logging while drilling: a three-year perspective · lwd is a subset of mwd services, which also...
TRANSCRIPT
MWD services
Short-normalresistivity
Gamma ray
Drillingmechanics
LWD•Compensated Dual Resistivity•Compensated Density Neutron
Direction&
Inclination
Real-time telemetry & recorded data
nThe relationship between MWD and LWD measurements. In this article, LWD is used to denote wireline-quality petrophysi-cal measurements made while drilling—resistivity, gamma ray,density and neutron porosities, calipers and photoelectric factor(Pe). LWD is a subset of MWD services, which also include sen-sors for drilling-related measurements, such as direction andinclination and equipment for data recording and telemetry.
4
Logging While Drilling: A Three-Year Perspective
Stephen BonnerBrian ClarkJacques HolenkaBernard VoisinSugar Land, Texas, USA
Jim DusangRandy HansenJim WhiteAberdeen, Scotland
Tom WalsgroveAmoco (UK) Exploration CompanyLondon, England
Logging while drilling (LWD) has been in commercial service since the late 1980s. Since then, the basic measurements
in hardware, processing and interpretation—some incremental, some monumental—has furthered understanding of the
FORMATION EVALUATION
Since the introduction of logging whiledrilling (LWD) in 1989, jobs in more than400 wells worldwide have propelled thetechnology through the first stage of matura-tion. Tools have improved, interpretationmethods can address more complex prob-lems, and challenges presented by highlydeviated and horizontal wells are beingovercome. Since 1989, worldwide expendi-tures for all measurement-while-drilling(MWD) services has more than doubled,from about $250 million in 1989 to a pro-jected $520 million in 1992.1
This article summarizes advances inSchlumberger’s LWD technology and appli-cations over the past three years (right ).Highlighted are tool technology, evolvingLWD applications and advances in dataprocessing and interpretation.
For help in preparation of this article, thanks to DaveBest, Keith Moriarty and Richard Rosthal, SchlumbergerLogging While Drilling, Sugar Land, Texas, USA; TrevorBurgess, Jean-Michel Hache, Marc Lesage, Bernard Mon-taron and Alan Starkey, Anadrill, Sugar Land, Texas,USA; Charles Flaum and Mike Gibbs, Etudes et Produc-tions Schlumberger, Montrouge, France; Ken Henry,Schlumberger International Coordination, Houston,Texas, USA; Martin Lüling, Schlumberger-Doll Research,Ridgefield, Connecticut, USA; Rod Nelson and FrankShray, Schlumberger Well Services, Houston, Texas, USA.
Oilfield Review
Tool TechnologyBasic LWD measurements—resistivity, neu-tron and density porosities and photoelec-tric factor—have not changed since theirintroduction (next page, below), but tooltechnology has undergone several refine-ments. These include a range of engineeringimprovements—from more robust sensordesign, to more secure mounting of connec-tors and integrated circuits. These improve-ments have led to increased tool durability.
Because the tools are contained in drillcollars, hardware takes up space inside thecollar, reducing the cross-sectional areaavailable for mud flow. This reduction inarea, and erosion of tool components by
sand and lost circulation material, meanmud flow rates are lower than in plain col-lars. However, demand for LWD measure-ments has increased in wells that requirehigher flow rates for effective hole cleaning.In response, the maximum mud flow rate forthe 61/2-in. tools was recently increasedfrom 450 to 600 gallons per minute [from28 liters/sec to 38 liters/sec]. This upgrademakes the tools practical in wells wheremud rate requirements might have pre-cluded them in the past. The upgrade wasaccomplished by redesigning internal toolparts to allocate more cross-sectional area
In this article, DWOB (Downhole Weight-on-Bit), M3(Measurement-while-drilling telemetry system), TLC(Tough Logging Conditions system), CDR (CompensatedDual Resistivity), CDN (Compensated Density Neutron)and EPT (Electromagnetic Propagation Tool) are marks ofSchlumberger.1. Modified from Hall GT: “Measurement While Drilling
Update,” First Boston Equity Research, July 31, 1991.In this report, MWD services include direction andinclination, porosity, resistivity and gamma ray mea-
nThe team that built the 8-inch CDN toolin Sugar Land, Texas, USA with (below)the 6 1/2-inch tool and its red, slide-on sta-bilizer, and (above) the 8-inch tool. Theblack plugs cover windows for the densitysource and gamma ray detectors. The 8-inch tool, because of its larger diameter,has the source and detector in the bladerather than in the tool body, which movesthem closer to the formation.
have remained the same, but evolution
best ways to use the measurements.
nMeasurementsand formationparameters fromCDR and CDN tools.
LWD ToolsCDR Tool
Parameter
Correlation
Rt
Rxo
Thin bedsInvasion
Shale volume
Porosity
Measurement/Computation
Dual resistivities (Rps and Rad)Gamma ray (total API)
Dual Resistivities
Natural gamma ray scintillationspectroscopy (Th, U, K)Computed gamma ray
Compensated neutron porosityCompensated spectral gamma-gamma density
Tool OD/Mud ratein./gal/min
61/2 / 60063/4 / 8008 / 85081/4 / 120091/2 / 1400
CDN Tool
Lithology Density-neutron crossplotPe
Ultrasonic caliper1
61/2 / 600
63/4 / 800 (under development)
8 / 850
Rugosity,detection offree gas
1. For 63/4- and 8-in. CDN.
to mud flow, and by increasing the tooldiameter. In 1991, the outside diameter ofthe CDR (Compensated Dual Resistivity)tool was increased from 61/2 and 8 in. to63/4 and 81/4 in., respectively. The 61/2-in.CDN (Compensated Density Neutron) toolwill be replaced in the near future by a 63/4-in. version. The 63/4-in. tools permit a maxi-mum mud flow rate of 800 gal/min [50liters/sec]; the 81/4-in. tool is rated to 1200gal/min [76 liters/sec].
The most fundamental change in thenuclear tool is detector design. The first gen-eration used a combination of 3helium (He)detectors, also used in wireline tools, andGeiger-Mueller detectors. Compared toGeiger-Mueller detectors, 3He detectors,have a broader dynamic range, do not needcorrection for spurious activation, are lessaffected by borehole salinity and have betterstatistics, permitting a higher rate of penetra-tion (ROP). But they were not thought to beas rugged as Geiger-Mueller detectors. Fieldexperience proved otherwise, and since1990 the CDN tool uses 3He detectors only.Older tools are being retrofitted.
The CDN tool has a 7.5-curie 241ameri-cium-beryllium neutron source and a 1.7-curie 137cesium density source, both con-nected to a source retrieval assembly. In thefirst version of the tool, the sources andretrieval head were connected with a flexi-ble steel cable. This has been replaced witha flexible titanium rod, giving more reliableretrieval and more accurate placement of
the sources. Also improved is the densitydetector shielding, which eliminates sensi-tivity to spurious signals from the mud.
The CDN tool uses a full-gauge stabilizerwith windows cut in the blades in front ofthe density source and gamma ray detectors(top). When the hole is in gauge, the blades
surements and charges for real-time data transmission.
wipe away mud from in front of the sensors,thereby minimizing borehole effects. Alocking mechanism is being retrofitted onthe stabilizer to increase its resistance toslippage under high torque and jarring. Newstabilizers with an integrated lock are underdesign and scheduled for release next year.A range of stabilizers, including undergaugesizes, has been added for use in horizontaland deviated holes.
6
A
B
in in7
12
17
22
12 22
Large Axis-Wireline
Large Axis-Ultrasonic
Small Axis-Wireline
Large Axis
Small Axis
Superpositionof Calipers
MWD UltrasonicCaliper
While DrillingSmall Axis-Ultrasonic
X300
X400
Dep
th, f
t
nComparison of LWD ultrasonic caliper madwireline caliper, made five days after drillingborehole has become larger with time due toAt zone B, the discrepancy between LWD anments is due to mudcake buildup in front of
A significant development in nucleartechnology was the introduction in 1991 ofthe 8-in. CDN tool, and, in the near future,of a 63/4-in. version. Other than being ableto operate in holes up to 123/4-in., the 8-in.tool has several features that distinguish itfrom the 61/2-in. tool. In the 8-in. tool, neu-tron and density detectors are both in stabi-lizer blades rather than in the tool body, asin the smaller tools. This design is preferred
in12 22
Large Axis
Small Axis
Wireline Four-ArmCaliper 5 Days
After Drilling
X3
X4
X5
X6
X7
ROP
150 0ft/hr
e while drilling and. At zone A, the sloughing shales.
d wireline measure-the permeable zone.
nA gas detection log mditions, using the ultrasics (next page). The ultblades of LWD tools. It rubber window of simi
in the larger tool because locating the detec-tors in the tool body would have placedthem too far from the formation and therebydegraded the measurement. The 8-inch toolalso includes an ultrasonic caliper (see“Unocal, Indonesia,” page 21).2 The 63/4-in.version will have new electronics toincrease the number of measurements perfoot by a factor of four and will also have anultrasonic caliper.
Oilfield Review
Mea
sure
d de
pth,
ft
LWD UltrasonicCaliper
While Drilling
00
00
00
00
00
Large Axisin Gas Indicator
0 750
8 18
gal/min
Flow In
0 25units
Baseline-no gas
Bit entersgas-
producingfracture
zone
Sensors pass
below gas-producing
zone
Gas-producing
fracture zones
ade while drilling in underbalance con-onic caliper (above), and caliper schemat-rasonic sensor is mounted in stabilizertransmits a pulse that passes through alar acoustic impedance to that of mud.
2. Orban JJ, Dennison MS, Jorion BM and Mayes JC:“New Ultrasonic Caliper for MWD Operations,”paper SPE/IADC 21947, presented at the SPE/IADCDrilling Conference, Amsterdam, The Netherlands,March 11-14, 1991.
3. Rewcastle SC and Burgess TM: “Downhole ShockMeasurements Increase Drilling Efficiency andImprove MWD Reliability,” paper IADC/SPE 23890,presented at the IADC/SPE Drilling Conference, NewOrleans, Louisiana, USA, February 18-21, 1992.
4. A g, for gravity, is a unit of force equal to the force ofgravity exerted on a body at rest. It is used to indicatethe force to which a body is subjected when acceler-ated. For example, 25g is 25 times the force exertedon the body as when it is at rest.
Two ultrasonic sensors are mounted 180°apart on stabilizer blades. The sensors func-tion in a pulse-echo mode that allows thedirect measurement of standoff, from whichshort and long axes of the borehole diame-ter are computed. The vertical resolution is1 in. [25 mm] and accuracy of the diametermeasurement is ± 0.1 in. [2.5 mm]. Thecaliper is used to correct the density andneutron porosity measurements for boreholeeffects and can be used as a borehole stabil-ity indicator (previous page, left). It can alsobe used for downhole detection of freegas—gas bubbles, not dissolved gas—through a combination of formation and“faceplate” echo signals (below and previ-ous page, right). The faceplate echo is mea-sured at the surface of the tool, at themud/window interface. It is affected by gascontent in mud, with echo amplitudeincreasing with gas content. The smallestamount of detectable gas is less than 3%volume of free gas. Real-time transmissionof this information can shorten the timeneeded to detect gas influxes while drilling.This can simplify kill operations.
Memory of the CDN and CDR tools wasdoubled to 1 megabyte in 1991, and withthe introduction of Anadrill’s second-genera-aaa
July 1992aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaa a aa aaaaa aaaaaaaaaaa aaaaMWD
Ultras
LWD toolwith ultra-sonic sensorin stabilizer A
mpl
itude
When gas bubbles are present in the mud, thless than that of the mud. Much of the pulse athe mud/window interface. When no gas bubthrough the window with no reflection, and agenerating a formation echo of higher amplit
tion MWD/telemetry system in 1991, down-link to the tools can be established whilethey are in the hole. Previously, tool opera-tion—such as data sampling rate—had to bepreset at the surface and was not adjustableonce the tool was downhole. The downlinkcapability permits, for example, the operatorto use one CDR sample rate during drillingand switch to a higher rate while trippingout, or to turn sampling off in front of casing,thereby saving memory. The current genera-tion mud telemetry system permits transmis-sion of data at up to 3 bits/sec.
Another advance is the introduction of adownhole shock measurement transmittedto surface. This measurement enhancesselection of the bottomhole assembly (BHA)and drilling parameters, and may increasesurvival of MWD/LWD tools. Many failuresof MWD/LWD tools result from high shockand vibration produced during drilling. Lat-eral vibration contains the most energy anddoes the most damage to downhole tools,the drillstring and drill bits. Traditionally,engineers predicted a rough running drillingenvironment from surface torque measure-ment and modeling of drillstring dynamics.But this is not a direct shock measurementand therefore does not properly account for
aaaaaaaaaaaaaaaaaaaaaaMud/formationinterfaceMud/window
interface
Window echo
Window
Mud Formation
Firingpulse
Formationecho
Formationecho
Firingpulse
sonicensor
Mudinterfaceecho
Time after firing
Clean mud, no gas Mud with gas
e acoustic impedance of the window ismplitude is therefore lost on the echo atbles are present, the pulse passes
stronger pulse reaches the formation,ude than the faceplate echo.
all causes of vibration and for frictional lossbetween the surface and BHA. Now thedriller can see the downhole environment inreal time and adjust rotary speed, weight-on-bit and flow rate to eliminate or reduceshocks. The drilling engineer can also usethis information to design BHAs less proneto vibration.
Anadrill’s downhole measurement ofshock is made by an accelerometer near thetelemetry electronics, 20 to 50 ft [6 to 15 m]above the bit.3 At a given time interval, typi-cally 60 seconds, the cumulative number ofshocks exceeding 25g4 is reported to thesurface (next page). This 25-g value is con-sidered the optimal limit for alerting thedriller to possible mechanical failure.
7
aa
5. Allen D, Auzerais F, Dussan E, Goode P, Ramakrish-nan TS, Schwartz L, Wilkinson D, Fordham E, Ham-mond P, Williams R: “Invasion Revisited,” OilfieldReview 3, no. 3 (July 1991): 10-23.
6. Shray F: “LWD Detects Changes in Formation Param-eters Over Time,” Petroleum Engineer International64, 4 (April 1992): 24-32.
LWD/Wireline ComparisonsIn the three years since LWD technologybecame available, the industry has foundfive main applications for these tools:•Insurance logging, in case the well is lost,
can’t be logged with wireline tools or willyield poor-quality wireline logs.
•Logging before invasion, which mayreveal hydrocarbon zones that might bemissed by wireline time.5 In some high-permeability formations, borehole fluiddisplaces hydrocarbon from the near-wellbore rock, making the well look like adry hole by the time of wireline logging.This effect may be more common in hori-zontal than vertical wells because thedrainhole is exposed to full hydrostaticmud pressure for the long period requiredto drill the lateral section.
•Geosteering and enhancement of drillingefficiency (discussed below).
•Savings in rig time in settings requiring theTLC (Tough Logging Conditions) systemand offshore.
•Multiple pass logging. Comparison logsmade at different times can help distin-guish pay from water zones, locate fluidcontacts and identify true formation resis-tivity and density.6
Tom Walsgrove of Amoco UK, who isresponsible for introducing new technologyto the company’s North Sea operations, hasdone a cost-benefit analysis of LWD vs.wireline logging that has wide application.He finds that LWD has a clear cost advan-tage when well deviation is 60° or more. Atthis deviation, triple combo wireline toolsrequire conveyance by drillpipe with theTLC system. But LWD can be beneficial inother settings that are not so easily identi-fied. It is for this gray area that Walsgrovehas developed general criteria to helpchoose between wireline and LWD forbasic formation evaluation.He has found LWD cost-effective when:•Rig cost is high. Rig cost high enough to
make LWD attractive exists almost entirelyoffshore and where time-consuming TLClogging would be required. Most of theadditional expense of a TLC operation is inrig costs, which run about $5000/hour inthe North Sea. Amoco is looking toMWD/LWD as a means to cut this cost.Use of LWD may also accelerate selectionof coring and casing points and perfora-tion intervals, which can contribute to sav-ings in rig time.
•Water-base muds are used. These typicallyyield poorer borehole conditions by wire-line time than oil-base mud (OBM). Wire-line tools may therefore be difficult to get
200
300
Start of 171/2-in.hole
400
95 RPM
120 RPM
Run
#1
Run
#2
ROP
SWOB
Surface Torque
Downhole Shock
RPM
sec-1kNm
m/hr
1000 kg20
75 0
0 0 20 0
50 150
150Average
Maximum
Minimum
340
368
Dep
th, m
nRelationship between shock levels measured downhole, and rotary speed in a NorthSea well. Shock increases with increasing rotary speed and as stabilizers pass changesin borehole diameter. In run #2, a pendulum assembly—with stablilizers only high inthe BHA in order to drop hole angle—was run with low weight-on-bit to control holeangle. This resulted in greater vibration of the BHA and failure of the MWD transmis-sion. High shocks below 340 meters corresponded to the stabilizers passing from the 24-in. rathole to the 17 1/2-in. hole. At this point, surface torque also increased, indicatingincreased drag from the stabilizers. Rotary speed was reduced at about 347 m inresponse to the shock levels, and 95 rpm maintained minimum shock. At 368 m, thedriller, constrained to maintain verticality, increased rotary speed to 120 rpm andshocks rose to 120 sec -1.
8 Oilfield Review
7. The intergovernment treaty organization that governsdumping in the North Sea, the Paris Commission, isexpected to pass a ruling that 1% to 2% oil on cut-tings is not a significant environmental hazard. Thiswill then be ratified by the Commission ministers andcome into force in January 1994. This would representthe minimum standard to be enforced by signatorycountries.
nWhat can happen when CDR resistivity is not transmitted in real time. CDR data wereacquired in memory mode only, since the data stream to the surface was confined todrilling optimization information. In this case, the memory logs (gamma ray, resistivityand density) showed the well trajectory went into the top of the pay zone at point A,only to go back out through the top of the pay zone at point B. The pay zone here wasmore than 10 feet [3 m] thick but the drill bit crossed only the top 10 feet. (Adapted fromWhite, and Hansen and White, reference 9.)
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaa a aa aaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaa a aHorizontal distance, ft
Poor sand
Thin shale
Mainreservoir
Gam
ma
Ray
CD
R R
esis
tivity
, sha
llow
CD
ND
ensi
ty
True
ver
tical
dep
th, f
t
100 800X50
X75
200 500
0.530
GA
PI
ohm
-mA
B
9July 1992
down the hole, may get stuck or yieldpoor logs. In the UK sector of the NorthSea, as OBM use continues declining forenvironmental reasons, Amoco expectsLWD use to increase. Use of OBM isbanned in exploration wells in Norwayand the Netherlands. The UK is expectedto enact other restrictions by 1994.7
•The well is in an area known for opera-tional difficulties—swelling shales or,most commonly in the North Sea, over-pressured zones. Under these conditions,LWD offers the advantage of insurancelogging. But more importantly, real-timedata can accelerate decisions about wellcontrol and how the well should bedrilled and completed. Amoco finds thatreal-time data provide “an effective meansof minimizing risk and reducing excessiveoperating costs” in exploration wells.Most of this savings is related to savings inrig time. To illustrate this point, Walsgrovecites an instance in which LWD reducedoperating costs by allowing more accu-rate location of coring points within asand/shale sequence.
“From our work on other wells in thisarea,” Walsgrove said, “I knew it washighly probable that the first core wouldbe mostly shale with only minor sandstringers. But with MWD/LWD we wereable to pick the coring point at the top ofa thick sand unit. We saved ourselves thecost of cutting that wasted core, whichtypically takes a day—about $120,000.”
In another instance, the company spentseveral days drilling a rathole in base-ment, including cutting a terminal corethat would confirm that the well was inbasement before logging. If real-timegamma ray and resistivity logs had beenused, the data would have indicated base-ment quickly, saving days of drilling.
•When real-time access to pore pressureanalysis (in undercompacted formations)and petrophysical data are needed fordrilling decisions and planning a wirelinelogging program and completion strategy.
Real-time access to petrophysical datahas led to a new application, calledgeosteering—using real-time measure-ment of geologic and reservoir parametersto help steer the bit along highly deviatedand horizontal trajectories. These parame-ters are added to conventional steeringdata—direction and inclination—and dataabout the shape and extent of geologicstructures, mainly from seismic profiles.The usual goal is to keep the well within aparticular bed or maintain an appropriatedistance from a bed boundary or fluidinterface (right, above and top).
Horiz. distance, ft
Sta
ndar
d D
ensi
ty
Den
sity
ØN
Nor
mal
ized
Sho
rt-S
paci
ngVa
rianc
e
p.u.
g/cm
3
30
0
2.2
2.7
0
4
X700 X800 X900 X1000 X1100 X1200
nDetails of CDN logs for the same well as above. The normalized short-spacing vari-ance is not a regular product but used here to compute the rotational density output.
When geosteering was introduced, theCDR resistivity log was used for detectinghydrocarbons and the gamma ray log forfinding marker beds and kick off points.Because the CDR measurement has higherresolution and precision than the typicalMWD short-normal resistivity measurement,thin beds can be more easily identified. It isalso more sensitive to boundaries betweenbeds of contrasting resistivity—indicated ina deviated well by the characteristic peakon the log curve, called a horn—and has agreater depth of investigation (see “Why aHorn?” below).
Although resistivity remains the mainheadlight, a handful of operators haverecently examined CDN density logs to seeif they can contribute to geosteering (previ-ous page, below). However, the CDN tool isalways farther from the bit than the resistiv-ity measurement so its contribution to steer-ing may be as a confirmation of gamma rayand resistivity logs.
