the lowdown on low-resistivity pay -...
TRANSCRIPT
4
The Lowdown on Low-Resistivity Pay
Austin BoydHarold DarlingJacques TabanouSugar Land, Texas, USA
Bob Davis Bruce LyonNew Orleans, Louisiana, USA
Charles FlaumRidgefield, Connecticut, USA
James KleinARCO Exploration and Production TechnologyPlano, Texas
Robert M. Sneider Robert M. Sneider Exploration, Inc. Houston, Texas
Alan SibbitKuala Lumpur, Malaysia
Julian SingerNew Delhi, India
For help in preparation of this article, thanks to JayTittman, consultant, Danbury, Connecticut, USA; Bar-bara Anderson, Ian Bryant, Darwin Ellis, Mike Herron,Bob Kleinberg, Raghu Ramamoorthy, Pabitra Sen, Chris Straley, Schlumberger-Doll Research, Ridgefield, Connecticut; David Allen, Kees Castelijns, Andrew Kirk-wood and Andre Orban, Schlumberger Wireline & Test-ing, Sugar Land, Texas, USA; Steve Bonner and TrevorBurgess, Anadrill, Sugar Land, Texas; Dale Logan,Schlumberger Wireline & Testing, Roswell, New Mexico,USA; and Pierre Berger, GeoQuest, Bangkok, Thailand.
Evaluating low-resistivity pay requires interpreters to discard the notion
that water saturations above 50% are not economic. Various tools and
techniques have been developed to assess these frequently bypassed
zones, but there are no shortcuts to arriving at the correct petrophysi-
cal answer.
■■Clays are the pri-mary cause of low-resistivity pay andcan form duringand after deposi-tion. They are dis-tributed in the for-mation as laminarshales, dispersedclays and struc-tural clays. Othercauses of low-resis-tivity pay includesmall grain size, asin intervals ofigneous and meta-morphic rock frag-ments, and con-ductive mineralslike pyrite.
Lamination of beds Shale clasts Clay-lined burrows
Pore fillings Pore linings Clay grains
Burrowed sand Ash shards Conductive pyrite
0.25 mm
0.5 in
When Conrad and Marcel Schlumbergerinvented the technique of well logging, low-resistivity pay was, practically speaking, acontradiction in terms. Their pioneeringresearch hinged on the principle that gas- oroil-filled rocks have a higher resistivity thanwater-filled rocks. Through the years, how-ever, low-resistivity pay has become recog-nized as a worldwide phenomenon, occur-ring in basins from the North Sea andIndonesia to West Africa and Alaska. With
low oil prices driving the reexploration ofmature fields, methods of interpreting low-resistivity pay have proliferated.
This article examines the causes of low-resistivity pay in sands, then explores thetools and techniques that have been devel-oped to evaluate such zones. A case studyshows how log/core integration helps pin-point the causes of low-resistivity pay in theGandhar field in India.
Generally, deep-resistivity logs in low-resistivity pay read 0.5 to 5 ohm-m. “Low
Oilfield Review
In this article, AIT (Array Induction Imager Tool), ARC5(Array Resistivity Compensated), CBT (Cement BondTool), CDR (Compensated Dual Resistivity tool), CMR(Combinable Magnetic Resonance tool), CNL (Compen-sated Neutron Log), DLL (Dual Laterolog Resistivity),ELAN (Elemental Log Analysis), EPT (ElectromagneticPropagation Tool), FMI (Fullbore Formation MicroIm-ager), Formation MicroScanner, GeoFrame, GLT (Geo-chemical Logging Tool), Litho-Density, IPL (IntegratedPorosity Lithology), MicroSFL, NGS (Natural Gamma RaySpectrometry tool), Phasor, RAB (Resistivity-at-the-Bittool), SFL (Spherically Focused Resistivity), SHARP (Synergetic High-Resolution Analysis and Reconstructionfor Petrophysical Parameters) and TDT (Thermal DecayTime) are marks of Schlumberger. Sun is a mark of SunMicrosystems, Inc.1. Moore D (ed): Productive Low Resistivity Well Logs
of the Offshore Gulf of Mexico. New Orleans,Louisiana, USA: Houston and New Orleans Geologi-cal Societies, 1993.
2. Scala C: “Archie III: Electrical Conduction in ShalySands,” Oilfield Review 1, no. 3 (October 1989): 43-53.
3. One milliequivalent equals 6 x 1020 atoms.
■■The most com-mon depositionalenvironments forlow-resistivity pay: A) Lowstand basinfloor fan complexes B) Deep waterlevee-channelcomplexes andoverbank deposits C) Transgressive-marine sandsD) Lower parts(toes) of delta frontdeposits and lami-nated silt-shale-sand intervals inthe upper parts ofalluvial and dis-tributary channels. (Adapted from Dar-ling HL and SneiderRM: “Productive LowResistivity Well Logs of the Offshore Gulf of Mexico: Causesand Analysis,” in reference 1.)
Lowstand basinfloor fan complex
A
B
C
D
Leveed channelcomplex
Transgressivemarine sands
Alluvialchannel
Distributarychannel
Delta front “toes” andshingled turbidites
Overbankdeposits
Overbank
contrast” is often used in conjunction withlow resistivity, indicating a lack of resistivitycontrast between sands and adjacent shales.Although not the focus of this article, low-contrast pay occurs mainly when formationwaters are fresh or of low salinity. As aresult, resistivity values are not necessarilylow, but there is little resistivity contrastbetween oil and water zones.
Because of its inherent conductivity, clay,and hence shale, is the primary cause oflow-resistivity pay (previous page).1 How
Autumn 1995
clay contributes to low-resistivity readingsdepends on the type, volume and distribu-tion of clay in the formation.
Clay minerals have a substantial negativesurface charge that causes log resistivity val-ues to plummet.2 This negative surfacecharge—the result of substitution in the claylattice of atoms with lower positive valence—attracts cations such as Na+ and K+ whenthe clay is dry. When the clay is immersedin water, cations are released, increasing thewater conductivity.
The cation exchange capacity, or CEC,expressed in units of milliequivalent3 per100 grams of dry clay, measures the abilityof a clay to release cations. Clays with ahigh CEC will have a greater impact on low-ering resistivity than those with a low CEC.For example, montmorillonite, also knownas smectite, has a CEC of 80 to 150meq/100 g whereas the CEC of kaolinite isonly 3 to 15 meq/100 g.
Clays are distributed in the formationthree ways:• laminar shales—shale layers between
sand layers• dispersed clays—clays throughout the
sand, coating the sand grains or filling thepore space between sand grains
• structural clays—clay grains or nodules inthe formation matrix. Laminar shales form during deposition,
interspersed in otherwise clean sands (left).In the Gulf Coast, USA, finely layered sand-stone-shale intervals, or thin beds, make up
5
Spherically FocusedResistivity
Evaluated Gas Pay Potential Gas Pay
-160 40Spontaneous Potential
0.2 20
Deep Induction0 150
Total Gamma RayGAPI ohm-m
0.2 20ohm-mCompensated
Neutron Porosityp.u.60 0
60 0
Density PorosityMDEN=2.68
Dep
th, m
X100
-160
X200
40
Spontaneous Potential
0.2 206FF40 Induction
Short Normal Resistivity
ohm-m0.2 20
■■Left: Induction Electrical Survey logs run in 1960 in a thinly bedded, gas-bearing section of the Vicksburg formation in south Texas,USA. Net pay is 7 ft. Right: Conventional triple combo—neutron, density and gamma ray tools—run in 1993 in a well offset 100 ftfrom the original 1960 well. Net pay is 14 ft.
about half the low-resistivity zones.4 Manylogging tools lack the vertical resolution toresolve resistivity values for individual thinbeds of sand and shale. Instead, the toolsgive an average resistivity measurement overthe bedded sequence, lower in some zones,higher in others.
Intervals with dispersed clays are formedduring the deposition of individual clay par-ticles or masses of clay. Dispersed clays canresult from postdepositional processes, suchas burrowing and diagenesis. The size differ-ence between dispersed clay grains andframework grains allows the dispersed claygrains to line or fill the pore throats betweenframework grains. When clay coats the sandgrains, the irreducible water saturation ofthe formation increases, dramatically lower-ing resistivity values. If such zones are com-pleted, however, water-free hydrocarbons
6 Oilfield Review
can be produced (see “Low-Resistivity Payin the Gandhar Field,” page 8).
Structural clays occur when frameworkgrains and fragments of shale or claystone,with a grain size equal to or larger than theframework grains, are deposited simultane-ously. Alternatively, in the case of selectivereplacement, diagenesis can transformframework grains, like feldspar, into clay.Unlike dispersed clays, structural clays actas framework grains without altering reser-voir properties. None of the pore space isoccupied by clay.
Other causes of low-resistivity pay includesmall grain size and conductive mineralslike pyrite. Small grain size can result in lowresistivity values over an interval, despiteuniform mineralogy and clay content. Theincreased surface area associated with finergrains holds more irreducible water, and, aswith clay-coated grains, the increasingwater saturation reduces resistivity readings.Intervals of igneous and metamorphic rock
fragments—all fine grained— mimic the logsignature of clays, featuring high gamma ray,low resistivity and little or no spontaneouspotential (SP). Unlike thin beds, this type oflow-resistivity pay can vary in thicknessfrom millimeters to hundreds of meters.
Finally, sands with more than 7% by vol-ume of pyrite, which has a conductivitygreater than or equal to that of formationwater, also produce low-resistivity readings.5This type of low-resistivity pay is consideredrare.
The challenge for interpreting low-resistiv-ity sands hinges on extracting the correctmeasurement of formation resistivity, esti-mating shaliness and then accurately deriv-ing water saturation, typically obtained fromsome modification of Archie’s law.6
Improved vertical resolution of loggingtools and data processing techniques arehelping to tackle thin beds. Nuclear mag-
4. Thin beds have a thickness of 5 to 60 cm [2 in. to 2ft] and laminae are less than 1-cm [0.4-in.] thick,commonly 0.05 to 1 mm [0.002 to 0.004 in.].Bates RL and Jackson JA (eds): Glossary of Geology.Falls Church, Virginia, USA: American GeologicalInstitute, 1987. Dictionary of Geological Terms. New York, NewYork, USA: Doubleday & Co., 1984.
5. Clavier C, Heim A and Scala C: “Effect of Pyrite onResistivity and Other Logging Measurements,” Transactions of the SPWLA 17th Annual LoggingSymposium, Denver, Colorado, USA, June 9-12,1976, paper HH.
6. In 1942, Gus Archie proposed an empirical relation-ship linking a rock’s resistivity, Rt, with its porosity, φ , and water saturation Sw :
.
Other terms in the equation are the formation waterresistivity Rw, and the cementation and saturationexponents, m and n. For further reading:“Archie’s Law: Electrical Conduction in Clean,Water-Bearing Rock,” The Technical Review 36, no. 3 (July 1988): 4-13.“Archie II: Electrical Conduction in Hydrocarbon-Bearing Rock,” The Technical Review 36, no. 4(October 1988): 12-21.For a discussion on the numerous versions ofArchie’s law that have been developed to handle avariety of shaly sand environments:Worthington PF: “The Evolution of Shaly-Sand Con-cepts in Reservoir Evaluation,” The Log Analyst 26(January-February 1985): 23-40.
7. Barber TD and Rosthal RA: “Using a MultiarrayInduction Tool to Achieve High-Resolution Logs withMinimum Environmental Effects,” paper SPE 22725,presented at the 66th SPE Annual Technical Confer-ence and Exhibition, Dallas, Texas, USA, October 6-9, 1991.Hunka JF, Barber TD, Rosthal RA, Minerbo GN,Head EA, Howard AQ Jr and Hazen GA: “A NewResistivity Measurement System for Deep FormationImaging and High-Resolution Formation Evaluation,”paper SPE 20559, presented at the 65th SPE AnnualTechnical Conference and Exhibition, New Orleans,Louisiana, USA, September 23-26, 1990.
8. FMI* Fullbore Formation MicroImager. Houston,Texas, USA: Schlumberger Educational Services,1992.
9. Olesen J-R, Flaum C and Jacobsen S: “WellsiteDetection of Gas Reservoirs with Advanced Wire-line Logging Technology,” Transactions of theSPWLA 35th Annual Logging Symposium, Tulsa,Oklahoma, USA, June 19-22, 1994, paper Y.
■■AIT Array Induction Imager Tool and IPL Integrated Porosity Lithology logs runin the same well as conventional triple combo on previous page. The improved vertical resolution of AIT logs and the enhanced sensitivity of the IPL-derivedneutron porosity have increased net pay to 63 ft.
Potential Gas PayD
epth
, m
X100
X200
AIT Resistivities10-90 in.
0 45
HNGS Thorium Content(HTHO)
ppm
0.2 20ohm-m
-20
60 0
Neutron PorositySandstone (APSC)
0 5p.u.
HNGS Potassium Content(HFK)
10 40c.u.
APS CaptureCross-Section (SIGF)
-160 40
Spontaneous Potential
MV
20
Differential Caliper
p.u.
60 0
Density PorosityMDEN=2.68 (DPO)
p.u.
Rt = Rw
φm Swn
netic resonance (NMR) logging showspromise for assessing irreducible water satu-ration associated with clays and reducedgrain size (see “Nuclear Magnetic Reso-nance Imaging—Technology for the 21stCentury,” page 19). And because the mostopportune time to measure resistivity occursduring drilling, when invasion effects areminimal, resistivity measurements at the drillbit also play an important role in diagnosinglow-resistivity pay. Thin Beds
Autumn 1995
One obvious method for resolving the resis-tivity of thin beds is to develop logging toolswith higher vertical resolution, deeper depthof investigation, or both. Two loggingdevices that have proved especially helpfulin evaluating thin beds are the AIT ArrayInduction Imager Tool and the FMI FullboreFormation MicroImager tool. The AIT tooluses eight induction-coil arrays operating atmultiple frequencies to generate a family offive resistivity logs.7 The logs have mediandepths of investigation of 10, 20, 30, 60 and90 in. and vertical resolutions of 1 ft [0.3 m],
2 ft [0.6 m] and 4 ft [1.2 m]. The FMI toolimages the borehole with an array of 192button sensors mounted on four pads andfour flaps.8 It has a vertical resolution of0.2 in. [5 mm].
Successive improvements in resolving thinbeds are strikingly visible in a series of logsmade 33 years apart in adjacent wells in thesouth Texas Vicksburg formation (previouspage and left ).9 In 1960, induction/ shortnormal logs indicated 7 ft of net gas pay andonly two beds with resistivity greater than 2ohm-m. In 1993, a new well was drilledwithin 100 ft [30 m] of the original well andlogged with conventional wireline tools.The induction/SFL Spherically FocusedResistivity logs doubled the estimated payto 14 ft [4.3 m], with seven beds above 2
(continued on page 11)
7
The Gandhar field, on the western coast of India,
is the largest on-land field in the country (left).
Most hydrocarbon production comes from deltaic
sands of the Hazad member, three of which con-
tain low-resistivity pay.
One of these sands, called GS-11, has resistiv-
ity values of 2 to 6 ohm-m, but contains wells
that produce clean oil on the order of 50 m3/d
[315 B/D] (next page). A detailed study of GS-11,
integrating core and log data, allowed inter-
preters to unravel the low-resistivity phenomenon
and formulate a reliable mineralogical model and
water saturation estimates.
Core Studies
Sixty core samples from three GS-11 wells pro-
vided thin sections for study of texture and miner-
alogy. Polished sections helped reveal the pres-
ence of metallic minerals. Scanning electron
microscope (SEM) and X-ray diffraction (XRD)
studies of cores identified clay minerals. In addi-
tion, laser and sieving methods were used to
analyze grain size.
The core investigations showed several mech-
anisms contributing to high conductivity.
Medium- to fine-grained sands ranged from gray
to green-gray, with green indicating chloritic
miles
Delhi
I N D I A
Gandhar
Dabka
Khambhat
Dhadhar River
Narmada River
Mahisaga River
G U L F O FC A M B A Y
0 km 25
0 15.5
Low-Resistivity Pay in the Gandhar Field
■■SEM photographs showing coated grains and clay matrix (left) and quartz overgrowth with chlorite coatingon quartz grains (right).
■■Gandhar field on the westerncoast of India.
8 Oilfield Review
Quartz OvergrowthClay Coating
20 µm200 µm
clays. Bioturbation created thin, fine clay lamina-
tions over clean sands. Quartz was the most
prominent mineral, with minute opaque
minerals—pyrite or magnetite—occurring in
bioturbated sections. Pyrite, which increases
the formation conductivity, was limited to the
clayey part of the matrix and constituted less
than 5% by volume.
Clay, primarily chlorite, coating the grain sur-
faces was indicated by SEM pictures and XRD
studies (previous page, bottom). Smaller grains
were coated more than larger grains. Laser
analysis of samples shows the GS-11 sand to
be in the silt range, with grain sizes averaging
22 to 32 microns.
Formation Evaluation
Logs were analyzed to identify clay types and
heavy minerals. Thorium-potassium crossplots of
the NGS Natural Gamma Ray Spectrometry logs
identified predominant clays as chlorite in the
sands and kaolinite/chlorite in the shales. The
density-neutron crossplot showed a trend toward
high density (low porosity) with little increase in
the neutron. The particles associated with this
behavior, which included fine-grained quartz and
heavy minerals such as siderite, pyrite and
ilmenite, were collectively called silt.
From core- and log-derived information, a min-
eralogical model of kaolinite, chlorite, quartz and
silt was chosen for the GS-11 sands. Validation
for the model came from geochemical analysis of
21 core samples from different wells. A few sam-
ples were analyzed to determine the weight per-
cent of oxides, such as silicon dioxide [SiO2],
using X-ray fluorescence (XRF) and the results
were interpolated between samples. The percent-
ages were then converted into weight percent of
elements using standard tables and processed
9Autumn 1995
■■Log response from Well Z shows an average resistivity reading of 3 to 4 ohm-mover the GS-11 sand, which produced clean oil during conventional testing.
GS-
11 s
and
Dep
th, m
XX80
XX90
X100
X110
1.95
45 -15
2.95
20000.26 16
-25 125
0 150
Deep Resistivity
Shallow Resistivityohm-m
Gamma Ray
SP
Caliper
GAPI
MV
in.
Density
Neutron Porosityp.u.
g/cm3
with a mineralogical model to give weight per-
cent of minerals. The model based on geochemi-
cal analysis was constrained to include only
quartz, kaolinite, chlorite and ilmenite. This
constraint allowed the weight percent of minerals
to be converted to volume percent using the
total porosity from log interpretation and the
mineral densities.
Comparison of the log and XRF mineral analy-
ses shows agreement between the total clay per-
centage and the relative volume of kaolinite and
chlorite (left). The silt and ilmenite percentages
do not agree, as might be expected since the silt
was defined to include finer grained quartz.
Conclusions
The composite results from the extensive log-
core analysis show agreement between core- and
log-derived parameters (next page). Water satu-
ration values computed from the Waxman-Smits
equation compare well with those derived from
capillary pressure measurements.1 Because little
water had been produced from existing GS-11
wells, the log-derived water saturation values
were considered to represent irreducible water
saturation values.
The core studies showed that the low-resistiv-
ity measurements in the GS-11 sand have two
sources. First, individual sand grains are coated
with clay. Second, the silt-sized formation grains
lead to higher irreducible water saturations in
the formation.
10 Oilfield Review
Free Water
Quartz
Silt
Bound Water
Chlorite
Quartz
Chlorite
Bound Water Bound Water
Ilmenite Ilmenite
Kaolinite
Quartz
Chlorite
KaoliniteKaolinite
Free Water
1:100 m
XX54
XX56
XX58
XX60
XX62
Sandstone Coarsesandstone
Bioturbatedsandstone
Silty carbon-aceous shale
Laminatedsilty shale Shale
Dep
th, m
Cor
eD
escr
iptio
n
Log AnalysisWeight % of Minerals from
XRF Analysis of Oxidesfrom Cores
Volume % of Mineralsfrom XRF Weight %
and Log Porosity
■■Comparison between log and XRF mineral analyses of Well Y. A mineralogical model of kaolinite, chlorite,quartz and silt was chosen.
1. Waxman and Smits modified Archie’s law to account forthe increased conductivity of shale by introducing a shali-ness parameter based on cation exchange capacity (CEC).See: Waxman MH and Smits LJM: “Electrical Conductivi-ties in Oil-Bearing Shaly Sands,” Society of PetroleumEngineers Journal 8, no. 2 (1968): 107-122.
ohm-m. Later the same year, the secondwell was logged with a combination of AITand IPL Integrated Porosity Lithology tools.10
The high resolution of the AIT tool—1 ft ver-sus 2 ft for the induction—and theenhanced sensitivity of the IPL-derived neu-tron porosity increased net pay to 63 ft [19.2m] and showed 13 beds with resistivitygreater than 2 ohm-m.
Resistivity Measurements at the Bit Improvements in measurements-while-drilling (MWD) technology have not onlyboosted the efficiency of directional drilling,but also enhanced thin-bed evaluation.11
Two tools‚ the RAB Resistivity-At-the-Bit tooland the ARC5 Array Resistivity Compen-sated tool—are especially useful in thin-bedenvironments by providing resistivity databefore invasion has altered the formation.
The RAB tool provides five different resis-tivity readings plus gamma ray, shock andtool inclination measurements. Configuredas a stabilizer or a slick collar, the RAB toolis run behind the bit in a rotary drillingassembly and above the motor in a steerabledrilling assembly.
One resistivity measurement, called “bitresistivity,” uses the drill bit as part of thetransmitting electrode. With the RAB toolattached to the bit, alternating current is cir-culated through the collar, bit and formationbefore returning to the drillpipe and drillcollars above the transmitter. In the case ofoil-base mud, which is an insulator, the cur-rent loop is complete only when the collarsand stabilizers touch the borehole wall. Thevertical resolution of the RAB bit resistivity isonly 2 ft and it gives the earliest possiblewarning of changes in formation resistivity.
Four additional resistivity measurements,with 1-in. vertical resolution for thin-bedapplications, are made with three buttonelectrodes and a ring electrode. The shallowdepths of investigation—3, 6 and 9 in. for
11Autumn 1995
Qv from Logs
Qv fromCo vs Cw
Qv fromWet Chemistry
m fromEPT/MicroSFL
m=f(Qv from Logs)
m fromCo vs Cw
BadHoleFlag
5.0 0.0
1:200 m 1.0 3.0φ from Core
Combined Model
Kaolinite
Chlorite
Bound Water
Silt
Quartz
Moved Water
Water
Oil
Fluid Analysisp.u.
MovedHydrocarbon
p.u.
Sw fromWaxman-Smits
p.u.
Sw fromArchie
Swirr fromCap Studies
2.00 100 0 50 100 100 0
0 100
1.0 3.0
1.0 3.0
2.00
2.00
100 0
100 0
x780
x790
x800
x810
GS-11
GS-10
■■Composite log-core analysis of Well X. Core results are shown for the cementation exponent m; the CECnormalized for pore volume, Q v, irreducible water saturation Swirr and porosity φ. Q v, the CEC per volume ofpore fluid, was calculated from cores, by measuring resistivity at different water salinities, and from logs.
