Confidence Limits Associated With Values

Susan Macmillan, British Geological Survey, and Steve Grindrod, SPE, Copsegrove Developments

of the Earth’s Magnetic Field Used for

Summary
This paper describes updated uncertainties for use with predicted
Introduction
geomagnetic
The Earth’s magnetic parameters fieldwithin
is a vectormagnetic measurement-while-
quantity, dependent on contributions from three sources: the field generated in the Earth’s
drilling
Directional Drilling
position and time, and may be expressed as themodels
(MWD) survey-tool-error models.
to define positional-error ellipsoids along the wellbore, which Background
assist in hitting geological targets and avoiding collisions with field
These vector are
sumused
of the
core, the crustal
fromBGS
In 1993, electrical
field from
and Error
investigated
local rocks, and a combined external
Distributions
currents the flowing in the in
uncertainty upper
BGGM atmosphere
geomagnetic-
and June 2010 SPE in Drilling & Completion
existing wellbores. field magnetosphere.
values (Macmillan Global
et al.models
1993). usedHowever, directional drilling,
the estimates of the
Copyright © 2010 Society of Petroleum Engineers
The declination, dip angle, and total field strength of the Earth’s for example the BGGM, generally represent a combination
crustal- and external-field contributions in the overall uncertainties of the
magnetic
This fieldwas
paper (SPE 119851) are used
accepted with magnetic-survey
for presentation at the SPE/IADC Drilling tools for surveying main sourcestoward
were biased (i.e., internal
the North field, crustal
Sea. field, and
In addition, thetheBGGMundisturbed
has
Conference and Exhibition, Amsterdam, 17–19 March 2009, and revised for publication.
the wellbore. These values are often obtained from mathematical external
been field arising
updated from ever-present
continuously since 1993 magnetospheric
to take account ofcurrents).
new data,
Original
230 manuscript received for review 27 January 2008. Revised paper received for
models
review 19 Junesuch as peer
2009. Paper the British
approved Geological
12 November 2009. The values computed for the Earth’s
Survey (BGS) global-geo- magnetic-survey-satellite missions, and improved field-modelingmagnetic field at a given loca-
magnetic model (BGGM). As the Earth’s magnetic field is continu- tion and time from the BGGM will differ from
methodology. The 1993 study formed the basis of the geomagnetic the actual values
ally varying with time, the BGGM is updated annually to maintain that
errorwould
termsbe measured
included in thebecause
ISCWSA ofMWD-error
a combination of model
model. The ISC-errors,
accuracy. However, a global predictive model cannot capture allcrustal fields, and
WSA MWD-error external
model fields. It to
is designed is the magnitude
quantify of uncertainty
wellbore-position
sources of the Earth’s magnetic field, which results in uncertainties that we investigate
uncertainty, here. in use today [see ISCWSA (2009) for
and it remains
of the predicted parameters. The Industry Steering Committee on Directional-drilling
details of revisions]. While and survey-management
other error terms in companies the model (therehave a
Wellbore Surveying Accuracy (ISCWSA) published an MWD- requirement
are approximately for accurate
136) have geomagnetic-field
been updated, estimates and knowl-
the geomagnetic error
error model in 2000 (Williamson 2000). The geomagnetic-field edge termsof the associated
remain those thatuncertainties
were derivedwhen magnetic-survey
by Williamson (2000) tools
from
uncertainties that are part of this model were derived from workare the used
originalin the
study surveying
by Macmillanof wellbores. Magnetic-survey
et al. (1993). Another reason tools for
conducted by the BGS in the early 1990s. Since then, more-accu- measure
the update theofdirection of the wellbore
the geomagnetic relative
error terms to the
is that direction
there of
are signifi-
rate data from magnetic-survey satellites have been introducedthe intolocal geomagnetic field. In addition,
cant differences between the terms from the MWD-error model the magnetic dip angle
the BGGM, and the uncertainty of the predicted geomagnetic-field and
and total
thoseintensity
from theof the geomagnetic
original study. Table field arethe
1 lists required for use
geomagnetic
parameters has been reduced. in algorithms
error terms inthat reduce the
the ISCWSA error error caused by the magnetic field
model.