The measurement suite for geosteeringmay be the same as for formation evaluation(see “MWD/LWD Data Frames, next page).The suite typically includes resistivity,gamma ray, short- and long-spacing densityand neutron porosity, downhole torque anddownhole weight-on-bit to determine for-mation properties at the bit—with the addi-tion of tool face, or direction and inclination(D&I). This tells the driller the orientation ofthe bent housing.8 Tool face is needed at ahigh update rate—typically, every 10 sec-onds but as fast as every few seconds isdesirable for accurate steering with poly-crystalline diamond compact bits. Withthese bits, more reactive torque is generatedat the motor, making them go off coursesooner than conventional bits. If a highupdate rate for tool face is used, the updaterate on other data drops. The lower ratereduces resolution of petrophysical data, butnot usually to a significant degree.
Walsgrove’s analysis indicates that wire-line measurements are advantageous when:•Equity decisions are a concern. Because
wireline measurements remain theaccepted standard, both from technicaland legal standpoints, partners may shyaway from relying only on LWD measure-ments when there is a possible equitydebate. Some operators view LWD as lessdesirable in equity decisions because ofthe need to depth match it with wireline.Depth matching LWD to wireline is oftenstraightforward, but may be complicatedwhen invasion or borehole conditionscause large differences between LWD andwireline logs, or when driller’s depthvaries nonsystematically with wirelinedepth. This can happen because of pipestretch, compression or yo-yoing, or mis-counted pipe stands. Absolute depth cor-rection of LWD logs is sometimes doneusing a gamma ray from the cement bondlog when casing is set—usually one tothree days after LWD logging. This delayin absolute depth correction does not stalldrilling-related decisions based on LWDinformation. The relative depths of fea-tures on the LWD log are used to refer-ence formation tester sample, core andcasing points.
•High temperatures (greater than 320°F[160°C]) are encountered. Wireline toolstypically have a higher temperature ratingthan LWD tools, which are limited mainlyby their use of batteries.
•Hole size is very large. The current maxi-mum hole size for the CDN tool is 133/4in. [35 centimeters (cm)], and for theCDR tool is 171/2 in. [44 cm]; or 26 in.[66 cm] if mud is made with freshwater(1.5 ohm-m).
Other variables may complicate the choicebetween wireline and LWD. Amoco did acost analysis for two wells in which bothwireline and LWD services were run. In one
well, discounting rig time, LWD cost 241%more than wireline. In this instance, thelarge difference was due to LWD tools sit-ting on the rig unused for 21 days becauseof unanticipated problems setting casing. IfLWD had been run promptly, its cost wouldhave been 100% more than wireline.
In the second well, LWD logs were madepromptly. Although the LWD cost wasslightly higher than wireline, when rig costwas added, LWD afforded a net saving. TheLWD service was less expensive than in thefirst case because the operator needed onlythree days to mobilize and demobilize thetools, which were on a nearby rig.
Taking cost analysis one step further,Amoco modeled typical North Sea wells todetermine the most cost-effective triplecombo logging program. Wireline wasmarginally less expensive in the first exam-ple, which had:•straight hole•casing set at 8000 ft [2440 m] and average
ROP of 30 ft/hour [10 m/hr] to reach totaldepth (TD) at 10,000 ft [3050 m]
•two days each to mobilize and demobilizethe tools
•total rig operating cost of $120,000/day.The second model was identical, except thewell was deviated to 45° at 4000 ft [1220m]. Although TD and the casing point werethe same true vertical depth (TVD) asbefore, the openhole footage and total mea-sured depth increased. Again, wirelineoffered a marginal savings. But when welldeviation reached 70°, rig time associatedwith wireline cost exceeded LWD costbecause TLC hardware would be needed toconvey wireline tools. Financial analysis byanother major oil company shows that lost-in-hole cost is always more expensive withLWD than with TLC methods, but that riskof losing the tools is lower with LWD thanwith the TLC system.
10 Oilfield Review
Why a Horn?
A polarization horn on a CDR log in a horizontal
well indicates the tool is crossing a bed bound-
ary. This horn occurs because of charge buildup
at bed boundaries. Because this charge is caused
by the signal from the logging tool, it therefore
oscillates at the same frequency, acting like a
secondary, weak transmitter along the bed
boundary. This effect doesn’t occur in a vertical
well, where bed boundaries are horizontal and
parallel to the current loops generated by the
tool. In this setting, the current loops do not cross
interfaces between formations of contrasting
resistivities and no secondary field is generated.
But in a horizontal well, as the tool moves along
the bed boundary, the current loops pass from
one formation into another, generating a substan-
tial secondary field. At each bed boundary, the
received signal increases, producing the charac-
teristic polarization horns. The magnitude of this
polarization increases with the relative dip angle
and the resistivity contrast between the beds.
11July 1992
Data Transmission: Real Time orRecorded?Combined improvements in MWD teleme-try and LWD technology now permit real-time transmission of any LWD data. But notall data can be sent at once, so a majordecision is what data to send in real timeand what to store downhole for retrievalwhen the BHA reaches the surface (see “ASecond-Generation MWD/Telemetry Sys-tem,” page 13). In the future, through-drillpipe wireline systems will be availableto unload tool memory downhole.
The selection of real-time or downholememory mode depends on the application,and may change from section to section,even in a single well. While real-time datahelp with drilling decisions, the more datasent uphole, the slower the update rate ofeach measurement, so the lower the resolu-tion. The proper balance between real-timeand recorded modes involves determiningnot only what data are needed when, butwhat data quality is needed. The goal is toget the right data at the right time.
The recorded-only mode is commonlyused in formation evaluation for reservoircharacterization in development wells. Real-time data are used mainly in formation eval-uation for well management decisions, suchas whether to run wireline logs and, if so,which ones; where to core and set casing;and when to plug the well, to adjust hole tra-jectory or mud weight, drill deeper or stopdrilling. A typical real-time suite consists ofdirectional surveys, resistivity measurements(resistivity from attenuation—deep, Rad andresistivity from phase shift—shallow, Rps ),gamma ray, short- and long-spacing den-sity, neutron porosity, downhole torque anddownhole weight-on-bit (right ). Most ofthese data are updated every 26.5 seconds,equivalent to a 6-in. [15-cm] sample rate atan ROP of 70 ft/hour [21 m/hr]. In practice,twice the ROP—140 ft/hour [43 m/hr]—gives 12-in. [30-cm] sampling, sufficient formost real-time formation evaluation.
Amoco UK bases its selection of recordedvs. real-time data on comparison of datacost and quality. In exploration wells, thecompany usually transmits the CDR log inreal time to improve drilling efficiency. Indevelopment wells, the company usuallyacquires LWD logs only in recorded mode.This reduces logging cost as a result of nothaving to run an MWD tool for directionand inclination data.
8. There are two tool face measurements. Magnetic toolface is in reference to magnetic north. Gravity toolface is in reference to the high side of the hole.
Typical Formation Evaluation Setup
DownholeMeasurement
MWD sync
Rps
Rad
LWD GR
SS density
LS density
TN porosity
DTRQ
DWOB
Rps
Rad
LWD GR
SS density
LS density
TN porosity
VALT
Rps 26.5
26.5
26.5
26.5
26.5
26.5
53
53
53
68
68
68
68
68
68
34
34
34
Rad
LWD GR
SS density
LS density
TN porosity
DTRQ
DWOB
VALT
TransmittedData
Update frequency(sec)
Equivalent ROP at two data points/ft
Typical Geosteering Setup
DownholeMeasurement
T/F sync
GY [T/F]
Rps
Rad
GZ [T/F]
SS density
LS density
GY [T/F]
TN porosity
LWD GR
GZ [T/F]
DTRQ
DWOB
GY [T/F]
VALT
GZ [T/F]
Gravity T/F 10.6
53
53
53
53
53
53
53
53
53
170
34
34
34
34
34
34
34
34
34
Rps
Rad
LWD GR
SS density
LS density
TN porosity
DTRQ
DWOB
VALT
TransmittedData
Update frequency(sec)
Equivalent ROP at two data points/ft
DTRQ: downhole torque; VALT: MWD turbine voltage output, which is proportional to mud flow, so VALT acts as a downhole flowmeter; T/F sync: a number identifying the sequence of measurement to be transmitted; GY and GZ (T/F) are values of theaccelerometer measurements along the Y and Z axes; the tool is the X axis. GY and GZ are used to compute the orientation ofthe bent sub with reference to the high side of the hole. The DWOB measurement is downhole weight-on-bit. Many downholemeasurements appear more than once in the left column, indicating that they are sent more than once per data frame.
MWD/LWD Data Frames
nSample data frames from the North Sea, used for formation evaluation and geosteering.
The application of LWD in North Seadevelopment wells is unique, having to dowith how they are drilled. Many operatorsuse MWD directional and gamma ray infor-mation to position the 95/8-in. [25-cm] cas-ing just above the target. Some also use theCDR resistivity for this task. Casing is setand the hole drilled out with a “locked”BHA, in which stabilizers are arranged tominimize drift of the borehole. From thispoint, MWD is not used because the drillerhas to drill a straight hole a short distanceand can’t miss the target. Once in the target,LWD measurements are used in memorymode to log the reservoir.
12 Oilfield Review
Mainreservoir
GammaRay
LoggedPhase
Resistivity
Rt Input forComputer
Model
SimulatedPhase
Resistivity
Poorsand
Thinshale
Dep
th, m
GAPI ohm-m ohm-m0 100 2 2000 2 2000
x100
x200
x300
x400
x500
nResistivity modeling of the uppermostresistivity horn shown on page 9, top.Here, formation dip, bed boundaries andresistivities were estimated from the geo-logic model and used to develop a modelof Rt and a synthetic resistivity log. Themodel is confirmed because the syntheticlog nearly matches the measured log.This modeling was done because of a dis-parity between nuclear and resistivitylogs. On page 9 (top figure), the densityand gamma ray logs show that the zonebetween the two peaks appears to be ashale. But the resistivity log does not showthe low value normally associated withshales in this area. Because the shale isthin (about 2 ft [60 cm]), it is possible thatthe large volume of rock investigated bythe CDR tool prevented it from seeing theshale. This presented an ideal opportunityto run a resistivity simulation across thiszone. The square log of Rt successfullysimulated the measured log, reinforcingthe model of the formation as a shalewith significant lateral extent. (Adaptedfrom Hansen and White, reference 9.)
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaA B
A B
A B
A B
Shaleaa Main reservoiraa Poor sand
Deepresistivity
Mea
sure
d de
pth,
ft
ohm-m1 10
100
0
200
300
400
A
B
nFour possibleinterpretations fora deep resistivityresponse in a hori-zontal well. Morelog data areneeded to con-strain the geologicmodel. There areno horns becausethe deep resistivitymeasurementtends not to showhorns and becausethe resistivity con-trast between bedsis small. (FromWhite, reference 9.)
9. White J: “Recent North Sea Experience in FormationEvaluation of Horizontal Wells,” paper SPE 23114,presented at the SPE Offshore Europe Conference,Aberdeen, Scotland, September 3-6, 1991.Hansen RR and White J: “Features of Logging-While-Drilling (LWD) in Horizontal Wells,” paper SPE 21989,presented at the SPE/IADC Drilling Conference, Ams-terdam, The Netherlands, March 11-14, 1991.
10. Leake J and Shray F: “Logging While Drilling KeepsHorizontal Well on Small Target,” Oil & Gas Journal89, no. 38 (September 23, 1991): 53-59. The samephysics, but for induction logging is detailed in KleinJD: “Induction Log Anisotropy Corrections,” Trans-actions of the SPWLA 32nd Annual Logging Sympo-
11. CDRDIP is now a module in ELMOD. For back-ground:Anderson B, Bonner S, Lüling MG and Rosthal R:“Response of 2-MHz LWD Resistivity and WirelineInduction Tools in Dipping Beds and Laminated For-mations,” Transactions of the SPWLA 31st AnnualLogging Symposium, Lafayette, Louisiana, USA, June24-27, 1990, paper A.Anderson B, Barber TD, Singer J, and Broussard T:“ELMOD—Putting Electromagnetic Modeling toWork to Improve Resistivity Log Interpretation,”Transactions of the SPWLA 30th Annual LoggingSymposium, Denver, Colorado, USA, June 11-14,1989, paper M.
A Second-Generation MWD/Telemetry System
Data Processing AdvancesMost data processing routines for logs weredesigned for low-angle holes, typically lessthan 40°. Assumptions in the verticalwell—azimuthal symmetry of formationsand invasion around the borehole, no lat-eral variation in formation properties, andformations becoming older and deeper withmeasured depth—break down for direc-tional and horizontal wells.9 The increasinguse of LWD in horizontal and highly devi-ated holes has stimulated progress in theprocessing of LWD log data.In directional and horizontal wells, dataprocessing and interpretation is a two-stepprocess. The first step, while the well isdrilled, is to determine whether, when andwhere the well trajectory enters the geologictarget and whether it stays there. This isdone by monitoring formation evaluationinformation—cores, cuttings, MWD andLWD data—as the well is drilled, and then,modifying the well path as needed. The sec-ond step, which takes place after the well isdrilled, is detailed appraisal of formationparameters for reservoir evaluation, wellcompletion and field development. To meetthese needs, and interpretation needs inconventional wells, several new data pro-cessing methods for LWD have been devel-oped. These include:•Prediction of CDR resistivity in high-angle
and horizontal holes with varying relativedip between the borehole and beds, usingthe RangDB program (“rang dee bee,” forrelative angles data base)
•Real-time phase-resistivity caliper andborehole correction to phase and attenua-tion measurements
•Correction of the CDR measurements forthe effect of resistivity anisotropy10
•Correction of the density measurement toaccount for elliptical holes.
Prediction of CDR resistivity in high-angleholes with varying relative dip. An objectiveof interpreting CDR logs in high-angle wellsis to build up a model of the formation sur-rounding the borehole. This interpretationmust be made with care because there maybe insufficient data to develop a unique geo-logic scenario (previous page, left). Simula-tion of CDR logs may help clarify this pro-cess (see “Electromagnetic Tool Modeling,”page 22). Often, CDR response in a horizon-tal well can be predicted before drilling. Thismay be critical for identifying log signaturesthat show when a drill bit is entering or leav-ing a target. This prediction of CDR responsein a well with constant, unchanging relative
July 1992
dip is achieved by using the CDRDIP mod-ule in the electromagnetic modeling pack-age, ELMOD11 (previous page, right). Butwhat about the typical high-angle well,where the well inclination changes continu-ously?
For this, a second-generation modelingprogram has been developed, RangDB. Thisprediction has several steps. The first step, aswith any simulation, uses local knowledgeand offset well data to provide a simple, lay-
sium, June 16-19, 1991, Midland, Texas, USA, paper T.“Formation Anisotropy: Reckoning with its Effects,”Oilfield Review 2, no. 1 (January 1990): 16-23.
ered model of the geology. This model isverified using ELMOD data to simulate theresistivity curve from the offset well.
The second step uses this model as inputto several runs of CDRDIP at differentangles (next page, left). The angles are cho-sen to represent the range of dips likely tobe encountered along the trajectory.RangDB then computes values of phase shiftand attenuation resistivities versus TVDinterpolated at 1° deviation steps between
In 1991, Anadrill introduced a second-generation
MWD telemetry system, which is used to convey
MWD and LWD measurements. The new system,
called the M3 MWD Telemetry tool (M for MWD; 3
for 3 bits/second), includes several notable
departures from the first-generation system. Here
are some highlights.
• A mud pulse modulator that resists jamming.
At the heart of mud pulse telemetry is a down-
hole modulator that generates pressure pulses
in the mud. It is placed in the path of the mud
flow and opens and closes to generate pres-
sure pulses. The first-generation modulator
was occasionally jammed by debris and lost
circulation material. The new modulator has
better resistance to jamming because it is self-
cleaning and has lower drag. It performs suc-
cessfully in muds having up to 40 lb/barrel of
loss circulation material.
• Reduction in transmission data error rate by an
order of magnitude. This was accomplished
two ways. The new modulator permits a higher
signal-to-noise ratio, and introduction of a
downhole microprocessor allows finding and
correcting errors in the data stream. After error
correction, the received data contain less than
1 bit error per 1000 bits—an error rate of 0.1%.
• Data compression. The data transmission rate
of the M3 system is 3 bits/sec. However, the
downhole microprocessor makes it possible to
minimize the number of bits in the data stream,
which increases the update rate while keeping
sufficient resolution. This minimization yields
an effective transmission 30% faster than the
first-generation system.
• Reprogramming while downhole. MWD tools no
longer have to be reprogrammed on the sur-
face only. Using surface-to-downhole commu-
nication, achieved by varying mud flow rate,
the engineer can reprogram data transmission
speed, vary in real time which data are
recorded, alter the sample rate and vary the
rate at which data are sent uphole in real time.
• New electronics packaging. The hand-wired
electronics of the first-generation system has
been redesigned using proven LWD technology.
This has resulted in significant improvement in
reliability.
13
Data Services Catalog. Houston, Texas, USA:Schlumberger Educational Services, 1990: 105-106.
True bed thickness, ft
6350 6390 6430 6470 6510
Ohm
-m1000
100
10
1
1
1
1
RelativeDip
85°
75°
45°
0°
RpsRad
nResistivity logs forbeds dipping 0° to85°, showingdevelopment ofhorns as a bedboundary is crossedat increasinglyshallow angles.The 0° example isa modeled CDRlog, based on aninduction log in avertical well. Theothers are modeledCDR logs using theCDRDIP program.These logs showthat variation indip angle, evenwhen adjusting fortrue bed thickness,gives logs of differ-ent character.Comparing curvesat the 4-ohm-m ref-erence line showsthat resistivitycreeps upwardwith increasing dip angle.
14
12 1/4 in.
26
22
18
14
10
Bor
ehol
e di
amet
er, i
n.
50 100 150 200 250 300 350
Depth, ft
LWD phase caliperFour-arm wireline caliper
the chosen dips. These resistivity values areentered in a data base, which allows fast pre-diction of resistivity log response for changesin well deviation or formation dip at anypoint in the well. In the near future, the pro-gram will then take the computed resistivityvalues and, in conjunction with real-timewell directional data, produce a provisional,simulated CDR log along the actual trajec-tory. During drilling, these models can becompared with the real-time CDR resistivitylog to determine, for example, if the bit isapproaching a calcite streak or a bed bound-ary. In particular, the log predictions give thedirectional driller enough information to cor-rect the course of the drill bit.
The RangDB program encourages cross-disciplinary communication by requiringcooperation from the geologist/petrophysi-cist and directional driller when planning adirectional well (next page, left ). Becausethe program integrates the well plan fromthe drilling department with formation topsand dips from the geology department, ittests whether the two are synchronized.
The program also runs at the wellsite,using the actual well trajectory to help mod-ify the trajectory in real time. This is doneby determining whether a differencebetween the modeled and measured CDRlog is due to the trajectory not going asplanned, or the formation depth or dipdiverging from the model.
nComparison ofphase and wirelinecalipers. At a fewplaces—140 to 170feet, 240 and 430feet—the caliperwhile drillingreports a largervalue than thewireline reading,perhaps due toswelling shales.The phase caliperat 60 ft readsbetween the twowireline valuesbecause it reportsan average bore-hole size. Limita-tions of the phasecaliper’s verticalresolution cause itto miss some of thethinner sands. Notethat the shales arealmost uniformlywashed out to 17in. [43 cm]. (From Rosthal et al,reference 12.)
Oilfield Review
-bit size
400 450
12. Rosthal RA, Best DL and Clark B: “Borehole CaliperWhile Drilling From a 2-MHz Propagation Tool andBorehole Effects Correction,” paper SPE 22707, pre-sented at the 68th SPE Annual Technical Conferenceand Exhibition, Dallas, Texas, USA, October 6-9, 1991.
Resistivity caliper and borehole correction.The phase caliper is computed from the rawphases of a 2-MHz electromagnetic wavemeasured at two receivers and from thephase shift between the receivers.12 The rawphase is strongly dependent on mud resistiv-ity and borehole size, whereas the phaseshift is only slightly affected. Since mud con-ductivity is known, the measurement can beused to calculate large changes in boreholediameter from bit size in conductive muds(previous page, bottom). This caliper can beused to monitor borehole stability duringdrilling or tripping. Calculation of the phasecaliper does not require special equipmentand does not compromise the tool’s standardmeasurements. Mud, however, must be atleast 1 siemens/m (S/m) [1 mho/m] more
CDR modeling for curved traject
Modeledlog #1
Modeledlog #2
Modeledlog #3
We
6500
7000
6000
5500
0
TVD
, ft
Anticipated Geologic Model
ELMOD(CDRDIP)
0
-40
-80
40
80
Dep
th, f
t
Well traject
Well traject
Well traject
Modeling of CDR logs at various
deviations from 0° to 90°
Formationdip andhorizon
TVD
Resistivity, ohm-m
•TVD•Relative dip between borehole and beds•Electromagnetic phase shift•Electromagnetic attenuation
RangDB data base:
July 1992
nInputs for CDR resistivity modeling of varioucome from both the drilling and geology dep
conductive than the formation. Accuracy ofthe borehole diameter measurement is ±1/2in. [1.3 cm], the vertical resolution is about36 in. [91 cm], and it will work in boreholediameters up to 24 in. Because the phasecaliper works best for large variations inborehole diameter, it complements the ultra-sonic caliper, which works best where bore-hole variation is no larger than about 6 in.
The calculated borehole diameter can beused to apply a borehole correction to thetwo resistivity measurements. This correctionis needed most in large boreholes filled withvery conductive muds.
Resistivity anisotropy code. Resistivityanisotropy—different resistivity values inhorizontal and vertical directions—may pro-duce markedly different resistivity logs of the
ories
Modeledlog #4
Well trajectory
#1
Plannedllbore Trajectory
400 1200800
Horizontal offset, ft
ory #2
ory #3
ory #4
Welltrajectory#1
s well trajectoriesartments.