10. “Neutron Porosity Logging Revisited, ”OilfieldReview 6, no. 4 (October 1994): 4-8.
11. Bonner S, Burgess T, Clark B, Decker D, Lüling M,Orban J, Prevedel B and White J: “Measurements atthe Bit: A New Generation of MWD Tools,” Oil-field Review 5, no. 2/3 (April/July 1993): 44-54.Allen D, Bagersh A, Bonner S, Clark B, Dajee G,Dennison M, Hall JS, Jundt J, Lovell J and RosthalR: “A New Generation of Electrode Resistivity Mea-surements for Formation Evaluation WhileDrilling,” Transactions of the SPWLA 35th AnnualLogging Symposium, Tulsa, Oklahoma, USA, June19-22, 1994, paper OO.
the buttons and 12 in. for the ring electrode—allow interpreters to characterize early-time invasion (left).
The recently-introduced ARC5 tool pro-vides five phase and attenuation resistivitymeasurements, like the AIT tool, with a verti-cal resolution of 2 ft. With a 43/4-in. diame-ter, it is especially useful for formation evalu-ation in slim holes typical of deviated drilling(next page, top).
The measurements and spacings of theARC5 and AIT tools are comparable,although not identical, making petrophysi-cal evaluation with either tool in the samewell or between wells seamless (below left).The multiple measurements of the ARC5tool also allow interpreters to radially mapout the invasion process. The additionalphase and attenuation measurements pro-vide a better characterization of electricalanisotropy than existing MWD tools.
Improving Thin-Bed Evaluation ThroughData ProcessingDespite the emphasis on developing high-resolution resistivity logging tools, manyopenhole tools still have a vertical resolutionof 2 to 8 ft [0.6 to 2.4 m]. Several data pro-cessing techniques have been developed toenhance the vertical resolution of these tradi-tional tools (next page, bottom). All methodsuse at least one high-resolution measure-ment to sharpen a low-resolution one andrequire a strong correlation between the two.An existing technique helpful in interpretinglow-resistivity pay is Laminated Sand Analy-sis (LSA), a computer program for evaluatingthe shaliness, porosity and water saturationin beds as thin as 2 in. [4 cm].12
A newer approach for identifying andevaluating thin beds is the SHARP Syner-getic High-Resolution Analysis and Recon-struction for Petrophysical Parameters soft-ware. SHARP processing improves theresolution of log inputs to the ELAN Elemen-tal Log Analysis module, thereby improvingsaturation and reserve estimates. Currently,SHARP software exists as an interactive,stand-alone prototype application for Sunworkstations but a second generation ver-sion will be incorporated into the GeoFramereservoir characterization system by the endof 1995.
12 Oilfield Review
12. Allen DF: “Laminated Sand Analysis,” Transactionsof the SPWLA 25th Annual Logging Symposium,New Orleans, Louisiana, USA, June 10-13, 1984,paper XX.
Res
istiv
ity, o
hm-m
100
0.2
Laminated wet sands
570 580 590 600
Distance, ft
AIT
MicroSFL
RAB ringafter drilling
RAB ringwhile drilling
Res
istiv
ity, o
hm-m
Res
istiv
ity, o
hm-m
101
102
101
102
400 450 500 550Depth, ft
90 in.60 in.30 in.20 in.10 in.
34 in.28 in.22 in.16 in.10 in.
ARC5 Phase Shift Resistivities at CAT Well
AIT Resistivities at CAT Well
■■Evaluating inva-sion with the RABtool. In laminatedwet sands, the RABlogs made afterdrilling and whiledrilling anticorre-late, showing pref-erential invasion.
■■Comparison of ARC5 log with the AIT log at 2-ft vertical resolution. The logs were runin the Customer Acceptance Test (CAT) Well in Houston, Texas, USA.
X800
X900
Dep
th, f
t
0 150
0500
ROP5
Gamma RayGAPI
ft/hr
0.2
ARC5 Phase Shift ResistivityBorehole Corrected
10 in. (P10H)
200ohm-m
34 in. (P34H)
28 in. (P28H)
22 in. (P22H)
16 in. (P16H)
-19
100
0
CCL
Borehole Sigma
c.u.
Gamma Ray
GAPI
cps
Far Detector Background
1.0
0
100 60
0.6
1500
TDT Porosity
Inelastic Count RateFar Detector
cps
0
0
Formation Sigma
c.u.
0
Near DetectorCount Rate
cps3000 0
Far DetectorCount Rate
cps1200 0
13Autumn 1995
■■ARC5 log run in wash down mode in front of thin gas stringers. Rough hole conditions precluded running wireline logs in the wellexcept for a CBT Cement Bond Tool log and a TDT Thermal Decay Time log. The TDT log confirmed the presence of gas indicated bythe ARC5 log.
Data Processing Methods for Enhancing Vertical Resolution
Technique
Enhanced PhasorProcessing (1988)
Enhanced ResolutionProcessing (1986)
Laminated SandAnalysis (1984)
Measurements
Phasor Induction log
Litho-Density log
CNL CompensatedNeutron Log
Triple combo(gamma ray, neutronand density), EPTElectromagneticPropagation Tool
Improvement in Resolution
From 7 ft to about 3 ft[2 to 1 m]
From 18 in. to 4 in.[45 cm to 10 cm]
From 24 in. to 12 in.[61 cm to 30 cm]
Down to 2 in. [5 cm]
Method
Medium-inductionmeasurement used to enhancedeep induction measurement
Near-detector measurementused to compensate for fardetector measurement
Computes bound water saturation(shaliness) from EPT tool, usedwith dual-water model toredistribute the measured inductionresistivity, yielding estimates ofthe resistivity of thin beds. Effectiveporosity, water saturationand permeability are computed.
Near-detector measurementused to compensate for fardetector measurement
1
■■Establishing bedboundaries andmodes (right) withthe SHARP programand a FormationMicroScanner log(left). SHARP analy-sis determines bedboundaries frominflection points on the secondderivative of anaverage FormationMicroScannerresistivity reading.Modes are estab-lished by groupingresistivity measure-ments on a his-togram (not shown).A square represen-tation of the aver-age FormationMicroScanner resis-tivity curve showsthe bed boundariesand modes.
1.0 100ohm-m
XX20
XX40
XX60
Square Average Resistivity
Average ResistivityFormation MicroScanner
Images
SHARP analysis relies on high-resolutioninputs, such as Formation MicroScanner,FMI or EPT Electromagnetic PropagationTool logs to define a layered model of theformation (right). The program looks at thezero crossings on the second derivative ofthe high-resolution log, where the slopechanges sign, to indicate bed boundaries. Inthe case of a Formation MicroScanner orFMI log, the SHARP program examines thesecond derivative of an average resistivityreading from all button sensors.
With bed boundaries established, SHARPanalysis plots a histogram of the frequencyof a particular resistivity value within thelogged interval of interest. By studying howresistivity values cluster, an interpreter cangroup the values into different populations,or modes. SHARP analysis assumes that allresistivity data in a particular mode comefrom the same kind of formation, and furtherthat the resistivity value in a particular modeis constant. In addition, SHARP evaluationassumes that petrophysical parameters such
4 Oilfield Review
Dep
th, f
t
XX40
XX60
XX80
X100
1.0 100ohm-m 1.0 100 1.0 100 1.0 100 1.0 100
SquareAverage
Resistivity
FormationMicroScanner
AverageResistivity
ohm-m
Original LLD
Resistivity Model
ohm-m
Original LLD
Reconstructed LLD
ohm-m
Original LLD
ohm-m
Original LLD
Reconstructed LLDRefined
Resistivity Model
■■Enhancing the reso-lution of deep lat-erolog measurements(LLD) of the DLL DualLaterolog Resistivitytool. The FormationMicroScanner aver-age resistivity curveis shown with itsSHARP-generatedaverage resistivitysquare log (far left).The bed boundariesand number ofmodes establishedby SHARP is used togenerate anenhanced-resolutionLLD curve. The mod-eled LLD measure-ment of the DLL toolis refined by compar-ing the original DLLlog with the recon-structed LLD log(middle) and interac-tively adjusting thebed boundaries andmode values toachieve a bettermatch (right).
13. Ramamoorthy R, Flaum C and Coll C: “GeologicallyConsistent Resolution Enhancement of StandardPetrophysical Analysis Using Image Log Data,paper SPE 30607, to be presented at the 70th SPE Annual Technical Conference and Exhibition,Dallas, Texas, USA, October 22-25, 1995.
14. Chapman S, Colson JL, Everett B, Flaum C, HerronM, Hertzog RC, La Vigne J, Pirie G, Quirein J,Schweitzer JS, Scott H and Wendlandt R: “TheEmergence of Geochemical Well Logging,” TheTechnical Review 35, no. 2 (April 1987): 27-35.
XX80
1
2
3
4
5
6
Mode number
-1 4in.DCAL
0 150GAPI
OriginalGamma Ray
EnhancedGamma Ray
Pad 1 Pad 2 Pad 3 Pad 4
FormationMicroScanner Images
0.2 200ohm-mOriginal AIT 60-in.
Enhanced AIT 60-in.
Original AIT 10-in.
60 0p.u.
Enhanced NeutronPorosity
Original NeutronPorosity
1.65 2.65g/cm3
Enhanced Density
Original Density
Bed Boundaries
1:120 ft
Dep
th, f
t
X1000
XX900
as density, neutron porosity and sonic veloc-ity are also constant in a given mode.
After establishing the number of beds andmodes in the logged interval—the “squarelog”—the SHARP program calculates a setof mode values that minimizes the differ-ence between the original and square logs.This model, a squared resistivity log of bedboundaries and mode values, is filtered withthe response function of a logging tool toproduce a synthetic, or so-called recon-structed, log (previous page, bottom). Themodel is refined by minimizing the differ-ence between the measured log and thereconstructed log. At a workstation screen,the log interpreter can interactively adjustthe boundaries and bed values of the modesto achieve a match.
When the synthetic and measured logsmatch, the model can be used as a high-res-olution input into the ELAN interpretation.To sharpen the resolution of other logs, suchas the gamma ray, the model of bed bound-aries determined previously is utilized toreconstruct other squared, enhanced logs forhigh-resolution formation evaluation. A low-resistivity example from the Gulf of Mexicoshows how SHARP analysis improvedreserve estimates by 28%, even whenapplied to AIT measurements and a high-res-olution triple combo of density, neutron andgamma ray logs (right and next page).
Rather than reconstruct logs using SHARPanalysis, Raghu Ramamoorthy and CharlesFlaum, of Schlumberger-Doll Research,Ridgefield, Connecticut, USA have devel-oped a simpler technique to enhance pro-ducibility and hydrocarbon content esti-mates made with conventional petrophysicalanalyses in thin beds.13 Working with logsfrom the GLT Geochemical Logging Tool,they picked a high-resolution clay indicator,either the FMI or EPT log, and calibrated it tothe clay volume derived from the GLT mea-surement. In addition to clay volume, theGLT tool combines nuclear spectrometrylogging measurements to determine mineralconcentrations and cation exchange capac-ity of the formation.14
15Autumn 1995
■■Comparison of original and SHARP-enhanced logs for a low-resistivity payexample from the Gulf of Mexico. Theinterval was logged with a high-resolu-tion triple combo. The original 10-in. and60-in. depth-of-investigation curves fromthe AIT log are shown with the enhancedAIT 60-in. log. Only the 60-in. curve wasenhanced because the SHARP prototypesoftware does not yet have the modeledtool response for other AIT measurements.An ELAN interpretation (next page) usingthe enhanced logs shows a 28% increasein estimated reserves.
10.00 1.66 0.20Resistivity, ohm-m
16
0 150
1:120 ft
Kaolinite
Bound Water
Quartz
p.u. 1000Volume Scale
Image 100 0 1000
Swp.u.
p.u.Volume Scale
GAPI
GammaRay
Montmorillonite
Hydrocarbons
Water
Moved Water
Irreducible Water
Illite
Kaolinite
Bound Water
Quartz
Montmorillonite
Hydrocarbons
Water
Moved Water
Irreducible Water
Illite
ELAN with High-Res Inputs ELAN with Enhanced InputsD
epth
, ft
XX900
X1000
respond to clays, the GLT-derived clay vol-ume—a low-resolution measurement—isenhanced by looking at local variations ofthe high-resolution FMI measurement. Thelow-resolution GLT clay volume is adjustedby the difference between the FMI-derivedclay volume and its value averaged over theresolution of the GLT tool, which is 2 ft:
With well data, an empirical relationshipis established between clay volume andporosity. This relationship is applied to theenhanced GLT clay volumes to derive high-resolution porosity values. Enhanced GLTclay volumes and porosity values are thenprocessed with calibrated FMI resistivity val-ues to boost the resolution of hydrocarbonsaturation estimates.
Applying this technique to GLT and FMIlogs from a well in Lake Maracaibo,Venezuela reveals overlooked reserves. TheFMI image shows the highly laminatednature of the formation, with beds on theorder of 1 ft. A comparison of standard-reso-lution and high-resolution ELAN interpreta-tions shows that potential pay zones havebeen completely masked in the conven-tional processing (next page, top).
Using Electrical Anisotropy to Find Thin-Bed PayJames Klein and Paul Martin of ARCOExploration and Production Technology inPlano, Texas, and David Allen of Schlum-berger Wireline & Testing in Sugar Land,Texas are modeling electrical anisotropy todetect low-resistivity, low-contrast pay suchas thin beds.15 The researchers found that awater-wet formation with large variability ingrain size is highly anisotropic in the oil legand isotropic in the water leg. They attributethe resistivity anisotropy to grain-size varia-tions, which affect irreducible water satura-tion, between the laminations.
They tested their theory by modeling thethin, interbedded sandstones, siltstones andmudstones of the Kuparuk River formationA-sands of Alaska’s North Slope, located 10miles [16 km] west of Prudhoe Bay. Themodel, based on a Formation MicroScannerinterpretation, contains layers of low-perme-ability mudstone and layers of permeablesandstone with variable clay content.
The simulated resistivity data aredescribed as either perpendicular—mea-sured with current flowing perpendicular tothe bedding—or parallel—measured withcurrent flowing parallel to the bedding.
V clay, high-res = V clay, low-res GLT +
[V clay, high-res FMI - < V clay, high-res FMI >].
Oilfield Review
■■Using a high-resolution measurement to enhance a low-resolution one. The clay vol-umes derived from a GLT log of well in Lake Maracaibo, Venezuela were enhancedwith an FMI log from the same well. The enhanced ELAN interpretation (track 3) fea-tures several pay zones that were missed on the standard ELAN interpretation (track 2).
■■Effect of saturation on electrical anisotropy in the Kuparuk River formation, Alaska,USA. Resistivity data taken in the oil column curve to the right, but resistivity datataken in the water leg are nearly linear. The position of data along the oil column arcindicates the lithology of the formation. Strong anisotropy may be present in the oil col-umn, depending on the saturation in the more resistive component. In the water leg,the same formation might display little or no anisotropy.
100
10
1
0.10.1 1 10 100
Per
pend
icul
ar re
sist
ivity
50% sand:50% shale
Oil Column Water Leg
100%shale
0.1 1 10 100
Parallel resistivity, ohm-m Parallel resistivity, ohm-m
100% sand
Increasingoil saturation
100% shale
100%
san
d
Formation MicroScannerImages
Dep
th, f
t
X80
X85
0 180 360
0 100 0 100p.u. p.u.
Water
Bound Water
Siderite
Orthoclase
Pyrite
Montmorillonite
Illite
Hydrocarbon
Calcite
Quartz
Rutile
Muscovite
Kaolinite
Plotting perpendicular versus parallel resis-tivity for a given interval shows how hydro-carbon saturation influences electricanisotropy (below left). Simulated resistivitydata in the oil column curve to the right,but simulated resistivity data in the waterleg are nearly linear. The position of dataalong the oil column arc indicates thelithology of the formation.
Today, this technique works only with2-MHz MWD tools such as the CDR Com-pensated Dual Resistivity tool. The CDRphase and attenuation measurements pro-vide a unique response to anisotropy thatallows the perpendicular and parallel resis-tivities to be determined. The techniquerequires that the logging tool be parallel tothe beds so that differences in the phase andattenuation of resistivity measurements canbe used to establish anisotropy. Althoughthe technique cannot yet be applied at otherangles, its originators believe some opera-tors will value it enough to tailor the devia-tion of their wells so that logging tools canrun parallel to beds of interest.
Nuclear Magnetic Resonance LoggingAlthough thin-bed evaluation is challenging,the tools and techniques described so farprovide answers in most cases. More trou-blesome to interpreters than thin beds isanother prominent cause of low-resistivitypay, reduced grain size, which contributesto high irreducible water saturations. TheCMR Combinable Magnetic Resonance toolshows potential for measuring irreduciblewater saturation and pore size.16
The CMR tool looks at the behavior ofhydrogen nuclei—protons—in the presenceof a static magnetic field and a pulsed radio
17
15. Anisotropy is the variation of a property with direction. In this case, it is variation of resistivity in the vertical (perpendicular) versus horizontal (parallel) planes. For a review of electrical anisotropy:Anderson B, Bryant I, Helbig K, Lüling M and Spies B: “Oilfield Anisotropy: Its Origins and Electri-cal Characteristics,” Oilfield Review 6, no. 4 (January 1995): 48-56.Allen DF, Klein JD and Martin PR: “The Petrophysicsof Electrically Anisotropic Reservoirs,” Transactions ofthe SPWLA 36th Annual Logging Symposium, Paris,France, June 26-29, 1995.
16. Morriss CE, MacInnis J, Freedman R, Smaardyk J, Straley C, Kenyon WE, Vinegar HJ and Tutunjian PN:“Field Test of an Experimental Pulsed Nuclear Mag-netism Tool,” Transactions of the SPWLA 34th AnnualLogging Symposium, Calgary, Alberta, Canada, June13-16, 1993, paper GGG.Chang D, Vinegar H, Morriss C and Straley C: “Effective Porosity, Producible Fluid and Permeabilityin Carbonates from NMR Logging,” Transactions ofthe SPWLA 35th Annual Logging Symposium, Tulsa,Oklahoma, USA, June 19-22, 1994, paper A.
Oilfield Review
frequency (RF) signal (right ). A proton’smagnetic moment tends to align with thestatic field. Over time, the magnetic fieldgives rise to a net magnetization—more pro-tons aligned in the direction of the appliedfield than in any other direction.
Applying an RF pulse of the right fre-quency, amplitude and duration can rotatethe net magnetization 90° from the static fielddirection. When the RF pulse is removed, theprotons precess in the static magnetic field,emitting a radio signal until they return totheir original state. Because the signalstrength increases with the number of mobileprotons, which increases with fluid content,the signal strength is proportional to the fluidcontent of the rock. How quickly the signaldecays—the relaxation time—gives informa-tion about pore sizes and, to some extent,the amount and type of oil.
A CMR log displays distributions of relax-ation, or T2 times, which correspond to pore
size distributions. The area under a spec-trum of T2 times is called CMR porosity.
Unlike previous NMR tools, the CMRtool is a pad-mounted device. Permanentmagnets in the tool provide a static mag-netic field focused into the formation. TheCMR tool’s depth of investigation, about 1 in.[2.5 cm], avoids most effects from mudcakeor rugosity. Its vertical resolution of 6 in.[15 cm] allows for comparison with high-resolution logs.
A low-resistivity example from theDelaware formation in West Texas showshow the NMR response allows log inter-preters to measure residual oil saturationdirectly from the CMR log (below). NMRmeasurements on core samples from theDelaware formation show that the NMRresponse will decay within the first 200 mil-liseconds (msec) if the pores are filled withwater. If the pores are filled with oil, how-ever, the signal decays after about 400 msec.
The T2 distributions in track 4 have beendivided into three parts. The area under theT2 curve to the left of the first cutoff, shownas a blue line at 33 msec, represents irre-ducible water saturation. The area under thecurve from 33 msec to 210 msec (red line)represents producible fluid. Above 210msec, the area under the curve representsoil, presented as a CMR oil show in track 3.This measurement of oil actually refers toresidual oil saturation since the CMR toollooks only at the flushed zone.
With the introduction of the CMR tool,log interpreters are gaining the upper handin the struggle to assess low-resistivity pay.Although there are no easy answers whenevaluating low-resistivity pay, the tools andinterpretation techniques are in place tomore efficiently find these frequentlybypassed zones. —TAL
18
Dep
th, f
t
X300
0 200
Gamma RayGAPI 0.1 1000
MicroSFL Logohm-m
Laterolog Shallow
Laterolog Deep
Mud Log Showgas units 10000
CMR Porosityp.u. -1030
CMR Free Fluidp.u.
CMR Oil Showp.u. -0.020.08
Bound Fluid Volume33-msec line
210-msecoil/water line
T2 Distribution
.003 1
T2 Logsec
6 16in.
sec
1.500.001
-1030Caliper fromLitho-Density tool
X350
X250
■■Principle behind the CMR tool. Permanentmagnets in the CMR tool create a staticmagnetic field B that gives rise to a netmagnetization among hydrogen nuclei (top).A pulsed radio frequency signal rotates thenet magnetization 90° away from the staticmagnetic field (middle). After the RF pulse isremoved, the protons precess back to theiroriginal state, emitting a radio signal whosestrength is proportional to the fluid contentof the rock (bottom).
■■Early field test of the CMR tool in the Delaware formation, West Texas, USA. Based onNMR measurements of core samples from the Delaware formation, the T2 distributionsin track 4 have been divided into three parts. The area under the T2 curves to the left ofthe 33-msec cutoff is irreducible water saturation. From 33 msec to 210 msec, the areaunder the curve represents producible fluid. Above 210 msec, the area under the curverepresents oil. In track 3, the mud log show curve was derived from the total gas mea-sured on the mud log. It indicates that there is an oil-water contact halfway throughthe interval, at about X320 ft.
BStaticmagneticfield
Net
mag
netiz
atio
n
BNet
magnetizationRadio frequency
pulse
B
Radiosignal
Nuclear Magnetic Resonance Imaging—Technology for the 21st Century
Although well logging has made major advances over the last 70 years, several important reservoir proper-
ties are still not measured in a continuous log. Among these are producibility, irreducible water saturation
and residual oil saturation. Nuclear magnetic resonance (NMR) logging has long promised to measure these,
yet it is only recently that technological developments backed up by sound research into the physics behind
the measurements show signs of fulfilling that promise.
Autumn 1995
Bill KenyonRobert KleinbergChristian StraleyRidgefield, Connecticut, USA
Greg GubelinChris MorrissSugar Land, Texas, USA
For help in preparation of this article, thanks to AustinBoyd and Billie-Dean Gibson, Schlumberger Wireline &Testing, Sugar Land, Texas, USA.In this article, CMR (Combinable Magnetic Resonancetool), ELAN (Elemental Log Analysis), Litho-Density (pho-toelectric density log) and NML (Nuclear Magnetism Log-ging tool) are marks of Schlumberger. MRIL (MagneticResonance Imager Log) is a mark of NUMAR Corporation.
