PS

Statistical Comparison of Hydrocarbon Gas Composition and Isotopic Ratios from Multiple Sampling Methods*

Benjamin T. Kirkland1

Search and Discovery Article #70298 (2017)**

Posted October 9, 2017

*Adapted from poster presentation given at AAPG 2017 Annual Convention and Exhibition, Houston, Texas, United States, April 2-5, 2017
**Datapages © 2017 Serial rights given by author. For all other rights contact author directly.

1
Nexen Petroleum U.S.A. Inc., Houston, Texas, United States (BenjaminKirkland@nexencnoocltd.com)

Abstract

A standard aspect of formation evaluation is an inferential analysis of hydrocarbon gases encountered in the subsurface. The nature of these gases can
indicate fluid saturation, phase, quality, provenance, and many other unknowns that are fundamental to understanding the petroleum system as well as
commerciality of the respective well. Currently, there are four principal sampling and analysis techniques employed. These include well site mud gas
analysis, offsite analysis of collected mud gas (IsoTubes®), headspace analysis from cuttings samples (IsoJars®), and analysis of flashed gas from down-
hole fluid-sampling tools (e.g. MDT). The interpretation of these data is imperative to any petroleum systems analysis, but, as is often the case in applied
exploration science, only one or two sampling methods may be prudent to collect during operations. Additionally, historic data may be incomplete or
limited, and an understanding of relationships and inherent biases in the sampling and analytic methods can help to increase confidence when dealing
with such limited datasets. This study offers a statistical comparison of these four methods, in the context of applied analysis of a deepwater dataset, to
quantify sampling and analytical uncertainty. It has been observed in limited case studies that normalized gas composition measurements are variable
between IsoTube® and MDT samples, but a statistical analysis on a large dataset across multiple hydrocarbon plays with both compositional and isotopic
variables has not been published. This comparison, combined with well-site GC and headspace gas analysis, creates a robust analytic tool that can help to
overcome the problems of data sufficiency and cost associated with running redundant analyses.

References Cited

Ablard, P., C. Bell, D. Cook, I. Fornasier, J.P. Poyet, S. Sharma, K. Fielding, L. Lawton, G. Haines, M. Herkommer, K. Mccarthy, M. Radakovic, and L.
Umar, 2012, The Expanding Role of Mud logging: Schlumberger Oilfield Review, v. 24/1, p. 24-41.

Ayan, C., P.Y. Corre, M. Firinu, G. Garcia, M. Kristensen, M. O'Keefe, T. Pfeiffer, C. Tevis, L. Zappalorto, and M. Zeybek, 2013, New Dimensions in
Wireline Formation Testing: Schlumberger Oilfield Review, v. 25/1, p. 32-41.
Bernard, B.B., J.M. Brooks, and W.M. Sackett, 1978, Light hydrocarbons in recent Texas continental shelf and slope sediments: J. Geophys. Res., v. 83,
p. 4053-4051.

Berner, U., and E. Faber, 1996, Empirical carbon isotope/maturity relationships for gases from algal kerogens and terrigenous organic matter, based on
dry, open-system pyrolysis: Organic Geochemistry, v. 24, p. 947-955.

Chung, H.M., J.R. Gormly, and R.M. Squires, 1988, Origin of gaseous hydrocarbons in subsurface environment: theoretical considerations of carbon
isotope distribution: Chemical Geology, v. 71, p. 97-103.

Haworth, J.H., M. Sellens, and A. Whittaker, 1985, Interpretation of Hydrocarbon Shows Using Light (C1-C5) Hydrocarbon Gases from Mud-Log Data:
AAPG Bulletin, v. 69/8, p. 1305-1310.

Mankiewicz, P.J., R.J. Pottorf, M.G. Kozar, and P. Vrolijket, 2009, Gas geochemistry of the Mobile Bay Jurassic Norphlet Formation: thermal controls
and implications for reservoir connectivity: AAPG Bulletin, v. 93/10, p. 1319-1346.

