SPE 89899

Depth Control: Reconciliation of LWD and Wireline Depths, Standard Practice and an
Alternative Simple but Effective Method
H. Wilson, SPE, Baker Hughes INTEQ UK, J. C. Lofts, G. C. Page, SPE, Baker Atlas/INTEQ UK,
A. G. Brooks, SPE, Baker Hughes INTEQ Houston, D. Walder, BP Aberdeen

Copyright 2004, Society of Petroleum Engineers Inc.

The measurement of depth by wireline and pipe results in
This paper was prepared for presentation at the SPE Annual Technical Conference and differences in apparent depth which are generally unquantified
Exhibition held in Houston, Texas, U.S.A., 26–29 September 2004.
and unexplained. It is common to work with a “wireline
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
depth” and a “driller’s depth”, which are generally accepted to
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to be different. From a formation evaluation perspective it is
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at desirable to integrate information acquired by both these
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
methods at a common datum in order to evaluate formation
for commercial purposes without the written consent of the Society of Petroleum Engineers is parameters. In turn it must be relatable to similar information
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous in other wellbores, and hence an absolute reference. This leads
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
to the requirement for what has been called FE depth
(formation evaluation depth), which is the best possible
estimate of absolute depth with all formation evaluation data
Abstract corrected to it. Present perception and generally standard
Depth is one of the most important formation evaluation practice weights wireline depth accuracy superior for FE
measurements, but it is also one of the most difficult to define depths.
accurately. LWD measurements, which rely on driller’s depth,
are becoming more prolific and widely used. Wireline Definitions of Depth
measurements, with their independent determination of depth, It is useful at this point to pause and define what we mean by
are still commonplace, while the use of both wireline and “depth”. The simplest definition is the length of the centre line
LWD measurements together is still largely unavoidable in a of the wellbore, independent of any deviation. This is the
full data acquisition program. Reconciliation of depth, along hole or Measured Depth (MD). This is the first
including survey error when converting to TVD’s, is still one important parameter to measure, relative to a defined fixed
of the key problems faced when evaluating and fully starting point, typically the rig floor or rotary table.
interpreting these data sets. A True Vertical Depth calculation (TVD) requires the input
This paper reviews the current common practices applied to of both measured depth and wellbore deviation. This is more
correction and correlation of depth. Generally, these practices useful as an absolute reference than measured depth alone
are intended to correct all other measured depths to a single when defining the location of subsurface features, as it is
definitive depth reference. The extent to which the intended independent of the wellbore geometry.
goal is achieved is assessed. The paper also describes an Relative Depth of sensors to formation features, or each
alternative approach which accepts all depths as valid, but other, is another parameter which has become important with
assigns uncertainties to each. A real well is used to compare the addition of multiple sensors at different points along the
the outcomes for each approach. One conclusion is that length of single wireline or LWD assemblies, and also in order
correction of drill pipe depths for stretch effects is desirable, to position sampling mechanisms at the correct point relative
and a simple but effective correction method is described. to formation features.
The integration of these measurements at a single depth
Introduction point in the evaluation of formation parameters requires that
The measurement of depth has developed slowly and with they are accurately depth matched. This is only possible by an
apparently lower priority than many other formation initial accurate determination of measured depth at all points,
evaluation measurements. Initially, depth was measured only and it may also require additional information such as
by the length of pipe used to drill a well until oil was seen to accelerometer data, time of measurement for memory data, or
flow out, thus defining the depth of the reservoir. With the weight on bit in the case of LWD, in order to refine accuracy.
advent of wireline sensors, a new measurement was available
that could read formation parameters and relate them to the
measurement of the length of cable run into the well. With the
introduction of LWD sensors, the reliance on the length of the
drillpipe was renewed.
2 SPE 89899