For nearly 70 years, the oil industry hasrelied on logging tools to reveal the proper-ties of the subsurface. The arsenal of wire-line measurements has grown to allowunprecedented understanding of hydrocar-bon reservoirs, but problems persist: a con-tinuous log of permeability remains elusive,pay zones are bypassed and oil is left in theground. A reliable nuclear magnetic reso-nance (NMR) measurement may change allthat. This article reviews the physics andinterpretation of NMR techniques, andexamines field examples where NMR log-ging has been successful.
Some BasicsNuclear magnetic resonance refers to aphysical principle—response of nuclei to amagnetic field. Many nuclei have a mag-netic moment—they behave like spinningbar magnets (next page, left ). These spin-ning magnetic nuclei can interact withexternally applied magnetic fields, produc-ing measurable signals.
19
2
■■Relaxation curves. Water in a testtube has a long T2 relaxation time of3700 msec at 40°C (upper curve).Relaxation times of a vuggy carbonatemight approach this value. However,water in rock pore space has shorter
Water in test tubeT2 = 3700 msec
Water in pore space of rockT2 = 10 to 500 msec
CMR porosity
Time, msec
Sig
nal a
mpl
itude
100% porosity
For most elements the detected signals aresmall. However, hydrogen has a relativelylarge magnetic moment and is abundant inboth water and hydrocarbon in the porespace of rock. By tuning NMR logging toolsto the magnetic resonant frequency ofhydrogen, the signal is maximized and canbe measured.The quantities measured are signal ampli-tude and decay (see “All in a Spin—NMRMeasurements,” below). NMR signal ampli-tude is proportional to the number of hydro-gen nuclei present and is calibrated to giveporosity, free from radioactive sources andfree from lithology effects. However, the
1. Murphy DP: “NMR Logging and Core Analysis—Simplified,” World Oil 216, no. 4 (April 1995): 65-70.
2. For examples of laboratory T2 core measurementsenabling direct comparison to log measurements andto laboratory T1 core measurements:Straley C, Rossini D, Vinegar H, Tutunjian P and Mor-riss C: “Core Analysis by Low Field NMR,” Proceed-ings of the 1994 International Symposium of the Soci-ety of Core Analysts, Stavanger, Norway, September12-14, 1994, paper SCA-9404.Kleinberg RL, Straley C, Kenyon WE, Akkurt R andFarooqui SA: “Nuclear Magnetic Resonance of Rocks:T1 versus T2 ,” paper SPE 26470, presented at the 68thSPE Annual Technical Conference and Exhibition,Houston, Texas, USA, October 3-6, 1993.
0
■■Precessing protons. Hydrogen nuclei—protoOnce disturbed from equilibrium, they precesame way that a child’s spinning top precess
Magnetic field
Spinning motion
Precessional m
decay of the NMR signal during each mea-surement cycle—called the relaxationtime—generates the most excitement amongthe petrophysical community.
Relaxation times depend on pore sizes(right ). For example, small pores shortenrelaxation times—the shortest times corre-sponding to clay-bound and capillary-bound water. Large pores allow long relax-ation times and contain the most readilyproducible fluids. Therefore the distributionof relaxation times is a measure of the distri-bution of pore sizes—a new petrophysicalparameter. Relaxation times and their distri-butions may be interpreted to give other
ns—behave like spinning bar magnets.ss about the static magnetic field (left) in thees in the Earth’s gravitational field (right).
relaxation times. For example, relax-ation times for a sandstone typicallyrange from 10 msec for small pores to500 msec for large pores. The initialamplitude of the relaxation curvegives CMR porosity.
otion
Gravitational field
Spinning motion
3. Kenyon WE: “Nuclear Magnetic Resonance as a Petro-physical Measurement,” Nuclear Geophysics 6, no. 2(1992): 153-171.
4. Kenyon WE, Howard JJ, Sezinger A, Straley C, Matteson A, Horkowitz K and Ehrlich R: “Pore-SizeDistribution and NMR in Microporous Cherty Sand-stones,” Transactions of the SPWLA 30th Annual Log-ging Symposium, Denver, Colorado, USA, June 11-14,1989, paper LL.Howard JJ, Kenyon WE and Straley C: “Proton Mag-netic Resonance and Pore Size Variations in ReservoirSandstones,” paper SPE 20600, presented at the 65thSPE Annual Technical Conference and Exhibition,New Orleans, Louisiana, USA, September 23-26,1990.Gallegos DP and Smith DM: “A NMR Technique forthe Analysis of Pore Structure: Determination of Continuous Pore Size Distributions,” Journal of Colloidand Interface Science 122, no. 1 (1988): 143-153.Loren JD and Robinson JD: “Relations Between PoreSize Fluid and Matrix Properties, and NML Measure-ments,” SPE Journal 10 (September 1970): 268-278.
5. Freedman R and Morriss CE: “Processing of Data Froman NMR Logging Tool,” paper SPE 30560, to be pre-sented at the 70th SPE Annual Technical Conferenceand Exhibition, Dallas, Texas, USA, October 22-25,1995.
petrophysical parameters such as permeabil-ity, producible porosity and irreduciblewater saturation. Other possible applica-tions include capillary pressure curves,hydrocarbon identification and as an aid tofacies analysis.1
Two relaxation times and their distribu-tions can be measured during an NMRexperiment. Laboratory instruments usuallymeasure longitudinal relaxation time, T1
and T2 distribution, while borehole instru-ments make the faster measurements oftransverse relaxation time, T2 and T2 distri-bution.2 In the rest of this article T2 willmean transverse relaxation time.
All in a Spin—NMR Measurements
NMR measurements consist of a series of manip-
ulations of hydrogen protons in fluid molecules.1
Protons have a magnetic moment and behave like
small bar magnets, so that their orientations may
be controlled by magnetic fields. They also spin,
which makes them act like gyroscopes.
A measurement sequence starts with proton
alignment followed by spin tipping, precession,
and repeated dephasing and refocusing. Trans-
verse relaxation and longitudinal relaxation limit
how long a measurement must last. Only after
completion of all these steps—which takes a few
seconds—can the measurement be repeated.2
Oilfield Review
0.1 1.0 10.0 100.0 1000.0
Sig
nal D
istr
ibut
ion
0.03
Sig
nal a
mpl
itude
0
0 100 200 300 400Time, msec
1
0.00
■■Signal amplitudeprocessed to give T2 distribution. TheCMR tool measuresdecaying NMR sig-nal amplitude (top),which is the sum ofall the decaying T2signals generatedby hydrogen pro-tons in the measure-ment volume. Sepa-rating out ranges ofT2 values by amathematicalinversion processproduces the T2 dis-tribution (bottom).The curve repre-sents the distribu-tion of pore sizes,and the area underthe curve is CMRporosity. Interpreta-tion of pore size dis-tribution and loga-rithmic mean T2 areused to calculatesuch parameters aspermeability andfree-fluid porosity.
NMR Applications and ExamplesThe T2 distribution measured by the Schlum-berger CMR Combinable Magnetic Reso-nance tool, described later, summarizes allthe NMR measurements and has severalpetrophysical applications:• T2 distribution mimics pore size distribu-
tion in water-saturated rock• the area under the distribution curve
equals CMR porosity• permeability is estimated from logarith-
mic-mean T2 and CMR porosity• empirically derived cutoffs separate the T2
distribution into areas equal to free-fluidporosity and irreducible water porosity.3Application and interpretation of NMR
measurement rely on understanding therock and fluid properties that cause relax-ation (see “NMR Relaxation Mechanisms,”page 26). With this foundation of the mech-anisms of relaxation, the interpretation of T2
distribution becomes straightforward.T2 Distribution—In porous media, T2
relaxation time is proportional to pore size.4At any depth in the well the CMR toolprobes a rock sample that has a range ofpore sizes. The observed T2 decay is thesum of T2 signals from hydrogen protons, inmany individual pores, relaxing indepen-dently. The T2 distribution graphically showsthe volume of pore fluid associated witheach value of T2 , and therefore the volumeassociated with each pore.
Signal processing techniques are used totransform NMR signals into T2 distributions(right ). Processing details are beyond thescope of this article.5
T2, msec
1. This article uses hydrogen proton when discussing NMRmeasurements. However, other texts use proton, nucleus,moment and spin interchangeably. For the purposes ofNMR theory they are all considered synonyms.
2. For a detailed description of NMR measurements:Fukushima E and Roeder SBW: Experimental Pulse NMR:A Nuts and Bolts Approach. Reading, Massachusetts,USA: Addison-Wesley, 1981.
Farrar TC and Becker ED: Pulse and Fourier TransformNMR: Introduction to Theory and Methods. New York,New York USA: Academic Press, 1971.
3. The SI unit of magnetic flux density is the Tesla—equal to1000 gauss.
B0 field
z
y
x
Net magnetizationalong z-axis
Precessingmagneticmoments
Proton alignment—Hydrogen protons are
aligned by application of a large constant mag-
netic field, B0. Alignment take a few seconds and
the protons will remain aligned unless disturbed.
The latest logging tools use elongated permanent
magnets, of about 550 gauss in the measurement
region—about 1000 times larger than the Earth’s
magnetic field. These are applied to the forma-
tion during the entire measurement cycle (right).3
Spin tipping—The next step is to tip the
aligned protons by transmitting an oscillating
21Autumn 1995
■■Proton alignment. The first step in an NMR mea-surement is to align the spinning protons usingpowerful permanent magnets. The protons precessabout an axis parallel to the B0 direction—the netmagnetization being the sum of all the precessingprotons. In logging applications B0 is perpendicularto the borehole axis.
22
■■Spin tipping. Aligned protons are tipped 90° by a magnetic pulse oscillating at the resonance, orLarmor frequency.
z
x
y
B0 field
B1 field
Porosity
Dolomite
Calcite
Anhydrite
Quartz
Bound Water
Illite
CMR Perm0.1 md 1000
Perm
1:240 ft
Perf
X350
X400
X450
CMR Porosity
25 p.u. 0
Hydrocarbon
Free Water
Cap Bound Water
Wiggle
Bound-Fluid Volume100-msec Line
0.1 msec 1000
750-msec Vug Line
0.1 msec 1000
T2 Distribution
In an example taken from a carbonatereservoir, T2 distributions from X340 ft toX405 ft are biased towards the high end ofthe distribution spectrum indicating largepores (left ). Below X405 ft, the bias istowards the low end of the spectrum, indi-cating small pores. This not only provides aqualitative feel for which zones are likely toproduce, but also helps geologists withfacies analysis.
Lithology-independent porosity—Tradi-tional calculations of porosity rely on bore-hole measurements of density and neutronporosity. Both measurements require envi-ronmental corrections and are influenced bylithology and formation fluid. The porosityderived is total porosity, which consists of
■■Pore-size distribution and free-fluidindex—carbonate example. In this well,the oil company was concerned aboutwater coning during production. Theinterval below X405 ft showed nearly100% water saturation by conventionallog interpretation (track 3). However, theCMR log showed low T2 distribution val-ues over this interval (track 4) indicatingsmall pores. Larger pores are indicatedabove X405 ft by higher T2 distributions.Applying a free-fluid index cutoff of 100msec to the distributions showed thatmost of the water was irreducible. Thisresult gave the oil company confidence toadd the interval X380 ft to X395 ft to itsperforating program.
magnetic field, B1, perpendicular to the direction
of B0 (right). For effective spin tipping:
f = γ B0
where f is the frequency of B1—called Larmor
frequency—and γ is a constant called the gyro-
magnetic ratio of the nucleus. For example, the
Larmor frequency for hydrogen nuclei in a field of
550 gauss is about 2.3 MHz.
The angle through which spins are tipped is
controlled by the strength of B1 and the length of
time it is switched on. For example, to tip spins
by 90°—as in most logging applications—a B1
field of 4 gauss is switched on for 16 microsec-
onds (µsec).
Precession and dephasing—When protons are
tipped 90° from the B0 direction, they precess in
the plane perpendicular to B0. In this respect
they act like gyroscopes in a gravitation field
(page 20, left).
At first all the protons precess in unison. While
doing so they generate a small magnetic field at
the Larmor frequency which is detected by the
antenna and forms the basis of the NMR mea-
surements. However, the magnetic field, B0, is
not perfectly homogeneous, causing the protons
to precess at slightly different frequencies. Grad-
ually, they lose synchronization—they dephase—
causing the antenna signal to decay (next page).
The decaying signal is called free induction
decay (FID) and the decay time is called
Oilfield Review
23Autumn 1995
■■Pulse sequence and refocusing. Each NMR mea-surement comprises a sequence of transverse mag-netic pulses transmitted by an antenna—called aCPMG sequence. Each CPMG sequence starts with apulse that tips hydrogen protons 90° and is followedby several hundred pulses that refocus the protonsby tipping them 180° (top). After each pulse theantenna becomes a receiver that records the signalamplitude (middle). The fast decay of eachecho—called free induction decay—is caused byvariations in the static magnetic field, B0. The decayof each echo amplitude is caused by molecular inter-actions and has a characteristic time constant ofT2—transverse relaxation time. The circled numberscorrespond to steps numbered in the race analogy(below).
Imagine runners lined up at the start of a race(bottom). They are started by a 90° pulse (1). Afterseveral laps, the runners are spread around the track(2, 3). Then the starter fires a second pulse of 180°(4, 5) and the runners turn round and head back tothe starting line. The fastest runners have the far-thest distance to travel and all of them will arrive atthe same time if they return at the same rate (6a).With any variation in speed, the runners arrive backat slightly different times (6b). Like the example ofthe runners, the process of spin reversal is repeatedhundreds of times during an NMR measurement.Each time the echo amplitude is less and the decayrate gives T2 relaxation time.
6b. Somespins dephased: echo amplitude
reduced
6a. Spinsreturn at the
same rate theyfanned out
90° 180°
1 2 3 4 600
Start next CPMGsequence
Time, µsec
CPMG Pulse Sequence
160 320 320
Spin echoesFree induction
decay
Am
plitu
de
Signal
3. Spinsfan out
4. 180° pulsereverses spinprecession
2. Spins inhighest B0precessfastest
1. 90° pulsestarts spinprecession
or
3 4 5
2
1 6a 6b
B1
ampl
itude
Time, µsec
5. Spins inhighest B0precessfastest
■■Transverse decay. As the protons precess aboutthe static field, they gradually lose synchronization.This causes the magnetic field in the transverseplane to decay. Dephasing is caused by inhomo-geneities in the static magnetic field and by molecu-lar interactions.
Dephasing inthe x-y plane
B0 field
z
x
y
T2*—the asterisk indicates that the decay is not a
property of the formation. For logging tools T2* is
comparable to the span of the tipping pulse—a
few tens of microseconds.
Refocusing: spin echoes—The dephasing
caused by the inhomogeneity of B0 is reversible.
Imagine a race started by a starting gun, analo-
gous to a 90° tipping pulse. The runners start off
in unison, but after several laps they become dis-
persed around the track—due to their slightly dif-
ferent speeds. Now the starter gives another sig-
nal by firing a 180° pulse. The runners turn
around and race in the opposite direction. The
fastest runners have the greatest distance to
travel back to the start. However, if the conditions
remain the same—which is never the case—all
the runners arrive back at the same time (above).
Similarly, the hydrogen protons—precessing at
slightly different Larmor frequencies—can be
refocused when a 180° pulse is transmitted. The
180° pulse is the same strength as a 90° pulse,
but switched on for twice as long. As the protons
rephase, they generate a signal in the
antenna—a spin echo.
Of course, the spin echo quickly decays again.
However, the 180° pulses can be applied repeat-
edly—typically several hundred times in a single
NMR measurement. The usual procedure is to
apply 180° pulses in an evenly spaced train, as
■■Typical decaying spin echo amplitude plot for a rock.Each dot represents the amplitude of a spin echo. Inthis example recorded time is less than 0.3 sec.
1.60
1.20
0.30
0.40
0.00
0.400.1 sec
CPMG Decay for a Rock
Sig
nal a
mpl
itude
24 Oilfield Review
φ Free-fluid
φC M R
φ Total
Clay-boundwater
Produciblefluids
Capillary-boundwater
Con
vent
iona
llo
gsC
MR
der
ived
■■CMR porosity. Hydrogen in rock matrix and in clay-boundwater has short relaxation times that are lost in the dead time ofthe CMR tool. CMR porosity includes only capillary-bound waterand producible fluids and is not influenced by lithology. Totalporosity—as measured by conventional logs—also includes clay-bound water. The dotted line indicates that CMR porosity doesnot include microporosity associated with hard shales.
producible fluids, capillary-bound waterand clay-bound water (below).
However, CMR porosity is not influencedby lithology and includes only produciblefluids and capillary-bound water. This isbecause hydrogen in rock matrix and inclay-bound water has sufficiently short T2
relaxation times that the signal is lost duringthe dead time of the tool.
An example in a clean carbonate forma-tion compares CMR porosity with thatderived from the density tool to show lithol-ogy independence (next page, top). Thelower half of the interval is predominantlylimestone, and density porosity, assuming alimestone matrix, overlays CMR porosity. AtX935 ft, the reservoir changes to dolomiteand density porosity has to be adjusted to a
dolomite matrix to overlay the CMR porosity.If the lithology is not known or if it is com-plex, CMR porosity gives the best solution.Also, no radioactive sources are used for themeasurement, so there are no environmentalconcerns when logging in bad boreholes.
Permeability—Perhaps the most importantfeature of NMR logging is the ability torecord a real-time permeability log. Thepotential benefits to oil companies are enor-mous. Log permeability measurementsenable production rates to be predicted,allowing optimization of completion andstimulation programs while decreasing thecost of coring and testing.
Permeability is derived from empiricalrelationships between NMR porosity andmean values of T2 relaxation times. These
relationships were developed from brinepermeability measurements and NMR mea-surements made in the laboratory on hun-dreds of different core samples. The follow-ing formula is commonly used:
kNMR = C(φNMR)4(T2, log)2
where kNMR is the estimated permeability,φNMR is CMR porosity, T2, log is the logarith-mic mean of the T2 distribution and C is aconstant, typically 4 for sandstones and 0.1for carbonates.
A cored interval of a well was loggedusing the CMR tool. The value of C in theCMR permeability model was calculatedfrom core permeability at several depths.After calibration CMR permeability wasfound to overlay all core permeability pointsover the whole interval (page 26). Over thezone XX41 m to XX49 m the porosity variedlittle. However, permeability varied consid-erably from a low of 0.07 md at XX48 m toa high of 10 md at XX43 m. CMR perme-ability also showed excellent vertical resolu-tion and compared well to that of core val-ues. The value of C used for this well will beapplied to subsequent CMR logs in this for-mation enabling the oil company to reducecoring costs.
Free-fluid index—The value of free-fluidindex is determined by applying a cutoff tothe T2 relaxation curve. Values above thecutoff indicate large pores potentiallycapable of producing, and values belowindicate small pores containing fluid that istrapped by capillary pressure, incapable ofproducing.
close together as possible. The entire pulse
sequence, a 90° pulse followed by a long series
of 180° pulses, is called a CPMG sequence
after its inventors, Carr, Purcell, Meiboom and
Gill.4 The echo spacing is 320 µsec for the
Schlumberger CMR tool and 1200 µsec for
NUMAR’s MRIL tool.
Transverse relaxation, T2 —The CPMG pulse
sequence compensates for dephasing caused by
B0 field imperfections. However, molecular pro-
cesses also cause dephasing, but this is irre-
versible. These processes are related to petro-
physical properties such as movable fluid poros-
ity, pore size distribution and permeability.
Irreversible dephasing is monitored by measur-
ing the decaying amplitude of the spin echoes in
the CPMG echo train (left). The characteristic echo
amplitude decay time is called the transverse
relaxation time, T2, because dephasing occurs in
the plane transverse to the static field B0.5
Longitudinal relaxation, T1 —After a period of
several times T2, the protons completely lose
coherence, and no further refocusing is possible.
After the CPMG pulse sequence is delivered the
protons return to their equilibrium direction paral-
■■Lithology-independent porosity. BelowX935 ft, the lithology is limestone withsome dolomitization (track 1), while aboveis dolomite. Two porosity curves (track 2)are derived from density measurements—one assumes a limestone lithology and theother dolomite. CMR porosity overlays thedensity limestone porosity in limestoneregions and overlays dolomite porosity indolomite regions—demonstrating thatCMR porosity is independent of lithology.
1:240 ft
0 100 30
30
30
0 p.u.
–10
–10
–10
–10
X900
X1000
Density Porosity, Limestone
p.u.
p.u.
p.u.
Zero Porosity
Combined Model
Illite
Bound Water
Quartz
Calcite
Dolomite
Oil
Water
Density Porosity, Dolomite
CMR Porosity
%
25Autumn 1995
■■Longitudinal relaxation, T1. When the CPMG pulsesequence ends, the protons gradually relax backtowards the static magnetic field. They do so withcharacteristic time constant T1, longitudinal relaxation.
x-y planedecay
z-axisbuildup
B0 field
z
x
y
4. Carr HY and Purcell EM: “Effects of Diffusion on Free Pre-cession in Nuclear Magnetic Resonance Experiments,”Physical Review 94, no. 3 (1954): 630-638.
5. Transverse relaxation is sometimes called spin-spin relax-ation in early NMR literature. Similarly, longitudinal relax-ation is sometimes called spin-lattice relaxation.
6. Kleinberg RL, Farooqui SA and Horsfield MA: “T1/T2 Ratioand Frequency Dependence of NMR Relaxation in PorousSedimentary Rocks,” Journal of Colloid and InterfaceScience 158, no. 1 (1993): 195-198.
Kleinberg et al, reference 2 main text.
lel to B0 (left). This process occurs with a different
time constant—longitudinal relaxation, T1. The
next spin-tipping measurement is not started until
the protons have returned to their equilibrium
position in the constant B0 field.
T1 and T2 both arise from molecular processes.
In many laboratory measurements on water-satu-
rated rocks, it has been found that T1 is frequently
equal to 1.5 T2.6 However, this ratio varies when
oil and gas are present in rock samples.
26 Oilfield Review
1. Kleinberg RL, Kenyon WE and Mitra PP: “On the Mecha-nism of NMR Relaxation of Fluids in Rocks,” Journal ofMagnetic Resonance 108A, no. 2 (1994): 206-214.
2. Kleinberg RL and Horsfield MA: “Transverse RelaxationProcesses in Porous Sedimentary Rock,” Journal of Mag-netic Resonance 88, no. 1 (1990): 9-19.
Sezginer A, Kleinberg RL, Fukuhara M and Latour LL:“Very Rapid Simultaneous Measurement of Nuclear Mag-netic Resonance Spin-Lattice Relaxation Time and Spin-Spin Relaxation Time,” Journal of Magnetic Resonance92, no. 3 (1991): 504-527.
NMR Relaxation Mechanisms
■■Comparison of log and core data. CMR porosity shows a good match with core poros-ity measurements. Computed CMR permeability has been fine-tuned to match corepermeability enabling CMR logs to replace conventional coring on subsequent wells.