Schoell, M., 1983, Genetic characterization of natural gases: AAPG Bulletin, v. 67, p. 2225-2238.
 Statistical Comparison of Hydrocarbon Gas Composition and Isotopic Ratios from Multiple Sampling Methods
Benjamin T. Kirkland 1
1
A CNOOC LIMITED COMPANY Nexen Petrole um USA, Inc. Houston, TX A CNOOC LIMITED COMPANY

ben jam in.kirkland@nexenusa ,net

Abst ract Data Sources Results Conclusions

There are three locations in which subsurface gases are sampled Though a relatively limited dataset has been used for t his st udy. en ough notable trends have
A Sl andard aspect of formation evaluation is an Infe rential analysis o f hy>drOl:<lrbon gases .
I

,
~ +1---CiII ~o=~I---------<
~
encountered In the subsurface. The nat ure of the gaSM ca n Indicate fluid sa turotlon, p hase, throughout the drilling process: t he wellbore, the shale shaker, and the Hu dsplce Gas emerged th at a tool for general correction can be developed. At this point, it is certainly
quali ty, p rovenance. and m any other u nknowns t hai are fu nd amenta l to und erstandlnB the

~
mud logging unit. Downhole wireline formation testers are generally the 11 - 16 most applicab le for conventional plays, and an area of future wo rk can be to broaden the
petrDleum system as we ll as commerclalit y of t he respectlye w ell. Currently, there arc four
pr fnclpal sampllna and analys1s tech n lque~ emploved . These include well site mud gas highest confidence, bu t the samples are extre mely expensive and often Isotub., basin and play types in the sam ple distributi ons.
analysis, offsile analysis of collected mud gas (lsoTubes). headspace analysis from cuttings Isotubes
.samples (Isolars). and analysis of nashed gas from down -hole nuld sampling tools le.B. only taken after minimum commercial or geologic success criteria. It is n- 19 " . /6 The following table is for correction of headspace gas, Isotube, and wellsite GC samples to
MDD. TI'le Interpretation of these data 15 imperative to any systems an alysiS, but, as it often uncommon to have more t han a few WFT sample point s in a we ll, so WFT samples. It is onlV for com positional correction as no isotopic correction is necessary
the case In applied exploratiot'l selence, only one or two sa mplifl& methods may be prudent W. lI sit. GC
much less expensive samples are generally taken at the shale sh ake r and between Isot ubes and WFTs. The recommend ed bulk correction is the median of the
to collect during operations. Additionally, historic data m.ay be incomple te or li mi ted, and I
an underS1:a odlng of relationships and Inherent biases In the sampling and inalytk: methods
can help (0 Increase mnfide nc.e when deali ng with such limited dalaset.5. This study offe rs
a statiS'fical compari5(ln of thest fOur methods. In the (OnteJc l of a deepwater datase t, to
in t he mudlogging unit with grea ter frequency. In fact, the wellsite gas
chro matograph is nea rl y always run continously fo r drilling safety and
.. so .. ., ., ., WH
I n- O
varian ce distribution for each co mpound. Because the entire distribu tion is reduced t o one
number, it is important to be familiar with t he ori ginal plo ts if a speCific analysis could
QUOIn tify sa mpling a nd a nal,.-tiCOlI uncertainty. It h.u bee n obs ervl!d in limited case stud ies monitoring reasons. We llsite GC analvsis is returned as a curve Hudspactl Gas. depend on small va riation.
that no,malb:ed Kas composition measurements iHe va,iable between I~rube and MDT
S<l mples. but a sratfsool analysis on a laree dataset iOIIeross m ultiple hydroc;arbon plays with commonly displayed on the mud log, and t he other samples are sent [,;:"'=:::.:~ n-16

both compositional and Isotopic variables has not been publ ished. This com parison.
combined with well·site has chromatography and hea d.s pace g'" ana lysl~. crl!ales a robust
analytic tool th at ca n hel p to overcome tile problems of data sufficiency and cos. associated
with r L,Jnning redundanr ana lyses.
offsite and returned as a spreadsheet with composition and isotope data.