The original approach to deriving the uncertainties involved associated with the
For a Gaussian ordrillstring. The uncertainties
normal distribution of errors,are theused to com-
1-standard-
separating the various error sources in the magnetic field and pute positional-error
deviation (1 ) error limitellipsoids along the
is equivalent wellbore
to being 68.3%to help hit the
confident
assessing them individually. This paper uses a simpler approachincreasingly that the estimatedsmall geological
value is within targets
that and miss other
amount of thewellbores.
true value.
where clean orientated magnetic downhole data are simulated using Additionally, for a Gaussian distribution, 2 is equivalent to being
geomagnetic-observatory data. Spot absolute measurements of95.4% the confident and 3 is equivalent to being 99.7% confident.
magnetic field made at observatories around the world are adjusted Although Williamson (2000) states that no restrictive assumptions
for the crustal magnetic field to make them more representativeare made about the statistical distribution of measurement errors,
of hydrocarbon geology. The adjusted observatory data are then all of the input error magnitudes are quoted as 1 values. To
compared with the predicted values from the BGGM to assess the obtain 95.4 and 99.7% confidence limits for the output values,
uncertainty. The uncertainties do not fit a normal distribution, so the computed final error estimates are multiplied by 2 and 3,
they are expressed as limits for various confidence levels. They respectively. vary It is, unfortunately, the case that for any other distri-
with location and, in their derivation, do not assume any underly- bution of errors, 2 or 3 times 1 is not equivalent to being 95.4
ing empirical error distribution. While they also vary with time, we or 99.7% confident. This is particularly true for geomagnetic data,
provide time-averaged look-up tables that should be valid for asand to obtain the 95.4 and 99.7% confidence limits, it is neces-
long as there are good-quality satellite data on which to base global sary to actually calculate them. The preferred confidence level in
magnetic-field models. Options to reduce the uncertainties further
using data from local magnetic surveys [in-field referencing (IFR)]
and observatories (interpolation IFR) are also described.
The use of the revised geomagnetic uncertainty values in the
MWD-error model will reduce wellbore-position uncertainty to
reflect the increased accuracy from recent improvements in geo-
magnetic modeling. This is demonstrated using results for the
ellipsoids of uncertainty output by an MWD error model for three
standard ISCWSA well profiles.
 TABLE 1—THE GEOMAGNETIC ERROR TERMS IN THE
ISCWSA ERROR MODEL*
Error Source
Declination (constant)
Declination (BH dependent)
Magnetic dip angle
Total field intensity
Code
AZ(G)
DBH(G)
MFD(G)
MFI(G)
Magnitude (1 )
0.36°
5000°nT**
0.20°
130 nT
* The code identifies the error term in the error model.
** An example of how the B H-dependent error works is at a site where BH =
the15,000
oil industry for well-planning
nT. The declination error is then (AZ2 +purposes
(DBH/15000)2is
) = that
0.49° it is equivalent
to 2 (i.e.,
Data and 95.4%),
Analysisand it is this level that is generally presented
Methodology
in
In this
this paper.
study, However,
we use because different companies
geomagnetic-observatory data use
to different
estimate
General
confidence Approach.
levels, For
the each BGGM,
detailed resultsthere
for are
the two
68.3, stages
90, in 95.4,
95,
BGGM
deriving uncertainty:
its confidence Thislimits.
is the First
next-best
of all, source
we of datathe
calculate short of
minute-
99,
Dataand
clean 99.7%
Sources.
magnetic levels are available
Geomagnetic-observatory
data measured online
within from
minute
a wellbore. BGS
means
We (BGS 2009)
are
compute thethemain
by-minute
or Copsegrove
data used differences
in this study; between
Developments Fig. 1a the
(ISCWSA
shows observations
2009).
the observatoryand the BGGM
locations. at
This
difference
Determining
each between
available the model
Crustal-Field
observatory predictions
andContribution
use theseand observatory
to the Uncertain-
differences data that
to1995–2007,
estimate
means
were
ties. not
The that
used in
global any
in given
the
survey BGGM
construction
data set yearof in
a
provided the
givena time
BGGM
total period
of model.
more than
the limits.
June 2010
there Second,
SPE Drilling40we adjust
& Completion the minute-by-minute differences to
We are
9,000
account definebetween
a “BGGM
observations.
for the crustal
and
The year” 50as
vector
field
million
the yeardatastarting
observations
at each
from
observatory. werea soon
maximum after
compared
This is
ofto134
the
necessary the
observatories
release
BGGM of a
values; available
BGGM
the in May for comparison
of
differencesare each year.
between with
For
the a given
example,
observed BGGM
the model.