Caliper125 375mm
Bit Size125 375mm
Gamma Ray0 150g API
SP-120 30mV 1:120 m
Core Permeability.01 md 100
CMR Permeability.01 md 100
Logarithmic Mean T21 msec 10000
Core Porosity0.2 m3/m3 0
CMR Porosity0.2 m3/m3 0
CMR Free-Fluid Porosity0.2 m3/m3 0
XX40
XX50
6. Straley C, Morriss CE, Kenyon WE and Howard JJ:“NMR in Partially Saturated Rocks: Laboratory Insightson Free Fluid Index and Comparison with BoreholeLogs,” Transactions of the SPWLA 32nd Annual Log-ging Symposium, Midland, Texas, USA, June 16-19,1991, paper CC.
7. Akbar M, Petricola M, Watfa M, Badri M, Charara M,Boyd A, Cassell B, Nurmi R, Delhomme JP, Grace M,Kenyon B and Roestenberg J: “Classic InterpretationProblems: Evaluating Carbonates,” Oilfield Review 7,no. 1 (January 1995): 38-57.
Chang D, Vinegar H, Morriss C and Straley C: “Effec-tive Porosity, Producible Fluid and Permeability inCarbonates from NMR Logging,” Transactions of theSPWLA 35th Annual Logging Symposium, Tulsa,Oklahoma, USA, June 19-22, 1994, paper A.
Many experiments have been made onrock samples to verify this assumption.6 T2
distributions were measured on water-satu-rated cores before and after they had beencentrifuged in air to expel the produciblewater. The samples were centrifuged under100 psi to simulate reservoir capillary pres-sure. Before centrifuging, the relaxation dis-tribution corresponds to all pore sizes. Itseems logical to assume that during cen-trifuging the large pore spaces empty first.Not surprisingly, the long relaxation timesdisappeared from the T2 measurement (nextpage, top).
Observations of many sandstone samplesshowed that a cutoff time of 33 msec for T2
distributions would distinguish betweenfree-fluid porosity and capillary-boundwater. For carbonates, relaxation times tendto be three times longer and a cutoff of 100msec is used.7 However, both these valueswill vary if reservoir capillary pressure dif-fers from the 100 psi used on the cen-trifuged samples. If this is the case, theexperiments may be repeated to find cutofftimes appropriate to the reservoir.
In a fine-grained sandstone reservoirexample, interpretation of conventional logdata showed 70 to 80% water saturation
There are three NMR relaxation mechanisms that
influence T1 or T2 relaxation times: grain surface
relaxation, relaxation by molecular diffusion in
magnetic field gradients and relaxation by bulk
fluid processes.1
Grain surface relaxation—Molecules in fluids
are in constant motion—Brownian motion—and
diffuse about a pore space, hitting the grain sur-
face several times during one NMR measurement.
When this happens, two interactions may occur.
First, hydrogen protons can transfer nuclear spin
energy to the grain surface allowing realignment
with the static magnetic field, B0— this con-
tributes to longitudinal relaxation, T1. Second,
protons may be irreversibly dephased—contribut-
ing to transverse relaxation, T2. Researchers have
shown that in most rocks, grain-surface relaxation
is the most important influence on T1 and T2. The
ability of grain surfaces to relax protons is called
surface relaxivity, ρ.2
Surfaces are not equally effective in relaxing
hydrogen protons. For example, sandstones are
about three times more efficient in relaxing pore
water than carbonates. Also rocks with a high
content of iron or other magnetic minerals have
larger than usual values of ρ and, hence, shorter
NMR relaxation times.
Pore size also plays an important role in sur-
face relaxation. The speed of relaxation depends
on how frequently protons can collide with the
surface and this depends on the surface-to-vol-
ume ratio (S/V) (next page, bottom). Collisions
27Autumn 1995
nGrain surface relaxation. Precessing protons move about pore space colliding with other protons and with grain surfaces (left). Everytime a proton collides with a grain surface there is a possibility of a relaxation interaction occurring. Grain surface relaxation is the mostimportant process affecting T1 and T2 relaxation times. Experimenters have shown that when the probability of colliding with a grain sur-face is high—small pores (center)—relaxation is rapid and when the probability of colliding with a grain surface is low—large pores(right)—relaxation is slower.
Small pore
Am
plitu
de
Time, msec
Large pore
Am
plitu
de
Time, msec
nFree-fluid porosity. Free-fluid porosity isdetermined by applying a cutoff to the T2distribution curve. The area underneaththe curve above the cutoff gives free-fluidporosity (top). This zone relates to thelarge producible pores of a rock sample.The value for the cutoff was determinedin the laboratory from a large number ofwater-saturated core samples. First, T2distributions were measured. Then thecores were centrifuged under 100 psi pres-sure to simulate draining down to typicalreservoir capillary pressures. The amountof fluid drained equals the free-fluidporosity, which is converted into anequivalent area on the T2 distributioncurve. The area—shaded from the right—determines cutoff values. A comparisonof T2 distributions taken before and aftercentrifuging shows the validity of thistechnique (top). Free-fluid porosity mea-sured in sandstone formations in twowells using the cutoff value of 33 msecobtained above shows good correlationwhen plotted versus centrifuge porosity(bottom).
Rock grain
Rock grain
Rock grain
H d t
5 10 15 20
centrifuge, p.u.
Fre
e-flu
id p
oros
ity a
t 33
mse
c, p
.u.
20
15
10
5
Well AWell B
0
Free-fluid cutoff33 msec
T2 original
T2 spun sample
Time, msec
Pore diameter, micronsS
igna
ldi
strib
utio
n
1 10 100 1000
0.01 0.1 1 10
Free-fluidφ
φ
■■Free-fluid index—sandstone example.In this predomi-nantly shaly-sand-stone (track 1), T2distributions (track 5)fall mainly belowthe 33-msec cutoffline, indicating capillary-boundwater. However, the ELAN interpreta-tion (track 3)—made without CMRdata—shows highwater saturation,implying water pro-duction. The ELANinterpretation withCMR data (track 4)clearly shows thatmost of the water isirreducible. Thiswell produced at30% water cut, validating the CMR results.
Clay
Sandstone
Salt
Limestone
Dolomite
Anhydrite
Hydrocarbon
CMR Perm0.1 md 1000
Perm
Effective Porosity
25 p.u. 0
Hydrocarbon
Water
Wiggle
Bound-Fluid Volume33-msec Line
.001 msec 3
T2 Distribution1:240 ft
X400
WaterSaturation
100 0p.u.
Effective Porosity
25 p.u. 0
CMR Bound-Fluid Volume
25 p.u. 0
Hydrocarbon
Water
Irreducible Water
are less frequent in large pores and they have a
small S/V—relaxation times are, therefore, rela-
tively long. Similarly, small pores have large a
S/V and short relaxation times.3
For a single pore, the nuclear spin magnetiza-
tion decays exponentially, so the signal ampli-
tude as a function of time in a T2 experiment
28
decays with characteristic time constant,
[ρ2(S/V)]–1. Therefore,
1/T2 = ρ2S/V.
Similarly,
1/T1 = ρ1S/V.
Rocks have a distribution of pore sizes, each
with its own value of S/V. The total magnetization
is the sum of the signal coming from each pore.
The sum of the volumes of all the pores is equal
to the fluid volume of the rock—the porosity. So
the total signal is proportional to porosity and the
overall decay is the sum of the individual decays,
which reflects pore size distribution. NMR mea-
surements of porosity and pore size distribution
are the key elements of NMR interpretation.
Relaxation by molecular diffusion in magnetic
field gradients—When there are gradients in the
static magnetic field, molecular motion can
cause dephasing and hence T2 relaxation. T1
relaxation is not affected. In the absence of such
gradients, molecular diffusion does not cause
NMR relaxation.
A B0 gradient has two possible sources: the
magnet configuration of the logging tool, and the
magnetic susceptibility contrast between grain
materials and pore fluids in porous rocks.
Keeping the CPMG echo spacing to a mini-
mum, and keeping the applied magnetic field
small reduce the contribution of diffusion to T2
relaxation to a negligible level.
Oilfield Review
8. Kenyon WE, Takezaki H, Straley C, Sen PN, Herron M,Matteson A and Petricola MJ: “A Laboratory Study ofNuclear Magnetic Resonance Relaxation and Its Rela-tion to Depositional Texture and Petrophysical Proper-ties—Carbonate Thamama Group, Mubarraz Field,Abu Dhabi,” paper SPE 29886, presented at the SPEMiddle East Oil Show, Bahrain, March 11-14, 1995.
9. For research into the effects of particulate invasion onNMR measurements:Fordham E, Sezginer A and Hall LD: “Imaging Multiex-ponential Relaxation in the (y, loge (T1) Plane withApplication to Clay Filtration in Rock Cores,” Journal
Fordham EJ, Roberts TPL, Carpenter TA, Hall LD andHall C: “Dynamic NMR Imaging of Rapid Depth Fil-tration of Clay in Porous Media,” American Instituteof Chemical Engineering Journal 37, no. 12 (1991):1900-1903.
10. Morriss CE, Freedman R, Straley C, Johnston M,Vinegar HJ and Tutunjian PN: “Hydrocarbon Satura-tion and Viscosity Estimation from NMR Logging inthe Belridge Diatomite,” Transactions of the SPWLA35th Annual Logging Symposium, Tulsa, Oklahoma,USA, June 19-22, 1994, paper C.
across a shaly sandstone formation. How-ever, on the CMR log most of the T2 distri-bution falls below the 33-msec cutoff indi-cating capillary-bound water. Interpretationincluding CMR data showed that most ofthe water was irreducible. The well hassince been completed producing economicquantities of gas and oil with a low watercut (previous page). The water cut may beestimated from the difference betweenresidual water saturation and water satura-tion from resistivity logs.
In another example, but this time in acomplex carbonate reservoir, the oil com-pany was concerned about water coningduring production. CMR log data showedlow T2 values below X405 ft indicatingsmall pore sizes. Applying the carbonatecutoff of 100 msec showed that nearly allthe water was irreducible, which allowedadditional perforation (see page 22, top). Todate no water coning has occurred.
Values for cutoffs can also be tailored toparticular reservoirs and help with faciesanalysis, as in the case of the Thamamagroup of formations in Abu Dhabi Oil Com-pany’s Mubarraz field offshore Abu Dhabi,UAE.8 In this field, classical log interpreta-tion showed water saturation of 10 to 60%.However, some zones produced no water,making completion decisions difficult. Per-meability also varied widely even thoughporosity remained almost constant. Labora-
Autumn 1995
tory measurements were performed oncores to determine whether NMR loggingwould improve log evaluation.
Cores showed a good deal of microporos-ity holding a large volume of capillary-bound water. Free-fluid porosity was foundin the traditional way by centrifuging thewater-saturated cores. For this reservoir,however, capillary pressure was known tobe 25 psi, so the core samples were cen-trifuged accordingly. This showed that NMRmeasurements could provide a good esti-mate of nonproducing micropores using aT2 cutoff of 190 msec. In addition, perme-able grainstone facies could be distin-guished from lower-permeability packstonesand mudstones with a cutoff of 225 msec.
Additional ApplicationsBorehole NMR instruments are shallow-reading devices. In most cases, they mea-sure formation properties in the flushedzone.9 This has some advantages as mud fil-
of Magnetic Resonance A113, no. 2 (April 1995): 139-150.
trate properties are well-known and can bemeasured at the wellsite on surface. Whenfluid loss during drilling is low, as in thecase of low-permeability formations, hydro-carbons may also be present in the flushedzone (see “The Lowdown on Low-ResistivityPay,” page 4). In these cases NMR tools maymeasure fluid properties such as viscosityand so distinguish oil from water.
A published example of the effects ofhydrocarbon viscosity comes from Shell’sNorth Belridge diatomite and Brown shaleformations, Bakersfield, California, USA.10
Both CMR logs and laboratory measure-ments on cores show two distinct peaks onthe T2 distribution curves. The shorter peak,at about 10 msec, originates from water incontact with the diatom surface. The longerpeak, at about 150 msec, originates fromlight oil. The position of the oil peak corre-lates roughly with oil viscosity. The areaunder this peak provides an estimation ofoil saturation. (continued on page 32)
3. Sen PN, Straley C, Kenyon WE and Whittingham MS:“Surface-to-Volume Ratio, Charge Density, Nuclear Mag-netic Relaxation and Permeability in Clay-Bearing Sand-stones,” Geophysics 55, no. 1 (1990): 61-69.
4. Morriss et al, reference 10 main text.
Bulk fluid relaxation—Even if grain surfaces
and internal field gradients are absent, relaxation
occurs in the bulk fluid. Bulk fluid relaxation can
often be neglected, but is important when water
is in very large pores—such as in vuggy carbon-
ates—and, therefore, hydrogen protons rarely
contact a surface. Bulk relaxation is also impor-
tant when hydrocarbons are present. The nonwet-
ting phase does not contact the pore surface, and
so it cannot be relaxed by the surface relaxation
mechanism. Also, increasing fluid viscosity
shortens bulk relaxation times.4 A bulk relaxation
correction must be made when the mud filtrate
contains ions of chromium, manganese, iron,
nickel or other paramagnetic ions. A sample of
mud filtrate can be measured at the wellsite to
calculate the correction.
Relaxation processes summary—Relaxation
processes act in parallel—their rates are additive:
(1/T2)total = (1/T2)S + (1/T2)D + (1/T2)B
where (1/T2)S is the surface contribution, (1/T2)D
is the diffusion in field gradient contribution, and
(1/T2)B is the bulk contribution. The corresponding
equation for T1 is
(1/T1)total = (1/T1)S + (1/T1)B.
There is no diffusion contribution to T1,
because that process results in a dephasing
mechanism. For the CMR tool, the surface relax-
ation mechanism will usually be dominant for
the wetting phase, and the bulk relaxation mech-
anism will dominate for the nonwetting phase.
29
3
History of Nuclear Magnetic Resonance Logging
As far back as 1946, nuclear magnetic resonance
(NMR) signals from hydrogen atom nuclei—
protons—were first observed independently by
Purcell and Bloch, and have since been used
extensively to characterize materials.1 Magnetic
resonance imaging instruments are commonly
used as diagnostic tools in medicine today.
Oil industry interest followed in the 1950s
with several patents for NMR logging tools filed
by companies such as California Research
Corporation, Schlumberger Well Surveying
Corporation, Texaco Incorporated and Socony
Mobil Oil Company.
The first NMR log was run in 1960. Developed
by Brown and Gamson of Chevron Research Com-
pany, the tool used the Earth’s magnetic field for
proton alignment—the principle underlying NMR
0
■■Principle of early NMR toformation are aligned to theprotons 90° (Borehole). Themeasures the decaying horwas extrapolated back to thtype of tool was that the bo
Borehole
Energize
Receive
E
z
Ear
logging tools for the next 30 years (below).2
Schlumberger ran two versions of this tool in the
1960s and early 1970s under license from Chevron
and later developed a third-generation tool—
NML-C—commercialized at the end of the 1970s.3
During this time, continuing research into NMR
interpretation produced some outstanding contri-
butions. Seevers developed a relationship
between relaxation time and permeability of sand-
stones in 1965, and Timur developed the concept
of free-fluid index and new methods to measure
permeability using NMR principles in 1968.4 The
relationship between pore size, fluid and matrix
properties was presented by Loren and Robinson
of Shell Oil Company in 1969.5
The 1970s and 1980s saw continuation of this
work by many oil companies in parallel with labo-
ols. The Schlumberger NML Nuclear Magnetism Logging Earth’s magnetic field—considered to be homogeneousy then start to precess about the Earth’s field and gradua
izontal magnetization as protons relax. The envelope of the start of the measurement to give NML porosity, assumrehole signal had to be eliminated by doping the mud sys
Fields Magnetization
y
xToolarth
th
Net
Rotatingframe ofreference
fL
ratory NMR techniques developed to characterize
core samples. Schlumberger also has a long tra-
dition of research into NMR techniques and pro-
duced important work on relaxation mechanisms
and pore-size distributions in the late 1980s.6
Much of this work continues today.
In the late 1980s the first experimental pulsed-
NMR logging tools were developed. A major dis-
advantage of early logging tools was having to
dope the mud system with magnetite to eliminate
the NMR signal from the borehole. To make the
technique more widely acceptable meant a radi-
cal design change to profit from advances in
pulsed-NMR technology—more commonly used
in the laboratory for core measurements.
A patent for a pulse-echo NMR logging tool
was filed by Jackson in 1978. This was followed
Oilfield Review
tool measured transverse relaxation, T2. Protons in the on these scales. A horizontally-mounted coil tips thelly relax back towards it (Magnetization). The same coile decaying signal gives T2 (NML Signal). T2 amplitude
ed equal to free-fluid index. One big drawback with thistem with magnetite—not very popular with drillers.
NML Signal
Tim
e
Free-fluidindex
Bpolarization
Bearth
S
N
NS
NS
Permanent magnet
Static magnetic field lines
Radio frequency coil
Radio frequency field lines
Sensitive zones
NS
N
S
S
N
1. Bloch F, Hansen W and Packard ME: “The Nuclear Induc-tion Experiment,” Physical Review 70 (1946): 474-485.
Bloembergen RM, Purcell EM and Pound RV: “RelaxationEffects in Nuclear Magnetic Resonance Absorption,”Physical Review 73 (1948): 679.
Jackson J and Mathews M: “Nuclear Magnetic ResonanceBibliography,” The Log Analyst 34, no. 4 (May-June1993): 35-69.
2. Brown RJS and Gamson BW: “Nuclear Magnetism Logging,” Journal of Petroleum Technology 12 (August 1960): 199-207.
3. Herrick RC, Couturie SH and Best DL: “An ImprovedNuclear Magnetism Logging System and its Application toFormation Evaluation,” paper SPE 8361, presented at the54th SPE Annual Technical Conference and Exhibition,Las Vegas, Nevada, USA, September 23-26, 1979.
Chandler RN, Kenyon WE and Morriss CE: “ReliableNuclear Magnetism Logging: With Examples in EffectivePorosity and Residual Oil Saturation,” Transactions of theSPWLA 28th Annual Logging Symposium, London, Eng-land, June 29-July 2, 1987, paper C.
Autumn 1995
■■Borehole NMR tool designs. Earth-field tool proposed1960 (top left) was the principle behind three generatiotools. Inside-out NMR uses permanent magnets and puright). This configuration was first proposed by Jacksonnetic Resonance Imager Log built by NUMAR Corporatithe first commercial pulsed NMR tool. A side-looking ininvented by Schlumberger is the basis for the CMR tool(bottom right). The top two figures are side views of thetwo figures are cross sections.
4. Seevers DO: “A Nuclear Magnetic Method for Determiningthe Permeability of Sandstones,” Transactions of theSPWLA 7th Annual Logging Symposium, Tulsa, Okla-homa, USA, May 8-11, 1966.
Timur A: “Effective Porosity and Permeability of Sand-stones Investigated Through Nuclear Magnetic Reso-nance Principles,” Transactions of the SPWLA 9th AnnualLogging Symposium, New Orleans, Louisiana, USA, June 23-26, 1968, paper K.
5. Loren JD and Robinson JD: “Relations Between Pore Size Fluid and Matrix Properties, and NML Measure-ments,” paper SPE 2529, presented at the 44th SPEAnnual Fall Meeting, Denver, Colorado, USA, September28-October 1, 1969.
Loren JD: “Permeability Estimates from NML Measure-ments,” Journal of Petroleum Technology 24 (August1972): 923-928.
6. Kenyon WE, Day PI, Straley C and Willemsen JF: “Com-pact and Consistent Representation of Rock NMR Datafor Permeability Estimation,” paper SPE 15643, presentedat the 61st SPE Annual Technical Conference and Exhibi-tion, New Orleans, Louisiana, USA, October 5-8, 1986.
in the late 1980s and 1990s by designs from
NUMAR and Schlumberger (left).7
The two tools currently available are the
Schlumberger second generation, pulse-echo
NMR tool—the CMR tool—and NUMAR’s MRIL
Magnetic Resonance Imager Log.8 Both use per-
manent magnets, instead of the Earth’s magnetic
field, to align protons, and a system providing
controllable radio-frequency (rf) magnetic pulses
allowing T2 measurements. The use of perma-
nent magnets means that the position of the mea-
surement volume can be controlled by tool
design—eliminating borehole doping.
31
by Brown and Gamson inns of Schlumberger NMLlsed NMR technology (top in 1984. The MRIL Mag-
on in 1990 (bottom left) wasside-out configuration commercialized this year borehole and the bottom
Kenyon WE, Day PI, Straley C and Willemsen JF: “AThree-Part Study of NMR Longitudinal Relaxation Proper-ties of Water-Saturated Sandstones,” SPE FormationEvaluation 3, no. 3 (1988): 622-636.
Kenyon et al, reference 4 main text.
7. Jackson JA and Cooper RK: “Magnetic Resonance Appa-ratus,” US Patent No. 4,350,955, August 1980.
Jackson JA: “Nuclear Magnetic Resonance Well Log-ging,“ The Log Analyst 25 (September-October 1984):16-30.
Kleinberg RL, Griffin DD, Fukuhara M, Sezginer A andChew WC: “Borehole Measurement of NMR Characteris-tics of Earth Formations,” US Patent No. 5,055,787, October 1991.
8. Kleinberg et al, reference 12 main text.
Miller MN, Paltiel Z, Gillen ME, Granot J and Bouton JC:“Spin Echo Magnetic Resonance Logging: Porosity andFree Fluid Index Determination,” paper SPE 20561, pre-sented at the 65th SPE Annual Technical Conference andExhibition, New Orleans, Louisiana, USA, September 23-26, 1990.
Borehole wall
Antenna
Wear plate
Permanent magnet
Permanent magnet
Bowspringeccentralizer
Electroniccartridge
CMR skid
Sensitive zone
14 ft
T2 distribution measurements were alsomade on crude oil samples having viscosi-ties of 2.7 cp to 4300 cp (below). Highlyviscous oils have less mobile hydrogen pro-tons and tend to relax quickly. The CMR logshowed the T2 oil peak and correctly pre-dicted oil viscosity. It also showed that theupper 150 ft of the diatomite formationundergoes a transition to heavier oil.
Capillary pressure curves, used byreservoir engineers to estimate the percent-age of connate water, may also be predictedfrom T2 distributions. Typically thesecurves—plots of mercury volume versuspressure—are produced by injecting mer-cury into core samples. Under low pressurethe mercury fills the largest pores and, aspressure increases, progressively smallerpores are filled. The derivative of the capil-lary pressure curve approximates the T2 dis-tribution. Some differences in shape areexpected as mercury injection measurespore throat sizes, whereas NMR measure-ments respond to the size of pore bodies.11
Other applications and techniques arelikely to follow with more complex opera-
nCMR tool. The CMR tool (left) is only 14 ft [4.3 m] long and is combinable with manyother Schlumberger logging tools. The sensor is skid-mounted to cut through mud cakeand have good contact with the formation over most hole sizes. Contact is provided byan eccentralizing arm or by powered calipers of other logging tools. Two powerful per-manent magnets provide a static magnetic field (right). By design the tool measure-ment volume is a region of about 0.5 in. to 1.25 in. [1.3 cm to 3.2 cm] into the formationand stretches the length of the antenna— about 6 in. [15 cm], providing the tool withexcellent vertical resolution. The area in front of the antenna does not contribute to thesignal, which allows the tool to operate in holes with a certain amount of rugosity, simi-lar to density tools. The antenna acts as both transmitter and receiver—transmitting theCPMG magnetic pulse sequence and receiving the pulse echoes from the formation.