Wi re line Formation Testers (WFTs)

Isolub.s
,,-lil Ul

o
813 C2%0 Hydrocarbon Gas Composition Correction Table
Aif values in m()l~ f or ~omp(J(iS() n with wirefine / ormCJtion fester samples

Downhole tools have the abilit y to sample pressurized formation fluid Wellslte GC
~
Wellsite GC e, ne. ie s nes
Motivation and Objecti ves
Analaysi s of su bsurface gases has routinely been performed during
direct ly fro m t he reservoir. Hyd ra ulic pistons push the probe, sea led wit h a
packe r, t hrough the mudca ke and and into the formati on. Fluid is then
.. .. " .. ] wn .,
IU
cr:
U
., ., .,
Isotube -7% +1%

drilling since t he earliest days of mud logging. It has evolved such pumped t hrough t he tool until co ntamination fro m invaded fi ltrate is Cl.
10 o Wellsite GC -5% +4% +1%
lowered to an acce ptable leve l. Then, a vessel in t he tool is fil led with th e

~ 13
that gas composition and isotopic ratios are powerful tools to
evaluate what the drillbit is encountering in the subsurface and
what t he implication s are for the petroleum system understand ing.
Headspace Gas (Isojars)
formation fl uid and t hen sent offsite for co nt rolled PVT analysis. IAy.n''''.. 20'',
"-19
Isotu b.. s
8 C3%0 Head spa ce +24% -3% -10% ·3% -4% -2% -2%
Wells IIGe
"'!"'1-'~DI~
o ~I---------<
WlIIIlIsit .. GC
Retu rn ing drilling fluid contains cutti ngs from the drilling process that are se parated before it is
1·' Ma nkiewicz et al., 2009
~~~~:!!!~ " -20 no. No correction recommended for isotopic roNos/
- " ,-----------=--::-.
l_ •·
and others have used
co mpound specifi'c
recirculated back down th e drillpipe. These cutti ngs contain fo rmation gases that will desorb in
ti me. By scooping a cutti ngs sample from the shale shakers into a small sealed cont ainer, such as
c
o
,. " 16 .. 12 10 1 WFT -, ., .J .0

an Isojar, a desorbed gas sample from formation rock fragments can be obtain ed f rom the emptv ~ Recommended workflow:
l _~ Hudspaca GIIS

~ I
isoptope analysis (CSIA)
space in t he top and analyzed in an offsite lab. This method is relatively inex pensive bu t is
.iii JI- }6
~ _" for fingerprinting/ o
subject to depth uncertain ty and samp ling (scooping) bias. Cl. 1. Check notes on indiv id ual samples and remove obvious outliers due to poor
~ =:: Isotubes

~
connectivity assessmen t Isotu be,

'0 .... and hydrocarbon Isotubes

E
o
"-19 ·-5 sa mpl e han d ling or sh ipping.
J.> ...
;;. ....
co
t hermal
determination .
matu rity While ret urned drilling fluid and suspended cuttings are arriving to t he shake shakers, a vacuum
line pulls t he libera ted formation gas fro m the fluid through a vacuum line to t he mud logging
u
Ul
II W. lI sita GC
J,- Jj
2. Observe isotopic values f or sample range - met hane carbon isotope ratios less
Ou ",mponent
unit . These gases, th at are dissolved In fl uid under high pressure during drill ing, are t hen
diverted at planned intervals into a small tube-shaped vessel. The depth (or de pt h interval) is
IU
e" II 16 12 10 2 Wfl -, -6 ... · 10 ·n ·14 , '6 · 1' .~ .5O .", -10 I. OJ . t han -60pp m may need less o r no co r rectio n (these are likely to be m icr obi al in
o rigin,.
c
Schoell, 1983 developed '~ t .~. calcula ted from the mud return lag tim e and written on the tube. The are placed in a box and, o 0 1 f-- Headspace Gas
1'1'· 14
met hods for determing _ when fu ll, shipped to an offs ite lab fo r analysis.
.0 Lege nd 3 . Apply the recomm ended correcti on fo r t he sample t ype and specific
L..
D Headspace Gas
/ ' ..-- !.~ b~
IU
origin (b,o- vs. thermo-! ·e ... - -
genic), maturity, and ~.~ ~-------
, r
~_
Wellsite Gas Ch romatography (Mud log) U
e
botu bes
" - 19 D
D
Isotubes ,---j~ /d-+-- -/"'7
'-'\r---=...\----"...."'<:::""'------7--:::~-'
com poun d and check resu lts (o bserve quarti le ranges as w e " ' .