Shale
Sand
same bed measured in a vertical well andmeasured in a well deviated more than 70°(below). Resistivity anisotropy is typicallyseen in formations with laminations less thanabout 3 in. [8 cm] thick, with alternatinghigh and low resistivities. Induction andCDR resistivity measurements in such a for-mation will read lower in a vertical well thanin a horizontal well. The opposite is true fora laterolog resistivity measurement. A codeto handle this problem, and correct the resis-tivity measured in the deviated well, was
nCoping with resis-tivity anisotropy.Measurement ofresistivity in a ver-tical well is alwaysof the horizontalvalue, whereas ina horizontal well, itis some combina-tion of vertical andhorizontal values.In the horizontalwell, where adja-cent shales distortthe current, thevalue of horizontalresistivity is neededto compare logvalues with thoseof offset verticalwells. The resistiv-ity anisotropy codeprovides this,using CDR toolmeasurementsand relative dipangle between thehorizontal welland formations. (From Leake andShray, reference 10.)
15
nResistivity modeling showing that resistivity anisotropy explained why pay zone resis-tivity in a horizontal well was higher than expected. The model demonstrates thatinequality between horizontal and vertical resistivities could reproduce the measuredlog results. The right inset shows that anisotropy is represented by a series of thin bedsof alternating high- and low- resistivity streaks. The two models, one vertical (0°) andthe other horizontal (89°), successfully reproduced the measured log responses—4-ohm-m in the vertical well and 9 to 10 ohm-m in the horizontal well. (From Leake and Shray, reference 10.)
16
Gam
ma
Ray
True
Ver
tical
Dep
thR
ps
Rad
0
120
0
20
GA
PI
ft
0
30
ohm
-m
Rps
Rad
Rt
0 7ohm-m 0 50
0 20
ohm-m
Top
of ta
rget
Measured depth, ft 100 200 300 400 500 600
Rps
Rad
Rt
Vertical Well Horiz. Well0° apparent dip 89° apparent dip
Sand
Shale
Rad
Rps
Estimateddielectric value
Res
istiv
ity, o
hm-m
1000
100
10
1
X450 X470 X490 X510 X530 X550
Depth, ft
developed to address resistivity anisotropy ina US Gulf Coast well. It has subsequentlybeen used in wells elsewhere in the US GulfCoast, Alaska and the North Sea.10
Jack Leake of Oxy and Frank Shray andRichard Rosthal of Schlumberger, comparedthe resistivity of a pay zone measured by theCDR tool in a horizontal well with thatmeasured by the wireline induction tool in anearby vertical well. A 4-ohm-m pay zoneon the induction log from the vertical wellappeared in a horizontal well as a 9-ohm-mzone on the Rad and 15 ohm-m on the Rpsmeasurements. At first glance, the separa-tion of the deep and shallow measurementscould have been due to invasion, but closerexamination showed this was not likely.Invasion was unlikely because the well waslogged while drilling, the hole was thoughtto be in gauge and the sensors were only 30minutes behind the bit. Although invasionhas been observed under these conditions,the amount of separation seemed too greatfor that explanation. The discrepancybetween vertical and horizontal logs couldhave been caused by neighboring bedswithin 2 vertical feet of the LWD sensor, butthere was no corroborating evidence thatthey were present. This left anisotropy as theprime suspect.
Evidence for anisotropy came from theELMOD resistivity modeling program thatwas able to simulate both the wireline (ver-tical) and LWD (horizontal) logs. The modelconfigured the top 3 ft [1 m] of pay as aseries of high- and low-resistivity streaks,which represents the effects of formationanisotropy. CDR logs were simulated in thissetting in a vertical well and a horizontalwell. Although these logs do not represent aunique solution, they reproduced the fieldresults (above, left).
Existing CDR software was modified tooutput a continuous log of horizontal andvertical resistivities. Inputs are Rad, Rps andthe angle between the hole and the bed inquestion. The vertical resistivity componentis used only for the calculation, not for trueresistivity. The horizontal component corre-lated closely with that measured by theinduction log in the vertical well.
nA pronounced dielectric effect causingan anomalously high, deep attenuationmeasurement. This example is from a vol-canic rock in a North Sea well. Theexpected dielectric constant for this for-mation was about 50, whereas the actualvalue was over 500. The same effect, butnot as dramatic, has been observed inshales in the US Gulf Coast, Alaska andthe Far East.
Oilfield Review
Another advance concerns correctingCDR measurements in conventional wells(deviated less than 40°) for dielectric effects.In formations with a high dielectricconstant13—typically highly resistive rockand some shales—the Rad value is anoma-lously high. Significant elevation of Radoccurs in most shales and some volcanicrocks. The problem is evident in thickshales, where there is no invasion and noborehole effect, yet Rad exceeds Rps (previ-ous page, bottom). A log processing routinehas been developed that uses Rad and Rps tosolve for the dielectric constant and correctresistivity for dielectric effects.
The routine uses information about dielec-tric constants collected from analysis of 300rock samples worldwide, performed at
13. A dielectric constant is a measure of the ability of amaterial to store electric charge for a given appliedfield strength. A dielectric is a material having a lowelectrical conductivity compared to that of metal.Fresh water, for example, has a dielectric constant ofabout 80; quartz is 4.65. The shale dielectric con-stant varies strongly with frequency of the electro-magnetic field and with mineral content. It typically
July 1992
Res
istiv
ity R
ad, o
hm-m
Att
enua
tion,
dB
Re
0 1 2 3
5.8
5.6
5.4
5.2
5.0
4.8
4.6
εR
10050
30
20
150
ε=300
5
7
10
20
50
100 50 30 20
Schlumberger-Doll Research, Ridgefield,Connecticut, USA. A data base was thencompiled, relating dielectric constant toconductivity as a function of frequency.
The data base provides an estimate of thedielectric constant as a function of conduc-tivity at the 2-MHz frequency of the CDRtool.14 An anomalously high Rad valueresults when the actual dielectric constant islarger than the estimated value. The dielec-tric processing program for the CDR toolcomputes from the two recorded CDR resis-tivities an apparent dielectric constant andits corresponding, dielectrically correctedresistivity, Reps, for epsilon, the symbol fordielectric constant. The program does thisautomatically, but a chartbook-type versionof this correction is shown (below).
has a dielectric constant of up to 10,000 at induc-tion log frequency (10 to 40 kHz), 100 to 200 atCDR log frequency (2 MHz) and 5 to 25 at the fre-quency of the EPT (Electromagnetic PropagationTool) (1.2 GHz). Measurement of the dielectric con-stant, in the case of the EPT tool, is used to distin-guish two fluids that have the same resistivity: freshwater vs. a mixture of saltwater and oil.
14. Clark B, Lüling MG, Jundt J, Ross M and Best D: “A Dual Depth Resistivity Measurement for FEWD,”Transactions of the SPWLA 29th Annual LoggingSymposium, San Antonio, Texas, USA, June 5-8,1988, paper A.
Phase shift, deg
sistivity Rps, ohm-m
4 5 6 7 8
ad=40
ad=19
Rps=11ε ps=45
15
10
7
Dielectrically correctedresistivity=12.5
Apparent dielectricconstant=180
15 10 7
Interpretation AdvancesMany advances in LWD pertain to log inter-pretation in horizontal wells, a setting inwhich LWD commonly has economicadvantage over basic wireline measure-ments. One interpretation considerationbeing given more attention, with the grow-ing number of horizontal wells, is gravitysegregation of fluids in the invaded zone(next page, top).
An example of this was seen in a NorthSea deviated gas well logged with LWDwhile tripping three days after drilling. Therewere two CDN density outputs: bulk densityand the rotationally corrected density. Agreater rotational density than bulk densitysuggested an oval hole.15 Amoco thoughtthe greater rotational density value was not
17
nCorrection for dielectric effect on 61/2-inch CDR logs. The red curve, just belowthe top of the fan, is the estimated dielec-tric value versus resistivity, derived fromlab studies. This curve predicts, for exam-ple, that a phase shift of 5° will producean Rps of 11 ohm-m and an attenuation of5.2 dB will produce an Rad of 19 ohm-m.These correspond to estimated dielectricvalues of about εps=45 and εad=40,respectively. The intercept of these tworesistivity values, however, gives thedielectrically corrected resistivity, whichis 12.5 ohm-m. The line on which theintercept falls indicates an apparentdielectric constant of 180, four to fivetimes greater than the predicted dielec-tric constant. The dielectrically correctedresistivity always falls between the othertwo resistivities, Rad and Rps, making it arobust computation.
15. For a discussion of rotational density:Best D, Wraight P and Holenka J: “An InnovativeApproach to Correct Density Measurements WhileDrilling for Hole Size Effect,” Transactions of theSPWLA 31st Annual Logging Symposium, Lafayette,Louisiana, USA, June 24-27, 1990, paper G.Betts P, Blount C, Broman B, Clark B, Hibbard L,Louis A and Oosthoek P: “Acquiring and Interpret-ing Logs in Horizontal Wells,” Oilfield Review 2,no. 3 (July 1990): 34-51.
due to ovalization but because invadedfluid, after three days, lay on the low side ofthe hole. “Since the introduction of LWD,failure to account for density variation asso-ciated with invasion in gas-bearing reser-voirs could lead to an erroneous interpreta-tion,“ Walsgrove said. ”Now I mentallymodel what’s going on in the formationbefore I begin log analysis.”16
Amoco has progressed in its interpretationof LWD logs in deviated holes. A case inpoint is a highly deviated well that pene-trated a known oil zone, but that LWD logsindicated anomalously high water satura-tions as determined from the deep resistivity(Rad) log. The zone in question was asand/shale sequence. Close examinationshowed that the deep resistivity was beingsuppressed so much by the overlying shalesthat the sands looked wet (see “Amoco,North Sea,” page 20). This suppressioneffect was proved by modeling the forma-tion and using tool response functions topredict log response.
Advances have also been made in inter-pretation of nuclear logs. Since the intro-duction of LWD, differences have some-times been noted between wireline andLWD nuclear logs in the same well. Recentwork has shed light on two mechanisms forthis disagreement: differences in how thetools make their measurements and changesin wellbore conditions in the time betweenlogging with LWD and wireline—namely,displacement of reservoir fluid by invadingmud filtrate of a different density, chemicalalteration of shales by drilling mud andincreased rugosity.17 How these effectsinfluence the preference for wireline orCDN logs is summarized (right).
For density measurements, the main dif-ference in response to wellbore conditionsis due to the different mechanical arrange-ment of the tools. Wireline density tools usepad-mounted sources and detectors that areheld against the formation, minimizing per-turbation of the signal by mud. The CDNtool makes its measurement through a full-gauge stabilizer (the same as bit size). Aslong as the hole remains in gauge, the stabi-lizer removes virtually all mud from in frontof the sensor, yielding a reliable measure-ment. The rotational density algorithm cancorrect for a small degree of standoff—1 to2 in. [3 to 5 cm], depending on mud weight
18 Oilfield Review
WirelineHole Condition
CDN HoleCondition
Smooth andin-gauge
Smooth, in-gauge
with mudcake
Smooth andenlarged
Smooth,enlarged
with mudcake
Smoothand altered
formationdensity
Enlargedand rugose
Enlarged,rugose
and alteredformation
density
Excellentagreement
Rotationalprocessing
required
Time-lapse densityinterpretation possible
CDN density preferred Sonic porositypreferred
nComparison ofhow wellbore con-ditions affect CDNand wireline den-sity measurements.(From Allen et al, reference 17.)
and mud chemistry. Greater standoffexceeds the algorithm’s ability to furnish ausable correction.
For neutron measurements, differentdesigns of wireline and LWD tools alsocause differences in how the tools respondto wellbore conditions (next page). Thewireline neutron measurement is made froma mandrel pressed against the borehole
wall, whereas the CDN measurement ismade from a tool centered in the borehole.The borehole effect for the wireline tool isnot important compared to the mudcakeeffect. On the contrary, the borehole sizeeffect on the CDN tool is large because thetool is centered in the hole. The wirelinetool standoff effect is comparable to theCDN borehole effect in amplitude.
Filtrate heavier thanformation fluid
Limit offiltrate
invasion
Pad tool
Volume seenby pad tool
Filtrate lighter thanformation fluid
nRadial asymmetry of the invasion profile in a horizontal wellthrough a homogeneous, permeable formation. A sufficient dif-ference in the densities of the filtrate and formation fluids willcause filtrate to form a teardrop oriented either up or down. Padand mandrel tools facing the low side of the well will see a dif-ferent invasion than those facing the upper side. These asym-metric effects are time dependent as the invasion profile devel-ops. (From White, reference 9.)
nComparison of wireline and CDN density and neutron porosity logs in a vertical southTexas well drilled with fresh mud. Above and below the bar, the hole was smooth andin gauge during CDN logging. The zone marked by the bar was logged with the CDNtool during a bit trip after being open several hours. Here, the CDN ∆ρ is high, consis-tent with a 9 pounds per gallon [lbm/gal] mud weight, and DCAL is close to zero. Thiscombination of readings indicates the hole was enlarged enough to prevent the CDNtool from maintaining contact with the formation. The combination of a well-stabilizedBHA and lack of hole deviation produced this standoff. By wireline time, the caliperindicates mudcake had formed. Arrows mark rugose intervals where wireline densityreads too high. In these intervals, the CDN measurement would be preferred, while inthe zone by the bar, wireline density would be preferred. (From Allen et al, reference 17.)
19July 1992
MWD Gamma Ray
Wireline Gamma Ray
CDN ∆
Wireline ∆
0
-0.9
0
-0.9
150
0.1
150
0.1
60
0
60
0
0
10
0
10
x700
x800
Dep
th, f
t
CDN ØN
CDN ØD
CDN Rotational DCAL
Wireline ØN
Wireline ØD
Wireline DCALρ
ρ
16. Others have noted the special case of invasion in ahorizontal well. See Woodhouse R, Opstad EA andCunningham AB: “Vertical Migration of Invaded Flu-ids in Horizontal Wells,” Transactions of the SPWLA32nd Annual Logging Symposium, Midland, Texas,USA, June 16-19, 1991, paper A.
17. Allen DF, Best DL, Evans M and Holenka J: “TheEffect of Wellbore Condition on Wireline and MWDNeutron Density Logs,” paper SPE 20563, presentedat the 65th SPE Annual Technical Conference andExhibition, New Orleans, Louisiana, USA, Septem-ber 23-26, 1990.
Charles Flaum of Schlumberger in Mon-trouge, France, studied CDN density logsand helped solve a key problem in horizon-tal drilling: knowing whether a formationboundary is crossed from above or below.
Flaum examined the rotational densityoutput in a horizontal well that both enteredand exited the top of a pay sand (page 9,top). The rotational correction makes use ofthe statistical variance of the near-detectorcount rate as the tool turns in the hole. Inthis computation, a normalized statisticalvariance of 1 indicates an in-gauge hole,assuming no radial variation in density.When the hole is in gauge, further excur-sions of the variance are not caused bychanges in standoff but by the tool crossinga density boundary at a low angle. Thesedeflections may be used to determinewhether the well is approaching a densitycontrast boundary from above or below.Flaum found that in a smooth hole drilledwith undergauge stabilizers, the variancewill have a positive shift in the absence of adensity contrast boundary. If a denser forma-tion approaches from below, the variancewill increase until the boundary is passed. Ifa denser formation approaches from above,variance will decrease. The opposite effectswill be observed if the approaching forma-tion is of lower density. The density “horns”may also help identify boundaries betweenbeds that have insufficient resistivity contrastto produce horns on the resistivity log.
These advances in LWD technology andtechnique are some of the innovations beingcombined in new ways with drilling-relatedmeasurements that are more quantitative.This synergy of LWD and drilling measure-ments may take geosteering, now a newtechnique, into the realm of the common-place. This will lead to the next step, whichmight be called geodrilling—real-timemerging of petrophysical and drilling data tofind more efficient and safe ways not only toposition the bit, but also to drill the well.
—JMK
Amoco UK first decided to use LWD when a 60°
development well had reached TD and the opera-
tor could not get wireline tools downhole. The
cause of difficulty was suspected to be post-
drilling degradation of the borehole. A triple
combo LWD suite made a single trip in the well.
The logs appeared valid, except for a density dis-
crepancy—0.03 g/cm3 too high.
In a subsequent well, drilled with oil-base
mud, Amoco determined that a comparison was
required between LWD and wireline measure-
ments. The LWD-wireline comparison showed
excellent agreement. The ρb from LWD and wire-
line almost match (right). The LWD neutron
porosity, however, was lower than wireline partic-
ularly in shales because the LWD measurement,
unlike the wireline measurement, accounts for a
combination of thermal and epithermal neutrons
and here was significantly affected by epithermal
neutrons.
Examination of the LWD logs shows that the
higher vertical resolution of the CDR tool in this
conductive environment (less than 2 ohm-m),
compared to the wireline medium induction
resistivity log, improved calculation of water sat-
uration in thin sands (right). This resulted in a
lower water saturation, increasing the deter-
mined oil in place.
nHigher thin-bedresolution on theCDR log, com-pared to that of themedium induction.
nClose agreementbetween wirelineand LWD densitylogs in a North Seawell. Divergence ofneutron porosities isprobably due todifferences in thephysics of wirelineand CDN measure-ments—wirelinedensity is a purelythermal measure-ment, whereas theCDN density mea-sures a combina-tion of thermal andepithermal neutrons.
20 Oilfield Review
-3 7
6 16
CDN Pe
Wireline Pe
CDN Caliper
Wireline Caliper
in
2.10 2.60 36 6
-0.8 0.2
CDN Density
Wireline Density
CDN ØN
Wireline ØN
g/cm3
g/cm3
p.u.
Dep
th, f
t
x000
x050
x100
Wireline ∆ρ
CDN ∆ρ
x700
x750
Dep
th, f
t
0 100 0.2 20 0.2 20GAPI ohm-m ohm-m
LWDGamma Ray
WirelineGamma Ray
LWD, Rps
Wireline,Medium Induction
Wireline,Deep Induction
LWD, Rad
Amoco, North Sea
LWD Case Studies
x700
x800
Dep
th, f
tShale Indicators Wireline Resistivity Porosities
Wirelinegammaray
LWDgamma ray
Diff.caliper
Deepinduction
Mediuminduction
Wirelinedensity
CDNneutron
Wirelineneutron
Wireline Caliper
LWD Density
Wireline Density Dep
th, f
t
7100
7150
7200
7250
6 in 16 1.7 g/cm32.7
nCorrugation of the wellbore wall from turbodrilling, evidencedby oscillation of the wireline caliper. In the gas sands, LWD den-sity is higher than the wireline value because the CDN toolmade better contact with the formation.
nGas sands in Southeast Asia less affected by excavationeffect—a decrease in apparent neutron porosity—at LWD timecompared to wireline time. Farther up the well, borehole condi-tions deteriorated by wireline time, preventing good pad contactof the nuclear tool, which masked other gas sands.
Unocal, Indonesia
Unocal has drilled a series of wells with MWD/
LWD tools in offshore Kalimantan, Indonesia. A
notable gas discovery was made in the second
exploration well in the Serang field. LWD was
planned over a long section.1 The objectives were
to evaluate the CDN measurements, obtain real-
time CDR measurements for comparison with
wireline logs in a nearby well and provide early
detection and evaluation of pay zones.
21July 1992
1. Norby P and Henry K: “Logging While Drilling (LWD)Results in Indonesia,” paper OTC 6862, presented at theOffshore Technology Conference, Houston, Texas, USA,May 4-7, 1992.
Use of OBM complicated evaluation with wire-
line logs because the mud masked many pay zones
on mud logs. The real-time CDR log, however,
identified pay zones in what turned out to be a sig-
nificant gas discovery. Downloading of CDN data
confirmed the top of the gas sand had been pene-
trated (above, left). These gas sands were later
verified by formation tester sampling. These zones
could have been mistaken for oil zones because
of deep invasion at the time of wireline logging.
The CDR log was 60 minutes or less behind the
bit and the CDN log was less than 3 hours behind
the bit. The CDN differential caliper, between
tracks two and three, indicates the hole was still
in good condition at this time, with no washouts
greater than 1/2 in. But at wireline logging, five
days later, washouts went off the scale—recorded
subsequently at 16 in. [40 cm].
In the lower gas sand, above 7200 ft, the wire-
line caliper shows a slight oscillation associated
with a regular corrugated effect on the borehole
from turbodrilling (above). In this interval, the
maximum density from the CDN tool is higher due
to better contact with the formation than that of
the wireline tool.
Barbara AndersonGerald MinerboMichael OristaglioSchlumberger-Doll ResearchRidgefield, Connecticut, USA
Tom BarberBob FreedmanFrank ShraySchlumberger Well ServicesHouston, Texas, USA
22
Resistivity modeling is shortening th
Although almost as old as logging its
from steering horizontal wells to inve
Modeling Electromagnetic Tool Response
WELL LOGGING
Resistivity modeling was used as far back as1927, when Conrad Schlumberger first rea-soned how current from an electrodespreads out into the formations around aborehole. But he would have called it “the-orizing.” Characteristics were assigned tothe formation (the formation model) and thelaws of physics, usually in idealized form,were used to predict analytically theresponse made by some electrode configu-ration (the sonde, or tool, model) to themodeled formation. Both theoretical andexperimental modeling have passed throughmany stages since then.
Early experimental modeling used smallelectrodes in “infinite” saltwater baths.
e learning curve in gaining understanding of the reservoir.
elf, resistivity modeling is an integral part of the latest developments,
stigating the effects of anisotropy.
Later, tool responses were studied usingmock-up sondes in more realistic environ-ments created by using thin impermeablemembranes to separate waters of differentsalinity. For a number of years, a resistornetwork was used at the Schlumberger-DollResearch laboratory in Ridgefield, Connecti-cut USA. This network, consisting of tens ofthousands of electrical resistors, simulatedresistivities in borehole, invaded zone andvirgin formation. In addition, theoretical cal-culations of sonde responses to layered andinvaded formations generated books ofdeparture curves. This theoretical approachwas especially important for tools that hadlarge depths of investigation or were notreadily adaptable to laboratory experiments.
