losses, if any, from
the mud at the bell nipple or in the flowline and to study the effects
of mud properties on gas extraction efficiency.
Obviously, field testing was required to gather information about
system gas losses. However, detailed information about the mud
property effects required more controlled laboratory test conditions.

Laboratory Test
Procedure. The laboratory test was designed in cooperation with
NL Baroid Logging and was conducted at their test flowloop facility in Houston, TX.
Drilling mud was circulated by a rig-size triplex pump through
a l00-bbl [15.9-m 3] flowloop. A hydrocarbon gas mixture was injected into the flowline upstream ofthe triplex pump. The gas mixture was injected through a sintered element to assist in dispersing
the gas into the mud stream. The mixture then passed through the
triplex pump to ensure homogeneous dispersion. The pressure between the triplex pump and the first orifice plate was held constant
at 1,500 psi [10 342 kPa] for each test. Fig. I is a schematic of
the test system. The gas mixture consisted of73% methane (C,),
20% ethane (C2), 4.5% propane (C 3 ), and 2.5% isobutane (iC 4 ).
This mixture was selected on the basis of prior field gas analyses
and safety considerations.
The gas was injected at an approximate rate of 1 % gas-in-mud
or at 10,000 ppm for the first six tests. This was a safe level and
above the minimum detection level of the chromatographs for all
hydrocarbon components. For the seventh test, the gas injection
rate was increased to 2.5% gas-in-mud or 25,000 ppm to compare
detection efficiency with a higher gas concentration in the system.
A modified 55-gal [0.2-m 3] drum served as the shaker box. Mud
flowed into the bottom of the drum from the flowline and exited
into the mud tank through a pipe on the opposite top side of the
barrel. The gas trap was mounted in this barrel. The flow rate of
mud through the gas trap was maintained at 3 gal/min [0.19
dm 3 Is]. Fig. 2 shows this arrangement.
Each of the first six tests lasted 15 minutes. The mud properties
were checked, and one mud sample from the tank was analyzed
for the amount of gas in the mud before test startup. This amount
was always quite low. The triplex pump was turned on and the gas
injection started. After 5 minutes, the gas had circulated around
the entire loop and could be detected at the gas trap.
During the test, the gas trap continuously extracted gas from the
mud in the shaker box. The gas was drawn off the top of the gas
trap by a vacuum line at a flow rate of 0.16 ft 3 /min [0.0045 m 3 /s].
The vacuum line was connected to two gas chromatographs and
one total hydrocarbon analyzer. Two FID chromatographs sampled the gas from the vacuum line in this test. The chromatographs
were computer-controlled and worked on alternating cycles. Because these chromatographs analyze gas on a batch-cycle process,
SPE Formation Evaluation, December 1988

GAS
INJECTED
MUD TANK
TRIPLEX PUMP

FLOW LOOP

ORIFICE PLATES TO CONTROL


PRESSURE DROP

Fig. 1-Laboratory flowloop system.

Fig. 2-Test shaker box.

use of two chromatographs with alternating cycles allowed analysis of the gas from the trap every minute. In normal field gas analysis, one chromatograph is used and C 1 through iC 4 are analyzed
every 3 to 4 minutes. NL Baroid's BLS-2000 computer system was
used to accumulate and store the data from the chromatographs,
which were then printed out with the time that each sample was
processed.
The THA also used a flame ionization detector to monitor continuously the total amount of hydrocarbons present in the gases from
the gas trap. The vacuum line from the gas trap had a split flow
stream into the chromatographs and the THA, which is the normal
arrangement in field mud-logging operations.
The mud was a water-based lignosulfonate. The two mud weights
used were 9.8 and 14.0 Ibm/gal [1174 and 1678 kg/m3]. At each
mud weight, three yield-point and plastic-viscosity values were tested. Table 1 shows the different combinations used for the seven
tests. Yield point is a relative measurement of the interaction between the solids in a mud system. Plastic viscosity is related to the
amount of solids in a mud system. They are independent but related measurements. At the conclusion of each test, the mud properties were checked to ensure that they had remained approximately
the same during the test.
To evaluate the gas-trap efficiency, gas concentration was measured in the flowline, gas-trap inlet and outlet, and vacuum line.
The gas concentration in the flowline, trap inlet, and trap outlet
was determined by using a steam still to remove the gas from the
mud. On the basis of separate gas desorption tests, it was determined that there was no appreciable absorption of gas on the solids
contained in the mud and that the steam still removed over 98 %
of the gas from the mud. Yet only 50 to 60% of the injected gas
was detected by the steam-still analysis in the flowline sample. Because of the good correlation between the flowline and gas-trap inlet value of gas concentration, sampling error can be eliminated
as a possible explanation for the loss. However, the loss was nearly constant in all tests. At this time, we cannot offer a definite ex-