2006
valuesand Fig. 1—Locations of (a) observatories, (b) repeat stations, and
andthe
because
The some
resolution observatories
of from
the data is 0.1 sited on
arc-minutes volcanic islands,
(approximately 0.002°)
(c) oil and gas fields with local magnetic data.
BGGM
BGGM2008 year values
crustal-field runs were
contributions 2006.5
assumedto 2007.5.
to their to beare
data The
dominated
notmajority by ofthe
representativedates
crustal
offield
the because the time-varying external field is removed in the
for declination
entered for (D) and magnetic
magnetic-field dip
computations angle (I),
using and
a 1 nTrelease
given for total of
magnetic fields experienced at oil fields. measurement and reduction process. Because the probability-dis- 231
field
theWe intensity
BGGM are (F). Measurement
expected to be withinaccuracy is closeBy
this period. tolooking
the resolution
at tribution plots of these differences had unexpected bumps toward
but doesinvestigate
vary from theonetemporal
observatory dependence
to another, of the confidence
depending on instru-
each revision
limits based on ofalltheofBGGM separately,
the differences in we
each can see how
BGGM yeartheand overall
more-extreme values, it was suspected that there were a number
ment stability
performance andchanged
has absolute baseline
over control.
the years.
theTwo
spatial dependence by computing the limits at each site of poor-quality observations. These had to be rejected to avoid
using
sources of data were used to estimate the crustal field:
recent BGGM years when data from modern magnetic-survey biasing the final results. The of the absolute value of the differ-
First, global repeat-station and magnetic-survey data from 1985 and
satellites are being incorporated ences was calculated for each field component, and any differences
onward. These are all land based.into the models.
Second, we used all local data
greater than 3 were rejected as outliers. This procedure resulted
sets gathered in the vicinity of oil fields for the purposes of IFR.
in approximately 31% of D, 17% of I, and 20% of F observations
These are dominated by aeromagnetic surveys and marine magnetic
being rejected before the confidence limits were calculated. The
surveys. The coverage provided is shown in Figs. 1b and 1c.
confidence limits caused by the crustal field derived from the
global survey data are shown in Table 2.
The IFR estimates provided spot estimates of the crustal field at
more than 200 sites, mostly offshore. Each site represents a distinct
oil field. We transform the local scalar data to obtain vector data
(Williamson et al. 1998), and we also downward continue them
to the maximum expected drilling depth. These data are also con-
sidered to be free of contamination by time-varying external fields
and along with the standard practice of employing a base station
nearby to correct for these signals, network leveling is widely used
to remove any remaining contamination. The set of IFR estimates
was not winnowed in any way before the limits were calculated.
The resultant limits are shown in Table 3.
The limits of the crustal magnetic field from these two sources
of data are combined by averaging and are then used to adjust the
minute-by-minute differences between the observatory data and
the BGGM. The aim of the adjustment at each observatory is to
make the results representative of those that the end user of the
 TABLE 2—UNCERTAINTY ASSOCIATED WITH THE CRUSTAL FIELD, AT SIX CONFIDENCE
LEVELS, CALCULATED USING GLOBAL SURVEY DATA SET
Confidence Level
68.3% (1 if Gaussian)
90%
95%
95.4% (2 if Gaussian)
99%
99.7% (3 if Gaussian)
Declination Limit
(degrees)
0.148
0.419
0.823
0.874
1.641
2.613
Magnetic Dip Angle
Limit (degrees)
0.072
0.126
0.147
0.150
0.173
0.179
Total Intensity Limit
(nT)
56
100
116
118
135
139
TABLE 3—UNCERTAINTY ASSOCIATED WITH THE CRUSTAL FIELD, AT SIX CONFIDENCE
LEVELS, CALCULATED USING DATA SETS LOCAL TO OIL AND GAS FIELDS
Confidence Level
68.3% (1 if Gaussian)
90%
95%
95.4% (2 if Gaussian)
99%
99.7% (3 if Gaussian)
Declination Limit
(degrees)
0.185
0.403
0.534
0.564
1.191
1.692
Magnetic Dip Angle
Limit (degrees)
0.081
0.163
0.208
0.223
0.575
0.703
Total Intensity Limit
(nT)
104
187
BGGM is likely to encounter. For example, some observatories are
222
located on volcanic rocks, and the local magnetic field is therefore
highly anomalous. These locations have very different geological 224
settings compared to those of typical oil fields. 355
The adjustment is achieved by first making independent esti-
mates of the observatory crustal-field contribution. The observa- 500
tory crustal biases are included as coefficients in an inversion of
satellite and observatory data for a magnetic-field model. This Variations in Uncertainties With Time. The global limits for the
combined with the new (smaller) crustal-field limits by squaring,
model is similar to the BGGM but with many more coefficients 95.4% confidence level for each BGGM year are shown in Fig. 2.