Large improvements in computing capa-bility have introduced qualitative changes in
the role of modeling during the last decade.Computerized modeling has reduced fromweeks to minutes the time required to cal-culate many effects of tool design changes.One can now systematically explore theeffects of environmental conditions such asborehole rugosity and caves, mudcake,invasion, dip, shoulder beds, and formationanisotropy on resistivity tool responses.
The latest stage in this evolution is aimedat providing rapid, low-cost log interpreta-tion through the use of fast computers withlarge, high-speed memories and efficientprograms. Recently, even massively parallelprocessing has been introduced to servethese goals. Some log interpretation by
interactive modeling is possible even onpersonal computers.1 Program packages forsimple one-dimensional (1D) modeling arecommonplace; two-dimensional (2D) andthree-dimensional (3D) modeling are practi-cal in many special cases, although theygenerally require the use of mainframes orsupercomputers. Two-dimensional modelingpermits examination of axially symmetricradial variations—for example, treatingzero-dip layering and coaxial invasionsimultaneously. Three-dimensional model-ing also handles azimuthal variations suchas circumferentially irregular caves or inva-sion, sonde eccentering and dipping beds.
Oilfield Review
For help in preparation of this article, thanks to CharlesFlaum, Services Techniques Schlumberger, Montrouge,France; Stan Gianzero, Austin, Texas, USA; MartinLüling, Schlumberger-Doll Research, Ridgefield, Con-necticut, USA; Bill MacGregor, Schlumberger Austin Sys-tems Center, Austin, Texas, USA; Richard Rosthal,Schlumberger Well Services, Houston, Texas, USA; LiangC. Shen, Well Logging Laboratory, University of Hous-ton, Houston, Texas, USA; Julian Singer, Schlumberger,New Delhi, India.In this article, AIT (Array-Induction Imager tool), CDR(Compensated Dual Resistivity), EPT (ElectromagneticPropagation Tool), Phasor and MicroSFL are marks ofSchlumberger. Cray is a mark of Cray Research, Inc.Connection Machine is a mark of Thinking MachinesCorp. VAX is a mark of Digital Equipment Corp.
1. Georgi DT, Phillips C, Hardman R: “Applications ofDigital Core Image Analysis to Thin Bed Evaluation,”Society of Core Analysts Transactions, paper 9206,June 14-17, 1992, Oklahoma City, Oklahoma, USA,(in press).
2.“Vertical Resolution of Well Logs: Recent Develop-ments,” Oilfield Review 3, no. 3 (July 1991): 24-28.The vertical response function describes the character-istic response of the tool as it passes, perpendicularly,an infinitely thin bed.
1800
1810
1820
1830
1840
0.2 2000 0.2 2000
Dep
th, f
t
Computed DeepInduction
Model Rt Profile Model Rt Profile
ohm-m ohm-m
Computed DeepInduction
nThe lack of uniqueness in forward mod-eling. Different formation models canlead to nearly identical induction logs.The ambiguity is resolved, however, byaddition of nearly any log with high verti-cal resolution.
Modeling Versus InversionThe distinction between modeling—fre-quently “forward” modeling—and inversionis sometimes muddied. The latter typicallyattempts to “back out” true resistivity, Rt,directly from the log with a minimum ofassumptions. The best known example ofthis approach is vertical deconvolutionthrough the use of inverse filters. In itspurest form, this method requires only thatthe vertical response function (VRF) of thetool be known accurately.2 In practice, VRFsare usually formation dependent, so approx-imations must be used. Nevertheless,deconvolution has been employed success-fully, running in real time on logging unitcomputers. Artifacts may appear, however, ifinverse filters overreach in trying to achievefine vertical resolution or if the 2D assump-tions implicit in the filter are violated.
In modeling, on the other hand, the ana-lyst suggests an environmental model. Thistrial model includes a description of theborehole and formation geometry and“parameter values”—numbers assigned tovariables such as borehole diameter andbedding dip, thickness and resistivity. Then,the tool physics —a model in its ownright—is used to compute an expected log,which is compared with the field log. If thematch isn’t good enough, the initial trialmodel is altered and the calculationrepeated. This process is iterated until thetwo logs match satisfactorily. Several criteriafor the quality of match are used, from sim-ple eyeballing to the more sophisticatedleast-squares and maximum entropy meth-ods described later. The model’s geometryor parameter changes are executed interac-tively, using the analyst’s intuition and expe-rience, or automatically, if computers andprograms of sufficient power are available.
Modeling intrinsically yields consistencywith the field log, even though the solutionisn’t unique. This nonuniqueness is seldom
July 1992
a serious problem, however, because therange of possible formation models can beseverely constrained by local knowledgefrom cores and logs. An extreme example ofthis condition shows two grossly differentmodels that predict the same deep induc-tion log (above). But in practice, almost anyadditional log with vertical resolution ofabout 1 foot [30 cm] or less (gamma ray,EPT Electromagnetic Propagation Tool, dip-meter or photoelectric factor, Pe) wouldresolve this ambiguity.
23
2
A
B
C
D
E
F
2 200 2 200ohm-m ohm-m
Measured Deep Induction Measured Deep Induction
Computed Deep Induction Computed Deep Induction
Rt Rt
Model One Model Two
2 200ohm-m
Measured Deep Induction
Computed Deep Induction
Rt
0
25
True
ver
tical
dept
h, ft
0
50
Mea
sure
d de
pth,
ftnThe importance ofmodeling. For awell inclined 56°,an initial trialmodel and its com-puted deep induc-tion log are com-pared with the fieldlog (model one).Modification of themodel leads to bet-ter agreementbetween computedand field logs, butsome discrepanciesremain (modeltwo). The finalmodel (right) pro-duces nearly per-fect agreement.Depths in modelsone and two are ona log-measuredscale, while thefinal results are pre-sented on a truevertical depth scale.The two scales aredifferent becausethe well is slanted.
Examples of Formation EvaluationThe economic importance of modeling isillustrated by a North Sea reserves calcula-tion based on induction log interpretationcarried out with a Schlumberger programcalled Induction Sonde in MultilayeredMedia (ISMLM).3,4 This is a 1D inductionmodeling code for layered media thatneglects borehole and invasion effects. Ithandles up to 150 parallel dipping layers.Invasion was considered negligible becausethe well was drilled with oil-base mud.
The measured deep induction log, initialtrial formation model and computed log areshown in the first model (above, left). High-resolution details of the trial model wereprovided by an EPT log. The effects of thefirst model revision are based on the ana-lyst’s experience. The analyst changes thethicknesses of conductive beds, adds layersto the sands, and improves the depth match
4
between the induction and EPT curves.Since visible discrepancies between thefield log and the modeled log remained, fur-ther model revisions were needed toachieve the final results (above, right). Thefinal model reduced the well’s estimatedaverage water saturation from 9.7% to7.2%. Because the hole is deviated 56°, thelog-measured depths (MD) are greater thanthe true vertical depths (TVD), and bedthicknesses are similarly magnified. ThisMD expansion of scale, obvious in highlydeviated wells, will be observed again in alater example.
Subsequent to this work, the ISMLM codewas made part of the Electromagnetic Mod-eling package, called the ELMOD program.The program consists of 1D and 2D codesthat compute the responses of electric log-ging tools to models of downhole environ-ment. The programs can be run at anySchlumberger Data Services Center or on aSchlumberger VAX workstation. Configura-
tions include induction and laterologs inmultiple horizontal beds with borehole andinvasion, and induction and CDR Compen-sated Dual Resistivity logs in multiple dip-ping beds without borehole or invasion.Other tool environments and fast inductioncodes are being evaluated for addition tothe package. Other codes in this programcan be used for dipping bed interpretation.
In the dipping bed interpretation using theELMOD program, bed boundaries were pro-vided by the Phasor deep induction log, andan apparent dip of 38° by the dipmeter log.5Discrepancy between the field induction logand initial computed log led to revision ofthe trial model and a recomputed log (nextpage, top). This improved the fit, but onemore iteration—fine-tuning the shapes ofsome beds and adjusting for overcompensa-tion—yielded an excellent visual match(ELMOD Simulation Three). Although thefinal model is not a unique representation of
Oilfield Review
nModeling simulation using the ELMOD program for interpretation of a deep induction Phasor log in a North Sea well withapparent dip of 38°. The initial trial model was refined in two steps, left to right, until agreement was reached between themodel-computed and field logs. Simulation Three was consistent with the log analyst’s knowledge of the field.
3. Fylling A and Spurlin J: “Induction Simulation, TheLog Analysts’ Perspective,” Transactions of the SPWLA11th European Formation Evaluation Symposium,Oslo, Norway, September 14-16, 1988, paper T.
4. Details of the ISMLM program: Anderson B and Gianzero S: “Induction SondeResponse in Stratified Media,” The Log Analyst 24,no.1 (1983): 25-31.Anderson B, Safinya KA and Habashy T: “Effects ofDipping Beds on the Response of Induction Tools,”paper SPE 15488, presented at the 61st SPE AnnualTechnical Conference and Exhibition, New Orleans,Louisiana, USA, October 5-8, 1986.Anderson B and Chew WC: “A New High Speed
ELMOD Simulation One ELMOD Simulation Two ELMOD Simulation ThreeD
epth
, ft
1000
1050
1100
1150
0.2 2000ohm-m
Resistivity
0.2 2000ohm-m
Resistivity
0.2 2000ohm-m
Resistivity
Measured deepPhasor inductionModeled deepPhasor inductionRt
Summary of Induction Modeling ProgramsSource
University ofHouston
Well LoggingLaboratory1
Schlumberger
Program
NLAYER
MLIND
DIPEX
WLAP
NDIP
IND9110
ISMLM2
(ELMOD)
TRIKHZ3
(ELMOD)
Radial Annuli
0
1
0
2
0
2
0
9
Horizontal Layering
Unlimited
30 feet [9 m]
3 layers
10, max 66 ft [20m]
100 layers
15 layers
150 layers
3 layers
Accounts for Dip?
No
No
Yes
No
Yes
No
Yes
No
CPU Time
10 sec4
11 sec4
40 sec4
12 min4
NA
NA
20 min5
3 min5
Method
Transmission
Perturbation
Analytical
Propagation
Analytical
Eigenmode
Analytical
Hybrid
1. Well Logging Technical Report, No. 7. Houston, Texas, USA: University of Houston Well Log-ging Laboratory, October 30, 1986. These programs are available to supporters of the labora-tory, a consortium including most of the major oil companies.
2. Anderson B and Gianzero S, reference 4.Anderson B, Safinya KA and Habashy T, reference 4.
3. Anderson B and Chew WC, reference 4.Anderson B and Chang SK, reference 13.
4. IBM 3090
5. VAX 11/780
5. Anderson B, Barber TD, Singer J, and Broussard T:“ELMOD—Putting Electromagnetic Modeling to Workto Improve Resistivity Log Interpretation,” Transactionsof the SPWLA 30th Annual Logging Symposium, Den-ver, Colorado, USA, June 11-14, 1989, paper M.
true resistivity, it was consistent with theanalyst’s knowledge of the field, and thepredicted water saturations were accorded ahigh degree of confidence.
Many more computer programs for a vari-ety of electrical logging measurements havebeen developed by service companies, oilcompanies and universities. Some are avail-able for commercial use, others only forresearch purposes (see “Summary of Induc-tion Modeling Programs,” above).
Technique for Calculating Synthetic Induction andDPT Logs,” Transactions of the SPWLA 25th AnnualWell Logging Symposium, New Orleans, Louisiana,USA, June 10-13, 1984, paper HH. 25July 1992
Deep inductionDeep laterolog Shallow laterolog
Medium induction
Resistivity, ohm-m Resistivity, ohm-m1 10 100 1 10 100
4
2
0
-2
-4
Dis
tanc
e fro
m b
ound
ary,
ft
nShoulder-bed effects on induction logs and laterologs in horizontal wells, calculatedby modeling. These curves show how traditional heuristic thinking does not apply: inhorizontal wells, laterolog tools are more sensitive to conductive than to resistive shoul-der beds, while the opposite is true for induction tools.
26 Oilfield Review
6. Gianzero S, Chemali R and Su S-M: “Induction, Resis-tivity, and MWD Tools in Horizontal Wells,” Transac-tions of the SPWLA 30th Annual Logging Symposium,Denver, Colorado, USA, June 11-14, 1989, paper N;also in The Log Analyst 31 (May-June, 1990): 158-170.Burgess T and Voisin B: “Advances in MWD Technol-ogy Improve Real Time Data,” Oil & Gas Journal 90,no. 7 (February 17, 1992): 51-61.
7. Chemali R, Gianzero S and Su SM: “The Dual Lat-erolog in Common Complex Situations,” Transactionsof the SPWLA 29th Annual Logging Symposium, SanAntonio, Texas, USA, June 5-8, 1988, paper N.
8. Anderson B, Barber TD and Lüling MG: “The Role ofComputer Modeling in Log Interpretation,” Transac-tions of the SPWLA 13th European Formation Evalua-tion Symposium, Budapest, Hungary, October 23-25,1990, paper L.
9. Anderson B, Bonner S, Lüling MG, and Rosthal R:“Response of 2-MHz LWD Resistivity and WirelineInduction Tools in Dipping Beds and Laminated For-mations,” Transactions of the SPWLA 31st AnnualLogging Symposium, Lafayette, Louisiana, USA, June24-27, 1990, paper A.
10. Leake J and Shray F: “Logging While Drilling KeepsHorizontal Well on Small Target,” Oil & Gas Journal89 (September 23, 1991): 53-59.
11. Clark B, Lüling MG, Jundt J, Ross M and Best D: “A Dual Depth Resistivity Measurement for FEWD,”Transactions of the SPWLA 29th Annual LoggingSymposium, San Antonio, Texas, USA, June 5-8,1988, paper A.
12. “Formation Anisotropy: Reckoning With its Effects,”Oilfield Review 2, no. 1 (January 1990): 16-23. Anderson et al, reference 8.
Modeling for Bit Guidance in Horizontal DrillingForward modeling is serving needs otherthan conventional log interpretation, suchas guiding the bit while drilling deviated orhorizontal wells and evaluating the influ-ence of adjacent (shoulder) beds on hori-zontal well logs.
Recent modeling studies in horizontalwells make clear that conventional rules-of-thumb don’t apply. Predictions have beenmade of responses of induction and focusedelectrode tools to shoulder beds,6 an impor-tant subject when logging thin beds or inhorizontal wells exhibiting vertical drift.Because the induction tool is relativelyunaffected by the borehole, its calculationswere carried out analytically. For thefocused electrode tools, however, the bore-hole is an essential part of the problem, soforward modeling was carried out with a 3Dfinite-element method (FEM) computation, amuch larger enterprise. This investigationconcluded that the electrode devices aremore sensitive to conductive than resistiveshoulder beds, indications of which hadappeared earlier,7 and that the opposite istrue for induction tools. This behavior is thereverse of that observed in vertical wells. Inthe calculated responses of these tools tothe boundary between two beds of 1 ohm-m and 10 ohm-m, the characteristic polar-ization horn is clearly visible on theinduction curve (above, right). This hornappears when surrounding beds have highresistivity contrast and the bed boundarydips more than about 45°.8 It is created byoscillating polarization charges induced atthe bed boundary.
Another modeling study explored theeffects of dipping beds and laminated for-mations on induction and CDR toolresponses.9 Since the CDR model assumes
point dipoles, the code was first qualified bycomparison with exact FEM calculations forhorizontal multiple beds and with test tankexperiments covering dips of 0° to 90°. Thisstudy predicted that both dip and shoulderscan cause shallow resistivity, Rps, and deepresistivity, Rad, to separate, with Rps readingcloser to Rt. Prominent polarization hornsappear at high-dip bed boundaries whenresistivity contrast is large, and the CDR toolmakes resistivity anisotropy apparent (Rps >Rad), the effect increasing with dip angle.
Oxy USA used forward modeling exten-sively while drilling horizontally into theCruse sand in La Salle Parish, Louisiana,USA.10 To avoid the problem of water con-ing they had observed in vertical wells, Oxyengineers planned to penetrate horizontallyinto the top 10 feet [3 m] of the 40-foot [12-m] thick Cruse. Resistivity models of markerbeds and of the pay sand itself were created
from induction logs in two nearby verticalwells. Then, logs expected in the horizontalwell were computed for the CDR tool usedin Logging While Drilling (LWD). This 2-MHz resistivity tool provides two outputs:Rps from a phase-shift measurement, andRad from attenuation.11 No one knew accu-rately in advance what the logs would looklike as the well curved through the markersand into the Cruse. In addition, actual bore-hole inclination is often not exactly asplanned. Therefore, prior to drilling,ELMOD’s dipping-bed code was used tocompute CDR logs expected at severalapparent dips.9 Thus, the right modeled logwould be available immediately for com-parison with the field log when the actualrelative dip became known while drilling.
Since the two CDR resistivity measure-ments have different depths of investigation,the curves are predicted to separate just
nModeled (left) and measured CDR resis-tivity logs as the tool curves into a paysand in a horizontal well. The measuredRps and Rad curves (right) show a charac-teristic crossover as the tool crosses fromthe cap shale into the sand. Just belowthe boundary, Rps exhibits a polarizationhorn created by induced oscillatingcharges at the interface. The logs com-puted from a model assuming an 85°inclination show the same characteristics.This inclination expands the MD scale bya factor of 11.47 relative to the TVD scale.(At 85°, 1 foot of TVD equals 11.47 feet ofMD [30 cm of TVD = 344 cm of MD].)
Mea
sure
d de
pth,
ft
x700
x800
2 2000ohm-m
Rad
Rps
Modeled CDR Resistivityat 85°
x750
Polarizationhorn, top of Cruse
Mea
sure
d de
pth,
ft
x800
2 2000ohm-m
Rad
Rps
x750
Polarizationhorn, top of Cruse
27July 1992
before the highly inclined tool leaves the capshale and enters the sand. Also, at the sand’supper boundary, the Rps curve predicts thecharacteristic polarization horn.8 This horn isexpected to be somewhat broader on thefield log than on the computed log becausemodeling calculations ignore the boreholeand the finite size of the tool’s transmitterand receiver dipoles. These features are visi-ble on logs modeled for the planned inclina-tion of the borehole (85°) at the top of thetarget sand, and on actual field logs (below).The boundary of the Cruse is located pre-cisely, confirming that the bit entered thesand at the desired depth.
Modeling played a role in interpreting anunexpected logging observation within thetarget zone. Differences between shallowand deep CDR readings seemed too large toexplain by invasion because CDR measure-ments are made shortly after a section isdrilled. Furthermore, the vertical-well induc-tion logs measured only 4 ohm-m, while inthe nearly horizontal well, the CDR deepattenuation measurement approaches 10ohm-m. In the vertical well, core analysisfrom the upper part of the Cruse suggestedmodeling it as a series of high- and low-resistivity streaks. Logs calculated on thismodel agreed with both the vertical deepinduction log and horizontal Rad measure-ments, confirming that the formation resis-tivity is anisotropic.12 This study led todevelopment of a CDR software packagethat provides continuous logs of calculatedhorizontal and vertical resistivities, usingmeasured apparent resistivity and relativedip angle as inputs.
How Modeling Calculations Are Carried OutPrior to starting the modeling process, manyanalysts apply chartbook corrections to thefield log. These corrections make the fieldlog more accurate—closer to Rt, for exam-ple—and allow the use of simpler modelsand faster computer programs for modeling.Then, the corrected field log and all otherconstraining information are used in settingup the initial trial formation model. Less fre-quently, the uncorrected field log is used,and the burden of accounting for featureslike borehole, invasion, shoulder and skineffect is borne by the formation model, toolmodel and computing code. In this case,the environmental corrections areaccounted for simultaneously, as preferred,rather than sequentially, as when applyingchartbook corrections. Unfortunately, thisapproach requires large programs and longcomputer times.
Although further advances in computingpower will likely change the picture in thefuture, current practice usually requires thelog analyst to construct an initial trial modeland propose changes at each iteration.“Feel” or “intuition” are frequently the basisfor doing this, often acquired from fieldexperience or published studies of toolresponses to specific environmental fea-tures. In the last five years there has been asurge of such studies that, themselves, were
28 Oilfield Review
nA small section of a simple grid used inaxially symmetric finite-element model-ing calculations of induction logs. In thecalculation, the electromagnetic field isdetermined at each node, or intersection,on the grid. In some cases the completegrid must extend hundreds of feet radiallyand vertically.
50
40
30
20
10
0
-10
-20
-30
-40
-50
z, in
.
0 10 20 30 40 50 60 70ρ, in.
Tool Responses to Environmental Features
Tool
Induction(standard, Phasor)
Array InductionImager Tool
Laterolog
High-resolutioninduction
EPT
CDR
Dip, shoulders, thin beds,anistropy, laminations, Rxo << Rt
Borehole, caves, shoulders,radial resistivity variation
Shoulder, high dip, caves,anisotropy, very high Rt
Magnetic mud and formationminerals, conductive caves
Mudcake, standoff, invasion,depth of investigation, laminates
Dip, shoulder radial resistivityvariation
1, 2, 3, 4
5, 6
2, 7
8
9
4, 10
Environmental Conditions Reference
1. Anderson B: “The Analysis of Some Unresolved Induction Interpretation Problems UsingComputer Modeling,” The Log Analyst 27 (September-October 1986): 60-73.
2. Chemali R, Gianzero S and Su SM: “The Effect of Shale Anisotropy on Focused ResistivityDevices,” Transactions of the SPWLA 28th Annual Logging Symposium, London, England,June 29-July 2, 1987, paper H.