planation for the gas loss in the flow loop. One possible explanation
is that a portion of the gas broke out of the mud as a result of the
rapid pressure drop at the orifice plates between the triplex pump
and the gas trap. This gas would stratify along the top of the flowline
and become difficult to sample or become trapped in dead space,
such as valves, gauges, or the piping manifolds.
The two types of efficiencies calculated in this study were
gas trap extraction efficiency
gas in trap ) _ ( gas in trap )
( inlet sample
outlet sample

x 100

(gas in trap inlet sample)


and
overall system efficiency
(

gas concentration as detected)


by gas chromatograph

- - - - - - - - - - - x 100.
(gas in trap inlet sample)

10.000

TEST

----2

--I

8.000

--3
--~

----5
----_. ,
6.000

E
a.

9-

u~.ooo

TABLE 1-MUD PROPERTIES FOR LABORATORY TEST

Mud Weight

Test
1
2
3
4
5
6
7

(Ibm/gal)
9.8
9.8
9.8
14.0
14.0
14.0
14.0

Plastic
Viscosity
(cp)

8
13
18
18

25
33
33

SPE Fonnation Evaluation, December 1988

Yield Point

(lbm/100 ft2)
2
8
12
5
10

2.000

10

15

TIME

20

25

30

(minute'

24

21

Fig. 3-Methane amount vs. time for Tests 1 through 6.


687

10.000 , - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
~c-----------------------------------------,

8.000
TEST

-----6

--7
6.000

I
~

(f

./

TEST
--I

_._-.

[!

--.

--- ,

-------5

"

I,

I,

, ,-.

4.000

---2
------1

2.000

I'
I'
I
I
I

_-...
THI. DATA

Fig. 4-Total hydrocarbon amount vs. time for Tests 1


through 6.

o~==~~----~----~----~----~--~
JO
15
20
25
o
10
TIME (minute)

Fig. 5-Methane amount vs. time for Tests 6 and 7.

Laboratory Test Results. The gas-trap extraction efficiency was


nearly constant in the first six tests. This result shows that plasticviscosity and yield-point changes in a 9.8- or 14.0-lbm/gal [1174or 1678-kg/m 3 ] lignosulfonate, oil-free, water-based mud did not
affect efficiency significantly. The result can be explained by the
fact that even though the values of plastic viscosity and yield point
were changed, the muds were still shear-thinning-Le., undergoing a reduction of viscosity at high shear rates. The shear-thinning
characteristic is indicated by the yield point value of the mud. Hence,
when the gases were extracted in the gas trap by a shearing action
of a paddle, the apparent viscosity of the mud in all the tests approached the same value. Therefore, a more-or-less constant extraction efficiency might be expected. However, gas extraction data
can be affected by changes in mud flow rate through the gas trap.
In this test, the flow rate of the circulating mud was held constant.
This is not necessarily the case during drilling operations.
For the first six tests, the system efficiency was about 25 %, while
the extraction efficiency of the trap was about 50 %. This implies
that 50% of the liberated gas removed by the trap was lost. The
gas can be lost from the mud outlet or from the other leaks in the
trap because of diffusion. This observation clearly demonstrates
a need to improve the gas-trap design.
Figs. 3 and 4 show the results of gas analyses over the six tests
for C 1 only and the THA, respectively. Although there are small
differences in the amount of gas measured, this difference is not
significant in interpreting the data for mud-logging purposes; Le.,
if these were six gas shows, the final result of the interpretation
would be the same. Fig. 4 shows the THA data on a typical mudlog scale of four-cycle log paper. This was done to put the test results in perspective when considering effects of the mud properties
on field gas shows.
We observed that the relative composition of the hydrocarbon
gases detected by gas chromatograph was the same as that of the
injected gas. This observation shows that the gas trap did not extract gas preferentially. However, the result may not be true for
a mud system containing diesel or oil.
Tests 6 and 7 were run using the same mud with approximately
the same properties, but a higher concentration of gas was injected
during Test 7. The overall system efficiency for Test 6 was 25 %
and for Test 7 was 17%. The trap efficiency for Test 7 also was
lower than for Test 6. This finding was the opposite of what would
be expected. Because only one test was performed at the higher
gas concentration, it is difficult to predict a trend. Additional testing in this area is necessary. Fig. 5 shows the gas chromatograph
data for Tests 6 and 7. As expected, the absolute readings are higher
for Test 7 because of the higher concentration of injected gas.
688

o.