3.9 msec865 cp
414 msec3.6 cp
T2
0.1 1.0 10.0 100.0 1000.0 10000.0
nT2 distributions for two oils of differentviscosities. When bulk fluid relaxationdominates, as in the case of fluid samples,viscosity influences relaxation time.Hydrogen protons in highly viscous fluidsare less mobile and tend to relax quickly.
tions that might involve comparing logs rununder different borehole conditions. Forexample, fluid may be injected into the for-mation that is designed to kill the waterNMR signal so that residual oil saturationmay be measured. This type of technique,called log-inject-log, has been used withother borehole logs to monitor injectivity orto monitor acid treatments.
32
Function of a Pulsed Magnetic Resonance ToolThe CMR tool is the latest generationSchlumberger NMR tool (see “History ofNuclear Magnetic Resonance Logging,”page 30). The measurement takes placeentirely within the formation, eliminatingthe need to dope mud systems with mag-netite to kill the borehole signal—a bigdrawback with the old earth-field tools. Ituses pulsed-NMR technology, which elimi-nates the effects of nonuniform static mag-netic fields and also increases signalstrength. This technology, along with thesidewall design, makes the tool only 14 ft[4.3 m] long and readily combinable withother borehole logging tools (above).12
The skid-type sensor package, mountedon the side of the tool, contains two perma-nent magnets and a transmitter-receiverantenna. A bowspring eccentralizing arm or
powered caliper arm—if run in combinationwith other logging tools—forces the skidagainst the borehole wall, effectively remov-ing any upper limit to borehole size.
An important advantage of the sidewalldesign is that the effect of conductive mud,which shorts out the antenna on mandrel-type tools, is greatly reduced. What littleeffect remains is fully corrected by an inter-nal calibration signal. Another advantage isthat calibration of NMR porosity is simpli-fied and consists of placing a bottle of wateragainst the skid to simulate 100% porosity.T2 properties of mud filtrate samples—required for interpretation corrections—mayalso be measured at the wellsite in a similarfashion. Finally, the design enables high-res-
Oilfield Review
33Autumn 1995
11. Morriss et al, reference 10.12. Kleinberg RL, Sezginer A, Griffin DD and Fukuhara
M: “Novel NMR Apparatus for Investigating anExternal Sample,” Journal of Magnetic Resonance97, no. 3 (1992): 466-485.Sezginer A, Griffin DD, Kleinberg RL, Fukuhara Mand Dudley DG: “RF Sensor of a Novel NMR Appa-ratus,” Journal of Electromagnetic Waves and Appli-cations 7, no. 1 (1993): 13-30.Morriss CE, MacInnis J, Freedman R, Smaardyk J,Straley C, Kenyon WE, Vinegar H and Tutunjian PN:“Field Test of an Experimental Pulsed Nuclear Mag-netism Tool,” Transactions of the SPWLA 34thAnnual Logging Symposium, Calgary, Alberta,Canada, June 13-16, 1993, paper GGG.
olution logging—a 6-in. [15-cm] long mea-surement aperture is provided by a focusedmagnetic field and antenna.
Two permanent magnets generate thefocused magnetic field, which is about1000 times stronger than the Earth’s mag-netic field. The magnets are arranged sothat the field converges to form a zone ofconstant strength about one inch inside theformation. NMR measurements take placein this region.
By design, the area between the skid andthe measurement volume does notcontribute to the NMR signal. Coupled withskid geometry, this provides sufficientimmunity to the effects of mudcake andhole rugosity. The rugose hole effect is simi-lar to that of other skid-type tools such asthe Litho-Density tool.
The measurement sequence starts with await time of about 1.3 sec to allow for com-plete polarization of the hydrogen protonsin the formation along the length of the skid(for measurement principles, see “All in aSpin—NMR Measurements,” page 20 ).Then the antenna typically transmits a trainof 600 magnetic pulses into the formationat 320-msec intervals. Each pulse inducesan NMR signal—spin echo—from thealigned hydrogen protons. The antenna alsoacts as a receiver and records each spinecho amplitude. T2 distribution is derivedfrom the decaying spin echo curve, some-times called the relaxation curve.
Logs for the Future?The prospects have always looked bright fornuclear magnetic resonance logging.Research has shown that interpretation ofNMR relaxation times provides a wealth ofpetrophysical data. The latest generationlogging tools—using pulsed NMR tech-niques—are building on that research andare providing powerful wellsite answersthat shed new insight into the basic ques-tion, “What will the well produce?” —AM
34
Advanced Fracturing Fluids Improve Well Economics
The oil and gas industry has witnessed a revolution in fluids technology
for hydraulic fracturing. Starting in the mid 1980s, focused research led
to major improvements in the performance of well stimulation fluids.
Today, new additives and fluids are extending these capabilities and
providing innovative solutions to nagging problems. The results are
more efficient and cost-effective treatments for enhancing well produc-
Kevin ArmstrongRoger CardReinaldo NavarreteErik NelsonKen NimerickMichael SamuelsonTulsa, Oklahoma, USA
Jim CollinsCalgary, Alberta, Canada
Gilbert DumontMichael Priaro (consultant)Neal WasylyciaPetro-Canada ResourcesCalgary, Alberta, Canada
Gary SlusherEnron Oil & GasCorpus Christi, Texas, USA
For help in preparation of this article, thanks to VicJoyce, Ken Nolte and Jon Mitchell, Dowell, Tulsa, Oklahoma, USA; Terry Greene, Dowell, Montrouge,France; Richard Marcinew, Dowell, Calgary, Alberta,Canada; and Mike Morris, Dowell, Corpus Christi, Texas, USA.In this article, HIGHSHEAR, CleanFLOW and PropNETare marks of Schlumberger.
Hydraulic fracturing is one of the oil and gasindustry’s most complex operations. Thistechnique has been applied worldwide toincrease well productivity for nearly 50years.1 Fluids are pumped into a well atpressures and flow rates high enough to splitthe rock and create two opposing cracksextending up to 1000 ft [305 m] or morefrom either side of the borehole (right). Sandor ceramic particulates, called proppant, arecarried by the fluid to pack the fracture,keeping it open once pumping stops andpressures decline.
What defines a successful fracture? It isone that:• is created reliably and cost-effectively• provides maximum productivity enhance-
ment• is conductive and stable over time.
This article describes today’s fracturingoperations and the pivotal role played bythe fracturing fluid. Then, it highlights fournew fluid technologies that are improvingfracture success and well economics.
Oilfield Review
1. For more on hydraulic fracturing: Waters AB: “Hydraulic Fracturing—What Is It?,” Jour-nal of Petroleum Technology 33 (August 1981): 1416.Veatch RW Jr, Moschovidis ZA and Fast CR: “AnOverview of Hydraulic Fracturing,” in Gidley JL,Holditch SA, Nierode DE and Veatch RW Jr (eds):Recent Advances in Hydraulic Fracturing, Monograph12. Richardson, Texas, USA: Society of PetroleumEngineers (1989): 1-38.
35Autumn 1995
nFracturing operation. Modern treatments rely on a variety of process-controlled mixing, blending and pumping equipmentand computer monitoring and recording systems, which permit real-time decision making.
Continuous fluid mixer Proppant blenderPump trucks
n Flow regimes. Radial flow profile for an unfractured well (top left). Linear flow into and along the fracture replaces radialflow with a more conductive path for hydrocarbons (top right). Inhigh-permeability formations, short fractures are used to reachbeyond the area of matrix damage near the wellbore (bottom).
Low-permeability, linear flow
High-permeability, linear flow
Unfractured, radial flow
The Rock, the Mechanics and the FluidHistorically, fracturing has been applied pri-marily to low-permeability—0.1 to 10 md—formations with the goal of producing nar-row, conductive flow paths that penetratedeep into the reservoir. These less restrictivelinear conduits replace radial flow regimesand yield a several-fold production increase(above). For large-scale treatments, as manyas 40 pieces of specialized equipment, witha crew of 50 or more, are required to mix,blend and pump the fluid at more than 50barrels per minute (bbl/min) [8 m3/min].Pumping may last eight hours with1,000,000 gal of fluid and 2,000,000 to4,000,000 lbm of proppant placed in thefracture (next page, left).
Until recently, treatments were performedalmost exclusively on poor producing wells(often to make them economically viable).In the early 1990s, industry focus shifted togood producers and wells with potential forgreater financial return. This, in turn, meantan increased emphasis on stimulating high-permeability formations.
The major constraint on production fromsuch reservoirs is formation damage, fre-quently remedied by matrix acidizing treat-ments. But acidizing has limitations, andfracturing has found an important niche.The objective in highly permeable forma-tions is to create short, wide fractures toreach beyond the damage. This is oftenaccomplished by having the proppantbridge, or screen out, at the end, or tip, ofthe fracture early in the treatment. This “tipscreenout” technique is the opposite of whatis desired in low-permeability formations
36
where the tip is ideally the last area to bepacked.
Why the different approach? The answer isfound in the relationship between fracturelength and the permeability contrastbetween the fracture and the formation.Where the contrast is large, as for low-per-meability reservoirs, longer fractures provideproportionally greater productivity. Wherethe contrast is small, as in high-permeabilityformations, greater fracture length providesminimal improvement. Fracture conductivityis, however, directly related to fracturewidth. Using short—about 100-ft [30-m]—and wide fractures can prove beneficial.2
High-permeability formation treatmentsare on a far reduced scale. Only a fewpieces of blending and pumping equipmentare required, and pumping times are typi-cally less than one hour, and often only 15minutes. Fluid is pumped at 15 to 20bbl/min [2.4 to 3.2 m3/min] with a totalvolume of 10,000 to 20,000 gal [37.9 to75.7 m3] and total proppant weight of about100,000 lbm [45,000 kg] (next page, left).This technique has been successful in theNorth Sea, Middle East, Indonesia, Canadaand Alaska, USA.
While fracturing treatments vary widely inscale, each requires the successful integra-tion of many disciplines and technologies,regardless of reservoir type. Rock mechanicsexperiments on cores, specialized injectiontesting and well logs provide data on forma-tion properties. Sophisticated computer soft-ware uses these data, along with fluid andwell parameters, to simulate fracture initia-tion and propagation. These results and eco-nomic criteria define the optimum treatmentdesign. Process-controlled mixing, blendingand high-pressure pumping units executethe treatment. Monitoring and recordingdevices ensure fluid quality and provide
permanent logs of job results. Engineerstracking the progress of the treatment usegraphic displays that plot actual pumpingparameters against design values to facilitatereal-time decision making. Production simu-lators compare treatment results with expec-tations, providing valuable feedback fordesign of the next job.
At the heart of this complex process is thefracturing fluid.3 The fluid, usually water-based, is thickened with high molecularweight polymers, such as guar or hydrox-ypropyl guar. It must be chemically stableand sufficiently viscous to suspend the prop-pant while it is sheared and heated in sur-face equipment, well tubulars, perforationsand the fracture. Otherwise, premature set-tling of the proppant occurs, jeopardizingthe treatment. A suite of specially designedchemical additives imparts important prop-erties to the fluid. Crosslinkers join polymerchains for greater thickening, fluid-lossagents reduce the rate of filtration into theformation and breakers act to degrade thepolymer for removal before the well isplaced on production.
The fracture is created by pumping aseries of fluid and proppant stages. The firststage, or pad, initiates and propagates thefracture but does not contain proppant. Sub-sequent stages include proppant in increas-ing concentrations to extend the fractureand ensure its adequate packing.
Fracturing fluid technology has also devel-oped in stages. Early work focused on iden-tifying which polymers worked best andwhat concentrations gave adequate prop-pant transport. Then, research on additivesto fine-tune fluid properties hit high gear.
Oilfield Review
Equipment Fluid Proppant
1,000,000 gal
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
Propp Propp Propp
40 vehicles
Low-Permeability Treatment
3 million lbm
Autumn 1995
nZones of leakoff in a fractured formation.In low-permeability reservoirs, theinvaded zone may be small or nonexis-tent, since polymer molecules are toolarge to enter the matrix. In these cases,an external filter cake controls fluid loss.In high-permeability reservoirs, signifi-cant matrix damage can occur becauseof particle penetration.
nScale of fracturing treatments. Equipment,long fractures in low-permeability formationthose needed for short fractures in high-perm
2. Hanna B, Ayoub J and Cooper B: “Rewriting the Rulesfor High-Permeability Stimulation,” Oilfield Review 4,no. 4 (October 1992): 18-23.
3. For background on fracturing fluids: Constien VG:“Fracturing Fluid and Proppant Characterization,” inEconomides MJ and Nolte KG (eds): Reservoir Stimu-lation, 2nd ed. Englewood Cliffs, New Jersey, USA:Prentice Hall (1989): 5-1–5-23.
4. Penny GS and Conway MW: “Fluid Leakoff,” in Gidley JL, Holditch SA, Nierode DE and Veatch RW Jr(eds): Recent Advances in Hydraulic Fracturing,Monograph 12. Richardson, Texas, USA: Society ofPetroleum Engineers (1989): 147-176.
5. Fluid efficiency, E, is defined as
in which VF is the fluid volume remaining in the fracture, VP is the total volume pumped and VL is the volume of fluid that has leaked into the formation.
External filter cakeBridging zone
Invaded zone
Uncontaminated formation
Flow
E = VF
VP = VF
VF +VL
Much was learned, but what finallyemerged was a huge array of complicatedfluids—difficult to prepare and pump—andan amazing assortment of single-use addi-tives (most had to be custom manufactured)that required expensive material inventories.
In the past ten years, a more productiveresearch direction has emerged. Oil compa-nies, service companies and polymer manu-facturers have concentrated on the basicphysical and chemical mechanisms underly-ing the behavior of fracturing fluids in anattempt to find improved approaches to fluiddesign and use. This initiative has led tomajor advances, including higher-performingpolymers, simpler fluids, multifunctionaladditives and continuous, instead of batch,mixing. These developments have had a sig-nificant, beneficial impact on the industry.
Recent innovations are extending the stateof the art in four areas: • controlling fluid loss to increase
fluid efficiency• extending breaker technology to
improve fracture conductivity• reducing polymer concentration to
improve fracture conductivity• eliminating proppant flowback to
stabilize fractures. Each provides new opportunities for
improving well economics, as described inthe remainder of this article.
Controlling Fluid LossA portion of the fluid pumped during a frac-turing treatment filters into the surroundingpermeable rock matrix.4 This process,referred to as fluid leakoff or fluid loss,occurs at the fracture face. The volume offluid lost does not contribute to extending orwidening the fracture. Fluid efficiency is oneparameter describing the fluid’s ability tocreate the fracture.5 As leakoff increases,efficiency decreases. Excessive fluid loss canjeopardize the treatment, increase pumpingcosts and decrease post-treatment well per-formance.
4 vehicles
20,000 gal 100,000 lbm
Equipment Fluid Proppant
Propp
Propp
High-Permeability Treatment
fluid and proppant requirements to creates (left) can be 10 to 50 times more thaneability reservoirs (right).
Typically, particulates or other fluid addi-tives are used to reduce leakoff by forming afilter cake—termed an external cake—onthe surface of the fracture face. Actingtogether with the polymer chains, the fluid-loss material blocks the pore throats, effec-tively preventing invasion into the rockmatrix (below).
This approach has been applied suc-cessfully for decades to low-permeability
37
■■Dynamic fluid-losstest apparatus.
Fluid inlet
Fluid outlet
Fluidslot
Leakofffluid
Oil-filled annulus
Side View
Core Sleeve
To differentialpressuretransducers
End View
Core
Sleeve
Coreholder
Oil-filled annulus
■■Dynamic fluid-loss cell and core holder. A specially designed core holder simulates a slot-flow geometry, the one most representative of the actual fracturingprocess. Pressure and leakoff measurements provide the data necessary to evaluate the depth of invasion and the impact of various additives on fluid loss.
Measuring Dynamic Fluid Loss in the Laboratory
38 Oilfield Review
Dynamic fluid-loss measurements were made in
the Dowell laboratory in Tulsa, Oklahoma, USA
using special fluid-loss cells with a slot-flow
geometry and a porous test surface on one of the
slot walls (above). Cylindrical cores of the same
type and dimensions as for static tests are used
to allow direct comparison with static fluid-loss
results. Cells are constructed from stainless steel
for operation to 3500 psi and 350°F [177°C]. The
inlet design ensures fully developed flow over the
test section. A backpressure regulator and a heat
exchanger, which cools the filtrate, prevent evap-
oration of the filtrate during operations above the
ambient boiling point of the fluid. As many as
three cells can be used simultaneously for testing
cores of differing permeabilities.
A special fluid-loss simulator was designed to
prepare the fluids under dynamic conditions, sub-
ject them to shear and temperature histories and
then measure fluid loss (right).
The two components of the apparatus are a
shear history simulator and a fracture simulator.
The first uses a static mixer and 800 ft [244 m] of
small-diameter tubing to simulate preparation
and shearing of the fluid in the well tubulars.
The second subjects the fluid to shear and temper-
ature conditions of the fracture, using two large,
floating-piston accumulators and coils of tubing
immersed in a temperature-controlled bath.
A computer-controlled valving arrangement
ensures that the fluid always travels in the same
direction across the core face.
■■Forces acting on a particle. A particle,such as a fluid-loss additive, flowinginside a fracture is subjected to shear, Fxand drag, FY, where vx is the velocity inthe flow direction as a function of the dis-tance, y, from the fracture face. The ratioof shear to drag is directly proportional tothe leakoff rate and inversely propor-tional to the particle size and the shearrate at the wall. High initial leakoff andoptimized particles can help ensure thatfluid-loss additives reach and remain onthe fracture face.
■■Shear rate history of a rock segment. As the tip of the fracture passes a particularlocation, in this case a point 50 ft [15 m] from the wellbore, the maximum shear rateoccurs. As the treatment progresses and the fracture widens at this location, the shearrate falls off rapidly initially and then more slowly later. The laboratory approximationand calculated curve show good agreement.
y
vx
Fy
Fx
39Autumn 1995
6. Navarrete RC, Cawiezel KE and Constien VG:“Dynamic Fluid Loss in Hydraulic Fracturing UnderRealistic Shear Conditions in High-PermeabilityRocks,” paper SPE 28529, presented at the 69th SPEAnnual Technical Conference and Exhibition, NewOrleans, Louisiana, USA, September 25-28, 1994.
7. Howard GC and Fast CR: Hydraulic Fracturing,Monograph 2. Richardson, Texas, USA: Society ofPetroleum Engineers, 1989.
8. Prud’homme RK and Wang JK: “Filter-Cake Forma-tion of Fracturing Fluids,” paper SPE 25207, pre-sented at the SPE International Symposium on Oil-field Chemistry, New Orleans, Louisiana, USA,March 2-5, 1993.
9. Shear rate is proportional to shear stress, the exactrelationship depending on the rheological modelused for the fluid.
10. Navarrete RC and Mitchell JP: “Fluid-Loss Controlfor High-Permeability Rocks in Hydraulic FracturingUnder Realistic Shear Conditions,” paper SPE29504, presented at the SPE Production OperationsSymposium, Oklahoma City, Oklahoma, USA, April2-4, 1995.
300
She
ar r
ate,
sec
-1
250
200
150
100
50
0 20 60 16040 14012010080
Time, min
350
CalculatedLaboratory approximation
Fracture height = 300 ftFracture length = 540 ftPump rate = 40 bbl/minPump time = 145 min
(< 0.1 md) formations in which polymerand particulate sizes exceed those of thepore throats. In high-permeability reservoirs,however, fluid constituents may penetrateinto the matrix, forming a damaging internalfilter cake. This behavior has promptedmechanistic studies to determine the impacton fracturing treatment performance.6
Classic fluid-loss theory assumes a two-stage, static—or nonflowing—process.7 Asthe fracture propagates and fresh formationsurfaces are exposed, an initial loss of fluid,called spurt, occurs until an external filtercake is deposited. Once spurt ceases, pres-sure drop through the filter cake controlsfurther leakoff. For years, researchers havedeveloped fluid-loss control additives undernonflowing conditions based on this theory.
The conventional assumptions, however,neglect critical factors found under actualdynamic—or flowing—conditions presentduring fracturing, including the effects ofshear stress on both external and internal fil-ter cakes and how fluid-loss additives movetoward the fracture face. In high-permeabil-ity formations, with an internal filter cakepresent, most of the resistance to leakoffoccurs inside the rock, leaving the externalcake subject to erosion by the fluid.
Analysis of fluid loss under dynamic condi-tions relates external cake thickness to theyield stress of the cake at the fluid interfaceand the shear stress exerted on the cake bythe fluid.8 These, in turn, depend on thephysical properties of the cake and the rheo-logical properties of, and shear rate inducedin, the fluid.9 Whether an external filter cakeforms, grows, remains stable or erodesdepends on the way these parameters varyand interact over time and spatial orientation.
Similarly, the effectiveness of additives tocontrol fluid loss depends on two factors:their ability to reach the fracture face quicklyand their ability to remain there. The formeris governed by the drag force exerted on theparticles and the latter by the shear forceexerted on them (right). The larger the ratioof drag to shear, the greater the chance thatthe particles will remain on the surface. Agreater leakoff flux to the wall, smaller parti-cle dimensions and a lower shear rate favorsticking. Promoting higher leakoff for betteradditive placement seems directly at oddswith controlling fluid loss! However, in prac-tice, higher initial leakoff can yield greateroverall fluid efficiency.
To confirm the controlling mechanisms,dynamic fluid-loss tests were conducted
using a slot-flow geometry, determined to bethe simplest representation of what occurs ina fracture. To completely describe the pro-cess, computer-controlled equipment wasconstructed to prepare and test fluids underdynamic conditions, subjecting them to thetemperature and shear histories found in afracture (see “Measuring Dynamic Fluid Lossin the Laboratory,” previous page). Cores ofvarious lengths were used in the tests to sim-ulate a fracture segment at a fixed distancefrom the wellbore.10 As the fracture tippasses a specific point, spurt occurs and theshear rate reaches a maximum (above ).Then, as the fracture widens, the shear stressdecreases. In the test apparatus, this is simu-lated by decreasing the flow rate with time.
40 Oilfield Review
6
Flui
d-lo
ss v
ol, m
l
5
4
3
2
1
0 2 4 6 16 181412108Time, min
Dynamic, 190 sec–1 shear rate
Temperature = 150°FPressure drop = 1000 psi
Dynamic, 40 sec–1 shear rate
Static
10 md
62 md
20 40 60 80 100 120 1400
2
4
6
8
10
12
14
16
She
ar r
ate,
sec
-1
0
50
100
150
250
200
300
350
Time, min
Flui
d-lo
ss v
ol, m
l
Shear rate
nDynamic versus static fluid-loss test results. For this crosslinked guar system, dynamicfluid-loss values are higher than static measurements. As shear rate increases, leakoffincreases. Here, rock permeability is 0.5 md.
nEffect of shear rate on dynamic fluid loss. Fluid loss increases with increasing perme-ability, illustrating the importance of controlling leakoff in high-permeability forma-tions. Maintaining high fluid efficiency is critical to creating a cost-effective fracturewith minimal formation damage.