3. Anderson B and Barber T: “Strange Induction Logs—A Catalog of Environmental Effects,”The Log Analyst 29 (July-August 1988): 229-243.
4. Reference 9, main text.
5. Grove GP and Minerbo GN: “An Adaptive Borehole Correction Scheme for Array InductionTools,” Transactions of the SPWLA 32nd Annual Logging Symposium, Midland, Texas, USA,June 16-19, 1991, paper P.
6. Hunka JF, Barber TD, Rosthal R, Minerbo GN, Head EA, Howard AQ Jr, Hazen GA andChandler RN: “A New Resistivity Measurement System for Deep Formation Imaging andHigh-Resolution Formation Evaluation,” paper SPE 20559, presented at the 65th SPEAnnual Technical Conference and Exhibition, New Orleans, Louisiana, USA, September 23-26, 1990.
7. Reference 7, main text.
8. Strickland R, Chemali R, Su SM, Gianzero S, Walker M, Klein J and Sakurai S: “New Devel-opments in the High Resolution Induction Log,” Transactions of the SPWLA 32nd AnnualLogging Symposium, Midland, Texas, USA, June 16-19, 1991, paper ZZ.
9. Anderson B, Liu Q-H, Taherian R, Singer J, Chew WC, Freedman B and Habashy T: “Inter-preting the Response of the Electromagnetic Propagation Tool in Complex Borehole Envi-ronments,” Transactions of the SPWLA 32nd Annual Logging Symposium, Midland, Texas,USA, June 16-19, 1991, paper XX.
10. Habashy T and Anderson B: “Reconciling Differences in Depth of Investigation Between 2-MHz Phase Shift and Attenuation Resistivity Measurements,” Transactions of the SPWLA32nd Annual Logging Symposium, Midland, Texas, USA, June 16-19, 1991, paper E.
carried out by modeling (see “Tool Responsesto Environmental Features,” below).
Even when forward modeling strives onlyfor formation description, most of the bur-den is on codes that implement the toolmodel—a representation of hardware andphysics for calculating how the toolresponds to its environment. This modelmay be approximate because it idealizestool hardware—treating finite-size coils orelectrodes as points, for example. Or it maysimplify the physics—as by using the geo-metric factor approach rather thanMaxwell’s equations in calculating induc-tion tool responses. Alternatively, the hard-ware or the physics, or both, may be treatedexactly. The decision is a trade-off betweenaccuracy and computing efficiency. Ineither case, some particular computationalgorithm becomes the vehicle for arrivingat the goal.
Purely analytical methods, using exactmathematical solutions, employ codes thatrun rapidly and require only modest com-puter memories. This makes them wellsuited to the small computers readily avail-able to most log analysts, but they are intrin-sically limited to simple geometries such asinvasion with no layering or layering withno invasion.
From this standpoint, numerical methodsare ideal. They break intractable mathemati-cal problems into smaller, more manageablepieces. Numerical methods, such as 2D-and 3D-FEM codes, can solve differentialequations in almost any geometry. The FEMis widely used in research and engineering,from the design of automobile bodies to thestudy of diffusion over corrugated surfaces.
A typical logging application is thenumerical solution of Maxwell’s equationsfor induction tools. This problem eventually
reduces to solving a (usually) large numberof simultaneous linear equations by matrixmethods.13 The immediate objective is tofind the electromagnetic field’s vector poten-tial at the nodes, or intersections, of a 3Dgrid, in the most general case. Simpler grids(above) can be used in solving axially sym-metric (2D) problems. The complete gridmay extend hundreds of feet vertically andradially to adequately cover the electromag-netic field. In one modeling exercise, thegrid was terminated where the vector poten-tial had fallen 15 orders of magnitude—tozero for practical purposes—from the start-ing point near the transmitter. Grid sizeincreases with distance from the transmitterin regions where both vector potential andgeneralized geometric factor are fallingslowly. This increases computational effi-ciency, with negligible loss in accuracy. In1982, a CDC CYBER 750 computer typicallytook five hours to compute 25 feet [7.6 m] ofinduction log.14 Today, the whole log takesonly 15 minutes on a Cray supercomputer.Most pure FEM codes need a fast vector pro-cessor or a Cray unit to run with reasonableturnaround times. Unfortunately, this meansinterfacing from a remote site, a capabilitywith only limited availability at present.
Hybrid techniques that retain the advan-tages of both purely analytic and purelynumerical methods have been developed.They typically break the problem into two
nHybrid modelingcalculation of deepand shallow lat-erologs in 25 beds.Calculating timeincreased to 12minutes (comparewith figures above),but most of theadditional timewas used in seg-menting the logand recombiningthe pieces, ratherthan in the compu-tation itself.
nComparison ofhybrid and finite-element modelingcalculations fordeep and shallowlaterologs. The twomethods yieldnearly identicallogs, but thehybrid calculationuses only one-eighth as muchcomputer time(21/2 minutes on anIBM 3081) as thecomplete finite-ele-ment calculation.
29July 1992
Res
istiv
ity, o
hm-m
100
10
1
0.1
Res
istiv
ity, o
hm-m
100
10
1
0.1-120 -90 -60 -30 0 30 60 90 120
Depth, in.
Hybrid methodFinite-element methodRt
Hybrid methodFinite-element methodRt
Res
istiv
ity, o
hm-m
Depth, in.
1000
100
10
1
-600 -400 -200 0 200 400 6000.1
RtDeep laterologShallow laterolog
13. Anderson B and Chang SK: “Synthetic InductionLogs by the Finite Element Method,” Transactions ofthe SPWLA 23rd Annual Logging Symposium, Cor-pus Christi, Texas, USA, July 6-9, 1982, paper M;The Log Analyst 23 (November-December 1982):17-26. Also published with more mathematicaldetail as “Simulation of Induction Logging by theFinite Element Method,” Geophysics 49 (November1984): 1943-1958.
14. Gianzero S, Lin YY and Su SM: “A New High-SpeedHybrid Technique for Simulation and Inversion ofResistivity Logs,” SPE Formation Evaluation 3 (March1988): 55-61.
15. Kaufman AA: “Theory of Induction Logging,”Siberian Dept. of Nauka Press, Novosibirsk, 1965.Anderson and Chew, reference 4.
parts, one of which is attacked analyticallyand the other numerically. This leads tocodes that can quickly solve, even on exist-ing workstations, some problems in fairlycomplicated geometries. The axially sym-metric problem of the laterolog in a bore-hole through horizontal, layered formationsprovides a simple example of a hybridapproach. The 2D Laplace’s equation thatgoverns the potential distribution around thetool is reduced to two 1D problems in radial(r ) and vertical (z) planes, respectively, bythe customary separation-of-variables tech-nique. Then, the vertical distribution of thepotential can be treated numerically—bythe FEM, for example—and the radial distri-bution treated analytically—by modal anal-ysis—or vice versa.14 Compare the dual lat-erolog results of this method with acomplete FEM calculation in a 5-foot [1.5-m] thick noninvaded bed (left). The logs areessentially identical, but the hybrid calcula-tion uses only one-eighth as much computertime as the complete FEM calculation. If thehybrid calculation is extended to 25 beds(below, left), the hybrid time is increasedfrom 2.5 minutes to 12 minutes on an IBM3081. Most of the additional time, however,was used in segmenting the log into man-ageable portions and then recombiningthem, rather than in the computation itself.
Earlier studies of hybrid techniquesdescribe other methods for modeling induc-tion tool responses.15 Numerous codes and
30 Oilfield Review
nA sampling of codes and computer timesused in modeling induction tool responses.Dashed boundaries indicate cases forwhich geometric factor theory was used;solid boundaries indicate use of Maxwell’sequations. Times shown are for 50-footsections of log, where boundaries exist,and for a single-point calculation whereone infinitely thick bed is indicated.
Bedboundaries
Invasionfront
Axis ofsymmetry
Axis ofsymmetry
Analytical Hybrid
Numerical
Geometrical factor theory:1 min VAX (Doll H, 1949) 1
Unlimited cylindrical boundaries:1-2 min VAX (Gianzero S and Anderson B, 1984)2
Unlimited bed boundaries:1-2 min VAX (Anderson B andGianzero S, 1983, reference 4)
Eccentricity: 5 min VAX(Gianzero S, 1978)3
Unlimited dipping beds:2-3 min VAX (Anderson B, Safinya KA and Habashy T, reference 4)
3D finite elements: < 2 hrs Cray (in progress)
2D finite elements: 20 min Cray(Anderson B and Chang SK, ref. 13)
Dip with borehole and invasion:15 min VAX (in progress)
Fast, semianalytic: 3 min VAX (Anderson B and Chew WC, reference 4)
Hybrid with/without skin effect:2 min VAX (Kaufman AA, reference 15)
Nonuniforminvasion
front
Known boundaries Approximated boundaries
1. Doll HG: “Introduction to Induction Logging and Applica-tion to Logging of Wells Drilled with Oil-Base Mud,”Transactions, AIME 186 (1949): 148-162.
2. Gianzero S and Anderson B: “Mathematical Theory forthe Fields Due to a Finite H.C. Coil in an Infinitely ThickBed with an Arbitrary Number of Co-Axial Layers,” TheLog Analyst 25, no. 2 (March-April 1984): 25-32.
3. Gianzero SC: “Effect of Sonde Eccentricity on Responsesof Conventional Induction-Logging Tools,” IEEE Transac-tions on Geoscience Electronics GE-16, no. 4 (October1978): 332-339.
computer times exist for various types ofmodeling (next page). Dashed boundariesindicate problems solved with geometricfactor theory; solid boundaries indicatesolutions developed using exact electromag-netic theory. The times shown are all for 50-foot [15-m] sections of log, where bedboundaries exist, and for a single-point cal-culation in the two cases where oneinfinitely thick bed is indicated (in this casethe log is constant as the tool moves verti-cally). These codes may be used for study-ing tool responses to specific environmentalperturbations, as well as in modeling for for-mation evaluation.
Least-Squares and Maximum EntropyMatching CriteriaSubjective eyeballing can usually evaluatethe fit between the field log and the mod-eled log. But two other matching criteria,least squares and maximum entropy (MEM),have received attention. They are of interest,however, more because they use algorithmsthat converge automatically to their respec-tive best-fitting models than because of theirinherent objectivity. No human interventionis needed to alter the model at each itera-tion. In application, these criteria lead tocomputational methods that are differentfrom one another and from the manualinteractive approach.
One application of the least-squares crite-rion picks the set of formation parametersthat minimizes the sum of the squares of dif-ferences (SSD) between the field log and thelog derived from the assumed formationmodel.16 In the language of statistics, themean square deviation, or variance,between the two logs is minimized. Forexample, with initial trial values of thicknessand resistivity assigned to each bed in themodel, the program computes the predictedlog and finds the SSD between it and the
24 in.
40 in.
32 in.
8 in.
25ohm-m
200 ohm-m
200ohm-m
5 ohm-m15 ohm-m
50 ohm-m
1ohm-m
Bor
ehol
e
nForward model-ing by applyingthe least-squaresmethod to a set ofsimulated logs.Synthetic inductionlogs and a lat-erolog (lower left)were computed forthe formationmodel shown.Then, using thedotted-line param-eter values (esti-mated by a stan-dard interpretation)as the initial guess,the least-squaresmethod convergedto the solid-line“final iteration”model (lower right).
Simulated Resistivity
Dep
th, f
t
0.2 2000ohm-m0
30
Laterolog
MediumInduction
DeepInduction
Computed Rt
0.2 2000ohm-m
ComputedInvasion Diameter
in.0 100
Initial guessFinal iteration
field log. Then, the model’s bed thicknessesand resistivities are automatically changedto reduce the SSD. This procedure repeatsuntil the SSD converges to a minimum.
Computing time depends, as in other iter-ative schemes, on how close the initialmodel log is to the field log. To keep thistime short, the user frequently minimizesthe number of parameters to be optimizedsince computing time increases with thisnumber. Also, if some parameters “interfere”with one another—produce similar effectson the log—their precision is adverselyaffected even though they are optimallydetermined in the least-squares sense.
Consider an application of the least-squares method to a set of simulated logs(right). The initial trial model used parame-ter values estimated by a standard interpre-tation. Because flushed zone resistivity, Rxo,values were assumed known from theMicroSFL resistivity tool, the modeling pro-gram was called upon to determine the payzone thickness in addition to Rt and inva-sion diameter in each of the three beds. Thisearly-1980s calculation took 20 minutes onan IBM 360, using FEM. Today it can bedone on a workstation in under 60 seconds.
The maximum entropy criterion comesfrom Shannon’s information theory and hasbeen used for reconstruction of blurredsatellite photographs and for extracting sig-nals from noisy data. It leads to a differentmethod of calculation17 although themethod bears some similarity to the least-squares approach—differences betweenfield log and modeled log are used in calcu-lation of χ2, a quantity related to the SSD.With respect to the use of χ2, the maximumentropy method can be considered anextension of the least-squares approach. Theactual algorithms employed reduce χ2 itera-tively, while maintaining entropy close to itsmaximum at each iteration. This procedure
16. Lin Y-Y, Gianzero S and Strickland R: “Inversion ofInduction Logging Data Using the Least SquaresTechnique,” Transactions of the SPWLA 25thAnnual Logging Symposium, New Orleans,Louisiana, USA, June 10-13, 1984, paper AA.
17. Dyos CJ: “Inversion of Induction Log Data By theMethod of Maximum Entropy,” Transactions of theSPWLA 28th Annual Logging Symposium, London,England, June 29-July 2, 1987, paper T.Freedman R and Minerbo GN: “Maximum EntropyInversion of Induction Log Data,” paper SPE 19608,presented at the 64th SPE Annual Technical Confer-ence and Exhibition, San Antonio, Texas, USA,October 8-11, 1989. Also published with minorchanges in SPE Formation Evaluation 6 (June 1991):259-268.
July 1992
selects the unique least-squares solutionhaving maximum entropy.
There is an important difference betweenthe MEM and least-squares methods. Thefinal model selected by the MEM is thesmoothest possible profile rather than theone that yields a minimum SSD. Of all pos-sible models, it is the one that has minimuminformation content (in the information-the-ory sense) consistent with the field log. Con-sequently, this criterion inhibits the appear-ance of artifacts that sometimes show upwhen using the least-squares method.Unfortunately, the MEM is computationallyexpensive. The cost is reduced if a prioriknowledge of bed boundaries is included inthe initial model, as with other methods, butcost remains the main impediment to widerMEM use at present.
31
-1000 -500 0 500 1000
Vertical position, in.
Oklahoma FormationR
esis
tivity
, ohm
-m
1000
100
10
1Medium inductionMEM without beds input
Synthetic deep inductionRt
Res
istiv
ity, o
hm-m
1000
100
10
1 Raw Doll ID signalMEM without beds input
Rt
Con
duct
ivity
, milli
Sie
men
s/m
0.1
1
10
100
1000
Cal
iper
, in.
7
9
11
13
15
17
8200 8250 8300 8350
Vertical position, ft
Measured raw deepinduction signal
MEM computed raw deep induction signal
Caliper
nForward model-ing of an inductionlog by the maxi-mum entropymethod (MEM)(left) and usinginformation fromthe medium induc-tion log to removeoscillations(below). Geometricfactor theory wasused to calculatethe synthetic deepinduction log fromthe assumed Rt pro-file. An advancedversion of the max-imum entropymethod, assumingno bedding struc-ture in the initialtrial model, thenconverged to thefinal Rt profile.Oscillations in thethicker beds resultfrom the Fourierspectrum of thedeep inductiontool’s vertical reso-lution function,and are unrelatedto use of the maxi-mum entropy mod-eling approach.
nA deep inductionfield example com-paring the logcomputed by themaximum entropymethod with themeasured log. Thepredicted final-model profile isalso shown.
32
Nevertheless, the MEM can provide a for-mation model from an induction log, with-out assuming knowledge of bed boundaries(left). The synthetic “field” log was calcu-lated by applying geometric factor theory(for simplicity) to the Rt distribution shown,and geometric factor theory was thus usedin representing tool response in the forwardmodeling. Oscillations visible in the thickerbeds are not a consequence of the MEM,but result from so-called blind frequenciesin the Fourier spectrum of the deep induc-tion’s VRF. These oscillations are readilysuppressed by adding information from themedium induction. Addition of the mediuminduction contributes a smoothing effect. Anactual example shows how well an MEM-predicted log matches its correspondingfield-measured induction (below, left).
The FutureFurther developments in modeling arealready in the pipeline. Generally, thesetake two forms, depending on the computa-tion size envisioned. One effort is aimed atproviding a computed log in about oneminute, using interfaces that are suitable forworkstations. These should make the pro-cess more analyst-friendly, through featureslike entry of parameters in graphical formatand windows that display the field log, thecomputed log and the model, including itsparameters, dips and boundaries.
The other development arena involvesextending the capabilities of large comput-ers. Primarily intended for use in tooldesign, these studies are deep into the 3Ddomain, using the FEM for modeling com-putations that take hours even on a super-computer. Still more exploratory are investi-gations of advanced algorithms andautomatic code generation for speeding upcomplex calculations. Some algorithms arebeing applied in commercial seismic pro-cessing using parallel computers like theCM-2 Connection Machine. —JT
Oilfield Review
Mixed-Terrain 3D Seismics in The Netherlands
3D seismics is an exploration technique that improves the success ratio of wildcat wells. Cost-efficient at
sea and on land, it has also proved its worth in mixed terrains. This article describes the planning, acquisi-
tion and processing of perhaps the most difficult 3D mixed-terrain survey ever made.
July 1992
Carsten PetersenAssen, The Netherlands
Horst BrakensiekMarkos PapaterposHannover, Germany
1. Bukovics C and Nooteboom JJ: “Combining Tech-niques in Integrated 3D Land, Shallow Water andDeep Channel Seismic Acquisition, ” First Break 8(October 1990): 375-382.
nAerial view of the Europort near Rotterdam, with 3D seismic survey grid superposed—distance between lines is 400 m. The gridindicates the ideal shooting and receiving positions. In practice, equipment has to circumvent obstacles. Additionally, the surveyhas to cope with the ever-changing terrain, from water to land and back again. The 15-km by 25-km survey took 6 months.
Since the late 1970s, the exploration andproduction company Nederlandse AardolieMaatschappij b.v. (NAM), jointly owned byShell and Exxon, has been systematicallyacquiring three-dimensional (3D) seismiccoverage of its Dutch concessions. Most ofthese surveys have been for exploration, apioneering trend when the surveys werebegun—3D seismic technology was initiallyheralded as a tool for improving reservoircharacterization. But the financial risks have
paid off. In 1988, 12 of NAM’s 15 explo-ration wells struck pay, an astonishinglyhigh proportion.1
The terrain has not been easy. The peopleof the Netherlands not only have harnessedevery square meter of their land for urban,agricultural or industrial use, but throughdyke building have steadily claimed vastareas of subsea land called “polder.” The
33
Biesbosch1986
Dordrecht1988
Beijerland1987
1
2
Amsterdam
3
7
6
5
1
2
Urban
GreenhousesFarmland
Wildlife reserveIndustrial/Refineries
4
Den Haag1985-6
Monster1990
area shown by thegreen rectangle inblock 4. Complex-ity of the survey isdemonstrated in thedetail of the port,which also showsthe 400-m by 400-m survey grid.
34
nLocation ofNAM’s 3D surveysnear The Hagueand Rotterdam.This articledescribes the 1990Monster survey, socalled because ofits complexity andthe fact that thetown of Monsterlies in the surveyarea. The adjoin-ing marine surveywas performed atabout the sametime. The surveycovered farmland,urban develop-ment, extensivegreenhouse culti-vation and theworld’s busiest oilport and refinerycomplex. At anygiven moment, theteam would bedeployed over an
2. “Company Profile: Shell,” First Break 9 (December1991): 566-567.
3. PRAKLA together with GECO now forms part of theSchlumberger seismic company, GECO-PRAKLA.
countryside is crisscrossed with canals, largeand small. Densely populated, the complexterrain provides endless challenge to theversatility of seismic crews as they coordi-nate complex shooting configurations.
NAM’s progress has been steady. Thou-sands of square kilometers of land and near-shore coast have been surveyed using 3Dtechniques. The work has concentrated intwo main areas, in the northeast of the coun-try including the giant Groningen gas field,and in the southwest, in and around thecities of Rotterdam and The Hague (left).
The southwestern area is particularly chal-lenging for both acquisition and processing.Some of the surveys have to contend with amix of near-shore marine areas, harbors,urban sprawl, sections of land covered ingreenhouses, and farmland. Typically, threedifferent sources are used—airguns, dyna-mite and vibrators—and two types ofreceivers—geophones and hydrophones.The several combinations of source andreceiver produce seismic traces that requirecareful phase correction during processing.
In addition, the processing must contendwith very difficult statics. Statics describesthe time correction that must be applied toeach individual trace to neutralize the effectof the unconsolidated weathering layer. Inthe Rotterdam area, the weathering layervaries dramatically in both thickness andacoustic velocity—two key controllingparameters—from one geophone station tothe next.
Despite these problems, 3D surveys herehave yielded valuable results. NAM reportsidentifying potential pay and then drilling adiscovery well beneath the Shell refinery inthe Europort, Rotterdam.2 This articledescribes the seismic acquisition of one ofthe last pieces of the puzzle for this south-west survey area by PRAKLA in 1990/91and accompanying processing challenges.3
PlanningThe survey area measures about 15 by 25kilometers [9 by 15 miles] and is composedof five types of terrain. The southwesternsection comprises mainly farmland. This isbordered to the north by the Europort withits web of canals through which about 200large boats, for example oil tankers, andnumerous smaller vessels ply daily, and thelargest conglomeration of refineries inEurope. The eastern section covers suburbsof Rotterdam. To the northwest is againfarmland, and then an area of land, 9 by 9km [6 by 6 miles], entirely covered ingreenhouses. Northwest of the greenhousesis beach, protected as a wildlife sanctuary. Ifthat were not already daunting enough, the
Oilfield Review
nGreenhouses cov-ering severalsquare kilometersof the survey. ThePRAKLA crewnegotiated thissensitive areausing explosive fora source—no glasswas shattered!