Fig. 6-Sample device for field tests.

SPE Fonnation Evaluation, December 1988

6000 r-

...~
'"...

5000 f-

60 r

Q,

50

~ ..000 f-

...a:~.

3000

2000-

r-

><

,:
z

G 40-

.....
....

.......
::::E

V1

>-

1000 -

30

r-

V1

o
;::
u

.......
I000O

20000

30000

..0000

50000

STEAM STILLED SAMPLE FROM SHAKER BOX, ppm

Fig. 7-Gas in mud at the shaker box vs. gas detected by the
mud logger.

The test had some limitations that should be kept in mind when
the results are reviewed. There were no hydrocarbons in the mud
before the gas injection. Many field muds will contain diesel or
other heavy hydrocarbons even though the mud is water-based. The
presence of the heavier hydrocarbon would very likely change the
way in which gases would dissolve and release from a mud. Additionally, the abrupt pressure drop that occurred when the mud flowed
through an orifice is different from the gradual pressure drop that
would occur in a well.

Field Test
Procedure. Although field testing is continuing, the results from
the first successful test are reported in this paper. Because each
rig configuration is different, the sample points were tailored to
the optimum location for accessibility and safety in our field testing. This field test was performed in northwest Colorado on a development well where potential production zones are considered
underpressured. The mud system was 8.7 Ibm/gal [1042 kg/m3]
KCI polymer. The plastic viscosity ranged from 7 to 15 cp [7 to
15 mPa' s] and yield point ranged from 3 to 20 Ibm/l00 ft2 [0.15
to 0.98 kg/m2]. The values were taken from the morning mud
report for the days during our testing and were not remeasured when
we took our mud and gas sample.
It was safe to place a sample tap on the casing valve because no
high-pressure gas was anticipated. This yielded a sample of the mud
10 ft [3 m] below the open bell nipple. Fig. 6 is a schematic of
the sample tap. A second tap was placed on the side of the flowline
about 5 ft [1.5 m] before the connection to the shaker box. As shown
in Fig. 6, the sample tap was designed to extract a sample from
these two points without exposing the mud to the atmosphere. The
mud flowed into the sample chamber through a 2-in. [5-cm] ball
valve, where it was extracted with a 1O-cm 3 syringe through the
rubber septum in the end of the sample chamber. This IO-cm 3 sample was injected into a reflux steam still to strip all the gas from
the mud. This gas was drawn off the steam still and analyzed by
the FID chromatograph in the mud-logging trailer.
Three samples were taken from the return mud system: the casing
valve sample, the flowline sample, and the sample taken just below the mud surface in the shaker box. They were taken in that
sequence on a lag basis to represent the same mud as it passed all
three points. Additionally, the gas-trap reading taken at this time
by the mud loggers was recorded for comparison.
Some problems with samples were encountered when lost circulation material that was added to the mud system blocked the needle of the syringe during sampling. Steel wool was used as a filter
in the sample chamber to overcome this complication.
Field Test Results. Eleven mud samples were taken during the field
test. Because of some sample loss or difficulty in obtaining a samSPE Formation Evaluation, December 1988

20 f-

V1

(j

10 l-

STEAM STILLED SAMPLE FROM SHAKER BOX, ppm

Fig. 8-Gas in mud at the shaker box vs. gas detection system efficiency.