summing, and square rooting.
Results
included to characterize the external-field variations. Second, these
It can be seen that the limits are decreasing with time. This is
The results
estimates are subtracted from the differences. The differences because for the
of the 95.4%inlimits
increase are presented
the number according
of high-quality to howdata,
satellite
are then ordered and the limits computed. Third, these limits are they vary with
especially sincetime,
2000then
whenwith location.
Ørsted data were first incorporated into
Fig. 2—The upper panel shows the 95.4%-confidence limits in declination, magnetic dip angle (dip), and total intensity (B total)
for BGGM1995–BGGM2006, and the lower panel shows the number of data used each BGGM year.
232
June 2010 SPE Drilling & Completion
the BGGM. There have also been various improvements in model
parameterization and inversion techniques since 1995, but again There are also variations in the limits with time of day. This is
especially since 2000. Because of these incremental improvements illustrated
in the BGGM from one year to the next, any solar-cycle effects are Variations in Fig. 4, again using
in Uncertainties With Eskdalemuir
Location. Thedata.limitsFrom
varythis, we
can see that
depending onfor high confidence
location of the oil levels,
field. theisBGGM
This because is less accurate
of the
masked
Fig. in the analysis
3—Variation by month presented here. Magnetic
of the external-field activity
contribution to is linked
(again the limits in Fig. 4 are the limits
external-field contributions only
with
the
June
the approximately
68.3% (lower), 95.4%, and
2010 SPE magnetic
Drilling & dip
11-year
99.7%solar
Completion
cycle,
(upper) had its last and have not been combined with the values from Tables 2core-
which limits
confidence external-field contribution to the and the associated
and 3)
in declination,
sunspot maximum during angle,
2000and total
(with intensity at
magnetic Esk- peakingfield modeling challenges at locations where the external fields
activity
dalemuir during the night when substorm processes in the tail of the mag-
in 2003)inand Scotland.
last minimum during 2008. Fig. 4—Variation by hour of the external-field contribution to
netosphere are particularly prevalent. Instabilities in the tail plasma
The variations in the external-field contributions to the 68.3, sheet of thethe 68.3% (lower), 95.4%, and 99.7% (upper) confidence limits
in declination, magnetosphere
magnetic dip angle, occur, andintensity,
and total the energy stored233
respec- there is
95.4, and the 99.7% limits during a typical BGGM year at Esk- tively, released when the magnetic-field
at Eskdalemuir in Scotland. lines relax from their stretched,
dalemuir (relevant for drilling in the North Sea) are illustrated intail-like configuration and “snap” back into a more dipolar configu-
Fig. 3. From this, we can see that for high confidence levels, theration. This process results in charged particles in the plasma sheet
BGGM is less accurate near the equinoxes (March and October)being energized and accelerated down to the polar ionosphere.
when the alignment of the Earth’s magnetic field in relation to However, for low confidence levels, the midlatitude ionospheric
the Sun’s magnetic field carried in the solar wind is favorable for dynamo current system is energized during sunlit hours, and, as a
efficient transfer of solar-wind energy into the magnetosphere. result, As the BGGM is less accurate during these periods.
a result, current systems within the magnetosphere and ionosphere Fig. 5 shows how the D limits compare with one another for
are energized more during these periods. For low confidence each year, and from this, it can be seen that the 95.4% limits
levels (e.g., 68.3%), the BGGM is less accurate during northern (equivalent to 2 if Gaussian) are more than twice the 68.3%
hemisphere summer when the ever-present auroral electrojet and limits (equivalent to 1 if Gaussian) and that the 99.7% limits
midlatitude ionospheric dynamo current system are enhanced. (equivalent It is to 3 if Gaussian) are more than three times the 68.3%
important to realize that the limits in Fig. 3 are the external-field limits. This clearly demonstrates that we are dealing with a non-
contributions to the overall BGGM limits only: they have not been Gaussian error source.
combined with the values from Tables 2 and 3. When this is done,
the variations in time of the external-field contribution are mostly
saturated by the internal-field contribution, although the baseline
contribution remains important.