Conversion GlossaryLand Acquisition
20m ⇔ 66 ft
40m ⇔ 131 ft
400m ⇔ 1312 ft
4.8 km by 1.2 km ⇔ 3 miles by 0.75 mile
2.4 km by 0.8 km ⇔ 1.5 miles by 0.5 mile
0.8 km ⇔ 0.5 mile
75 cm ⇔ 30 inches
2 m ⇔ 6.6 ft
1800 m ⇔ 5906 ft
25 m ⇔ 82 ft
Bin size
Reciever/shot spacing
Grid spacing
Reciever array
Rectangle of coverage
Swath width
Marine Acquisition
Airgun pontoon draft
S.V. Solea draft
Streamer length
Group spacing
survey also had to move out to sea and pro-vide overlapping coverage with a marinesurvey previously commissioned by NAMand completed by GECO.3
The first planning step was to decidewhere to start and the order in which to sur-vey these areas, each having its own specialaccess restrictions and safety aspects. Thesurvey was planned to begin in June 1990,and the following restrictions applied. Workin water areas was ruled out during winterbecause of the rapid onset of death in theevent of anyone falling overboard. Work onthe beaches was ruled out in spring duringthe bird breeding season and was also notdesirable during summer because of touristsand sunbathers. Work in farmland is bestduring winter when cattle are confined tosheds, but tolerable during any season.Once you have figured how to do it, seismicsurveying in greenhouses is possible any-time during the year.
The overall planning therefore took shapeas follows: The survey party would begin ineasy territory in the farmland in Block 1.They would then move up to the harborarea at the top of Block 1, gaining widerexperience with an increasingly large arrayof equipment. Flexibility was to prove thekey to accomplishing the difficult stages toschedule, and this early training periodproved essential for party members to gainfamiliarity with all types of equipment.
During the summer, the survey wouldthen move to Block 2 and survey the mouthof the port and its beach area. Towardautumn, the party would split, with asmaller group surveying the beach andadjoining offshore area in Block 3 and themajority of the crew deployed in Blocks 4and 5, the remaining harbor area and theRotterdam suburbs. As winter approached,the crews would recombine to survey thefarming area of Block 6 and then finish withthe greenhouses in Block 7 (above, right).
Once the overall sequence was estab-lished, detailed arrangements could begin.Consistent with earlier surveys, the 3D sur-vey was to map the subsurface with an arealresolution of 20 m—that is, the survey areawas subdivided into a mosaic of bins, eachmeasuring 20 m by 20 m (see “ConversionGlossary, right). This coverage is obtainedby arranging source rows and receiver lineson a much coarser grid measuring 400 m by400 m. A grid this size was therefore super-posed on the map.
The coarse grid provides the ideal posi-tioning of sources and receivers arrangedalong the vertical and horizontal lines of thegrid respectively. In reality, of course, obsta-cles of all kinds stand in the way. The grid
35July 1992
must be bent and squeezed to get aroundthem, but the new grid must also satisfy theenvironmental and safety concerns ofmunicipalities, private companies, landown-ers and the public. Three months before thesurvey begins in a particular area, NAMcontacts the various state authorities toobtain permission to conduct the survey anddiscuss these matters. Simultaneously, a rep-resentative of the survey contractor, called apermit man, contacts the private compa-nies—there were over 50 major companiesin the refinery area alone.
The last group to be approached, perhapsone month before the survey reaches theirland, is the farmers. To defuse tension thatmay arise when geophysicists deploy equip-ment and explosives on valuable farmland,NAM and the Dutch farmers’ union haveagreed to a third party performing farm per-mitting. Farmers are an important group tomaintain good relations with—NAM surveystouch up to 20,000 farmers every year. A
nActual shooting locations compared with the ideal 400-mby 400-m grid. If an obstacle prevents shooting on the grid,the location is shifted orthogonal to the shooting line. Vibra-tor locations (blue) follow roads, and explosive locations(rpink) have to avoid the middle of cultivated fields. The air-guns (orange) follow the grid best.
36
Vibrator
Explosive
Airgun
Offset shooting
Receiver line
farmer’s working life is constantly beingadjusted to accommodate the weather, andas a result seismic crews have to do likewise.A string of geophones laid across a field oneday may have to be relocated to the edge ofthe field the next because the weatherchanges and potatoes have to get planted.
Throughout the planning and even duringthe survey, all participants from the individ-ual farmer to the major corporation requireconstant liaison. Videos are shown to harborofficials explaining how seismic surveyswork, demonstrating the need for their par-ticipation to ensure contractor boats do nottangle with supertankers; mayors must bereassured that vibrating trucks will leavestreet surfaces as they found them.
Slowly, deviations from the ideal grid areagreed upon by all parties and a realisticsurvey plan takes shape. Generally speak-ing, sources are more of a problem thanreceivers. Unless an obstacle is truly insur-mountable, organizations and landownerstolerate placement of hand-sized geophoneswhere the grid stipulates. Sources on theother hand require relatively large equip-ment and usually cannot be constrained tothe grid (left). An exception is in the water,where airguns carried on pontoons can getto any location once shipping has beenaccounted for. A team of four vibratortrucks, used in urban areas at night, has tostick to roads. And explosives, used in thefields and—yes, in greenhouses—suffer theirown special restrictions.
All the while, surveyors must keep trackof every relocation of a source or receiver.Moving source locations from their idealposition raises the issue of how one ensuresan even coverage in each 20-m by 20-mbin. But this begs the more basic question ofhow sources and receivers arranged on anideal 400-m by 400-m grid provide therequired 20-m by 20-m coverage. Let usanswer the last question first.
Three-dimensional land seismics are usu-ally performed with receivers laid along sev-eral widely spaced parallel lines and shoot-ing points distributed along a line orthogonalto the receivers (next page, top). In the Rot-terdam survey, 480 receiver groups weredistributed along four lines, 120 groups perline, each group comprising several geo-phones or hydrophones connected in seriesto boost signal at the expense of noise.Shots were fired sequentially at ten equallyspaced stations between the two middlereceiver lines.
This configuration provides 4800 seismicrecords, whose reflection midpoints definethe rectangle of coverage. To get the requi-site coverage, that is putting a midpoint in
Oilfield Review
nThe survey teaminserting geo-phones in a streetin the Rotterdamsuburbs.
nStandard 3D land shooting configura-tion. 480 receiver groups are spreadalong 4 lines, 120 per line, with a 40-mspacing between groups. Shooting is per-formed at ten equally spaced locationsalong an orthogonal line between the twoinner lines. This provides rectangular cov-erage of an area equal to one-half thesurface deployment area.
Twelvefold coverage is obtained bystepping the shooting position in thedirection of the receiver lines, providing aswath of coverage comprising overlap-ping rectangles. More overlap is obtainedas the survey moves back creating anadjacent swath.
37
4.8 km
1.2 km
Source
Receiver
nEnsuring coverage when obstaclesrequire relocation of the source. The solu-tion is first to move the source, then tomove the receivers an equal distance butin the opposite direction. In the field,receivers do not have to be moved, butsimply reconnected to simulate movement.
July 1992
each 20-m by 20-m bin, the spacingbetween successive receiver and sourcepositions has to be twice 20 m, or 40 m.This means that the distance betweenreceiver lines, equal to ten source points,has to be 400 m and each receiver line is4.8 km long. At any moment, the surveyteam is therefore deployed over a 4.8-km by1.2-km area, shooting ten positions betweenthe two middle lines (below). The rectangleof coverage measures 2.4 km long by 0.8km wide.
Once the ten shots are complete, the sur-vey then steps either right or left, and theentire process is repeated. This provides asimilarly sized rectangle of coverage over-lapping the previous one and creating aswath of coverage 0.8 km wide. The logis-tics of stepping are relatively simplebecause most of the receivers can remain in
place. Just a few groups must be added tothe advancing end of the four receiver lines,while those on the receding end are pickedup and transported ahead for future use.
When the survey reaches the edge of thesurvey area, the entire top or bottomreceiver line is picked up, and relaid theother side of the remaining three lines.Shooting then recommences in the middleof the new configuration and the surveyprogressively steps in the opposite direction.This provides another swath of overlappingcoverage that overlaps the previous swath.Every point on the ground gets covered byjust two swaths.
NAM stipulated a fold of 12, so everypoint on the ground had to be covered bysix overlapping rectangles in a given swath.Since each rectangle measures 2.4 km long,the stepping distance therefore had to be
2400/6 = 400 m. This explains the coarse400-m by 400-m survey grid. Now for theoriginal question: How does the surveyteam allow for shot point deviations fromthe ideal grid?
The solution follows the principle thatwhen a source is moved left or right from itsproper position, reflection points remainunchanged if receivers are moved an equaldistance in the opposite direction (below).In practice, receivers are not moved at all.The source is moved in 40-m incrementsorthogonal to the shooting line or parallelto the receiver lines. The receivers can thenbe reconnected at the recording truck tosimulate physical relocation by the requi-site number of 40-m steps. Each of thesemanipulations is worked out in advancefrom the detailed planning and performedwithout interrupting the day’s schedule.
nSeismic equip-ment worth $20million, 137 person-nel, 88 vehiclesand over a dozenvessels were neededto conduct NAM’s3D mixed-terrainsurvey. The 6-month operationwas completedwithout accident.
nUndershootingused whenextremely largeobstacles preventshooting anywherenear the requiredgrid position.
38 Oilfield Review
Seismic EquipmentVehicles and VesselsPersonnel
Explosive trucks 2
Light trucks(4WD) 4
Drill rigs 4
Ram hammers 7
Chase boats 2
Recording units 2
Geophones20,400
Field station units1700
Power units316
Spread cables1700
Bay cables 66
Motorship Karin Cat
Pontoon
Sara Maatje
Party chief 1
Seismologists 3
Supervisors 13
Recording team 17
Cable crew 19
Cable support 16
Drilling team 20
Dynamite crews 4
Administration 1
Maintenance 7
Additional helpers 24
Recording trucks 2
Liaison vehicles 6
Light trucks 55
Mobileworkshop
Vibrators 6
Holders 8
Survey vessel Solea
Auxiliary boats 5
A variation of this technique, calledundershooting, solves the problem of beingunable to shoot in extensive areas thatvibrators cannot access or where dynamiteis prohibited (left ). In these cases, sourcepositions are shifted to an adjacent 400-mblock in the direction of the source line andshooting takes place during the previous orfollowing swath when the receiver lines arealso effectively shifted by one 400-m block.Undershooting ensures full coverage in themost complex areas.
39July 1992
nA pontoon carry-ing airguns used inthe shipping chan-nels and shallow-marine areas. Itsspecially designedairguns can oper-ate in water asshallow as 75 cm.
nBay cables—weighted hydro-phone cables—awaiting deploy-ment across a busyshipping channelof Rotterdam’s Euro-port. Bay cablesposed many prob-lems for the surveyteam. Althoughdesigned to lieacross the channelbottom, they weredisplaced by cur-rents and large-ves-sel movement andfrequently cut byanchors. Their posi-tion was determinedusing transponders.The yellow instru-ment on the tripodwas used to fix thelocation of the pon-toon equipped withairguns.
ExecutionThe physical reality of recording swaths of12-fold coverage across harbor, urban envi-ronment and farmland amounts to a platoonof men and equipment, trained for peak effi-ciency (previous page, below). One hun-dred and thirty-seven men headed by aparty chief included seismologists, record-ing teams, shooters, drilling teams, dyna-mite crews, cable support groups, mainte-nance and administration. But the crewcould not be constrained by labels. Every-one worked at several jobs to maintain theflexibility the mixed terrain demanded.
To get around on land, the crew logged400,000 hours driving 2 recording trucks, 2explosive trucks, 59 light trucks, 6 liaisonvehicles, a mobile workshop, 6 vibratorsand 12 small drill rigs. Specifying a parkingspace at PRAKLA’s base for each vehiclerepresents a trivial yet significant example ofthe organization required to keep track ofthe vehicle population and ensure its effi-cient use.
At sea and in the harbor, there was a shal-low-water survey vessel, the S.V. Solea, twovery shallow-draft airgun poppers—the SaraMaatje and a specially constructed pontoonthat could turn on its axis—and a smallflotilla of chase and supply boats. The shal-low-draft poppers can operate in only 75
centimeters of water, eliminating the use ofexplosive, a boon to the local fish popula-tion (above ). Explosives can kill fishbecause the high-frequency content of theexplosion excites natural oscillations in theirair bladders destroying their buoyancy con-trol. Airguns, on the other hand, have beenshown not to harm fish.
To record the survey, the party deployedequipment worth $20 million: 1700 fieldstation units—compact, weatherproof boxesof electronics that digitize the seismic signalfrom a group of receivers and on commandtransmit the data to the recording
truck—316 miscellaneous electronics pack-ages such as power units and repeater units,1700 spread cables used to connect thefield station units to the recording truck, 66bay cables—weighted hydrophone stream-ers for laying across the bottom of water-ways crisscrossing the survey area—and20,400 geophones.
Essential to the survey was simultaneousdeployment of marine and on-land equip-ment—the shooting area of 4.8 km by 1.2km frequently covered both shipping lanesand the bordering land. Activity in the waterhad to be coordinated with the harborauthorities and police, to avoid shipping.Often both airgun poppers were deployedto save the time just one would have takento negotiate the maze of shipping docks.
Bay cables, however, were the worstproblem (left). Easily shifted from their origi-nal locations by tides or the hefty wake of asupertanker, they were also cut by boats set-ting anchor to aid maneuvering in tightspaces. Practically none of the 66 baycables was serviceable at the end of the sur-vey. Because of bay cables being cut andalso because of the generally complex ter-rain, two recording trucks linked to differentsegments of the receiver lines and operatingin synchronization offered a recording ver-satility and speed that would have beenimpossible using the usual one.
Despite the extreme complexity of thesurvey, no accident marred the 174-dayoperation. Both NAM and PRAKLA certainlydid not stint on attention to safety proce-dures. In the refinery area, for example, thesurvey team had to learn and adhere to
40
Drill hammer
Sand
Clay/peat
Make hole Place explosive Insert bentonitetubes
Pull out casing Hole sealed
nMethod of drilling shot holes to avoid water contamination. The hole is created byhammering drillpipe into the earth rather than by conventional drilling with water cir-culation. As soon as the explosive is inserted, the hole is filled with cardboard tubescontaining bentonite clay. After the drillpipe is removed, the tubes fill with groundwater, swelling the bentonite and sealing the hole.
nClad in cleanboots and coats toavoid introducingbacteria andmicrobes into theclean environmentof the greenhouses,three members ofthe PRAKLA surveyteam insert tubesof bentonite into ashot hole.
each refining company’s separate in-housesafety code. Equally, the survey team had tomaintain safety standards of subcontrac-tors— for example, pilots hired to help laybay cables. Safety would always be on theagenda of the daily morning briefing ses-sions when plans for the next 24 hourswould be firmed up.
Environmental concerns were also everpresent, particularly in the farmland andgreenhouses where the crew used explosiveas seismic source. Being reclaimed land,most of this area averages an elevation afew meters below sea level, and shot holesdrilled conventionally for explosive wouldhave flowed water and possibly contami-nated the surrounding land.
The solution was to create holes withoutwater circulation, by pushing or hammeringdrillpipe to a depth between 12 and 20 mand using a disposable aluminum pointattached to the bottom of the drillpipe tofacilitate penetration into the earth (below).The explosive charge was then inserteddown the drillpipe, followed by cardboardtubes filled with bentonite clay. As thedrillpipe was pulled out of the ground, thecardboard tubes remained in the hole grad-ually filling with ground water. This swelledthe bentonite, sealing the hole from theenvironment and as a bonus providing per-fect tamping for detonation.
Further precautions were required to pro-tect the sterile growing environment of the
greenhouses. To avoid tramping in unwantedmicrobes and bacteria, the crew wereobliged to don white surgical coats andclean boots (below).
A critical part of the operation was pro-viding overlapping coverage with GECO’smarine survey. This was carried out in stagesthat covered progressively shallower water(next page, top). First came GECO’s surveyusing deep-sea vessels that could approachthe shoreline to almost 10-m water depth.Then, PRAKLA deployed the S.V. Solea, ashallow-water survey boat with a draft ofonly 1.3 m. Towing a 1800-m streamerequipped with 72 hydrophone groupsspaced every 25 m, the S.V. Solea continuedcoverage to around a 2-m [7-ft] water depthand also in an area of sea near a long break-water that GECO’s larger boats with longerstreamers could not negotiate.
Finally, the shallowest water up to thebeach was surveyed by laying geophonecables along the beach and using the pon-toon to air-pop along orthogonal shootinglines in the water. This provided coverageup to the conventional land survey. Asmuch as possible, the crew limited activityto high tide so the pontoon could get asclose as possible to the beach area.
Coverage and data quality are the keyingredients of a successful survey. The evencoverage of the PRAKLA survey, both at seaand on mixed terrain, attests to thoroughplanning and execution (next page, bot-tom). Thereafter, it is up to data processorsto create order in the millions of records andthrough exquisite manipulations of deconvo-lution, stacking and migration form an accu-rate, focused picture of the subsurface.
nBinning mapsshowing abuttingcoverage in thetwo main surveyedareas—the 15-kmby 25-km mixed-terrain area and ashallow-marinearea in the north-west that links witha deep-marine 3Dsurvey conductedby GECO.
nEnsuring cover-age in the near-shore area whereGECO’s deepmarine surveyabutted PRAKLA’smixed-terrain sur-vey. The transitionzone was surveyedin four stages mov-ing from sea to land.First the deep seasurvey, then a shal-low water surveyusing the shallow-draft S.V. Solea,then very shallowwater coverageusing an airgunpopper in the surfand geophonelines spread on thebeach, and finallythe land survey.
41July 1992
1–8 fold9–1213–20
Above 1212–foldBelow 12
GECO survey
S.V. Solea
Pontoon popping
Land survey10 m
nLocations of refraction surveys conducted to evaluate the weathering layer and per-form the statics correction. Each short line represents a 100-m spread used to make arefraction survey, essentially a small-scale survey designed to estimate the depth andacoustic velocity of the weathering layer only.
ProcessingThis survey posed two special challenges forprocessors: correctly assessing the effect ofthe weathering layer, in a series of processescalled statics correction, and accounting forthe different source signatures of airguns,dynamite and vibrators. It was decided thatdifferences between geophone and hydro-phone responses were too small to have anoticeable effect and were ignored.
The goal of statics correction is to time-shift every trace so the entire survey looks asthough it was made from a depth datum sit-uated in consolidated rock somewhat belowthe weathered layer.4 This removes theeffect of surface topography and travel timein the weathered layer which often variesgreatly on land. Without statics corrections,land seismic sections often have a choppyappearance, with reflectors showing poor orno continuity. Applying the time-shiftsrestores continuity.
The essentially flat topography of the Rot-terdam area poses no problems, but theweathered layer is complex (below). Nearthe shore, at least two layers must be con-sidered, dunes of dry sand with low velocitylying on top of wet sand with high velocity.Farmland often sits on peat that has verylow velocity and varies unpredictably in
42
Sea Dunes Refinery
Rubble Peat
Channel Farmland
nProfile of weathering layer showing difficulties in the near-shore environment withlow-velocity, dry sand dunes, in the refinery area with high-velocity rubble, in the ship-ping channels with no weathering layer at water depths greater than 10 m, and in thefarmland with low-velocity peat varying unpredictably in thickness.
thickness, up to 10 m [33 ft] thick. The ship-ping channels appear to have eroded alltrace of the weathering layer where waterdepth exceeds 10 m. In the refinery areabuilt on reclaimed land using rubble fromWorld War II bombing, the surface layer isfast, but there may be underlying weather-ing material of much lower velocity.
How does the crew go about measuringthe weathering layer’s velocity and thick-ness? Several approaches must be usedtogether. One is the refraction survey, aminiature seismic survey designed to catchonly energy that is reflected from the base ofthe weathering layer or refracted along it.Suitably plotted, this information can yieldthe required velocity and thickness. In thecountryside, a small explosive charge set ina hole 2 to 3 m deep provides the energy,and a geophone array comprising 24 geo-phones spread over about 100 m [328 ft]picks up the signal. The entire operation isconducted by a specialized crew of fourwho manage perhaps four or five surveys aday, about two for every square kilometer ofthe main survey (above). In urban areas,explosive is ruled out so PRAKLA devised asmall truck-conveyed hammer drop as analternative source, essentially a large weight
Oilfield Review
4. For a review of statics correction, seeDobrin MB and Savit CH: Introduction to Geophysi-cal Prospecting, 4th ed. New York, New York, USA:McGraw-Hill Book Company, 1988.Yilmaz O: Seismic Data Processing. Tulsa, Oklahoma,USA: Society of Exploration Geophysicists, 1987.
dropped from a height of 2 m onto a baseplate placed on the ground. In water, refrac-tion surveys are ruled out.
Weathering information is also obtainedat every shot point in the main survey thatuses explosive. Shot holes are usuallydrilled to just below the weathering layer, soplacing a geophone near the top of the holeautomatically measures the travel timethrough the layer. Additional data comefrom picking first breaks on every recordedtrace in the main survey, a Herculean taskthat is simplified by semiautomatic pickingprograms functioning on workstations. Afirst break indicates the first reflecting inter-face, presumed to be the bottom of theweathering layer.