~
•• , I Th e vacuum line (typ,cally the same line t hat feeds the Isotube manifold) al so run s to a gas Wellsite GC \
degrees of genetic "0 Wetl site GC 4. Compare data with WFTs or w ith each other_
Maximum 3rd Quarti le Median Mean 1st Quartil e Minimu m
mixing using carbon and I- ch romat ograph In t he mud logging Unit . This 's th e only gas sample collectio n and analys is t hat is done >-
hydrogen isotopes and
on-site. Th e typ,cal wellSi te GC will report methane through pe ntanes, bu t some advanced units now
have in·line mass spectro meters t o report through decanes as we ll as simple aromatics and alkenes.
I
,. II 1& .. 12 10 ... .a .10 -n -1" ." - 1& ..}O
- In general, wel lsite GC a nd Isotube sam ples are compositional ly more con si stent with WFTs t han
gas composition.
, Analyses are reported continu ously while drilling and are regularly calibrated. Interpretation of the real - Hu dspaee Gas headspace gas sam ples.
References
Abl ard, P., Be ll, C., Co ok, D., Forn asie r, 1.,Poyel, J.P., Sharma, S., Field ing, K' J Lawto n, lo,
I ~ ti me data ' 5 we ll established an d ,s commonly reported on petrophYSICal logs. 10bl"d."I.. 2012: " .warth ...,. , . ."
11· 16
Haines, G., Herkommer, M ., M cCarthy, K., Rada kovic, M., Umar, l ., 2012. The Expanding
In 1996, Berner and Faber published a method - Isot ube and wellsite GC samples are consist ent w ith each other; Isotub es m ay have a slightly tighter Role of Mud Logging. Schlumberger Oil field Review 24, no. 1. 24-41
Dat aset an d St atistical Methods Ayan, C., Co rre, P.Y., Firin u, M., Garcia, G., Kristense n, M" O' Keefe, M" Pfeiffe r, T., Tevis, c.,
for determing t he Ro% of the source ro ck at dist ributio n but t he mean is no t con sistentl y closer to W FT samples .
Zappalorto, L., Zeybek, M ., 2013 . New Dimensions in Wireline Formation Testing.
th e time of expUlsion of associated gases bV A proprietary dataset of deepwater wells wa s use d for thi' study. A tota l of 20 individual sample points were incl uded We ll site GC
Schlumberger Oilfield Review 25, no.l. 32-41
,,-IS
cross·plotting ca rbon isotopic ratios of across 12 di fferen t organ ic compaunts/isotopic ratios resulting in 240 individual samples. Nine individual wells were - By far, the great est compositional sam ple bias i s in m etha ne concentratio n (note scale' . Bernard. B.B., J.M . Brooks. W.M . Sackett, 1978.Light hydrocarbons in recent Texas
methane, ethane, and propane. This included in the dataset. -I · 10 ·11 _1.1 · 16 _11 conti nental shelf and slope sediments. J. Geophys. Res., 83, 4053-4061
-Headspa ce gas under-represents m et hane concentration by a median value of 24%.
An assumption was made that downhole sampling tools (or wirel ine formation testers · "WFT") produce the most Berner, U., Faber, E., 1996. Empirical ca rbon isotope/m aturity relationshi ps fo r gases from
'''1 ,t,' ~ methodologV can prove especially useful fo r -Isotubes and w e"sit e GC over-represent meth ane by m edian values of 5% and 4%, respectively. algal kerogen s and terrigenous orga ni c matter, based on dry, open -system pyrolysis, Organic
representative samples of true formation fluids (see Data Sources section for additional information). In order to protect
....~"l calibrati ng basin and petro leum syst em
the proprietary nature of t he data and to ca librate the representativeness of t he additio nal three sampling methods, the
Geochemistry 24, 947- 955
models, bu t the isotopic ratios of t he kerogen Chung, H. M ., J. R. Gormly, and R. M . Squires, 1988, Origin of gaseous hyd rocarb ons in
depth and associated va lues of t he downhole samples were normalized to zero and the additional th ree sampli ng b otubes - Becau se met hane is over- o r under-represented, rep resentation of ano t her co mpou nd w ill be skewed.