 Fig. 5—The declination 68.3% (lower trace), 90%, 95%, 95.4%, 99%, 99.7% (upper trace) limits for BGGM1995–BGGM2006.
are significant. It is assumed that there are no significant spatial
patterns
234 in the crustal-field contributions to the limits. However, it
is conceivable that the crustal field is more significant at locations
where the core field is strongest (i.e., at high latitudes, becauseFig. 6—Spatial variation in the 95.4% limits for BGGM2004 in
part
(a) declination, (b) magnetic dip angle, and (c) total intensity.
of the crustal field arises from magnetization that is induced by Larger
the circle diameters indicate larger
June 2010 SPE Drilling & Completion
limits.
core field). We expect any variations caused by changing inducing-
field strength to be small if they are detectable. The variations of
the limits with location for BGGM2004 can be seen in Fig. 6.
As one might expect, the limits are higher at high latitudes
where either the magnetic-field lines are open to the solar wind
and come under its influence or the auroral current systems are
present. Charged particles of origin external to the Earth’s magne-
tosphere can be accelerated down the open field lines and energize
auroral and polar current systems. These particles can interact with
atmospheric molecules and produce stunning auroral displays.
The auroral electrojets are not centered on the geographic poles
but on the geomagnetic poles. They are oval-shaped, extending
further from the geomagnetic poles and covering wider zones on
the nightside than on the dayside. They vary in intensity accord-
ing to time of day and also expand and contract according to
activity. The BGGM limits, especially for F in the region of the
dip equator where the main magnetic field is horizontal, are also
larger. Because the dip equator is not well defined by geomagnetic
coordinates based on a tilted dipole, we use corrected geomagnetic
(CGM) coordinates (Gustafsson et al. 1992; Tsyganenko et al.
1987) to order the limits. CGM coordinates of a point in space are
computed by tracing the magnetic-field line through the specified
point to the dipole geomagnetic equator, then returning to the same
position along the dipole field line and assigning the obtained
dipole latitude and longitude as the CGM coordinates to the start-
ing point. At the near-equatorial region, where the magnetic-field
lines may not reach the dipole equator and where, therefore, the
standard definition of CGM coordinates fails, a different approach
is applied, based on a minimum intensity value along the given
magnetic-field line. Essentially, CGM coordinates are better than
straightforward geomagnetic coordinates (based on tilted dipole)
because they take account of the higher spherical harmonic terms
in the estimation of the main magnetic field.
Fig. 7 shows how the 95.4% limits vary with CGM latitude
for BGGM2000–BGGM2006. Also plotted are splines (cubic B
splines with 10 knots) that are used to interpolate between the
CGM latitudes of the observatories. Organizing the limits in this
 95.4% D confidence limit & spline
Limit
(degrees)
1.0
1.5
2.0
2.5
–50
0
CGM latitude
50
95.4% I confidence limit & spline
0.40
Limit
(degrees)
0.20
0.25
0.30
0.35
–50
0
CGM latitude
50
Fig. 8—Global 95.4% limits for BGGM in (a) declination, (b)
magnetic dip angle, and (c) total intensity. The grey areas in
95.4% F confidence limit & spline
8a are where declination is poorly defined.
240 260 280
300
480
close vicinity of the dip equator is also missing because there are
–50 very few observatories directly affected by the equatorial electrojet
at the moment. However, a larger-scale feature in the F limits at
0 low latitudes is present, and this is likely to be associated with the
CGM latitude
50 equatorial Appleton Anomaly (Lühr et al. 2003).