The Measurement of Depth speed as the pipe, to which the logging tools are attached, and
Wireline depth is generally measured by one of two therefore depth matched to it. Mismatches between the hoist
methodologies, which are described in detail by Sollie1: and pipe movement can introduce depth errors which are
1) The traditional method uses a calibrated wheel which difficult to quantify.
is clamped onto the cable to be measured, and PCL logs are recorded with neutral weight on the drillstring,
measures depth by the number of rotations of the and can be recorded both running in and pulling out of the
wheel. This measurement is further corrected by the hole. LWD is generally recorded while drilling with weight on
use of magnetic reference marks, which provide a the bit. This results in additional shifts between the PCL up,
reference to detect any slippage or mis-calibration in down, and LWD datasets.
the measurement system, and for cable stretch due to
tension by the use of “stretch charts”. TVD - Logs might initially be annotated with MD, but vertical
2) A more recent methodology uses dual calibrated depth (TVD) is normally the defining parameter for formation
wheels which are compared at all times to minimise features. TVD is obtained by interpolating the well path
slippage. No magnetic marks or stretch charts are used survey for the MD of interest. The well path survey consists of
as it is assumed that while running into the well the measured data (MD, inclination and azimuth) from which co-
weight of tool plus cable under each section measured ordinates (northing, easting and TVD) are calculated. It is MD
will remain constant, and therefore the cable stays the and inclination that primarily determine TVD.
same length. When pulling out of the well the depth is
then corrected back to this reference. Measurement Accuracy
Each of these systems has its own benefits and drawbacks, There are many factors which affect the accuracy of depth
and both can work well in vertical wells. Both can be in error measurements. These can be grouped into three main
in deviated wellbores however, due to varying cable tensions categories: the reference point, the measurement system itself,
caused by drag and doglegs. In deviated wellbores wireline and changes in length after measurement. Some errors take the
systems will usually overestimate depth. form of biases, which displace the measurement in a known
direction, while others are uncertainties which may act in
Driller’s depth is defined by measuring the length of sections either direction. Estimates of typical values attributable to
of pipe, and then adding them together as they are run into the various sources are presented in Table 1.
hole in a “pipe tally”. It is not common to apply any additional
corrections, e.g. for pipe stretch, to this depth. Reference Point. It is usually assumed that the reference point
LWD depth uses this as a reference with the addition of a does not move. This may be a reasonable assumption if the rig
measurement system to define interval depth along each stand. is on land, but floating rigs can have significant variation in
This is generally accomplished by one of two systems: the position of the drill-floor (the common reference point),
1. An encoder is attached to the drum of the drawworks, over the timescales involved. Some of these are detailed in
and measures rotation of the drum, and hence the Table 1. Tides and wave action can produce movements of
length of drill-line used. This system is generally several metres, and it is not generally considered that wind and
used on fixed platforms or land rigs, and is the most changes in barometric pressure can also produce movement.
accurate. It might be possible to track and correct for these movements
2. An alternative system uses a separate geolograph line using GPS. If using such a method, care would have to be
attached to the traveling block which goes up over a taken to remove the effect of earth tides, which cause the
pulley and down to an independent measuring entire crust/reservoir/sea/rig to move up and down together
system. This has the advantage that it can encompass with amplitude of about 0.5 m and a period of about 12 hours,
movements of the wave compensation system on as seen by the reference satellites.
floating offshore rigs, but it can suffer from other If depths are referred to sea level, the datum must be clear.
inaccuracies such as the line bowing in the wind Mean sea level (MSL) is the most common datum used in the
affecting the measured length . oilfield, which is not the same as the lowest astronomical tide
There are also many other systems in use around the world (LAT) used on many navigation charts.
such as liquid/pressure based systems. All these systems are
only used to provide interval depths along the length of Measurement System. The second category of errors is
individual stands of pipe. At the end of each stand it is caused by the measurement systems themselves. These
common practice to reset the depth to match the driller’s tally. sources of error can be minimized by system design and
LWD depth cannot therefore be considered an independent procedures, and generally are not as large as the depth errors
measurement. caused by changes in length of the cable or pipe which occur
after measurement.
Pipe conveyed logging (PCL) uses a combination of both
wireline and pipe measurements, but essentially the primary After Measurement. The main category of concern is
depth reference is driller’s depth. At the latching point, accurately defining and correcting for changes in the length of
typically just above the reservoir section to be logged, the either the cable or drillpipe after measurement. The largest of
wireline depth is set to match either previously recorded LWD these sources of error are due to mechanical stretch, and
or tied to pipe depth. During the logging process the wireline environment (pressure/temperature). Wireline depth
winchman attempts to keep the cable moving at the same measurements generally take account of these to a greater or
SPE 89899 3