Pressure sensors along the core monitor theprogress of the polymer front.
Laboratory tests show that, for compara-ble fluids and rocks with permeabilities ofup to 50 md, fluid loss is greater underdynamic conditions than static conditions(left). Further, examining the impact of shearstress and permeability on the magnitude offluid loss and the effectiveness of leakoff-control additives in high-permeability for-mations led to five key conclusions.11
First, high shear rates can prevent the for-mation of an external filter cake and resultin higher than expected spurt (below, left).Second, an internal filter cake controls fluidloss, especially near the fracture tip. Third,the effectiveness of fluid-loss additivesincreases with formation permeability anddecreases with shear rate and fluid viscosity.Fourth, reducing fluid loss means reducingspurt, particularly under high shear condi-tions and in high-permeability formations.Finally, at high shear rates with no externalfilter cake, efficient spurt control must beachieved by plugging the pore throats at thesurface of the rock.
The effect of shear depends on the type offluid and the formation permeability. Typi-cally, above a threshold shear level, noexternal filter cake is formed. The magni-tude of fluid loss is dependent on the type ofpolymer and whether it is crosslinked. If thepermeability is high enough and the fluidstructure degrades with shear, polymer maybe able to penetrate the rock matrix.
Dynamic tests revealed that commonlyused additives were less effective in control-ling fluid loss than static tests had previouslyindicated. Also, a direct link between fluidefficiency and shear rate was demonstrated.The higher the fraction of fluid lost underhigh shear early in the treatment, the higherthe total leakoff volume and the lower theefficiency. Spurt has a dominant effect onefficiency and the volume of fluid pumped,particularly for the short pumping timesencountered in fracturing high-permeabilityformations. If spurt is not controlled quicklyand effectively during the high-shear period,fluid efficiencies drop dramatically.
These observations prompted researchersto develop a superior additive system that:• controls spurt under high shear rates in
high-permeability formations• minimizes the influence of permeability
on leakoff12
• limits the invasion of polymer into the matrix
• reduces the overall amount of polymerpumped into the fracture.
41Autumn 1995
Time, min
010 20 30 40 50
4
8
12
16
Flui
d-lo
ss v
ol, m
l
15 lbm/1000 gal HIGHSHEAR
25 lbm/1000 gal best available
No fluid-loss additive
nEffect of additives on fluid loss. The HIGHSHEAR additive system provides substantialimprovement in controlling leakoff compared to the best materials available today.
11. Navarrete et al, reference 6.12. In high-permeability formations, permeability can
vary widely. Using an average value in job designcalculations can present execution problems duringthe treatment. An additive that dampens the effect ofpermeability variation can be beneficial.
13. Brannon HD and Pulsinelli RJ: “Evaluation of theBreaker Concentrations Required to Improve the Per-meability of Proppant Packs Damaged by HydraulicFracturing Fluids,” paper SPE 19402, presented at theSPE Formation Damage Control Symposium,Lafayette, Louisiana, USA, February 22-23, 1990.
14. In instances where the impact is significant, the dam-age can be countered by lowering the polymer con-centration since fluids of lower viscosity can be usedduring the treatment.
15. Hawkins GW: “Laboratory Study of Proppant-PackPermeability Reduction Caused by Fracturing FluidsConcentrated During Closure,” paper SPE 18261,presented at the 63rd SPE Annual Technical Confer-ence and Exhibition, Houston, Texas, USA, October2-5, 1988.
16. Above this level, the fluid breaks with time due to thethermal degradation of the guar molecules.
Evaluation of several additive types led toa combination of particulate materials, theHIGHSHEAR system, that achieves theabove goals. One agent moves rapidly tothe fracture face during the early stage offluid loss when leakoff flux is high. This typeof particle seals a major portion of theexposed surface quickly and adheressecurely to the surface, resisting high shearforces. The second material plugs theremaining gaps in the developing filter cakeas the shear rate drops, significantly reduc-ing filtrate losses and sealing the surface sothat polymer particles cannot penetrate thematrix. Laboratory tests show a 25 to 75%reduction in spurt compared to the bestavailable products today (right). Fluid effi-ciency improves, meaning less fluid topump, less fluid to break and easiercleanup. Since this process is aided by hav-ing a less viscous fluid—one that promoteshigh initial leakoff—there is a further oppor-tunity for cost-savings and improvedcleanup by reducing the polymer concen-tration in the fluid (page 46).
One concern with fluid-loss additives hasbeen the possibility of their presence reduc-ing proppant-pack conductivity. A key ques-tion is: Does the reduction in matrix dam-age brought about by additives outweigh thepotential damage they may inflict within thefracture itself? As is well-known, the amountof conductivity damage is highly dependenton the type of fluid used.13 Flowback testswith different fluids and leakoff controladditives confirm that particulate fluid-lossadditives do limit matrix damage by mini-mizing fluid invasion. Conductivity testsshow that these agents are actually lessdamaging to the proppant pack than previ-ously thought. In most fracturing fluids, theirimpact on conductivity is minimal whenused in low concentrations.14
With these encouraging laboratory results,the next step was to test whether the HIGH-SHEAR system could reduce treatment costsand improve well productivity better thanconventional products. Extensive field trialsin Canada show an overall 15 to 20% costsavings (see “Canadian Treatments Demon-strate Cost Savings and Productivity Gains,”next page). Productivity improvements aver-aged 460% compared to 260% on offsetwells without the additive. Additional welltests are planned in areas such as the USGulf Coast where a variety of fluids are usedand larger-scale, high-permeability fractur-ing treatments are performed.
Understanding and Improving FractureConductivitySimply creating a fracture does not guaran-tee better well performance. The fracturemust provide a conductive flow path for for-mation fluids. For decades, poor treatmentresults were blamed on insufficient fracturelength or inadequate proppant transport andplacement. Studies in the late 1980srevealed that substantial fracture damage,and resulting impairment to flow, could becaused by polymer residue blocking thepore spaces between proppant particles.15
Once the proppant pack is placed andpumping stops, fluid filtrate leaks off into therock matrix and the pressure declines. Thefluid remaining in the fracture must then beflowed back to allow production of hydro-carbons. Termed cleanup, this process is crit-ical to the success of the treatment. One wayto aid cleanup is to ensure that the polymerresidue has been reduced to a minimum. Asecond is to use as little polymer as possibleto start with. The next two sections look atrecent developments in both areas.
Extending Breaker TechnologyAt the end of a treatment, the fluid left in thefracture has been partially dehydrated due tofiltrate loss. The effective polymer concentra-tion can be an order of magnitude higherthan that originally pumped, reaching 300to 600 lbm/1000 gal [36 to 72 g/cm3]. If thepolymer stays intact, an ultrahigh viscosity,gelled mass results that blocks the porespace and cannot easily be flowed back intothe well.
To prevent this, the polymer is attacked byfluid breakers—oxidizers or enzymes thatsever the polymer chain at its weakestpoints, degrading it into smaller, moremobile fragments. This reduces the viscosity
of the residual fluid, thereby allowing moreefficient cleanup. Breakers are used at reser-voir temperatures below about 325°F[163°C].16 If breaking is insufficient, con-centrated polymer remains in the proppantpack, reducing the conductivity and treat-ment effectiveness.
Historically, active chemical breakerswere dissolved in the fluid during surfacemixing. As a result, the fluid was beingattacked even as it was being pumped. Carehad to be taken to use a sufficiently lowbreaker concentration. Otherwise, viscositywould decrease too quickly, and proppantwould settle. With this low breaker concen-tration, only partial degradation of the poly-mer occurred. The result was impaired frac-ture conductivity.
Research during the past decade hasfound ways to increase breaking efficiency.
42 Oilfield Review
A B C D E F G H I J K L M N O P Q R S
Pro
duct
ion
rate
, m3 /
day
0
5
10
15
20
25
Wells
30
Rate before refrac
Rate after refrac
Pro
duct
ion
rate
, m3 /
day
20
18
16
14
12
10
8
6
4
2
0Well
1Well
2Well
3Well
4Well
5
Rate before refrac
Rate after refrac
■■Field trial of fluid-loss additive and low-guar fluid in Alberta, Canada.
■■Comparison of pro-duction rates beforeand after treatment.Five wells were refrac-tured in 1993 usingconventional fluidstechnology, resulting in an average 260%improvement in pro-duction rate (top). Nineteen wells refrac-tured in 1995 using theHIGHSHEAR fluid-lossadditive and the low-guar fluid showed anaverage 460% improve-ment (bottom).
Canadian Treatments Demonstrate Cost Savings and Productivity Gains
More than 19 operational trials of the HIGHSHEAR
fluid-loss control additive have been conducted
in Alberta, Canada in clean sandstone reservoirs
(right). Here, permeabilities vary from 50 to
150 md and porosities are 20 to 24%. Well depths
typically range from 2300 to 2500 ft [700 to
760 m] with bottomhole temperatures of 85 to
95°F [30 to 35°C].
Earlier wells fractured in the field in 1993, as
part of a refracturing program, yielded proppant
placement efficiencies as low as 70%. Because of
the relatively high permeability, its variability
within the reservoir and high-concentration prop-
pant schedules, an error of 10% in the fluid-loss
design calculation could lead to 30% of the prop-
pant not being placed. In addition to verification
of laboratory test results with the additive, objec-
tives for a new refracturing campaign were:
• reduced treatment cost
• improved job design and execution
• decreased volume of polymer pumped
• reduced well cleanup times
• increased production rates.
To achieve these goals, the strategy was to
replace the 30 lbm/1000 gal borate-crosslinked
fracturing fluid normally used with a reduced-
polymer, 22 lbm/1000 gal system (page 46) to
promote higher fracture conductivity, to speed
cleanup (due to less damage from residual
polymer) and to lower wellsite costs. With its
inherently lower viscosity, the new fluid meant
higher leakoff. To counteract this, the HIGHSHEAR
additive was used to limit spurt, seal the
fracture face and smooth out the effects of
permeability variations.
Pad volumes of 2000 to 2500 gal [7.6 to 9.5m3]
were used with total slurry volumes of 4000 to
6000 gal [15.1 to 22.7 m3] in four stages. Typical
rates were 15.7 bbl/min [2.5 m3/min]. Maximum
sand concentrations ranged from 16.5 to more
than 18.5 ppa.
The field trials were highly successful. Test
wells showed consistent proppant placements of
95 to 100%. Over time, as experience with the
system grew, pad volumes were maintained and
Autumn 1995
■■Retained permeability as a function ofporosity. The theoretical correlation relat-ing the two properties of the proppantpack has been confirmed in laboratorytesting. Small decreases in porosity canlead to significant reductions in retainedpermeability, resulting in decreasedhydrocarbon production.
Fractional reduction in porosity
Frac
tiona
l ret
aine
d pe
rmea
bilit
y
0.2 0.4 0.6 0.8 10
0.2
0.4
0.6
0.8
1
Theoretical model1 Layer2 Layer, test 12 Layer, test 2
By encapsulating breakers so they do notinteract with the fluid until released by rup-ture of the protective coating with time orstress, higher concentrations can be used,degrading more of the polymer. This hasbeen a major innovation for improvingcleanup and proppant-pack conductivityand is a common industry practice today.17
The mechanisms of proppant-pack dam-age have been the subject of much studyand much controversy. Recent findings haveled to a more consistent and reproducibleapproach to testing proppant-pack damage.By understanding the controlling processes,scientists have been able to develop new-generation additives to improve fluidcleanup and fracture conductivity.
A keystone of the investigation was devel-oping a relationship between retained per-meability and porosity reduction in prop-pant packs. A graph of this correlationshows that, for example, a 10% change inporosity (from 30 to 27%) can reduce per-meability by 35%. Laboratory tests verify thetheoretical relationship (above, right). Small,random blockages by residual polymer or athin, concentrated polymer filter cake at the
50% retained permeability 25% re
75% by filter cake
75% re50%25%
17. Gulbis J, King MT, Hawkins GW and Brannon HD:“Encapsulated Breaker for Aqueous Polymeric Flu-ids,” paper SPE 19433, presented at the SPE Forma-tion Damage Control Symposium, Lafayette,Louisiana, USA, February 22-23, 1990.
fracture face lead to large reductions inretained permeability (below).
Another part of the study involved identi-fying the parameters and test procedureswith the most significant effects on mea-sured conductivity. Tests were performed atthe Dowell laboratory in Tulsa, Oklahoma,USA and two independent laboratories. The
■■Pore blockages.Retained perme-ability can bereduced by anexternal filter cake (upper left). The larger the filtercake, the lower the permeability.Retained perme-ability can also bereduced by random,undegraded poly-mer fragments fill-ing the void spacesbetween proppantparticles. As theresidue increases,the permeabilitydecreases (upperright, lower left, lowerright). Even smallincreases in residuecan have a dra-matic effect and in both cases, the permeabilitydecrease can dra-matically impactproduction rates.
tained permeability
tained permeability
higher sand concentrations were achieved as a
result of increasing placement success. Less
fluid pumped meant less fluid to break. Cleanup
times were slashed from several days to one or
two days. Post-treatment production increased
from 260% in the 1993 campaign to 460% (previ-
ous page, middle). Overall treatment costs were
15 to 20% lower.
The changeover to the lower polymer loading
was successful due to the fluid-loss additive.
In separate trials without the material present,
proppant placement efficiency was erratic.
According to Gilbert Dumont, production engi-
neer for Petro-CanadaResources, “Fluids are
coming back crystal-clear with this system,
something we haven’t seen before. The wells are
cleaning up quicker, in one to two days rather
than three to ten days. This indicates we are get-
ting faster and more complete breaking of the
system. We believe our conductivities are higher
as supported by increased well productivity.”
Michael Priaro, engineering consultant for
Petro-Canada Resources adds, “Our success rate
is greater because of more consistent placement.
We’ve been able to obtain higher proppant con-
centrations with cleaner fluid. The cleanup has
been better and quicker and well productivity
after stimulation is definitely higher, averaging
4.6-fold improvement, with some wells showing
almost 7-fold improvement. Ten-fold production
improvements may be possible with further
advances in treatment design.”
From Neal Wasylycia, production engineer for
Petro-Canada Resources, “Comparisons of
results are difficult with slant, deviated and
directional wells with varied pays being frac-
tured. We definitely know that our results on ver-
tical wells are dramatically better, and 100%
fracture placement can be repeated with confi-
dence. Our overall refracturing costs are reduced
due to lower fluid cost, quicker cleanup and
reduced fluid disposal cost. The successful
placement of fractures in deviated and slant wells
represents our next challenge, as this is where
our next large refracturing opportunity exists.”
43
4
Lab 1 Lab 2 Lab 3
Test 1Test 2
Ret
aine
d pe
rmea
bilit
y, %
35
30
25
20
15
10
5
0
results confirm that polymer concentration,for example, is critical. Its influence is non-linear (below).
The relationships of other parameters totest results and reproducibility were alsoquantified, leading to a refined testing pro-cedure. Testing by these laboratories usingthe new guidelines demonstrates that reli-able results can be achieved, provided testconditions are closely controlled. Testingwithin the same laboratory is expected tovary by 15 to 20% while lab-to-lab varia-tions have a 25 to 30% relative deviation(right ) . Before uniform procedures, testresults often deviated by 100% or more.
4
0 100 200 300Polymer concentration, lbm/1000 gal
Ret
aine
d pe
rmea
bilit
y, %
100
80
60
40
20
0
Borate crosslinked guarTemperature = 160°F
+
+ an
nEffect of polymer concentration onretained permeability. As the residualpolymer concentration remaining in theproppant pack increases, the permeabilitydecreases, emphasizing the need for com-plete polymer degradation and removal.
nEnhanced proppantresults in polymer fragfrom the fracture. Thethe breaker (bottom) tocleanup and improve
nInterlaboratory comparison of conduc-tivity results. Major improvements inreproducibility stem from consistency intesting methods and close control overtest conditions, removing a major road-block in correlating data between labs.
The thorough investigation of conductivitytesting provided a road map of how toimprove breaker effectiveness. The problemwas attacked on two fronts: selection of anoptimal breaker for a given temperaturerange and development of an additive toassist in removal of the degraded polymer.
Each type and composition of breaker areeffective over a certain limited temperaturerange, based on performance and cost crite-ria. This requires tailoring of chemical prop-erties and encapsulation technology so thatthe breaker remains active in the proppantpack long enough to do its job. A suite ofmaterials is needed to cover the range oftemperatures that are encountered in frac-turing operations.
But a breaker by itself is often not enough.Conductivity testing shows that residualfragments left after the primary polymerstructure has been degraded can still causesignificant pore blockage (below). Worse,under certain conditions, the fragments cancoagulate—or bind together into a viscousmass—and reduce conductivity further.Sometimes, adding more breaker can aggra-vate coagulation.
To avoid this, a special blend of anticoag-ulants was developed. This blend, theCleanFLOW additive, works synergisticallywith the breaker to reduce the size of thepolymer fragments and prevent their ten-dency to coagulate. The dispersive action ofthe additive increases mobility and available
breaker
CleanFLOWd breaker
-pack cleanup. Use of a breaker alone (top)ments that may be difficult to remove
CleanFLOW additive system works with prevent fragment coagulation, facilitate
fracture conductivity.
flow paths. Pore blockage is reduced andproppant-pack permeability increases.
Laboratory testing in standard conductiv-ity cells shows that the breaker plus addi-tive system outperforms breaker alone,yielding almost 40% higher proppant-packpermeabilities at low-end additive concen-trations (next page, top). Retained perme-abilities improve to as much as 90% atincreased concentrations.
In field testing, treatments with theCleanFLOW additive showed higher levelsof polymer returns during flowback thanprevious treatments on offset wells. Higherreturns are a direct measure of improvedproppant-pack cleanup. Over a given flow-back period in a well in western Wyoming,USA, 62% of the polymer was returnedwhen the additive was used with breaker,compared to 31% for an offset well with thebreaker alone. At the end of the test phase,the well treated with the CleanFLOW addi-tive was still returning polymer, while theoffset well was not (next page, bottom).
Net present value (NPV) calculations onvarious formation types show that conduc-tivity-related production increases realizedfrom the use of the additive can significantlyimprove well economics.
Oilfield Review
18. Gulbis J: “Fracturing Fluid Chemistry,” in Econo-mides MJ and Nolte KG (eds): Reservoir Stimulation,2nd ed. Englewood Cliffs, New Jersey, USA: PrenticeHall (1989): 4-1–4-14.
Reducing Polymer ConcentrationToday, over 70% of the fracturing treatmentsconducted use guar- or hydroxypropyl guar-based fluids.18 The rheology of such fluidshas been studied for years. When added towater, guar molecules hydrate and swell,increasing in diameter and length. Thehydrated strands overlap and hinder motion,giving rise to an increased viscosity of thesolution. For adequate proppant transportand placement, a viscosity of 100 centipoise(cp) at a shear rate of 100 sec-1 is generallyaccepted as a minimum guideline.
To minimize fluid leakoff and counteractthe inherent thinning of guar systems at ele-
Autumn 1995
0200 400 600 800 1000
40
35
30
25
20
15
10
5
Cumulative flowbac
Pol
ymer
con
cent
ratio
n, lb
m/1
000
gal
31% polymer retur
CleanFsyste
2 gal/10
No breaker
Ret
aine
d pe
rmea
bilit
y, %
Breaker only0
10
20
30
40
Temperature = 225°F
vated temperatures, however, higher viscosi-ties have historically been used. In non-crosslinked (linear) systems, this meansadding more polymer, resulting in polymerconcentrations of 40 lbm/1000 gal [4.8g/cm3] or higher. This approach is expensiveand often results in fluid mixing and han-dling difficulties. Fracture conductivity canbe impaired, since more polymer must bebroken and produced back. Crosslinkinghas become a common means of enhancingviscosity at lower polymer levels. Whileeffective in building viscosity, this practicecan lead to complicated, hard-to-breakstructures. Using crosslinkers, polymer lev-
1200 1400 1600 1800 2000 2200 2400
k volume, bbl
CleanFLOW system and breaker
Breaker only
62% polymer return
n
LOWm00 gal
CleanFLOWsystem
4 gal/1000 gal
nConductivitycomparison. Thenew additive sys-tem provides sub-stantially greaterretained perme-abilities comparedwith no breaker orwith use of abreaker alone, con-firming that theadditive enhancesremoval of residualpolymer and limitspore blockages.
els can be reduced to about 25 lbm/1000gal [3 g/cm3]. The goal of recent researchhas been to formulate a reliable fluid usingeven less polymer to reduce cost andimprove fracture conductivity.
Various metal ions, including titaniumand zirconium, have been used for decadesas crosslinking agents. In recent years,boron [as B(OH)4
-] has grown in popularityand is by far the most common elementtoday. Various boron salts and compounds
45
nPolymer flow-back comparisons.For a test well inwestern Wyoming,USA, 62% of thepolymer wasreturned duringflowback. For anoffset well withbreaker alone, dur-ing a similar flow-back time, only31% of the polymerwas returned.
nBorate ion fraction as a function of tem-perature and pH. The amount of borate ionin solution directly impacts the crosslinkingprocess. As the pH increases past 8, theamount of borate available for crosslinkingincreases dramatically. The solubilityeffect gives rise to the reversible crosslink-ing exhibited by borate fluid systems.
75°F100°F150°F200°F250°F300°F
1
0.8
0.6
0.4
0.2
05 6 7 8 9 10 11 12 13 14
pH
Bor
ate
ion
fract
ion
can be used (below). If the pH19 of the fluidis above 8, crosslinking occurs almostinstantaneously when borate ions are addedto hydrated guar (right). A unique feature ofthe resulting fluid is that the crosslinking isreversible. As temperature increases, the pHfalls and the solution thins because there isless borate ion available in solution. Viscos-ity recovers as temperature decreases andthe pH rises.
If crosslinking is too rapid, high frictionpressure in wellbore tubulars and shearthinning may occur during pumping. Thismay not be a severe problem in shallow,low-temperature wells where pumpingtimes are short, but is of major concern fordeep, higher temperature treatments requir-
46
Borate Crosslinking of Guar
Structure of Guar
B(OH)3 + OH- B(O
B(OH)4- +
B(OH)4- +
R
2R
OH
OH
OH
OHR
O
O
RO
O
H
CH2
H H
OH OH
H H
O
H
HO
H
HO
H
O
CH2OHCH
2
H
OH OH
H H
OH OH
H
Galactosubstitu
Mannback
R= Mannose backbone
H
CH2OH
HHO
H
OH H
OH
O
H
CH2OH
HOH
OH
OH
H
OHO O
ing extended pumping times. For theseapplications, it is necessary to delay thecrosslinking process until the fluid has trav-eled through most, or all, of the well tubu-lars. This can be accomplished by varyingthe chemistry and additives in the fluid or byencapsulating the crosslinker, which permitstimed release of the active material. Delayedcrosslinking reduces friction pressure andhorsepower requirements and provides forhigher injection rates.