Initially, NAM hoped to interpolateweathering zone thickness and velocityacross the survey area from informationderived from refraction surveys. But this
July 1992
Stack with Field Statics
Stack with Field and Residual Statics
failed to adequately describe the weatheringzone’s extreme variability, and the resultingstatics corrections failed to pull reflectorsinto line. The first-break picking methodwas then employed to fill in detail.
Two steps in the processing chain areactually required to complete the staticscorrection. The first, called field statics, usesall the available data on the weatheringlayer described above to perform an initialnormalization of the traces to a datum. Thisimproves reflector continuity. But becauseof inherent uncertainty about the weather-ing layer, choppiness in reflector continuityoften remains. This must be eliminated in asecond step called residual statics, in whichadjacent traces in a gather are cross-corre-lated and then shifted to improve continuityfor reflectors judged to represent continuousinterfaces in the subsurface. Statics correc-tion is the single most important step in pro-
nStatics correctionon a section from aneighboring surveyin the Rotterdamarea. The field stat-ics that uses refrac-tion survey infor-mation and first-break picks on themain survey pullsreflectors some-what into line, butnevertheless leavesa choppy appear-ance. The secondstep, residual stat-ics, creates aninterpretableimage. Statics cor-rection is a crucialstep in this mixed-terrain area of TheNetherlands.
cessing the data from the Rotterdam area, asthese results demonstrate (below, left).
The second processing challenge wasmatching data recorded using three differenttypes of sources. This is necessary becauseeach source type has a distinct signature. Anexplosive source provides the sharpest pulseof acoustic energy. The airgun’s signaturemay contain small reverberations due to thebubble oscillating as it rises to the surface.In practice, several airguns are used togetherand tuned to eliminate the bubble effect.
Vibrators sweep through the seismic fre-quency spectrum and produce records thatmust be convolved with the sweep to pro-
43
duce a usable trace. The vibrator’s effectivesource signature differs from the other twoin that it is zero-phase, meaning that it issymmetrical about time zero, the effectivetime of shooting. The other two are mini-mum phase—nothing before time zero butthen as much energy as possible immedi-ately afterward (below).
Matching sources amounts to convertingthe minimum-phase traces produced byexplosive and airgun to the zero phase oftraces produced by vibrators, the approachfavored by NAM, or conversely converting
44
Zero phase
Minimum phase
Time0
nZero-phase and minimum-phasewavelets. The vibrator technique resultsin a zero-phase wavelet, while explosiveand airgun produce minimum-phasewavelets that have additional small dif-ferences. During processing, seismic tracesproduced by different sources must bematched to simulate what would haveresulted from using only one type of source.
the vibrator traces to minimum phase. Ineach case, the match has to account for thedetailed differences in source signature,even between dynamite and airgun, andalso the filtering effect of the recordinginstrumentation. If the processing proceedsin minimum phase, the results are generallytransformed to zero phase right at the end.This maximizes the resolution of the seismicimage. Comparing the difference betweenresults obtained from a section surveyed inthe southeast corner of the survey area andprocessed with and without matching
nA section from a neighboring survey in thematching, showing superposed reflectors obthree different sources—dynamite, airgun ansuperpose poorly. After matching, reflectors increased black in the section.
Without Source Matching
With Source Matching
Airgun Vibrator
sources attests to the importance of sourcematching (below).
Achieving successful acquisition and pro-cessing in this difficult terrain did not hap-pen overnight. The Dutch PRAKLA crewprogressively gained experienced in com-plex environments over several years, andNAM has been forthright in its help, adviceand encouragement. In the future, acquisi-tion will ease with the introduction ofadaptable telemetry systems that permit theelimination of cabling where needed, sim-plifying logistics on the ground. —HE
Oilfield Review
Rotterdam area before and after sourcetained from processing data obtained withd vibrators. Before matching, reflectors
superpose well, as evidenced by the
Dynamite
Mick AckersDenis DoremusSugar Land, Texas, USA
Ken NewmanMontrouge, France
nA coiled-tubingunit with diagramsof the injectorhead (left) andblowout preven-ters (center).
An Early Look at Coiled-Tubing Drilling
Interest in drilling slimhole wells with coiled tubing is high. So far, only a few experimental wells have been
drilled and many technological issues remain unresolved. But if these challenges are met, coiled-tubing
drilling could become the medium that finally delivers slimhole wells across the industry.
In this article, FSTS (Formation Selective TreatmentSystem), SideKick and CoilLIFE are marks of DowellSchlumberger. SLIM1 is a mark of Anadrill.For help in preparation of this article, thanks to BruceAdam, Dowell Schlumberger, Rosharon, Texas, USA.
1. For details of coiled-tubing hardware and its applica-tions to workover and logging:Ackert D, Beardsel M, Corrigan M and Newman K:“The Coiled Tubing Revolution,” Oilfield Review 1,no. 3 (October 1989): 4-16.
2. Littleton J: “Coiled Tubing Springs into HorizontalDrilling,” Petroleum Engineer International 2 (Febru-ary 1992): 20-22.
In recent years, workover and logging usingcoiled tubing has become increasinglywidespread (above ).1 During workoveroperations, coiled tubing has been usedsuccessfully to drill out cement plugs andremove scale—in most cases harder to drillthan formation. Now attention is focused oncoiled-tubing drilling as a technique todeliver cost-effective slimhole wells for bothexploration and production.2
Slimhole wells are normally defined ashaving at least 90% of their diameter lessthan 7 in. They are drilled using rotary rigsthat are much smaller than normalrigs—about 20% of their weight, requiringabout a quarter of the drillsite area. Overhalf of drilling costs depend on factorsother than drilling time, such as construct-
45
New wells Re-entry
Disposable exploration wells
Deviated development wells
Well deepening intonew producing zone
Horizontal extensioninto producing zone
Multiple radial drainholes
Straight holes
Lateral holes
Original
New
nPotential applica-tions for coiled-tub-ing drilling.
ing the drill pad and access roads, movingthe rig, and the cost of casing and consum-ables like mud.3 A coiled-tubing unit (CTU)is even smaller than a slimhole rig, is easierto mobilize and requires less equipmentand personnel. Its smaller site requirementleads to lower civil engineering costs. Thesmaller, quieter CTUs have a reduced envi-ronmental impact.
There are also particular benefits offeredby use of continuous tubing. It avoids theneed for connections, speeding up trip timesand increasing safety—many drill floor acci-dents and blowout/stuck-pipe incidentsoccur when drilling is stopped to make aconnection. CTUs have pressure controlequipment designed to allow the tubing tobe safely run in and out of live wells. Thestripper above the blowout preventers(BOPs) seals the annulus during drilling andtripping. This offers increased safety duringdrilling—similar to having a conventionalrig’s annular preventer closed all the time.This safety feature also facilitates underbal-anced drilling, in which drilling is carriedout while the well is flowing.
A range of different uses has been pro-posed for slim holes drilled by a CTU(right). So far, lateral production and verticalre-entry wells have been drilled. Theseexperimental wells were designed to provethat the technique can effectively meetdesign specifications.
Three re-entry horizontal production wellshave been drilled in the Austin chalk, Texas,USA, using 2-in. directionally-controlledcoiled tubing with 37/8-in. bits. In an effortto prove the efficacy of coiled-tubingdrilling for exploration, a vertical well wasdeepened in the Paris basin, France, using11/2-in. coiled tubing with 37/8-in. bits. Thiswas also a re-entry, but a new vertical wellis also planned.
This article reviews one of the Austinchalk wells and the Paris basin well. Then itwill look at the technological challengesarising from these experiences.
46
Lateral Re-Entry for ProductionLast year, Oryx Energy Company re-entereda vertical well in the Pearson field, Texas,USA, completed in Austin chalk. Horizontaldrilling in Austin chalk using mud com-monly encounters almost total lost circula-tion. To reduce mud losses, formation dam-age and costs, water is often used as drillingfluid. This decreases bottomhole hydrostaticpressure to less than formationpressure—underbalanced drilling. To com-bat annular pressure from formation flowduring drilling, conventional rigs use a rotat-ing stripping head or rotating BOPs to sealthe annulus. The wells are killed each timea trip is made.
By using a CTU, which has its annulussealed throughout drilling by the stripper,Oryx was able to run in and out of holewithout killing the well. This improved safety
3. Randolph S, Bosio J and Boyington B: “SlimholeDrilling: The Story So Far...” Oilfield Review 3, no. 3(July 1991): 46-54.
4. Ramos AB, Fahel RA, Chaffin M and Pulis KH: “Hori-zontal Slim-Hole Drilling With Coiled Tubing: AnOperator’s Experience,” paper IADC/SPE 23875, pre-sented at the 1992 IADC/SPE Annual Drilling Confer-ence, New Orleans, Louisiana, USA, February 18-21,1992.Wesson HR: “New Horizontal Drilling TechniquesUsing Coiled Tubing,” paper SPE 23951, presented atthe 1992 Permian Basin Oil and Gas Conference,Midland, Texas, USA, March 18-20, 1992.
and avoided the expense and potential dam-aging effects to the formation of pumpingbrines to kill the well prior to tripping.
To prepare the well, Oryx used a conven-tional service rig to remove the existingcompletion hardware, set a whipstock andsidetrack out of 41/2-in. casing at a true verti-cal depth of 5300 ft [1615 m]. Drilling wasthen continued using 2-in. coiled tubing,downhole mud motors, wireline steeringtools, a mechanical downhole orienting tooland 3 7/8-in. bits. An average buildup rate of15°/100 ft [15°/30 m] was achieved and ahorizontal section drilled for 1458 ft [444m].4 The main bottomhole assembly (BHA)components were:
Drillstring—Oryx employed a reel com-prising 10,050 ft [3060 m] of 2-in. outsidediameter coiled tubing with 5/16-in. mono-conductor cable installed inside the tubing.
Oilfield Review
5. Traonmilin E, Courteille JM, Bergerot JL, Reysset JLand Laffiche J: “First Field Trial of a Coiled Tubing forExploration Drilling,” paper IADC/SPE 23876, pre-sented at the IADC/SPE Annual Drilling Conference,New Orleans, Louisiana, USA, February 18-21,1992.Traonmilin E and Newman K: “Coiled Tubing Used forSlim Hole Re-entry,” Oil & Gas Journal 90 (February17, 1992): 45-51.
6. Ackert et al, reference 1.
Normal drilling60.9%
Waiting/repairs20.5%
Directional7.1%
Fishing11.5%
nFinal analysis of the Oryx well in Austin chalk, Texas, USA. With a steerable bottom-hole assembly, the horizontal section was drilled within its 50-ft vertical window.
5100
5300
5500
5700
5900
61000 400200 600 800 1000 1200 1400 1600 1800
Displacement, ft
True
ver
tical
dep
th, f
t
Coiled tubing and finalwell trajectory
Wireline connector
Downhole orienting sub
WhipstockDirectional survey tool
Check valve
Positive displacementmud motor
Fixed cutter bit
Tubing reel Power supply
Tubing injector
Bent sub
50 ft
Orientation tool—Because coiled tubingcannot be rotated from surface to alterdrilling direction, a downhole method ofchanging tool face orientation is needed. Toachieve this, Oryx deployed a mechanicaltool that converts tubing reciprocation intorotation—compression rotated the tool faceto the right, extension to the left. Onceadjusted, the tool face was locked in placeusing a minimum 250-psi differential pres-sure across the tool.
Directional survey tool—The survey toolinside a nonmagnetic collar relayed direc-tional information to surface via the wireline.
Directional BHA—Two assemblies wereused, depending on the build ratesrequired—a double-bend assembly consist-ing of a conventional 27/8-in. bent housingmud motor coupled to a single bent sub, ora steerable assembly comprising a single-bend motor.
Bit—Thermally stable diamond bits wereused to drill the curve and build sectionsand polycrystalline diamond compact(PDC) bits to drill the lateral section.
Oryx’s motive for drilling this well was toprove that coiled tubing could be used todrill a lateral well in a controlled manner.This was achieved—the final wellbore tra-jectory came within a 50-ft [15-m] verticalwindow along the horizontal section (above).
Because this well was the first of its kind,new techniques had to be developed, andmuch of the drilling equipment had to beadapted from existing conventional hard-
July 1992
ware. Orienting the tool face was not diffi-cult, but maintaining it was hard because ofthe unpredictable reaction of the coiled tub-ing to the torque generated by the drillingmotor’s rotation. Drilling was also slowedby failure of BHA components, particularlythe orienting and directional survey tools.
These difficulties affected the final costanalysis. Total cost was estimated by Oryxat twice that of using a conventionalrig—nondrilling time was responsible fornearly 40% of this (below). However, aspurpose-designed equipment becomesavailable and drilling procedures arerefined, coiled tubing should deliver morecost-effective, slimhole, lateral wells.
nCost breakdown for the first Oryx well.
Vertical Exploration WellLast year, Elf Aquitaine embarked on a seriesof trials to determine whether coiled tubingcould be used to drill slimhole wells, cuttingexploration drilling costs. The goal of thefirst well was to demonstrate that a CTU candrill a vertical well sufficiently fast, cut coresand test formations. Elf envisions initiallydrilling these slimhole wells with a singleopenhole section—avoiding the need forcasing—with the surface casings set usinglow-cost, water well rigs.
This first trial involved the re-entry of wellSaint Firmin 13 in the Paris basin.5 The planwas to use the CTU to set cement plugsacross the existing perforations at 2120 ft[646 m] and then drill a 2105-ft [642-m]vertical section of 37/8-in. diameter. Direc-tional measurements using a coiled-tubing-conveyed survey were to be taken every500 ft [150 m]. Then a 50-ft interval was tobe cored and logged. Finally, a zone was tobe flow tested by measuring pressurebetween two straddle packers.6
The trial was carried out by DowellSchlumberger using a trailer-mounted CTUwith a reel of about 6000 ft [1830 m] of11/2-in. tubing. To avoid the need for costlymodifications, standard surface hardware,like injector head with stripper and BOPstack, were used. A workover rig substruc-ture was installed over the existing wellheadto act as a work platform.
The operation encountered difficulties atthe outset—not with the drilling but with theintegrity of the well’s 30-year-old casing.After cement plugs were set, the well wouldnot hold the 360 psi above hydrostatic pres-sure required to withstand the anticipatedformation pressures. Because of this, drillingdepth was limited to 2955 ft [901 m] whichallowed limestone coring but did not extendto a high-pressure aquifer.
The drilling BHAs employed a high-speed, low-torque motor with PDC bits. Forcoring, a high-torque motor was used. Thedrilling and coring assemblies were made tohang vertically by incorporating heavy drillcollars into the BHA, creating a pendulumassembly. At the start, the deviation at thecasing shoe was 2° and, as expected, theBHA did not build angle—at 2362 ft and2795 ft [720 m and 852 m], the deviationangles were 23/4° and 21/4° respectively.During drilling, the rates were comparableto those drilled by conventional rigs at workin the area. This showed that a CTU candrill vertical wells at commercial rates. Twocores were cut and retrieved with goodrecovery—meeting the second objective ofthe trial.
47
Packer
Formation tobe tested
Packer
FSTS setting tool
nThe FSTS Formation Selective TreatmentSystem. The coiled-tubing conveyed FSTSstraddle packers were set across the for-mation to allow reservoir fluids to flow,proving that drillstem tests can be carriedout using coiled tubing.
Blind rams
Shear rams
Slip rams
Pipe rams
Kill line Choke line
Wellhead, casing or christmas tree
Annular preventer
Injector head
Stripper
Drill floor
Mud returns
BOP stack 41/16 in. 10,000 psi
nBlowout preven-ter configurationfor a well drilledwith a hole sizeless than 4-in.diameter.
Because the program had to be revised toavoid high-pressure zones, no oil-bearingformation could be tested. To prove the test-ing technology and meet the third objective,a drawdown test was carried out on a zonebetween 2221 ft and 2231 ft [677 m and680 m]. The FSTS Formation Selective Treat-ment System was deployed with its twopackers straddling this zone. The formationwas successfully isolated and, if it had beena reservoir, would have produced into thecoiled tubing (above).
48
Looking to the FutureIn addition to proving that coiled tubing canbe used to drill wells, the trial pointed outhow procedures could be changed andwhere future hardware development isrequired. For example, rate of drilling couldbe increased by incorporation of measure-ment-while-drilling tools to make direc-tional surveys, improving surface handlingand weight-on-bit (WOB) control tech-niques and better optimization of the BHA.
To address these issues, Dowell Schlum-berger has assembled a multidisciplinarytask force with Sedco Forex and Anadrill. Itswide-ranging agenda covers equipmentneeds, operational and safety procedures,tubing limits and personnel requirements.
Equipment needs—The Elf job utilized aworkover rig substructure. In the future, apurpose-built substructure will be employed.Standing 10 ft [3 m] off the ground and overthe wellhead, this substructure will act asthe drill floor to make or break the BHA andalso to support the injector head.
The BOPs will be mounted below theinjector head directly on top of the well-head, casing or christmas tree. If the holediameter is less than 4 in. [10 cm], 4 1/16-in.,10,000 psi coiled-tubing BOPs will be used.If the hole is larger, a standard set of 71/16-in., 5000 psi drilling BOPs will be usedinstead. In both cases an annular preventerwill also be incorporated into the stack(below and next page, left).
In the directional wells drilled so far usingcoiled tubing, BHA direction has beenaltered using reciprocation of an orientingtool. This technique has the dual disadvan-tages of interrupting drilling and requiringmanipulation with pressure in the tubing,which has a severe fatiguing effect. The taskforce has therefore designed BHAs thatincorporate an orienting tool controlled byusing mud flow rate.
Directional information can be sent tosurface either using wireline or mud-pulsetelemetry. Wireline offers real-time transmis-sion of high volumes of information. How-
Oilfield Review
Ground
BHA straight hole
BHA for buildup and
horizontal sections
Connector
Check valve assembly
Pressuredisconnect
Orienting tool
Coiled tubing
SLIM1 MWDin nonmagnetic
drill collar
Mud motor
Adjustablebent housing
Mud motor
Drill collars
nBlowout preven-ter configurationfor a well drilledwith a hole sizegreater than 4-in.diameter.
nStraight hole and buildup/horizontalbottomhole assemblies (BHAs). In both,fullbore check valves prevent backflow tosurface. The pressure disconnect suballows recovery of the coiled-tubing stringif the BHA gets stuck. A ball is droppedand pumped through the tubing until itseats in the disconnect sub. Internal tub-ing pressure is then applied that shearspins in the sub and releases the string.
The straight hole BHA includes drill col-lars, so that the string acts as a pendulumand tends to the vertical. Buildup andhorizontal BHAs include an adjustablebent housing to facilitate deviateddrilling. The housing angle is fixed at sur-face before the BHA is run, and its effecton the drilling angle is controlled byusing the orienting tool to rotate the toolface. Progress is monitored by the SLIM1MWD tool, which relays the information tosurface via mud pulses.
A CTU employs a low WOB, 2000 lb[900 kg] compared to 4000 to 6000 lb[1800 to 2700 kg] for conventional slim-hole drilling. So high-speed—700 rpm—low-torque drilling motors are expected tobe most effective. Polycrystalline diamondcompact or thermally stable bits are used.
Injector head
Stripper
Drill floor
Ground
ChokeKill
Mud returns
Annular preventer
Blind rams
Shear rams
BOP Stack 71/16 in. 5000 psi
Wellhead, casingor christmas tree
49July 1992
ever, having wireline in the tubing requiresa high level of maintenance and cuts downpumping options—like acidization treat-ments. To avoid the need for a cable link,the task force has adapted Anadrill’s SLIM1measurement-while-drilling system—whichuses mud pulse telemetry—so that it fitsinside a 31/16-in. diameter nonmagnetic drillcollar (right).
The chemistry of muds used when drillingwith a CTU is not expected to be signifi-cantly different from muds used in conven-tional wells. However, the technique doeshave some special rheological require-ments. In a re-entry well, the coiled tub-ing /casing annulus may be relativelylarge—perhaps 2 in. inside 7 in.—slowingthe annular velocity of the fluid and possi-bly compromising the cuttings-carryingcapacity of the mud. Further, because thefluid is pumped through small-diameter tub-ing, friction must be kept to a minimum byusing low solids muds with low viscositiesand yield points. To mix and treat drilling
nComparison of gas kicks in 5000-ft wellsdrilled using coiled-tubing and conven-tional methods with 3 1/8-in. and 61/2-in.BHAs, respectively. SideKick software wasused to compare the effects of influxesthat gave similar annular heights. In bothcases, the driller’s method was used to cir-culate out the kick, during which casingshoe pressures were about the same.Because of its smaller annular volume,the well being drilled by CTU experi-enced much smaller pit gains.
50 Oilfield Review
7. Ackert et al, reference 1.8. White D and Lowe C: “Advances in Well Control
Training and Practice,” paper, presented at the ThirdAnnual IADC European Well Control Conference,Noordwijkerhout, The Netherlands, June 2-4, 1992.
9. Newman KR: “Coiled-Tubing Pressure and TensionLimits,” paper SPE 23131, presented at the OffshoreEurope Conference, Aberdeen, Scotland, September3-6, 1991.
10. Newman KR and Newburn DA: “Coiled-Tubing-LifeModeling,” paper SPE 22820, presented at the 66thSPE Annual Technical Conference and Exhibition,Dallas, Texas, USA, October 6-9, 1991.Newman K: “Determining the Working Life of aCoiled Tubing String,” Offshore 51 (December1991): 31-32, 34-36.
25
40 80
15
Pit
gain
, bbl
Time, min
Coiled tubingConventional drilling
5
110 130
600
800
1000
Cas
ing
shoe
pre
ssur
e, p
si0
fluid, a trailer-mounted pumping and treat-ment unit has been constructed.