lesser extent, whereas drillpipe depth measurements do not. on bit is typically absent when taking surveys off bottom, but
This results in the commonly observed differences between present when acquiring LWD data.
driller’s and logger’s depths, with wireline depths generally
being deeper than driller’s. The Challenge
Additionally there are sources of gross human error, such as The challenge for FE depth is best seen as 4 individual stages,
incorrect addition of a pipe tally, or picking the wrong piece of as summarized below and illustrated in Fig. 1:
pipe. These can only be minimized by taking care in the
measurement procedure. Stage 1. Provide valid MD uncertainty estimates as an integral
For drillpipe, a number of factors contribute to depth bias, component of log data.
including2,3:
• Mechanical stretch. Stage 2. With the quantification of uncertainty comes the
• Thermal expansion. ability to quality control one depth measurement vs. another
• Hydrostatic pressure. (i.e., LWD and wireline).
• Differential pressure and mud circulation.
• Weight on bit. Stage 3. Valid estimates of uncertainty allow proper
evaluation and implementation of proposals for improved
Mechanical Stretch due to gravity tends to be the largest depth accuracy.
individual component of depth bias. The pipe is supported at
its top end and always elongates under tension, therefore the Stage 4. Working with valid quality controlled data; it is
bias is positive. The elongation increases as the well possible to calculate a weighted average depth from several
approaches vertical, and for a uniform drillstring in a straight independent depth measurements.
well the stretch can be modelled as a constant multiplied by
measured depth and by TVD. These stages are now considered in turn.

Thermal Expansion may also be a significant contributor to Stage 1 - Valid quantification of MD uncertainty. One
depth bias. The normal practice today is to measure pipe approach is to look at uncertainty estimates for MD that are in
length on the surface at the drilling location, prior to inserting everyday use in directional surveying.
it in the hole. As the pipe descends in the hole it is subjected to Directional survey systems measure inclination, azimuth
increasing temperatures, which depend on the TVD and the and measured depth (MD). Most survey systems have a tool
local geothermal gradient. Once again, the bias is positive, and error model, the inputs to which are estimates of the
the form of the stretch is approximately proportional to uncertainty associated with each of the three measurements.
measured depth, TVD, and geothermal gradient. The output is positional uncertainty, which is used to
determine the probabilities of collision with offset wells and of
Hydrostatic Pressure. When a drillstring is immersed in missing the geological target.
drilling fluid, it is subjected to increased hydrostatic pressure, Where two surveys are run over the same hole section, the
which causes shrinkage in its linear dimensions. This effect two uncertainty estimates are used to quality control one
therefore produces a small negative depth bias which is survey against the other. If the separation between the two
approximately proportional to measured depth, TVD, and mud surveys exceeds what could reasonably be expected from their
density. combined uncertainty, an investigation is carried out to
determine which is in error. If separation is within
Differential Pressure. When drilling fluid is circulated, the specification, no action is taken other than to define the more
internal pressure within the pipe exceeds the annular pressure accurate survey as definitive. Although rarely done in practice,
outside. This differential pressure causes the length of the pipe it is possible to merge the two surveys into a third, more
to extend. It also causes the diameter of the pipe to increase, accurate, survey5,6. The drive for greater accuracy will
known as ballooning, which tends to moderate the increase in probably result in more frequent application of this technique.
length4. These effects are always present for LWD The notable features of the above process are:
measurements, absent for surveys taken with no circulation • Tool specification is explicit.
and PCL wireline runs, and present with reduced amplitude for • Tool specification is matched to the objective.
surveys taken while circulating slowly off bottom. When mud • Actual performance is checked for conformance with
is circulated, however, there are some further effects which specification.
tend to lift and shorten the drillstring, including bit nozzle • Redundant data are used to quality control definitive
thrust and annular viscous drag. In addition, when the bit is on data.
bottom the pump-off force can also tend to shorten the • Discrepancies are resolved.
drillstring slightly. • Weighted averages of multiple data sets are possible.
• Uncertainty estimates are archived with the data.
Weight on Bit. The application of weight to the bit supports Although not yet adopted universally, this is accepted as
the drillstring and reduces the mechanical stretch. It therefore best practice and most elements are commonly applied.
provides a negative correction to the depth bias, which is The TVD uncertainty for a particular MD is available as an
proportional to measured depth and to weight on bit. Weight output of the survey tool’s error model. This information is
4 SPE 89899