pie, not all II had representative casing-valve, flowline, and shakerbox syringe samples. However, there were sufficient data for the
analysis.
Seven of the 11 samples had a casing-valve sample. In six of these
seven samples, there was more gas in the mud at the casing valve
than at the shaker box, with the difference ranging from 10 to 59 %.
We cannot adequately explain this wide range at this time. Gas appears to be lost from the mud system between these two points,
but more data from other field tests are required to make a conclusion about the reason for the large range.
The overall detection efficiency was checked by comparing the
amount of gas found in the mud from the shaker box with the amount
detected by the chromatograph in the mud-logging trailer. Nine of
the 11 samples had these values to compare. The efficiency ranged
from 6 to 56 %. The gas trap actually works as a gas concentrator.
The mud flow through the trap is at a greater rate than the rate
at which the vacuum line draws the gas out of the trap. This difference can be corrected if these two rates are measured, but the rates
were not measured in this field test. They were measured in the
laboratory test, however, and the overall detection efficiency was
corrected in the laboratory test to reflect this.
Fig. 7 shows that even when significant changes were detected
in the total amount of gas in the mud at the shaker box, the mudlogging system did not effectively extract and transport the gas to
the gas chromatograph. The total concentration of gas from the
shaker-box sample exceeded 45,000 ppm for some samples, but
the mud-logger's chromatograph reading remained at about 5,500
ppm. This result implies that if the increase to 45,000 ppm of gas
in the mud is considered a gas show, the mud logger may miss the
show. A plot of the system efficiency of each sample vs. the total
gas amount in the mud shows that the efficiency apparently decreases
as gas concentration increases. Fig. 8 shows this correlation.
In some samples, the amount of gas in the shaker-box sample
exceeded the amount of gas in the flowline sample. Because it is
unlikely that gas would re-enter the mud system in the shaker box,
there are at least three possible explanations. First, the flowline
on this rig had < 50 slope from the bell nipple to the shaker box,
allowing cuttings to build up in the flowline. The sample tap was
placed on the lower side of the flowline to ensure a liquid sample;
however, this also caused difficulty in sampling because cuttings
blocked the sample port. As a result of this problem, the flowline
sample was not always obtained at the proper lag time. The second
689

reason for less gas in the flowline sample could be that flowing
the mud through a 2-in. [5-cm] valve into the sample chamber
created turbulence and broke some of the gas out of the mud. The
third possibility is human error in sample analysis.

Conclusions
Laboratory Tests
1. The average extraction efficiency of the gas trap, as measured from steam-still samples taken at the trap suction and trap discharge, was about 50% in Tests 1 through 7. Therefore, mud
weight, plastic viscosity, and yield point did not significantly affect gas-trap extraction efficiency over the range tested for this particular mud.
2. The overall system efficiency, based on the comparison of the
amount of gas in the mud at the trap suction with the amount detected by the gas chromatograph, was about 25 % for Tests I through
6 and 17 % for Test 7, which involved a higher concentration of
gas in the mud.
3. The ratios of the extracted gas (ClfC2, C l fC 3 , and C l fiC 4 )
remained essentially in the same proportion as the injected gas over
all the tests; therefore, the gas trap did not extract any hydrocarbon component preferentially for these particular muds.

Field Test
1. The current hydrocarbon gas extraction system is inefficient
and needs improvement. The mud-logging readings do not correspond to the amount of gas transported from the drilling interval
because of gas losses in the lines and inefficient gas-trap design.
More tests are required to understand where losses occur and to
improve the gas-extraction system.

690

2. There was a 10 to 59% gas loss from the mud before it reached
the shaker box. Additional field tests will be required to quantify
the amount more accurately and to explain the wide range.
3. The efficiency of the gas-detection system for the field test
ranged from 6 to 56 %. These values are not corrected for the difference in flow rate through the trap and in the vacuum line.
4. There appears to be a correlation between the amount of gas
in the shaker box and the efficiency of the gas-detection system.
In general, as the amount of gas in the system increased, the gas
detection system efficiency decreased.

Acknowledgments
I would like to thank Texaco U.S.A. for permission to publish this
paper. I express my appreciation to Stewart Musgrove and NL
Baroid Logging System for their assistance in design and implementation of the laboratory test described in this paper. I also thank
Texaco Central Exploration and Denver Operations Divs. for their
cooperation and assistance in the field test.
51 Metric Conversion Factors
cp x 10*
E-03
ft2 x 9.290304*
E-02
gal x 3.785412
E-03
Ibm x 4.535 924
E-Ol
'Conversion factor is exact.

Pa's

m2
m3
kg

SPEFE

Original SPE manuscript received for review March 15. 1987. Paper accepted for publication Nov. 12, 1987. Revised manuscript received April 28, 1988. Paper (SPE 16158) first
presented at the 1987 SPE/IADC Drilling Conference held in New Orleans, March 15-18.

SPE Formation Evaluation, December 1988

You might also like