In a vertical hole, the setdown weightread at surface is equivalent to the WOB.However at high angles, the tubing com-presses inside the wellbore. If too muchweight is set down, the tubing may lockagainst the walls of the well, failing to trans-fer any further weight to the bit.7
Experience from the Paris basin wellshowed that while manual control of set-down weight was possible, it was tediousand required absolute concentration fromthe operator. To improve drilling efficiency,the CTU has been fitted with a system thatautomatically maintains a setdown weight.With this autodrilling system, the operatorcan monitor progress without having tomake continual minute weight changes.
The task force is reviewing three otherareas of equipment development underreview. All involve handling tubulars:removing the existing production tubing inre-entry wells, deploying the BHA into alive well, and running casing.
Operational and safety procedures—Procedures for controlling a slimhole wellwhen drilling with a CTU differ from thoseneeded when drilling with a conventionalslimhole rig. At the heart of this is the differ-ence in annuli. Conventional slimholewells have a narrow annulus and the mudtraveling up it creates a back pressure,called the equivalent circulating density(ECD). The ECD increases with pump rateand raises bottomhole hydrostatic pressure.This provides the option of dynamickill—increasing the rate to increase thepressure—but also a potential disadvantageof losing mud due to ECD exceeding theformation fracture gradient.
When drilled with a CTU, the annulus islarger—particularly in re-entry wells—ECDis not a factor and dynamic kill cannot beapplied. To evaluate other well-kill strate-gies, the task force used SideKick softwareto model gas influxes in a full size wellbeing drilled conventionally and a slimholewell being drilled by a CTU.8
The SideKick model was used to assessthe significance of the volume of influx.First, it modeled influxes that gave compara-ble heights of gas in conventional andcoiled-tubing annuli (about 7.5 and 3 bar-rels, respectively). The shut-in casing pres-sure (SICP) at surface and the casing shoepressure (CSP) were broadly similar in both
wells (left). But in modeling an influx of 7.5barrels in the slim and conventional annuli,the SICP and CSP in the coiled-tubing wellwere much higher—double or more.
Therefore, early detection of gas influxesduring coiled-tubing drilling is vital. TheCTU’s stripper seals the annulus and ensuresthat the mud return line is full, improvingthe reliability of delta flow measurements—the difference between mud flow rate in andout of the well. Delta flow is measurabledown to 10 gal/min [0.8 liter/sec], permit-ting rapid detection of kicks after allowingfor the volume increase due to cuttings. Inthe mud pits, resolution of conventionallevel sensors is improved by having mudtanks with a smaller base area than is normal.
All drilling operations are subject to safetyregulations limiting operational equipmentto zones—in Europe, Zone I allows only themost stringent explosion-proof equipment,Zone II the next most, and so on. Ironically,the compactness of a CTU complicatescompliance with these regulations.
In the Paris basin well, the Zone II classifi-cation was specially reduced by the authori-ties from a 100-ft to a 50-ft radius from thewellhead. If the radius had been any larger,it would have extended the zone’s require-ments to the cars on the edge of the lease(next page). Changes in local regulationsand in equipment classification may berequired in the future.
Tubing limits—Coiled tubing had a slowstart as a workover service because of unre-liability and propensity for unpredicted fail-ure. To combat this, Dowell Schlumbergerhas developed a better understanding of thefactors governing tubing fatigue; this is nowbeing applied to drilling operations.9
Repeated use of coiled tubing has threetypes of limitation:
nLayout of Elf’scoiled-tubingdrilling site in theParis basin,France.
51July 1992
Portakabin Portakabin
Access road
CT unitGen
erat
or
Cat
frac
unit
back
up p
ump
Bin Bin Bin
Substructure
Crane truck
Mud productsstorage area
Tool rack
50 ft safety perimeter
Water tank
Mud treatment/pumping unit
Fuel tank
Chokemanifold
• Pressure and tension limits—the burst andcollapse pressures and the maximum ten-sion and compression at various pres-sures. These are analogous to the limitsexperienced by drillpipe and can be cal-culated through tests and carefullyavoided during operations.
• Diameter and ovality limits—the degreeto which the pipe is collapsed, balloonedor mechanically damaged. This also hasan analogy in drillpipe where damagedpipe and couplings have to be detected.With coiled tubing, the physical shape ofthe tubing can be continuously monitoredduring the job to detect damage.
• Life limits—primarily due to bending inthe pipe at the gooseneck and on the reelas it is spooled on and off, often with thetubing pressured. Anticipating life limits oftubing has proved difficult, but is vital toavoid catastrophic failure. At its crudest,the fatigue of a reel of tubing can beequated to the number of times it is runinto and out of the well—termed cycles.
After extensive research, Dowell Schlumber-ger has developed a way of assessing coiled-
tubing fatigue that is more sophisticated thansimply counting cycles—the CoilLIFEmodel.10 During jobs, all tubing movementand pressures are monitored and recorded.The CoilLIFE software then calculates theamount of life remaining in the string. Ittakes into account the relative severity ofeach cycle, the nature of the fluids that havebeen pumped and the sequence in whichthe cycle occurred—which affects the accu-mulated damage.
Personnel requirements—The number ofpersonnel required for coiled-tubing drillingis likely to be about 50% of that needed forconventional operations. Not only are day-to-day operational requirements lower, butthe number of service personnel can also bereduced. For example, when running cas-ing, the mud system could be employed tomix and pump cement—eliminating theneed for a cementing engineer. All thedrilling information, along with basic mudlogging data and general surface data, willbe centralized in a computerized informa-tion system, eliminating the need for a full-time mud logger. —CF
52
Taking Advantage of Shear Waves
F O C U S
The introduction of the DSI Dipole Shear Sonic
Imager tool, which measures shear waves in
all types of formations, has stimulated
development of new applications for measurements of
shear wave speed, Vs, and its reciprocal slowness.
Slowness measurements—shear, compressional and
Stoneley—can help improve identification of rock
lithology and fluid content, and assist borehole and
surface seismic interpretation. In addition, shear wave
slowness provides the critical link in calculating rock
shear stiffness and bulk compressibility, two quantities
invaluable for planning fracturing operations and con-
trolling production to avoid sanding. A new log of the
pore-fluid bulk modulus K f, based on DSI measure-
ments, shows promise as an indicator of hydrocarbons
in sandstone formations.
Before the development of shear wave logging tools,
shear waves created by traditional tools using
monopole sources were measurable only in fast, or
hard, formations. Consequently, geophysicists based
their interpretation schemes largely on measurements
of the more accessible compressional speed, Vp. To
begin extending interpretation of both Vp and Vs mea-
surements, researchers at Schlumberger-Doll
Research in Ridgefield, Connecticut, USA drew from
three sources: Biot-Gassmann theory, laboratory mea-
surements of gas-saturated, quartz sandstones and
contact theory.
The theoretical groundwork for the K f log lies in the
Biot-Gassmann equations, which describe how com-
pressional and shear waves travel through a fluid-filled
porous medium such as oil-saturated rock:
and
In these equations, the composite density ρc is made
up of density contributions from water, hydrocarbon
and rock. N is the rock frame shear modulus—shear
modulus describes how a body deforms under a shear
stress. Kb and Kp are the bulk moduli of the rock frame
and pore space, respectively—bulk modulus measures
a body’s resistance to a change in volume under pres-
sure. The Biot-Gassmann theory also shows that Kp
and Kb are related via pore-fluid bulk modulus K f , the
bulk modulus of the rock grains Km , and porosity f :
where
Readily available shear wave measurements from the
DSI tool combined with the Biot-Gassmann relations
gave researchers a handle on N, but they still needed
to know more about Km, Kb and Kp for evaluating K f.
Laboratory data—ultrasonic measurements of shear
and compressional velocities plus porosity measure-
ments on gas saturated, pure quartz sandstones—
yielded the necessary information on these elastic
properties. Plots of Kb and N, calculated from the data,
α = 1 - K b
K m .
Kp = α 2
α - φKm
+ φKf
,
ρc Vs2 = N .
ρc Vp2 = Kp + Kb + 4N
3
Oilfield Review
Kb and N as a function of porosityfor pure quartz sand-stones.
Porosity, p.u.0 0.1 0.2 0.3 0.4
0
10
She
ar M
odul
us N
, GP
aB
ulk
Mod
ulus
Kb,
GP
a
20
30
40
50
0
10
20
30
40
50
Kb
/N
Porosity, p.u.
0
1
2
0 0.1 0.2 0.3 0.4
Laboratory data show that Kb/N is aconstant 0.9, inde-pendent of porosity.
versus porosity reveal a nearly linear porosity depen-
dence (top). In addition, the data establish that the
ratio Kb /N is a constant 0.9, independent of porosity
(middle). Finally, the lab data show that Km for quartz
equals 36 Gigapascals (GPa).
The laboratory findings on Kb and N were consistent
with Hertz-Mindlin contact theory, which models
unconsolidated rock as a random packing of spherical,
elastic grains. According to contact theory, any force
applied to the rock is transmitted at the grain contacts,
where one grain touches another. The grain contacts
consequently govern the rock’s bulk and shear moduli,
and hence its compressional and shear velocities.
Each grain contact can be thought of as two coil springs
that move in directions tangential and normal to the
contact surface. The stiffnesses of these contacts
July 1992
depend in part on the porosity of the rock because the
grain contact area, and thus the force transmitted
increases as porosity decreases. Mathematical expres-
sions describing the theory confirm experimental
results that Kb and N are nearly linear functions of
porosity and that the ratio of moduli for the dry rock,
Kb/N, is independent of porosity and equals 0.9 for
quartz sandstones.
With all parameters in the Biot-Gassmann equations
either known or derived from log measurements, inter-
pretation in shaly sands proves fairly straightforward.
First, the shear frame modulus N, calculated from
measurements of Vs (measured N), is compared to N
calculated from the porosity-dependent equation based
on lab data (predicted N). Porosity for the latter equa-
tion comes from independent measurements, such as
the neutron-density log. If the measured N matches the
predicted N, the formation is interpreted as a clean
sand; if it doesn’t match, the formation is considered a
shale. Next, setting Kb = 0.9N in the Biot-Gassmann
equations allows one to first solve for Kp, then for K f .
A K f log from the Gulf of Mexico shows the correla-
tion between K f value and the presence of hydrocar-
bons (next page). A K f value of 0 corresponds to gas,
and water has a higher K f value than oil. The log indi-
cates gas from 2343 to 2350 meters; oil from 2350 to
2365 meters and from 2368 to 2373 meters; and water
throughout other intervals.
Side-by-side comparisons of logs from the Gulf of
Mexico well demonstrate the consistency of the K f log
with neutron-density, resistivity and gamma ray logs.
Shales were detected at the depths marked using two
techniques. First, the measured (Vp/Vs)2 is greater than
the predicted value. Second, the measured N is signifi-
cantly below the predicted N for water-saturated sands.
Tracks seven through nine compare the measured Vp
with theoretical predictions assuming the formation is
filled with water, oil and gas, respectively. The match
between measured and predicted Vp for gas reveals a
partial gas saturation from 2344 to 2350 meters. This
agrees with the K f log, which has values of less than 1
at the same depth intervals. According to the veloci-
53
The introduction of the DSI Dipole Shear Sonic
Imager tool, which measures shear waves in
all types of formations, has stimulated
development of new applications for measurements of
shear wave speed, Vs, and its reciprocal slowness.
Slowness measurements—shear, compressional and
Stoneley—can help improve identification of rock
lithology and fluid content, and assist borehole and
surface seismic interpretation. In addition, shear wave
slowness provides the critical link in calculating rock
shear stiffness and bulk compressibility, two quantities
invaluable for planning fracturing operations and con-
trolling production to avoid sanding. A new log of the
pore-fluid bulk modulus K f, based on DSI measure-
ments, shows promise as an indicator of hydrocarbons
in sandstone formations.
Before the development of shear wave logging tools,
shear waves created by traditional tools using
monopole sources were measurable only in fast, or
hard, formations. Consequently, geophysicists based
their interpretation schemes largely on measurements
of the more accessible compressional speed, Vp. To
begin extending interpretation of both Vp and Vs mea-
surements, researchers at Schlumberger-Doll
Research in Ridgefield, Connecticut, USA drew from
three sources: Biot-Gassmann theory, laboratory mea-
surements of gas-saturated, quartz sandstones and
contact theory.
The theoretical groundwork for the K f log lies in the
Biot-Gassmann equations, which describe how com-
52 Oilfield Review
Taking Advantage of Shear Waves
pressional and shear waves travel through a fluid-filled
porous medium such as oil-saturated rock:
and
In these equations, the composite density ρc is made
up of density contributions from water, hydrocarbon
and rock. N is the rock frame shear modulus—shear
modulus describes how a body deforms under a shear
stress. Kb and Kp are the bulk moduli of the rock frame
and pore space, respectively—bulk modulus measures
a body’s resistance to a change in volume under pres-
sure. The Biot-Gassmann theory also shows that Kp
and Kb are related via pore-fluid bulk modulus K f , the
bulk modulus of the rock grains Km , and porosity f :
where
Readily available shear wave measurements from the
DSI tool combined with the Biot-Gassmann relations
gave researchers a handle on N, but they still needed
to know more about Km, Kb and Kp for evaluating K f.
Laboratory data—ultrasonic measurements of shear
and compressional velocities plus porosity measure-
ments on gas saturated, pure quartz sandstones—
yielded the necessary information on these elastic
properties. Plots of Kb and N, calculated from the data,
α = 1 - K b
K m .
Kp = α 2
α - φKm
+ φKf
,
ρc Vs2 = N .
ρc Vp2 = Kp + Kb + 4N
3
F O C U S
depend in part on the porosity of the rock because the
grain contact area, and thus the force transmitted
increases as porosity decreases. Mathematical expres-
sions describing the theory confirm experimental
results that Kb and N are nearly linear functions of
porosity and that the ratio of moduli for the dry rock,
Kb/N, is independent of porosity and equals 0.9 for
quartz sandstones.
With all parameters in the Biot-Gassmann equations
either known or derived from log measurements, inter-
pretation in shaly sands proves fairly straightforward.
First, the shear frame modulus N, calculated from
measurements of Vs (measured N), is compared to N
calculated from the porosity-dependent equation based
on lab data (predicted N). Porosity for the latter equa-
tion comes from independent measurements, such as
the neutron-density log. If the measured N matches the
predicted N, the formation is interpreted as a clean
sand; if it doesn’t match, the formation is considered a
shale. Next, setting Kb = 0.9N in the Biot-Gassmann
equations allows one to first solve for Kp, then for K f .
A K f log from the Gulf of Mexico shows the correla-
tion between K f value and the presence of hydrocar-
bons (next page). A K f value of 0 corresponds to gas,
and water has a higher K f value than oil. The log indi-
cates gas from 2343 to 2350 meters; oil from 2350 to
2365 meters and from 2368 to 2373 meters; and water
throughout other intervals.
Side-by-side comparisons of logs from the Gulf of
Mexico well demonstrate the consistency of the K f log
with neutron-density, resistivity and gamma ray logs.
Shales were detected at the depths marked using two
techniques. First, the measured (Vp/Vs)2 is greater than
the predicted value. Second, the measured N is signifi-
cantly below the predicted N for water-saturated sands.
Tracks seven through nine compare the measured Vp
with theoretical predictions assuming the formation is
filled with water, oil and gas, respectively. The match
between measured and predicted Vp for gas reveals a
partial gas saturation from 2344 to 2350 meters. This
agrees with the K f log, which has values of less than 1
at the same depth intervals. According to the veloci-
53July 1992
versus porosity reveal a nearly linear porosity depen-
dence (top). In addition, the data establish that the
ratio Kb /N is a constant 0.9, independent of porosity
(middle). Finally, the lab data show that Km for quartz
equals 36 Gigapascals (GPa).
The laboratory findings on Kb and N were consistent
with Hertz-Mindlin contact theory, which models
unconsolidated rock as a random packing of spherical,
elastic grains. According to contact theory, any force
applied to the rock is transmitted at the grain contacts,
where one grain touches another. The grain contacts
consequently govern the rock’s bulk and shear moduli,
and hence its compressional and shear velocities.
Each grain contact can be thought of as two coil springs
that move in directions tangential and normal to the
contact surface. The stiffnesses of these contacts
Kb and N as a function of porosityfor pure quartz sand-stones.
Porosity, p.u.0 0.1 0.2 0.3 0.4
0
10
She
ar M
odul
us N
, GP
aB
ulk
Mod
ulus
Kb,
GP
a
20
30
40
50
0
10
20
30
40
50
Kb
/N
Porosity, p.u.
0
1
2
0 0.1 0.2 0.3 0.4
Laboratory data show that Kb/N is aconstant 0.9, inde-pendent of porosity.
AVO, the change in seismic reflection amplitude as a
function of incident angle or offset, is used to distin-
guish hydrocarbon prospects from lithologic effects.
The K f log will allow researchers to isolate the effects
of formation fluid on the seismic response while logs
of the frame moduli, Kb and N, will isolate the effects
caused by lithology. The next chapter in K f log research
involves characterizing K f behavior in other formations.
—TAL
Acknowledgements and Further Reading
For help in preparation of this article, thanks to Andy Reischer, LarrySchwartz and Bill Murphy, Schlumberger-Doll Research, Ridgefield,Connecticut, USA. In this focus, the DSI (Dipole Sonic Imager) tool is a mark of Schlumberger.Murphy WF, Reischer A and Hsu K: “Modulus decomposition of com-pressional and shear velocities in sand bodies,” Geophysics (in press).Hornby BE, Murphy WF III, Liu H-L and Hsu K: “Reservoir sonics: ANorth Sea case study,” Geophysics 57 (January 1992): 146-160.
ties, oil occurs in the sands from 2350 to 2352 meters
and 2354 to 2368 meters. The K f log also indicates oil.
Finally, agreement between measured and predicted
velocities shows water in the sands from 2372 to 2379
meters. The K f log shows salt water at those depths. In
all cases, these findings mirror an interpretation of the
gamma ray, neutron-density and resistivity logs.
The new information offered by the K f log has
spawned several potential applications. The K f log
may help identify low resistivity pay in unconsolidated
sands where resistivity contrast is small. Because the
acoustic signals and responses of the DSI tool are rela-
tively unaffected by steel, the log may identify hydro-
carbon type behind casing. Most importantly, the K f
log represents a key component in integrating wellbore
DSI interpretation with near-well seismic measure-
ments such as amplitude variation with offset (AVO).
54 Oilfield Review
Consistency of inter-pretation from con-ventional and soniclogs taken in a Gulf ofMexico well.
Dep
th, m
GammaRay
Caliper
DensityPorosity
NeutronPorosity
DeepInduction
MediumInduction
1 20,0000 600 18
0 120
in
GAPI
p.u.
Ø
ohm-m
0 60
100 200
ShearSlowness
µs/ft
Comp.Slowness
0 10
TheoryGas
TheoryWater
(Vp/Vs)2
1 30
GPa
Measured
Predicted
N
0 4
Vp
Water
km/sec
0 4
Vpkm/sec
Oil
0 4
Vpkm/sec
Gas
0 10
Kp
Kf
GPa
0 10
KfGPa
0 10
2340
2350
2360
2370
2380
2390
Shales
Gas
Oil
Wat
er
Predicted
Measured
Predicted
Measured
Predicted
Measured
Acknowledgements and Further Reading
For help in preparation of this article, thanks to Andy Reischer, LarrySchwartz and Bill Murphy, Schlumberger-Doll Research, Ridgefield,Connecticut, USA. In this focus, the DSI (Dipole Sonic Imager) tool is a mark of Schlumberger.Murphy WF, Reischer A and Hsu K: “Modulus decomposition of com-pressional and shear velocities in sand bodies,” Geophysics (in press).Hornby BE, Murphy WF III, Liu H-L and Hsu K: “Reservoir sonics: ANorth Sea case study,” Geophysics 57 (January 1992): 146-160.
Consistency of inter-pretation from con-ventional and soniclogs taken in a Gulf ofMexico well.
Dep
th, m
GammaRay
Caliper
DensityPorosity
NeutronPorosity
DeepInduction
MediumInduction
1 20,0000 600 18
0 120
in
GAPI
p.u.
Ø
ohm-m
0 60
100 200
ShearSlowness
µs/ft
Comp.Slowness
0 10
TheoryGas
TheoryWater
(Vp/Vs)2
1 30
GPa
Measured
Predicted
N
0 4
Vp
Water
km/sec
0 4
Vpkm/sec
Oil
0 4
Vpkm/sec
Gas
0 10
Kp
Kf
GPa
0 10
KfGPa
0 10
2340
2350
2360
2370
2380
2390
Shales
Gas
Oil
Wat
er
Predicted
Measured
Predicted
Measured
Predicted
Measured
ties, oil occurs in the sands from 2350 to 2352 meters
and 2354 to 2368 meters. The K f log also indicates oil.
Finally, agreement between measured and predicted
velocities shows water in the sands from 2372 to 2379
meters. The K f log shows salt water at those depths. In
all cases, these findings mirror an interpretation of the
gamma ray, neutron-density and resistivity logs.
The new information offered by the K f log has
spawned several potential applications. The K f log
may help identify low resistivity pay in unconsolidated
sands where resistivity contrast is small. Because the
acoustic signals and responses of the DSI tool are rela-
tively unaffected by steel, the log may identify hydro-
carbon type behind casing. Most importantly, the K f
log represents a key component in integrating wellbore
DSI interpretation with near-well seismic measure-
ments such as amplitude variation with offset (AVO).
54
AVO, the change in seismic reflection amplitude as a
function of incident angle or offset, is used to distin-
guish hydrocarbon prospects from lithologic effects.
The K f log will allow researchers to isolate the effects
of formation fluid on the seismic response while logs
of the frame moduli, Kb and N, will isolate the effects
caused by lithology. The next chapter in K f log research
involves characterizing K f behavior in other formations.
—TAL
Oilfield Review