commonly used in anti-collision monitoring and target sizing. Stage 3. Improvements in depth accuracy. The application
It is less commonly used to associate TVD uncertainty with of uncertainty estimates to all depths not only defines the
log features. With minor reworking, the error models could accuracy of current methods, it also allows evaluation of
output the TVD uncertainty associated with an LWD MD. proposed improvements. We advocate two approaches to this:
The survey model emerging as a de facto standard is that • For drillpipe depths, the model indicates that stretch
described by Williamson7. This model is sanctioned by the correction offers the greatest scope for improvement,
Industry Steering Committee for Wellbore Survey Accuracy and we recommend the adoption of a simple and
(ISCWSA), and is commonly referred to as the ISCWSA robust correction as suggested by Brooks and
model. It deals only with Measurement While Drilling Wilson9.
(MWD) directional systems, but a sub-committee of the • We also recognize the potential of instrumented
ISCWSA has been working on a gyro tool error model, which solutions to stretch correction and surface
deals with wireline conveyed survey tools8. measurements, but discussion of such techniques is
beyond the scope of this paper.
Adoption of a new standard for wireline and LWD depths.
Adoption of the MD component of this or a similar model by A simple correction for drillpipe stretch. Within the
the logging industry would add discipline to the depth ISCWSA model, the DST term is problematic, since
accuracy debate. If all suppliers defined depth accuracy using uncorrected stretch results in a bias; a non-zero mean value.
the same MD uncertainty model, it would allow customers to Williamson7 acknowledged this and offered the following
evaluate competing accuracy claims and validate field results. alternative uncertainty estimate:
The current set of MD uncertainty terms available in the
ISCWSA MWD model, with values given at 1 standard (
σ MD = 0.35 2 + 2.4 × 10 − 4 ⋅ MD )2
,
deviation, is: −7
with a bias of + 4.4 × 10 ⋅ MD ⋅ TVD m . . . . . . . . . . . . . . . (2)
• Depth reference (DREF): value 0.35 m. The implementation of this biased model results in a
• Scale factor (DSF): value 2.4×10-4. probability distribution that is not centred on the nominal MD.
• Stretch (DST): value 2.2×10-7 m-1. This version of the model is widely used in well planning
These terms are combined to produce the following estimate software. If the biased position of the probability distribution
of uncertainty for the measured depth of a survey station: is taken into account when making planning and drilling
σ MD = DREF 2 + (DSF ⋅ MD )2 + (DST ⋅ MD ⋅ TVD )2 . . (1) decisions, the resulting wellpath will be very similar to that
which would be drilled by correcting pipe tally measurements
For logging depths it may be necessary to modify some of by an amount defined by the bias term in Equation (2). The
these terms, to account for term is therefore working as a drillpipe stretch correction.
• Time to depth correlation uncertainty for LWD. With sufficient understanding and co-ordination, the well
• A wireline specific DST term (wireline is measured planner and directional driller might plan and drill the well
while in tension). such that the most probable wellpath, rather than the nominal
• Transient effects associated with changes in wellpath, passes through the geological target. It is not
drillstring tension as WOB is drilled-off, and for however, intuitive, and it is unlikely that it encourages
wireline tool bounce, particularly while logging open consistent or optimum depth management practice.
hole. In directional surveying, measurements are Reported drillpipe depths, such as those assigned to LWD
normally either taken off-bottom on drillpipe or in data, will not be stretch corrected. The TVD assigned to
cased hole on wireline. wireline and LWD picks will be derived from uncorrected
directional MWD data.
Stage 2. Standard depth error comparisons. With the It was decided to investigate the validity of using a term
quantification of uncertainty comes the ability to quality similar to the IWCSWA bias to develop a simple and robust
control one depth vs. another. The difference between two drillpipe stretch correction.
depths should not be improbable when compared to their Questions that needed to be answered were:
relative uncertainty, derived from their individual • Is it a valid correction?
specifications. • Can we quantify the extent to which it reduces MD
When depth measurement accuracy is unspecified, as it uncertainty for drillpipe depths?
currently often is, the analyst’s intuition is relied on. This can • If so, is it a worthwhile improvement?
lead to acceptance of discrepancies that exceed the actual
• Could it be used as a cheap and robust method of
specification, and that therefore indicate an error in one or
correcting LWD depths?
both measurements. Alternatively, it can result in wasteful
• Would such a correction result in better agreement
investigation of discrepancies that are acceptable, based on
between wireline and LWD logs?
actual specifications. In the latter case, if a discrepancy,
In line with general error model practice, any proposed
although within specification, is not acceptable this indicates
performance improvement would have to be measured in
that the uncertainty of the measurements is too great. Since the
terms of its residual uncertainty. This provides a quantification
uncertainty is predictable in advance, it is possible to
of the benefit of the method, allowing cost-benefit analysis.
determine in advance that the measurement methods are not fit
A set of simple mathematical models was built to estimate
for purpose, and so specify other methods.
the magnitude of the depth biases and the amount by which
SPE 89899 5