For a stable and reliable borate crosslinkedfluid to be prepared at a given temperature,sufficient polymer chains must be present forentanglement to occur. For most guars, solu-tions containing less than 20 lbm/1000 galcannot effectively be crosslinked. In addition
H)4-
BOH
OH+ 2H2O
BO
O
R + 4H2O
H
H
HO O
H
OH OH
H H
O
CH2
OH
seents
osebone
nStructure of guarand crosslinkingby boron. Guar iscomposed of amannose back-bone with intermit-tent galactose sub-stituents (top).Boron, in the formof B(OH)4-, reactswith the hydroxyl(OH) groups on thepolymer in a two-step process to linktwo polymerstrands together(bottom).
to having sufficient polymer in solution, twoother criteria must be satisfied: the chainsneed to have enough active crosslink sitesand the proper number of borate ions mustbe present to build a network structure.Careful balance between the two has to bemaintained to produce a stable system.
In the past, borate crosslinked fluids at thelower threshold of 20 lbm/1000 gal guarconcentration have been formulated in thelaboratory and even tested in the field.Operational experience, however, showedthem to be unreliable and overly sensitive tosmall variations in fluid chemistry.
How can the threshold polymer level nec-essary for crosslinking be reduced success-fully? Based on earlier work with zirconatesystems, scientists evaluated what types ofmaterials associate with guar molecules toincrease their solubility. After studying thebehavior of a variety of materials, researchersidentified a combination of chemicals thatpermits stable borate crosslinking at polymerconcentrations as low as 15 lbm/1000 galand reliable field formulations to be mixed inthe 15- to 20 lbm/1000 gal range (next page,bottom).
These systems, referred to as low-guar flu-ids, can be used at fluid temperatures up to175°F [80°C]. They exhibit viscosities nor-mally measured for borate crosslinked fluidswith polymer concentrations 5 to 10lbm/1000 gal higher.
Proppant suspension properties are excel-lent, allowing use of proppant concentra-tions as high as 17 to 20 pounds of proppantadded (ppa). The fluid can be continuouslymixed and then crosslinked. It can be effec-tively degraded using available breaker sys-
Oilfield Review
nRetained conduc-tivity of low-guarsystems. By reduc-ing the polymerconcentration,cleanup is easierand the retainedconductivity of theproppant pack isgreater than forconventional fluidsthat require higherpolymer concen-trations.
Fluid Temperature (°F)Permeability
(darcies)Percent Retained
Conductivity
Proppant Type: Sand
Proppant Concentration: 2 lbm/gal
2% KCI20 lbm30 lbm
175175
125
32
17561
-4926
2% KCI 125 216 -15 lbm 125
125130106
604920 lbm
125 63 2930 lbm
2% KCI 100 244 -15 lbm 100 148 61
100 103 4220 lbm
tems for rapid, efficient cleanup followingthe treatment. Conductivity tests on proppantpacks placed with low-guar fluids show 150to 200% higher pack permeabilities thanthose obtained with conventional polymerconcentrations (right).
Over 500 treatments have been pumpedsince the introduction of the fluid in 1994.Because of the success of these treatments,the low-guar system is rapidly becoming thefluid of choice for applications to 175°F.
Reduced polymer loading can lead toincreased fluid leakoff. This can be avertedby the adding fluid-loss agents, particularlyin higher-permeability formations. Field stud-ies, however, have shown that this may notalways be necessary, depending on the typeof formation and its permeability. In morethan 130 jobs performed in Kansas, USA in5- to 50-md formations, no fluid-loss addi-tives were used. Fluid efficiency was notadversely affected, and conductive fractureswere obtained. Cleanup times were reducedby 50% compared to previous treatments.Proppant placement efficiencies—the per-centage of proppant effectively placed in thefracture—met or exceeded 97%.
In even higher-permeability formations,as tests in Alberta, Canada (page 42) con-clusively demonstrate, low-guar systemscan be used effectively with the newHIGHSHEAR fluid-loss additive to improvefracture conductivity, well cleanup and wellperformance.
Autumn 1995
Vis
cosi
ty a
t 100
sec
-1, c
p
Visc
osity
at 1
00 s
ec-1
, cp
180
160
140
120
100
80
60
40
20
0
400
350
300
250
200
150
100
50
01 2 30 0Time, hr
100°F125°F150°F
15 lbm/1000 gal
Controlling Proppant FlowbackFlow of proppant into the wellbore follow-ing a fracturing treatment is of major con-cern. This phenomenon may occur duringinitial cleanup or sometime after the well isput back on full production. Termed prop-pant flowback or proppant backproduction,it can lead to expensive, time-consumingremedial operations and safety concerns. Inlow-rate wells, proppant may settle in thecasing, requiring periodic wellborecleanouts. Loss of near-wellbore fractureconductivity can result, and production maycease entirely if the productive zone is fullycovered. In high-rate wells, erosion occursto tubulars, control valves and wellheadequipment. Disposal costs for producedproppant may be substantial.
The frequency of this problem hasincreased markedly as greater fracture
nViscosity of low-guar fluids overtime. Now, reliablefluids can be for-mulated with poly-mer concentrationsof between 15lbm/1000 gal (left)and 20 lbm/1000gal (right). Even atthe lower polymerconcentration, vis-cosities are abovethe 100-cp levelrequired for ade-quate proppantsuspension andtransport.
1 2 3Time, hr
20 lbm/1000 gal
widths and higher proppant concentrationshave become the norm. In areas such asAlaska and the North Sea, up to 20% of theproppant may be produced back, whilesome instances of up to 50% have beenreported.20 This can translate into anywherefrom 1000 to 100,000 lbm [454 to 45,400kg] of proppant per treatment. Although theflowback may stop with time, many wellsproduce proppant throughout their life-times. Wells often must be placed onrestricted chokes to limit pressure dropsand stabilize the proppant pack.21
While changes in fracture design and exe-cution can sometimes alleviate the problem,a typical solution has been the use of resin-coated proppant (RCP). RCPs are pumpedinto the fracture near the end of the treat-ment, referred to as tailing in. The well may
19. The measure of acid intensity equal to the logarithmof the reciprocal of the hydrogen ion concentrationof the solution. pH 7 is neutral; below 7 is acidic,above alkaline.
20. Martins JP, Abel JC, Dyke CG, Michel CM and Stew-art G: “Deviated Well Fracturing and Proppant Pro-duction Control in the Prudhoe Bay Field,” paperSPE 24858, presented at the 67th SPE Annual Tech-nical Conference and Exhibition, Washington, DC,USA, October 4-7, 1992.
21. Vreeburg R-J, Roodhart LP, Davies DR and PennyGS: “Proppant Backproduction DuringHydraulic Fracturing—A New Failure Mechanismfor Resin-Coated Proppants,” Journal of PetroleumTechnology 46 (October 1994): 884-889.
47
48 Oilfield Review
Crit
ical
flow
rate
, lite
r/m
in
Weight fraction fiber, %
9
8
7
6
5
4
3
2
1
0 0.5 1 1.5 2 2.5
nPropNET proppantflowback control.Random glass fibers pumped inthe fracturing fluidform a net-likestructure that stabi-lizes the proppantpack while allowinghigh productionrates of formationhydrocarbons.
nEffect of fiber concentration on criticalflow rate. As fiber is added, up to about1.5% by weight, the flow rate necessary tofail the pack increases. Beyond 1.5%,additional fibers do not improve packstrength significantly.
be shut in for a period of time to allow theresin to cure as its temperature rises, bindingthe proppant particles together at theirpoints of contact. Ideally, a consolidated,high-conductivity matrix is formed.
RCPs, however, are not universally appli-cable and have certain severe limitations.Performance is sensitive to shear, tempera-ture, closure pressure and shut-in time. Con-ductivities are frequently lower thanexpected. In low-temperature wells, anexpensive activator must be added to theRCP at a 0.5 to 2.0% concentration for theresin to cure. Shut-in times may be as longas 24 hours. Resin coatings can interactchemically with fracturing fluid additives.Cyclic loadings, caused by the well beingshut in and put back on production overtime, can cause the pack to fail. In extremecases, gelled masses of resins can be pro-duced back.22
Because of these drawbacks, there is amajor incentive to introduce a more consis-tent means of controlling proppant flowbackthat also improves well cleanup efficiencyand maximizes well productivity.
Recent research has helped define themechanisms underlying proppant-packdegradation and has led to the invention ofa physical, rather than a chemical, solutionto the problem. This innovation, calledPropNET technology, uses fibers to hold theproppant in place. Pumped together withthe proppant in the fracturing fluid, the
material forms a web, or network, whichstabilizes the proppant-fiber pack andallows high production rates of oil or gas(above). The technology is based on theprinciples of fiber reinforcement commonlyused in a variety of industrial and commer-cial applications as a strengthening method.For example, natural and synthetic fibers areused to protect dams and other concreteand soil structures from erosion. The inher-ent ability of fibers to stabilize highlyporous, particulate-containing materialsprovided a basis for these investigations.
A comprehensive set of tests has beenapplied to determine the applicability offiber reinforcement and to characterize theproperties and performance of fiber-contain-ing packs. Conductivity test cells measuredproppant-pack permeability. Special pack-mobility tests simulated conditions before,during and after well cleanup. Three config-urations of test cells were developed andconstructed to evaluate the key parame-ter—pack resistance to proppant flow-back—as measured by the maximum flowrate that the pack can withstand or the max-imum pressure drop across the pack beforeproppant is produced.
A variety of fiber types were investigated,including polymer, glass, ceramic, metaland carbon. Based on several evaluation cri-teria, a special, flexible glass fiber was cho-sen for its performance, cost and availability.The material has a 2.55 g/cm3 [21.3lbm/gal] bulk density, similar to that of mostproppants. Selection of the particular glassfiber was based on its stability to fluids anddownhole conditions. Fibers must meetminimum size criteria to be effective. Short,small diameter fibers are less effective instrengthening the proppant pack. Glass fiberdiameters of 10 to 20 microns and lengthsof 10 mm or more provide optimum packstability and ease of handling.
How the fibers work is open to debate,but it is thought that they interweave among
the proppant particles, providing increasedstrength, or that they stabilize and distributestress, aiding bridging, within a significantarea of the pack. The fiber structure is moreflexible than cured RCPs, allowing the prop-pant-fiber pack to shift without failing.
Laboratory tests show that the ability ofthe pack to resist proppant flowback is afunction of fiber concentration. Stabilityincreases with fiber content until a plateauis reached (below). While laboratory datashow that use of 1.5% fibers by weight canreduce permeability by up to 30% com-pared to packs without fibers,23 field resultsshow less reduction. Conductivity values forpacks with fibers are superior to those mea-sured for postcured RCPs.24
Despite the low concentration of 1.5%,the fiber level is about 30% by number ofparticulates, or about one fiber for every twoproppant particles. Fiber length is an orderof magnitude more than proppant diameter,so, for example with sand proppants, eachfiber touches approximately five particles(next page, top).
Single-phase flow tests using water andtwo-phase tests with water and gas were
conducted without a confining stress andwith a stress of 1000 psi. Results show thatpack stability increases with pack width to acertain limit and that packs with morespherical and uniform ceramic proppant aregenerally less stable than irregular grainsand packs. Packs can withstand pressure
49Autumn 1995
Max
imum
pre
ssur
e dr
op b
efor
e fa
ilure
, psi
/ft
140
120
100
80
60
40
20
00 20 40 60 80 100
Nitrogen content, % total flowing
nPhotomicrographof fiber-reinforcedproppant pack.The glass fibers areabout an order ofmagnitude longerthan the diameterof a typical prop-pant particle,allowing each fiberto contact aboutfive particles. Thefibers stabilize thepack by inter-weaving amongparticles, provid-ing increasedstrength and sta-bility. They mayalso promotebridging and stressdistribution withinthe pack.
nResistance of sand-fiber packs to prop-pant flowback. At a fiber content of 1.5%,the pressure drop the pack can withstandis a maximum when no nitrogen is pre-sent, exceeding 100 psi/ft. As the nitro-gen content increases, pack strengthdecreases until a minimum is reached atabout an 80% gas volume.
22. Almond SW, Penny GS and Conway MW: “FactorsAffecting Proppant Flowback with Resin CoatedProppants,” paper SPE 30096, presented at the Euro-pean Formation Damage Conference, The Hague,The Netherlands, May 15-16, 1995.
23. This observation is consistent with calculations inwhich fibers represent 5 to 6% of the pore volume.
24. Card RJ, Howard PR and Feraud J-P: “A Novel Tech-nology to Control Proppant Backproduction,” paperSPE 31007, SPE Production & Facilities (in press).
gradients in excess of 100 psi/ft, but typicallevels are 45 psi/ft (right). Worst-case flowconditions (80%/20% gas-to-water ratio)were used to establish maximum flow-rateguidelines. Under these conditions, packstability reaches a minimum at about 40%of the maximum pack strength in single-phase fluid flow. Based on laboratory data,maximum cleanup rates of 30 bbl/day/per-foration [4.8 m3/day/perforation] for sandand 20 bbl/day/perforation [3.2 m3/day/per-foration] for ceramic proppants are beingused in the field.
Tests results show that no minimum con-fining stress, shut-in time or reservoir tem-perature are required for the fibers to beeffective, overcoming some of the severestlimitations of RCPs.
Packs were cycled under stress to simulateshut-in and production periods, with packssubjected to alternating stress levels of 1000psi and 4000 psi. No pack failures wereobserved, even after more than 30 cycles,for both sand and ceramic proppant packs.Aging studies were also conducted, sinceglass fibers can be dissolved by formationwaters. The solubility rate depends on sev-eral factors, including temperature, pH and
the type of minerals present in the water (sil-ica being the most important). Results showthat fibers are expected to retain 50% oftheir effectiveness for at least two years at300°F [149°C] when in contact with a sil-ica-saturated formation brine. Lifeexpectancy could be greater, depending ondownhole conditions.
Glass fibers do not interact with commonfracturing fluid systems or additives, a keyconcern with RCPs. Their presence in thefluid slurry also reduces proppant settling,aiding proppant transport and placement.Glass fibers have certain limitations to beconsidered during treatment design. Theyare not effective at temperatures above300°F and under certain conditions: if for-
50 Oilfield Review
Economic Benefits of Fiber Reinforcement: Field Results
■■Fluid returns in asouth Texas well.Faster flowback ratesand larger cumulativeflowback volumeswere found with thePropNET system thanwith RCPs. Gas break-through was muchearlier and productsales were realizedsooner.
■■Decrease in flowbackcosts for a south Texaswell. With a muchshortened flowbacktime, well monitoringand associated costsare substantiallyreduced, providing anincreased return oninvestment for thetreatment.
■■Fracture treatment in south Texas.
400
300
200
100
0
Cum
ulat
ive
fluid
retu
rned
, bbl
7010 20 30 40 50 60Flowback time, hr
Gas
RCP
PropNET
Gas
10,000
0
Flow
back
cos
ts, $
142 4 6 8 10 12
Flowback time, days16
20,000
30,000
PropNET flowback
RCP flowbackFlowback costsdecreased by$12,600
More than 150 hydraulic fracturing treatments using PropNET fiber-reinforced packs have been conducted in
the USA (top) and Venezuela. Here is a look at three typical examples.
Texas, USA:
A 10,000-ft [3080-m] gas well in south Texas,
USA with a bottomhole temperature of 275°F
[135°C] was fractured using a borate-crosslinked
fluid. A fiber-reinforced pack was placed with
1.5% fiber by weight of proppant. Fibers were
added during the entire proppant stage. Pumping
pressure levels were similar to treatments with-
out fibers present. Of the nearly 16,000 lbm
[7300 kg] of proppant placed, less than 0.07%
was produced back. The well was later cycled
through four shut-in and production periods, with
a shut-in closure stress of 1900 psi and a flowing
tubing pressure of 4200 psi. Production was
essentially proppant free and remains so today
well over a year later. Productivity from the well
exceeded that from offsets.
In another south Texas gas well, having two
sandstone layers separated by a shale layer,
226,000 lbm [102,500 kg] of proppant were
placed using a borate-crosslinked guar fluid.
Here, fibers were tailed in with the last 15% of
the proppant. Flowback started as soon as the
treatment was completed. The initial flowback
rate of 500 bbl/day [80 m3/day] was increased
to 1000 bbl/day [160 m3/day]. Less than 0.05%
of the proppant was produced back over a
four-day period.
How fiber reinforcement compares with RCPs
was evaluated in an offset well. An upper produc-
tive zone was fractured with a 15% fiber tail-in
treatment, while a lower zone used a 23% RCP
tail-in. The PropNET zone had a much higher ini-
tial flowback rate, earlier gas breakthrough and
more rapid fluid returns (middle). Cleanup costs
were reduced and fracturing fluid recovery was
maximized, allowing the well to be placed on
production sooner (bottom). Conductivities were
higher with PropNET treatments than for RCPs.
Autumn 1995
25. Fibers are resistant to hydrochloric acid [HCl], theacid commonly used, but not to hydrofluoric acid[HF] which is known to dissolve glass.
mation water is not silica-saturated, if con-fining stress exceeds the crush strength ofthe proppant, or if the fracture will betreated later with certain types of acid.25
While fibers could be distributed through-out the entire length of the proppant pack,field experience supports laboratory testsshowing that only the tail-in portion of thetreatment, typically the last 15 to 20% of theproppant pumped, normally requires fiberaddition. If, however, there are concernsthat the entire height of the producing zonecannot be covered or that sand control willbe a problem, it is prudent to use fibersthroughout the proppant pack.
Extensive field experience in the USA(see “Economic Benefits of Fiber Reinforce-ment: Field Results,” previous page) hasdemonstrated the superior performance offiber-reinforced packs, including reducedshut-in times, faster cleanup rates andgreater execution efficiency. This translatesdirectly into rapid well turnaround anddecreased wellsite costs.
On average, less than 0.2% of the prop-pant pumped was produced back in over150 treatments. Cleanup rates as high 4000bbl/day [636 m3/day] water and gas rates ofup to 10 MMscf/day were observed, 10times those used by local operators forRCPs. Since cleanup can represent 10 to60% of the total cost of a fracturing treat-ment, this time savings has a dramatic eco-nomic impact.
With the concept proven both in the labo-ratory and field, this technology will rapidlybe applied beyond the Western Hemi-sphere. To support this expansion, alterna-tive fibers are being tested for applicationabove 300°F in areas such as the North Sea.
New Solutions with Novel TechnologyThe impressive developments in fracturingfluids technology from 1985 to 1993 havebeen reinforced by innovations during thepast two years. Advanced fluid-loss addi-tives are improving fluid efficiency at lowerfluid viscosities. Combined with new low-guar systems, this means reduced costs andreduced damage to the fracture. Enhancedbreaker and additive systems are speedingwell cleanup, giving more complete fluiddegradation and cleaner, more conductivefractures. And, the fracture is now more sta-ble, thanks to innovations in proppant flow-back control.
Individually and collectively, these newtechnologies are benefiting oil and gas oper-ators by reducing treatment and wellcleanup costs, increasing well productivityand speeding product sales to market. In thefuture, the increased emphasis on more effi-cient and effective fluids and the synergisticapplication of low-cost innovations willyield further economic gains. —DEO
According to Gary Slusher, project production
engineer for Enron Oil & Gas, in Corpus Christi,
Texas, “We have had to live with RCPs. They can
be effective, but they are expensive and require
long cure times, and once you start pumping
them you’re committed. When we bring a well
back, we have to do it gingerly to avoid prob-
lems. With PropNET, we can start and stop pump-
ing as necessary and aggressively flow back the
wells. The system is more effective in stopping
proppant production than RCPs.
“We’ve seen both direct and indirect benefits
with PropNET. We’re getting much higher fluid
recoveries and more rapid cleanup. This reduces
our total job costs substantially, by 12 to 15% or
more. We believe our effective frac lengths are
longer, too. But, most importantly, we get the gas
to market sooner by turning the wells around
faster, sometimes in about 24 hours instead of
seven to ten days. We’ve also been able to use
simpler fluids with less additives. That’s saving
us additional money and reducing friction pres-
sures and horsepower requirements. When the
job has been executed according to design and
the material has been placed where it’s needed,
the results have been excellent.”
Indiana, USA:
In low-temperature gas wells in Indiana, USA,
RCPs are commonly used when fracturing multi-
ple zones. This requires an activator and an
extended shut-in period for curing of each zone
after treatment, or a total time of four to five days
per well. When fiber technology was used
instead, the initial zone could be flowed back
only 10 min after the job was completed. Follow-
ing a limited flowback period, the next zone could
be fractured immediately, completing operations
in one day, a 75 to 80% savings in rig labor, rig
time and equipment costs.
51
The Road to High-Density Seismic
Geophysicists work increasingly with complex reservoirs that often
suffer from poor quality seismic data—sometimes so poor, the reser-
voirs eluded detection in older surveys. One example is the Wandoo
field in offshore Western Australia. Here, an innovative method of
acquiring seismic data has helped replace noise with visible structure
and turned a challenging discovery into a commercial field.
52
Malcolm BoardmanAmpolex LimitedPerth, Western Australia, Australia
Robin WalkerStavanger, Norway
Designing a seismic acquisition scheme islike choosing equipment for a photo shoot.Both require proper illumination of a sub-ject—one with light, the other with acousticwaves. Both are concerned with the focus ofenergy. And both strive to obtain thesharpest possible image.
Before a photo shoot, a photographermust choose the proper lens for the sub-ject—macro lens, wide-angle, mediumfocal length or telephoto. The telephoto dis-cerns distant objects but typically can’tfocus on those nearby. The macro lensmakes the sharpest image of objects withina few inches, but can’t see those far away.
A seismic survey is designed to functionlike a telephoto lens. The “optical properties”of a seismic survey—the depth within theearth that the survey is tuned to see mostclearly—are determined by two interlinked
variables. Resolution is controlled by thebandwidth of the sources and receivers,mainly by the high-frequency content. Depthof investigation, and to some degree band-width, is controlled by the size of the sourcesand the acquisition geometry—the spacingof sources and receivers and, in a marine sur-vey, the depth at which they are towed.
Most reservoirs are at least 1000 m deep[3280 ft] and so acquisition equipment andgeometry are usually optimized to meet thisdemand. At the depth of a typical pay zone,the phenomenon of overlying rock layersattenuating higher frequencies—so-calledearth losses—means that the recoveredbandwidth is normally expected to be from8 Hz to no higher than 60 or 70 Hz. Adapt-ing survey equipment and geometry toinvestigate a shallower interval and higherfrequencies is more complicated than sim-ply twisting a different lens on the camera. Itinvolves the equivalent of designing anentirely new lens.