is needed for t he most effective analysis. subsurface environment: theoretical considerations of carbon isot ope distribution: Chemica l
,,· /9
methods are represe nted as positive or nega tive distributions of values about this normalized downhole sample value. -Headspace gas over-represent s propan e the most, by a median value of 10%. Geology, v. 71, p. 97-103
A generalized depth uncertai nty has been ascribed to the various sampling methods that incl udes +/- 20 reporte d feet I I WeUsite GC -Isotubes and w e"site GC under-repre sent ethane t he most by median va lues of 4%. Haworth, LH., Sellens, M ., and Whittaker, A. 1985. Interpretation of Hydrocarbon Shows
Using Light (Cl ·C5) Hydrocarbon Gases from Mud-Log Data. AAPG Bulletin 69 (8): 1305-1310
These are onIV a few of t he published t ools for hvdrocarbon gas both sha llower and deeper t han the downhole sample. This is intended to address wireline stretching or misreporting of ,,·15
Mankiewicl, P.J., Pottorl, R.J., KOlar, M.G., Vrolijket, P., 2009. Gas geochemistry of the
analysis (notable others include Bernard et aI., 1979 and Chung et the downhole ,ample as well as uncertainty of the lag time associated with mud returns. There is no weighting of the 20 II 1& 11 10 , wtT -2 "" -I -10 ·n -1" . 16 - 1' .10
- A" carbon isotopic ratios appear to be represented with great accuracy by Isotubes. Mobile Bay Jurassi c Norphle t Formation: t hermal controls and implications for reservoir
aI., 1988). The mi ssing piece of t he ana lyses lie in the inferential individual samples within the 40 foo t window. Due to the great var iation in rate of penetration, mud weight balance, co nnectivity. American Associati on of Pet roleum Geologist s Bulletin 93, 1319-1346
Hud spaee Gas
un certainty associated with di fferen t sampling methods. Those reservoir permeability, and other fact ors between wells and individ ual samples, the following data condition ing st eps Scho ell, M., 1983, Genetic characterization of natu ral gases: AAPG Bulletin, v. 67, p. 2225-
n - 16 - Isotop ic rati os f ro m headspace ga s may also be consist ent with W FTs, but more samples are needed. 2238
engaged in petroleum systems anlavses are often working in areas were taken:
of extremely limi ted and poorly documented data, but the scarCity -Wellsite GC: A 40 foot moving average was taken from the .Ias fi le. The downhole sample depth was used. Isotub.,

of t he data necessitate inclusion of all types. This study seeks to - Isotu bes: Individua l reported depths within +/. 20 feet of downhole samples were included . Interval depths were
,,-19 - Hydrogen isotopi c r.l tio s may be gene rall y accurat e w it h slightly more uncerta inty. Ac know ledgements
overcome the individual limitations of such data by taking a averaged and included as samples whe n the averaged depth was within +/. 20 feet. Wells' II GC
The au thor would li ke to than k Ka isoon Tan and Nat han Cartw right for thei r t houghts on
- Beca use outliers due t o container failures m ay skew t he mean, med ian val ues appear to be more formation evaluation and Nexen Petroleum USA, Inc. for permission to take on t his stu dy and
statistical approach and documenti ng t he uncertainty from a -Headspace Gas: The same technique was used as Isotube samples. For slightl y broader interval spacing (i.e. <SO feet). present the data . Additionally, thanks is always in order for the aut ho r's coworkers and
dataset of different plays, ages, depths, and fluid qualiti es. the values from the in terval coveri ng the downhole sample were used. 20 IS 16 12 10 -6 -I . 10 .n . 1-1 .'6 11 .20 repre sentative as a central tenden cy. family who inspire, teach, and make allowances,