the depth uncertainty might be reduced by applying such • Scenario 3, large unresolved discrepancies might lead
biases. These models are described in Appendix A. a stakeholder to contest the wireline depth and favour
These models were used to calculate an estimated MD bias the LWD depth for equity reasons.
with a form similar to the DST term, a value of 4.3×10-7 m-1, In these and similar situations, the validity of the depth
and a residual uncertainty of approximately 1.3×10-7 m-1. The measurement is usually questioned some time after the data
proposed MD uncertainty is therefore are acquired, when evidence is more difficult to gather. This

( ) (
2
σ MD = 0.35 2 + 2.4 × 10 − 4 ⋅ MD + 1.3 × 10 −7 ⋅ MD ⋅ TVD )2
,
alternative practice encourages validation of the data at the
time of acquisition.
after applying a correction of + 4.3 × 10 −7 ⋅ MD ⋅ TVD m . . . (3) Applying the simple stretch and WOB corrections to the
scenarios used in Table 2. results in the data shown in Table
Stage 4. Honouring and using all depth related data. If 3. The comparison in scenario 2 is still out of specification; an
stages 1 and 2 ensure that only valid measurements are expected outcome if the discrepancy is the result of gross
accepted for consideration, a combination (a weighted error. In scenarios 1 and 3, MD differences decrease and
average) of depths is possible. A weighted average value agreement improves. These examples are designed to illustrate
contributes most when both measurements are of poor, but the intended outcome of applying drillpipe stretch correction
similar accuracy. Where two measurements have very to LWD MD’s. The method must be tested on real examples.
different accuracies, a weighted average value will tend to be
very close to the more accurate of the two, delivering little Field Example
benefit over the simpler alternative of accepting the more In this example a field has been developed by a combination
accurate measurement. However, the redundant measurement of high angle extended reach producers and injectors. The
always contributes, even if only by validating the prime development wells have been drilled up to 5880 m in total
measurement. For two valid measurements MD1 and MD2, the length. In the majority of cases the 12 ¼” section has been
weighted average depth is drilled in or just above the top of the target sand. The 8½”

MDcombined =
(MD ⋅σ 1 2
2
+ MD2 ⋅ σ 1 2 ) . . . . . . . . . . . . . . . (4) producing sections are predominantly horizontal and drilled at
angles from 87 degrees up to 93 degrees. The injectors are
(σ +σ 21
2 2
) drilled to penetrate the reservoir in the 8 ½” section at angles
and the uncertainty associated with this value is less then 30 degrees.
FE Strategy adopted. In the majority of cases LWD
σ 1 ⋅σ 2
σ combined = . . . . . . . . . . . . . . . . . . . . . . . . . . (5) gamma/resistivity and density/neutron logs have been run over
σ 12 + σ 2 2 both the 12 ¼” and 8 ½” sections. Wireline conveyed on
drillpipe has generally been run only in the 8 ½” section.
where σ1 and σ2 are the uncertainties associated with the
Depth Strategy adopted. For depth reference the LWD logs
measurements MD1 and MD2.
are considered to be the master set, and the wireline logs have
been depth-shifted to match these logs at the start of the 8 ½”
Illustrative example. Table 2 presents 3 typical scenarios in
section. No further shifts have been applied to the wireline
which both LWD and wireline logs identify a common feature.
logs at acquisition. The final logging suite has been shifted to
These are fabricated examples, designed to illustrate how
match the wireline to the LWD during preparation of the
outcomes might vary between the current common practice
composite log.
and an alternative approach. LWD uncertainty is calculated
In the case well in Figures 2 and 3, an injector, the total depth
using equation (1) and the standard ISCWSA values. Wireline
drilled was 3583 m. The 9 5/8” shoe was set at 3284 m and the
uncertainty is calculated using the same equation, but with
wireline logs were depth matched to the MWD in the first 20
inputs of 0.2 m for DREF, 1.2×10-4 for DSF, and 1.3×10-7 m-1 m of the 8 ½” section. The well penetrated a regional OWC at
for DST, selected to conform to the assumption that wireline MD 3352 m, a TVD of 2062 m (Fig 2). This is 2 m TVD
depths are more accurate than drillpipe depths. The validity of shallower than in all of the offset vertical wells. Based on the
these input values, particularly those used for wireline, is open ISCWSA model, the uncertainty associated with this TVD was
to question. However, they are adequate for the purpose of
± 2.1 m (1σ). Although a predictable outcome based on known
illustrating typical depth discrepancies and how it is possible
system performance, the TVD discrepancy was however still
to apply a more reasoned approach to resolving those
significant in terms of field development assumptions.
discrepancies.
Applying the drillpipe stretch correction, described herein, the
The difference in outcomes might not appear dramatic, but
TVD difference with offset vertical wells was reduced to only
the “alternative practice” does deliver more valid depth data,
0.2 m. The uncertainty associated with the new TVD was also
and in doing so avoids erroneous decisions and wasteful
reduced, to ± 1.3 m (1σ). This TVD difference could support
disputes.
compartmentalisation and change the current assumptions
Examples might be :
about connectivity. Compartmentalisation within the field has
• Scenario 2, if it happens that the wireline depth is out
had an impact on field development. In some areas producing
of specification (has suffered a gross error), it would
wells have been shut-in due to the lack of pressure support.
still be accepted as definitive in preference to the
Uncorrected, this depth error might have had a significant
valid LWD depth.
impact on present and future field development plans.
6 SPE 89899