This was the requirement for surveying afield off the coast of Western Australia.Conventional seismic methods nearly over-
Oilfield Review
0 km 50
0 miles 31
WesternAustralia
Carnarvonbasin
0 km 1
0 miles 0.62
Wandoo A W1
A1
A3W5
W2
A2
A4
W6
W6 Hor st1
W3
W7
Oil-watercontact
Gas-oilcontact
Wandoo B
EnderbyFault
Maitland
Bambra
TanamiHarriet
Sinbad
Rosette
Campbell
BarrowDeep
DampierBurrup
Wandoo
Angel
CossackWanaea
NorthRankin
Dixon
Goodwyn
Echo/Yodel
I N D I A NO C E A N
1000
m 200 m
Gas field
Oil field
Pipeline
Pipeline
Pipeline
Carnarvon basin
Autumn 1995
nWandoo A mono-pod, in 52 m [170 ft]of water, offshoreWestern Australia.All Wandoo produc-tion wells have been drilled fromthis platform. Workis underway todevelop the WandooB production plat-form to accommo-date eight horizon-tal wells. (Courtesy ofAmpolex Limited.)
nOil and gas fields inthe northeastern flankof the Carnarvonbasin, Western Aus-tralia (left). Detail ofthe Wandoo field(above) shows theexisting Wandoo Awells and the plannedWandoo B wells.
looked the Wandoo field, with 75 millionbarrels of reserves (left). Even when old 2Dsurveys were reprocessed with the latesttechniques, they failed to clearly illuminatethe sandstone play. Everything about theplay kept it elusive. The oil leg is thin—22m [72 ft]—placing it at the threshold of whatcan be resolved by relatively long-wave-length seismic energy.1 It is shallow—600 m[1970 ft]—which made imaging it conven-tionally akin to photographing a sparrowjust outside your window with an unwieldy600-mm telephoto. High-angle faults riddlethe zone, dispersing seismic energy. And, asicing on the cake, the unconsolidated paysand is overlain by a layer of dense carbon-
53
For help in preparation of this article, thanks to Kelli Burton, Mandy Coxon and Dieter Ramcke, Geco-Prakla,Perth, Western Australia; Robyn Gallagher, GISolutions,Canberra, New South Wales, Australia; Jakob Haldorsenand Doug Miller, Schlumberger-Doll Research, Ridge-field, Connecticut, USA; Tore Karlsson, Geco-Prakla, Stavanger, Norway.1. The shortest wavelength in a typical marine survey is
usually no smaller than 30 m [100 ft]. The top andbottom of a feature that can be detected is typically nosmaller than one-quarter wavelength, or, for a 30-mwavelength, about 7 m [23 ft]. Detectability means afeature can be seen, but can’t be discerned clearly.
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaSeabed multiplesPrimaryreflections Pegleg
Source Hydrophone array
Dense limestone3000 m/sec
Poorly consolidatedsandstone2000 m/sec
nTypes of multiples in a marine setting. In Wandoo, multiples tend to develop where alow-velocity layer overlays a high-velocity layer, such as at the interface of the sea andthe limestone.
ate that generates a signal processing night-mare called multiples (above).2
Today Ampolex Limited, the operator ofthe Wandoo field, produces 19,000 barrelsof oil per day (BOPD) from an early pro-duction system with five horizontal wells.By 1997, Ampolex plans to drill approxi-mately eight more wells to complete thedevelopment of the field. A key to this suc-cess is an innovative approach to 3D seis-mic acquisition.
What Makes Wandoo Work?The Wandoo field, named after an Aus-tralian hardwood tree, lies 60 km [37 miles]offshore within a 40- by 100-km [25- by62-mile] trend along the northeastern flankof the Carnarvon basin. As the preeminentbasin in Western Australia, the Carnarvoncovers 650,000 km2 [250,000 square miles],an area about the size of Germany plusItaly, or slightly smaller than Texas. Thirtyyears of exploration in the basin have turnedup 50 discoveries, holding 1.4 billion bar-rels of oil and 28 Tcf of gas. Drilling so farin the Carnarvon has achieved a 1-in-4 suc-cess ratio.3
Like the Permian Basin of West Texas, theCarnarvon comprises many smaller basins.The Wandoo area is heavily faulted and
54
includes unconformities and sudden lithol-ogy changes significant enough to obscurestandard seismic signals. The depositionalenvironment grades from marine to shallowmarine to fluvio-deltaic and nearshore shelf.
The Wandoo pay zone is interpreted as ashelf sandstone deposited in the Early Creta-ceous during a brief regression within atransgression.4 The reservoir contains highconcentrations of electrically conductiveglauconite, an iron-rich mica. Glauconiteraises havoc with resistivity log interpreta-tion, presenting one of the frustrations oflow-resistivity pay (see ”The Lowdown onLow-Resistivity Pay,“ page 4). The reservoirwas characterized using core analysis, for-mation testing and nuclear magnetic reso-nance logs.5 The Wandoo pay zone pro-duces 19° API oil, the heaviest oil onproduction in Australia.
Interest in the Wandoo area stretches backto 1965, but for much of that time, the shal-lowness of the reservoir led it to be under-rated.6 The latest chapter opened in 1990,when Ampolex acquired acreage from aprevious operator that saw little prospect forthe region. Wandoo lies on the Enderby ter-races, an Early Mesozoic complex listricfault trend.7 In the Wandoo area, vintage 2Dseismic data showed badly imaged faultblocks, and structure was not recognized atthe Early Cretaceous level. Drilling in theregion had turned up mainly dry holes anda few disappointing gas and oil shows.
The main barrier to understanding the
field was poor seismic data, suffering fromlow fold and multiples. After careful analysisof subtle clues in the data, Ampolex, how-ever, believed in the possibility of a shallow,Early Cretaceous play roughly mapped bythe previous operator. After acquiring theacreage, Ampolex reprocessed the 2D datawith modern techniques. This yielded aclearer picture that boosted optimism for theplay.
“It was lousy quality data,” said MalcolmBoardman, an Ampolex geophysicist on theproject, “but gave us enough courage todrill.” Wandoo-1 came on stream in May1991, testing at up to 4500 barrels per dayof 19° API oil, produced from a 22-meteroil column, which was overlain by a smallgas cap.
The next step was a field-wide hydro-graphic site survey, a routine procedure tolocate drilling rigs safely.8 A site survey usesshallow penetrating 2D seismic to probe forseabed lithology and shallow gas pockets,which can result in a blowout or unstablesupport of the rig. It also uses side-scansonar to investigate the seabed for obstruc-tions such as rock outcrops and pipelines.
The seismic site survey tested the potentialof high-resolution seismic methods.Although typically a low-tech procedure(minimal processing is performed, andsource depth is often not rigorously moni-tored), the site survey contained enoughhigh frequencies and depth penetration topresent a picture of the shallow interval thatwas surprisingly clearer than that of thereprocessed 2D survey (next page).
The site survey demonstrated that the sig-nal was there to record. However, it wasclear that 2D techniques would be unableto resolve the complex structure adequately.Without full understanding of the locationand displacement of the faults, potentialreservoir compartmentalization wouldremain unquantified.
The Wandoo discovery had developedinto an imaging problem and thereforerequired a 3D survey that could image theshallow zone and the complex pre-Creta-ceous below.9 Geco-Prakla was awardedthe job after an extensive exercise of surveyrequirement specification, subsequent ten-der and bid analysis. “Geco-Prakla submit-ted a combination of the most technicallycomprehensive and sophisticated solu-tion—and the cheapest,” Boardman said.
aa
Oilfield Review
Shallow, Mean and NastyAt the simplest level, Wandoo has two char-acteristics that give geophysicists indiges-tion: a shallow target and a complex target.Shallowness alone is not a problem for mostseismic methodologies. A shallow target canbe imaged relatively easily if it is a simplelayercake structure with reasonable acousticcontrast between layers. But lace those lay-ers with many fractures, and make theacoustic contrast between layers huge—inthis case, a velocity of 1500 m/sec for sea-water overlaying 3000 m/sec in the lime-stone—then conventional methods can behindered by multiples and ray bending.
Autumn 1995
500
1000
1500
2000
0
Enderbyfault
T
Enderby 1 Wandoo 1
Gam
ma
ray
Gam
ma
ray
W
Dep
th, m
nStratigraphy near Wandoo 1, based onenhanced the confidence of Ampolex thrate fault block.
The foremost challenges of imaging Wan-doo are:• preserving sufficient frequency bandwidth
to achieve the resolution needed to delin-eate the shallow producing horizon.
• mitigating the effect of multiples, gener-ated by a large contrast in acousticimpedance between dense limestone out-cropping at the seabed and far less denseunderlying sands.
• visualizing structure beneath the majormid-Jurassic unconformity, associatedwith continental breakup.A common link in all three challenges, to
extend the photography analogy, is design-
Bathonian-Bajocian
Toarcian
North Rankin formation
Brigadierformation
Mungaroo formation
Hamptonfault
(Base Cretaceous)Lower Muderongshale
op M. Australis sandstoneTop Muderong shale
Hampton 1
Enderby terrace
Gam
ma
ray
E
reprocessed 2D seismic. This viewat Wandoo 1 was located high in a sepa-
ing a lens that focuses close and does notgenerate unwanted reflection of light. Inseismic terms, here is how each of thesethree challenges was met (see “How Wan-doo Was Different,” next page).
The first challenge, preserving bandwidth,required rethinking seismic technique. Inconventional 3D seismic, frequenciesabove about 60 Hz contribute to higher res-olution, but are attenuated during their pas-sage to and from the reflector, and so arenot recorded. The target is usually imagedwith low frequencies, which are deep pen-etrating but result in a lower resolution.(The propagation properties of high and lowfrequencies explains why you can hear thelow-frequency bass of the jazz combo nextdoor, but not the high-frequency tenor sax-ophone.)
Preserving high frequencies meant firstcreating them. This was done four ways.
55
2. A multiple is seismic energy that has been reflectedmore than once, and persists through the arrival ofdesired signals. The Wandoo survey was affectedmainly by short-path multiples, which arrive shortlyafter the primary event and obscure structural and strati-graphic detail. The effect is similar to hearing a publicannouncement in a large railway station, in which mul-tiple echoes make the message unintelligible.
3. Williams P: “Oil and Gas Down Under: A DevelopingFrontier,” Euroil 6, no. 4 (April 1995): III-XIV.
4. Boardman M and Delfos E: “Wandoo—A New Trend,” Proceedings of the Australian Petroleum Exploration Association Conference, Sydney, Australia,March 20-23, 1994; also in APEA Journal 34, pt 1(1994): 586-601. Regression is a retreat of the sea from the land, broughtabout by a fall in sea level, uplift of land, or both. Trans-gression is the opposite: the spread of the sea over theland, brought about by a rise in sea level, subsidence ofland, or both.
5. Delfos E and Boardman M: “Wandoo—A New Trend,”APEA Journal 34, pt 1 (1994): 586-601.
6. Ampolex 1994 Annual Report: 20.7. A terrace, in this context, is a local shelf or step-like
flattening in otherwise uniformly dipping strata.8. A hydrographic survey is usually a low-fold, analog
seismic survey for soil engineering purposes.9. For an introduction to seismic imaging:
Farmer P, Gray S, Hodgkiss G, Pieprzak A, Ratcliff D,Whitcombe D and Whitmore D: “Structural Imaging:Toward a Sharper Subsurface View,” Oilfield Review 5,no. 1 (January 1993): 32-41.
6- to 10-mtowing depth
18.5-m subarraysix gun positions per subarray,with three-gun cluster leading
Typical Configuration
4.5-m subarraytwo gun positions per subarray,each a three-gun cluster
4-mtow depth
Wandoo
15- to 20-msubarrayseparation
nComparison ofa typical config-uration (top) andhigh-resolutionair gun arrays.
56 Oilfield Review
How Wandoo Was DifferentTechnique Typical Survey Wandoo Main Benefit
Streamer separation
In-line group interval
In-line offset
Source bandwidth
Shot-point interval
Air gun array
75 to 150 m[246 to 492 ft]
12.5 to 25 m[41 to 82 ft]
150 to 300 m[492 to 984 ft]
8 to 90 Hz
25 m
20 to 40 guns arranged ina swath 20 m [66 ft] wideand 18 m [60 ft] long, firedin alternating groups.
6 to 10 m [20 to 33 ft]
25 m[82 ft]
6.25 m[21 ft]
43 m[141 ft]
30 to 150 Hz
12.5 m
6 guns in 2 in-line rows 4.5 m [15 ft] apart and 4.5 m long, functioning asa single point source.
4 m [13 ft]
Improves ability to resolve steeplydipping, shallow reflectors bypreserving high-frequency content in energy diffracted by them
Improves near-trace coverage
Greater bandwidth preserves high-frequency content
Increases signal-to-noise ratio with higher fold
Compact array provides a moreuniform wavefront with broader band width, maintaining high fold at shallow depths
Preserves high frequenciesTowing depth
Air gun displacement 4000 in.3 735 in.3
57Autumn 1995
nHow a shallow, point-source air gunarray preserves high-frequency content.Polar plots (middle) compare the relativeintensity versus direction of a modeledoutgoing seismic wave at 100 Hz, for con-ventionally deployed and Wandoo airguns. Amplitude (blue is low, red high) isshown for an angle of incidence of thewavefield, in which 0° is vertical and 90°is horizontal. Differences in the wavefieldare caused mainly by the dimensions ofthe air gun array and the water depth atwhich the guns are deployed. If the plotswere modeled for 40 Hz, the conventionalair guns would show nearly the sameresponse as the Wandoo plot at 100 Hz. In the source amplitude plot (bottom) thenotch in the curve for the conventionalairgun array results from tow depth. Airguns are normally towed at a depth of 6 m [20 ft] or more, which improves bothlow-frequency content and operationalease. However, when energy from thesource is reflected back from the sea sur-face, destructive interference causesnotches in the frequency plot to appear at progressively lower frequencies withdeeper tow. At Wandoo, the shallow towensured this notching was pushedbeyond the useful frequency spectrum.
0 100 200
-12
0
-24
-36
-48
-60
90°80°60°40°20°
0
Tow
dire
ctio
n
Source Array Configurations
Frequency, Hz
Mic
ropa
scal
/Hz
at 1
m
Conventional
Wandoo
Notch
Conventional Wandoo
0 -42
Decibels
3397 in.3-air guns at 6-m depth 735 in.3-air guns at 4-m depth
18.5
m
18.5 m
4.5
m
First, air guns were gathered into a tightlyspaced array to act as a point source (previ-ous page, top). The length of the gun arrayshad to be reduced to match the overallshorter, denser sampling of the hydrophones.Concentrating the guns as a point sourceresulted in a more uniform wavefront(above). This contributed to producing aclean, high-frequency signature.
Second, guns were tuned to achieve ahigh “peak-bubble ratio,” an indicator ofhow sharp a source signature is in the timedomain. Air guns emit a high-pressure burstof air. The resulting bubbles expand and
contract as they rise in the water, sendingsecondary acoustic pulses that can be diffi-cult to eliminate during processing. In awell-tuned array, firing of the guns is timedto stagger the bubble oscillation periods.This produces a destructive interference ofthe secondary pulses. In effect, they canceleach other to produce a single, cleanpeak—one peak when the bubbles expand,another when they contract—that preservesthe high-frequency content.
Third, the sources were towed at a shallowdepth—4 m rather than the usual 6 to 10 m.Surprisingly, the sources were not suscepti-ble to excessive operational problems,despite their shallow depth.
A fourth element to increase high-fre-quency content was rewiring conventional
streamers for a shorter group interval, whichis the distance between the centers ofgroups of hydrophones. A group is a num-ber of hydrophones that feeds a singlechannel. By shortening the group length from16 m to 10 m [53 to 33 ft], and the groupinterval from 12.5 m to 6.25 m, geophysicistswere able to avoid destructive interference ofwavefronts reflecting off shallow targets at
80.6 m264 ft
1057 m3470 ft
Paravane
In-line offset
Single-sourcearray
Activetail buoy
1057 m
125 m25 m
Wandoo Configuration
3000 m
500- to 750-m
total spread
Dual-source array
100- to 150-m
streamer separation
Typical Array
nThe Wandoo survey spread compared with a conventional spread (above). Streamersin the Wandoo survey (below) were 1057 m [3470 ft] long, about one-third the typicalstreamer length.
10. For a discussion of destructive interference of seismic signals: Waters KH: Reflection Seismology (2nd ed). NewYork, New York, USA: John Wiley and Sons, 1981.
11. In a typical survey, a pair of source arrays is fired inalternating sequence about every 10 sec, with thevessel moving at about 4.5 knots [8.3 km/hr; 5.2 miles/hr]. This means a shot is fired about every 25 meters. In the Wandoo survey, the vesselslowed to 4 knots [7.4 km/hr; 4.6 miles/hr] and a sin-gle-source array was fired every 6 sec, at about therecycling threshold of the system with the required 3-sec record. This resulted in a shot every 12.5 m.
higher angles and with higher frequenciesthan desired (next page).10
A secondary contributor to high-frequency content was a smaller total gundisplacement. Displacement is the volumeof air released by the guns, which affects thefrequency content and depth penetration ofthe signal. The smaller the displacement, thehigher the frequency and the shallower thepenetration. Despite a displacement fivetimes smaller than usual, data wereacquired to 3 sec, or to a depth of about3750 m [12,300 ft]. (In seismic speak, depthof penetration is measured by recording
time. Three seconds is about half the usualtime.) The final step that boosted high-fre-quency content took place during process-ing, and consisted of adjustment in arrivaltimes to account for the effect of tides. Tidalfluctuations can influence resolution bychanging the vertical distance between thesource-receiver spread and the reflector.
Use of smaller gun displacement alsocontributed to solving the second two chal-lenges, coping with multiples and the effectof the breakup unconformity. Smaller gundisplacement helped cut noise generated bymultiples by producing a smaller signal in
the first place. “We didn’t want a big, loudsignal,” said Malcolm Boardman. “Thatwould induce too many reverberations, andthe earth would just ring like a bell.”
But the main tool to pare down the effectof multiples and the unconformity was toincrease fold. Fold is a measure of the multi-plicity of source-receiver pairs that probe agiven point in the earth. The higher the fold,the more traces are added together toenhance the coherent signal while reducingthe amplitude of random noise. Normalsampling is 30-fold coverage over an area of6.25 x 25 m. Wandoo was 20-fold over anarea 3.125 x 12.5 m [10 x 41 ft]. This meansthat in Wandoo, within 39 m2 [420 ft2], thesame reflection point in the subsurface wassampled by 20 source-receiver pairs—almost triple what is normally done.
Increasing fold in this survey relateddirectly to sampling density, which is deter-mined by the number of receivers and bythe number of times the source is fired persquare kilometer. In this case, it was firedabout four times more frequently than istypical.11 To date, the Wandoo survey is thedensest in the industry, at about 16 timesthe typical value. This density was madepossible by the redesign of the acquisitiongeometry to allow sampling at a high ratewhile still remaining economically viable(left and below ). Using six streamers,instead of three or four, was a key to keep-ing the technique economic.
A challenge of high-density sampling isachieving accurate positioning information,which means measuring and recording thespread—the arrangement of hydrophones inrelation to the source, in both in-line andcrossline directions. Accurate positioningwas crucial because of the tight geometry ofthe spread and the use of in-sea elements asshort as 7 m. Typical positioning precisionof 10 m could not be tolerated.
3000 m1.9 miles
59
Conventional
Wandoo
Single conventional group, 24 individual hydrophones16.12-m
group length
12 individual hydrophones
6.25-mgroup interval
12.5-mgroup interval
16.12-mgroup length
9.88-mgroup length
9.88-mgroup length
0.5
0
1.0
Am
plitu
de
100 200 300 400Cutoff frequency, Hz
16-mgrouplength
10-mgrouplength
16 m
A
A'
Streamer10 m
Streamer
16-m group length, 12.5-m interval length
10-m group length, 6.25-m interval length
Steeply dipping reflector
How group length affects cutoff frequency
nAnatomy of a group length and groupinterval in a hydrophone array (above), and how group length and interval affectfrequency content of signals from shallowreflectors (left and below). Hydrophones(above) are grouped to form a manageablenumber of channels and to cancel outsources of noise. Group length is selected toreduce the amplitude of returning energy atfrequencies and in-coming angles that areinadequately sampled. Group interval—thedistance between the centers of neighbor-ing groups—is selected to sample returningenergy that is adequately sampled forimaging of subsurface structures. For sim-plicity, the schematic (left) shows groupsthat do not overlap. In this view A and A’represent the time lag for a wave to propa-gate along the length of a hydrophonegroup. Wave periods greater than A or A’are cut off. A shorter group length (A’)means a shorter time lag and the retentionof short-period (high-frequency) wave data.
60 Oilfield Review
nA dual-purpose thin paintbrush and thick paintbrush. The multi-3D acquisition scheme can provide high-resolution images of ashallow target while simultaneously imaging a deeper target. The small shaded area depicts the zone investigated by the shortstreamer traces, and the large shaded area is that area investigated by the long streamer traces.
Positioning software was adapted toaccommodate the tight geometry. Moreover,the limit was pushed using conventionalpositioning technology: the satellite-basedglobal positioning system, laser ranging,acoustic networks and land-based radioranging. As a result, the spread was posi-tioned to within 5 m [16 ft] and at times towithin 2 m [7 ft], contributing to the accu-racy of the final product.
Where does Wandoo Lead?In Wandoo, the most valuable contributionof high-density seismic was precisiondrilling that hit the target the first time inappraisal and development wells. In drillingeach of five wells, Ampolex was able toland precisely in the pay zone, and keep thewell trajectory within 2 m vertically for a1000-m [0.6-mile] horizontal section.Ampolex was able to predict where the wellwas stratigraphically on the first try. None ofthe development wells had to be pluggedoff and redrilled. “Good seismic meant that
we had no surprises and therefore, noadded drilling expenses,” Boardman said.
For Ampolex, a key lesson was how totune acquisition parameters for optimalresults. “Any target shallower than 1200meters [4000 ft] would profit from some ofthese parameters, such as closer streamerspacing,” Boardman said. “Our work inWandoo has made us think about the possi-bilities of tuning regular 3D surveys.”
For Geco-Prakla, an additional benefitwas the opportunity to develop a shallow-looking 3D survey, leading to evolution ofwhat is called multi-3D: simultaneous useof interleaved short and long streamers, withconventional penetration and high-resolu-tion sources, and simultaneously acquiringhigh- and normal-resolution data to imagetargets at multiple depths (below). It is like acamera that uses two lenses simultaneously,a macro and a telephoto.
The multi-3D approach now under devel-opment is seen to provide three key bene-fits. First, it can be used for detection ofshallow gas, replacing the lesser quality 2D
seismic component of the site survey. Sec-ond, it provides high-resolution imaging ofshallow traps and secondary targets thatmight be missed by a conventional survey.And third, multi-3D can provide high-reso-lution, interpretable images of the near sur-face to build a detailed velocity model,needed to model ray behavior and convertdata from the time domain to depth. Animproved near-surface velocity modelimproves the entire seismic section, sincemost velocity problems are compounded byshallow velocity errors. It would also reducethe risk of artifacts at depth resulting fromnear-surface velocity anomalies, such asriver channels, which do not show up prop-erly on surveys of conventional resolution.These artifacts are often disappointingdrilling targets.
As 3D seismic continues to mature anddiversify into new niches, multi-3D may fillan expanding need for high-density, high-resolution data. The lessons learned in Wan-doo may help multi-3D pioneers get themost from this latest evolutionary step. —JK