Limit (nT)
These spline functions are then used to derive time-averaged
look-up tables arranged by geographic latitude and longitude for
200 each confidence level. The ISCWSA error model is in the process
220 of being upgraded to have the option to read these tables. The tables
also can be used to make global plots of the spatial variation of
Fig. 7—The 95.4% confidence limits in declination, magnetic
dip angle, and total intensity for BGGM2000–2006 arranged by
a particular limit, and in Fig. 8, the 95.4% limits are shown. The
manner
correctedallows one tolatitude.
geomagnetic see theAlso
effects
shownofare
thespline
auroral tables are available from BGS (2009) and ISCWSA (2009).
fits. and equatorial
currents
June 2010 more clearly
SPE Drilling in the results. While there are some unrealistic
& Completion
features in the results, for example the hemispherical asymmetry for
the magnetic-dip-angle results, we choose to use them as they Comparisons are With Existing Magnetic-Field Uncertainty Esti-
rather than manipulate them by some arbitrary means to mates. It is interesting
Implementation to see how
and Well-Profile these estimates
Comparisons Usingcompare
Revised with
compensate those in the present
Magnetic-Field ISCWSA
Uncertainty error model,
Estimates. Theas detailed in Table
geomagnetic error 1.
for the inadequacies of observatory coverage at low latitudes and Fig.in9 shows the differences at the 95.4% confidence level235 in the
the southern hemisphere. The expected increase in the F limitssense in (ISCWSA estimate – new estimate). In deriving these differ-
ences, we compute the 95.4% confidence limits as 2 times the 1
values as given in Table 1, including of course the BH dependence
for declination. This is the only method available for deriving
these limits from the existing published literature (Williamson
2000). The new geomagnetic error estimates are a considerable
improvement on the existing estimates. The same applies for the
68.3% confidence level.
terms in the ISCWSA error models (Table 1) were updated using
values from the
Conclusions andlook-up
Discussion tables to derive new wellbore positional
uncertainties
Revised for threeto
uncertainties typical well profiles.
be associated withSpecifically,
magnetic-field testsvalues
236
were conducted using the standard MWD (Rev.These MWD +
2) anduncertain-
computed from the BGGM have been derived.
axial correction
ties are more robust (Rev.than2) models and the three
any computed standard
previously and ISCWSA
should be Fig. 9—Differences in 95.4% confidence limits in the sense (IS-
well profiles: ISCWSA#1
incorporated extended-reach
into implementations of thewell
ISCWSAin the North
error modelSea, CWSA estimate – new estimate) for (a) declination, (b) magnetic
using aeromagnetic
dip angle, and (c) total data overThe
intensity. the UKareas
grey landmass andwhere
in 9a are comparing the
ISCWSA#2 fish-hookby
as soon as possible well
meansin the ofGulf of Mexico,
the look-up and ISCWSA#3
tables. They will bethe transformed
differencesdata
are with
off vectorJune
scale. observations
2010 SPE made&on
Drilling the ground.
Completion
designer
valid as long wellas in good-quality
the Bass Strait (Williamson
satellite data are 2000). The standard
available to help Early work on this is reported in Williamson et al. (1998). IFR
MWD model assumes a
build global magnetic-field models. magnetically clean drillstring, whereas the
reduces the contributions from the crustal field to the uncertain-
MWD + axial
Options correction
exist to reduce model corrects for drillstring
the uncertainties further usingmagnetiza-
data ties by amounts similar to those in Tables 2 and 3, although some
tion
fromusing
magnetic magnetic-dip
surveys local and total-intensity
to the drilling site estimates.
(IFR) to estimateuncertainty is added back in depending on measurement or trans-
theThe
localmagnetic-field
crustal field and, erroradditionally,
magnitudes from are reduced to 1 equiv-
nearby observatories formation method.
alents by first extracting the limits for
(interpolation IFR) to estimate the local external field. the location of the well
If the site While the crustal field presents a bias error for all MWD-survey
and
is onfor theitfinal
land, confidence
is relatively easy level desired,
to make thenobservations
vector dividing them in by
themeasurements in a given well and dominates the uncertainty asso-
the equivalent
vicinity number
of the drilling of (but
site if it isatassumed
a sufficientthatdistance
the error todistri-
ensure ciated with any global magnetic-field model, the external field also
bution is Gaussian. This is so that the magnetic-field errors can
no contamination from the drilling rig). However, at sea this is apresents a bias error but affects individual survey measurements
be combined
much greaterand propagated
challenge because withofother error terms
the difficulties in expressed
establishing or sequence of measurements taken within a few hours of one
as standard
azimuth of adeviations. The divisorssystem
vector magnetometer are 1 for onthe 68.3% survey
a moving limits,
another. The external field can be much more significant in terms
1.64 for the 90% limits, 1.96 for the 95% limits, 2 for the 95.4%
platform and of removing the magnetic-effect of that platform. of amplitude than the crustal field, especially at high latitudes. In
limits, 2.58
However, for the
using 99% limits, and
a combination 3 for GPS,
of gyros, the 99.7% limits. For
and accelerometers fact, the external field dominates the pattern in the BGGM limits
example,
for attitude fordetermination
ISCWSA#1 well and profile at 60°N,
performing 2°Eturns
360° and aat95.4%locations seen in Figs. 6 through 8. If the surveys are from a critical part of
confidence in the final error ellipsoids, the
where the magnetic-field strength is already known, the accurate standard geomag-
the well path, the effects of magnetic activity on drilling decisions
netic error terms
estimation of the AZ(G),
magnetic DBH(G), MFD(G),
field vector at and MFI(G)
sea has been (Table
shown to can be crucial. For these reasons, techniques such as interpolation
1)
be were
feasiblereplaced
(Lesurwith et al.0.375°,
2004).0°, 0.12°, and 96 nT. These are IFR have been developed to estimate the magnetic field downhole
derived from the values in the 95.4% look-up
One drawback of direct measurements of the geomagnetic-field table for 60°N,
on a minute-by-minute basis incorporating data from nearby
2°E, then
vector candividing
be that by the2.poor data coverage (caused by the time taken magnetic observatories (Reay et al. 2005; Bowe and McCulloch
to The
make error
vector models are then run,
observations) and the
precludes ellipsoid
making an parameters
extensive map at
2007). Because the next maximum in solar activity is expected in
total depth and at the confidence level desired are output. These
of the crustal field and estimating the field at depth. However, it2011 and magnetic activity is expected to increase from now till
are compared with those from the existing
is relatively easy to make observations of the strength of the mag-error models. For ISC-
a couple of years after this, the benefits of using real-time geo-
WSA#3
netic field,well, theifcomparison
and, made from is anat 3000 mit measured
aircraft, is relatively depth
quick. not
Anthe
magnetic data for correcting downhole surveys will increase over
total depth at 4030 m because the axial correction
extensive area can be covered in a few days, providing sufficient fails at 90° hole
the next few years.
inclination. Fig. 10 summarizes the results for
data for potential field transformations, such as directional deriva- lateral uncertainty
(radius
tives and of downward
the ellipsoid major axis).This
continuation. The method
percentage reduction
has been is
validated
highest for low levels of confidence and lowest for high levels of
confidence. For the 95.4% level, the percentage reduction varies
from 2% for ISCWSA#2 well profile and the standard MWD
model, to 27% for ISCWSA#1 well profile and the MWD + axial
correction model.
 Lateral Position Uncertainty for ISCWSA#1 Well Profile at 8000 m MD
MWD
600
MWD+New Mag Errors
MWD+Axial Corr
MWD+Axial Corr+New Mag Errors
500
Major Axis
Radius (m)
400
300
200
100
0
65
70
75
80
85
90
95
100
Confidence Level (%)

Lateral Position Uncertainty for ISCWSA#2 Well Profile at 12500 ft MD

MWD
120
MWD+New Mag Errors
MWD+Axial Corr
MWD+Axial Corr+New Mag Errors
100
Major Axis
Radius (ft)
80
60
40
20
0
65
70
75
80
85
90
95
100
Confidence Level (%)

Lateral Position Uncertainty for ISCWSA#3 Well Profile at 3000 m MD

MWD
45
40
35
Major Axis
Radius (m)
30
25
20
15
10
5
0
65
70
75
80
85
 90
95
100
Confidence Level (%)
MWD+New Mag Errors
MWD+Axial Corr
MWD+Axial Corr+New Mag Errors

Fig. 10—Lateral-position uncertainties at total depth for 4 ISCWSA-error-model (Rev. 2) implementations and 3 well profiles.
June 2010 SPE Drilling & Completion
237
Acknowledgments
We would like to thank Allan McKay (formerly BGS, now
References
Petroleum GeoServices) for his contributions at an early stage
Bowe,
of thisJ.work.
and McCulloch, S. 2007.
The scientists Theinstitutes
and Value of Real-time
operating Geomagnetic
geomagnetic
Reference Data to the Oil and Gas Industry. In Space Weather: Research
observatories and undertaking magnetic surveys around
towards Applications in Europe, No. 344, ed. J. Lilensten, Chap. 5.2,
the world impact on geomagnetic field modeling. Geophys. Res. Lett. 30 (17):
are thanked
289–298. for theirThe
Dordrecht, tireless efforts.Astrophysics
Netherlands: Global standards andScience
and Space dis- Susan
1906. doi: 10.1029/2003GL017407.
Macmillan is aM.D.,
geophysicist with BGST.D.G.,
specializing in
semination of observatory data benefit greatly from the efforts of Macmillan, S., Firth, Clarke, E., Clark, and Barraclough,
Library, Springer. modeling theError
geomagnetic field. She has worked closely with
238
Intermagnet andSurvey
the World D.R. 1993. estimates for geomagnetic field values computed from
British Geological (BGS).Data
2009.Centers.
BGS Global Global magnetic-field
Geomagnetic Model, the drilling industry for 20 years developing services and mod-
the BGGM. Technical Report WM/93/28C, British Geological Survey,
models are also dependent on data from magnetic-survey satel-elsKeyworth,
http://www.geomag.bgs.ac.uk/bggm.html. to improve the accuracy
Nottingham, UK. of drilling
June with
2010 magnetic
SPE DrillingMWD
& Completion
Gustafsson,
lites, and theG., Danish-led
Papitashvili, N.E., andand
Ørsted Papitashvili,
German-led V.O. 1992.
CHAMPA revised
sciencetools. Macmillan has a degree in mathematics, a PhD, and
1990. Reay, S.J., Allen, W., Baillie, O., Bowe, J., Clarke, E., Lesur, V., and
corrected
teams geomagnetic
are thanked coordinate The
in particular. system for magnetic
local Epochs 1985 andaround
data is Macmillan,
a fellow of the RoyalSpace
Institute of Navigation and the Royal
S. 2005. weather effects on drilling accuracy in the
J. Atmos. Terr. Phys. 54 (11–12): 1609–1631. doi: 10.1016/0021- Astronomical Society.Geophysicae
Steve Grindrod 23 is
oil9169(92)90167-J.
fields have come from a wide variety of sources, and we thankNorth Sea. Annales a consultant engineer
(9): 3081–3088.
with Copsegrove Developments Ltd. He obtained a PhD in
the survey
ISCWSA. companies
2009. MWD ErrorHalliburton
Model Rev.3,Sperry Drilling Services, Baker
http://copsegrove.com/ Tsyganenko, N.A., Usmanov, A.V., Papitashvili, V.O., Papitashvili,
engineering from Lancaster University in the UK andofstarted his
N.E.,
Hughes INTEQ, Schlumberger, and Tech21 for helping make some and Popov, V.A. 1987. Software for Computations the Geomagnetic
MWDModel.aspx. career in the oil industry with Shell in the drilling research sec-
of them Field and Related Coordinate Systems. Special Report, Soviet Geo-
Lesur, V., available. Finally,
Clark, T., Turbitt, C.,the
and two reviewers
Flower, S. 2004.are thankedfor
A technique for their
tion of the Committee,
laboratories Moscow,
in The Netherlands. After various jobs
physical Russia.
comments,
estimating which helped
the absolute improve
vector the manuscript.
geomagnetic field from a marine vessel.
with Shell, Grindrod
Williamson, H.S. 2000.became a consultant
Accuracy Predictioninfor 1991 specializing
Directional Measurement
J. Geophys. Eng. 1 (2): 109–115. doi: 10.1088/1742-2132/1/2/002. in While
directional surveying
SPE Drillissues. He 15
is also
(4): chairperson of the
Drilling. & Compl 221–233. SPE-67616-PA.
Lühr, H., Rother, M., Maus, S., Mai, W., and Cooke, D. 2003. The diamag- ISCWSA (SPE-Well Positioning Technical Section) Error Model
doi: 10.2118/67616-PA.
netic effect of the equatorial Appleton anomaly: Its characteristics andMaintenance Subcommittee.
Williamson, H.S., Gurden, P.A., Kerridge, D.J., and Shiells, G. 1998.
Application of Interpolation In-Field Referencing to Remote Offshore
Locations. Paper SPE 49061 presented at the SPE Annual Technical
Conference and Exhibition, New Orleans, 27–30 September. doi:
10.2118/49061-MS.