In addition, the PCL wireline depth error increased over the References
8 ½” section (Fig. 3). At the 9 5/8” shoe the two runs were 1. Sollie, F. and Rodgers, S. “Towards Better Measurement of
depth shifted together, but by TD the wireline was 2 m TVD Logging Depth”, Paper D, SPWLA 35th Annual Logging
Symposium, Tulsa, 1994
deeper than the LWD. This error was introduced over 300 m
of drilled section, and demonstrates the difficulty in 2. Ekseth, R. “Uncertainties in Connection with the Determination of
maintaining depth synchronization between the wireline winch Wellbore Positions” dissertation, Norwegian University of Science
and the drillpipe movement as discussed earlier. Ideally a final and Technology, Trondheim, March 1998
data set could be generated which ties the recorded data to
stretch corrected drillpipe depth as a best estimate of true FE 3. Kirkman, M. and Seim, P. “Depth Measurements with Wireline
depth. and MWD Logs” SPWLA (Houston Chapter), 28 April 1992

Conclusions 4. Roark, R.J. and Young, W.C. “Formulas for Stress and Strain”,
1. The general principles applied to directional survey McGraw-Hill, fifth edition, 1975
measurement accuracy and quality control can be 5. Berger, P.E. and Sele, R. “Improving Wellbore Position accuracy
applied to wireline and LWD log measured depths. of Horizontal Wells by Using a Continuous Inclination
2. This would result in more accurate depths and more Measurement from a Near Bit Inclination MWD Sensor” SPE
disciplined use of depth measurements. 50378, SPE International Conference on Horizontal Well
3. Specific benefits would be: Technology, Alberta, November 1998
• Improved overall quality of wireline depth, by
providing a valid quality control check 6. Chia, C.R., Phillips, W.J. and Aklestad, D.L. “A New Wellbore
Position Calculation Method”, SPE 85111, SPE Drilling and
• Reduced incidence of gross error
Completion, September 2003
• Identification of gross error when it does occur
• Common standard to quantify benefit of 7. Williamson, H.S. “Accuracy Prediction for Measurement While
proposed depth enhancement systems. Drilling” SPE 67616, SPE Drilling and Completion, December
4. Correction of drill pipe depths for stretch effects is 2000
desirable to help increase LWD depth accuracy.
5. A simple drill pipe stretch correction provides a cost- 8. Torkildsen, T., Havardstein, S.T., Weston, J.L. and Ekseth, R.
effective improvement in depth accuracy. “Prediction of Wellbore Position Accuracy when Surveyed with
Gyroscopic Tools” SPE 90408, SPE Annual Technical Conference
6. The potential for instrumented independent depth
and Exhibition, September 2004
solutions should also be addressed in parallel.
9. Brooks, A.G. and Wilson, H. “An Improved Method for
Nomenclature: Computing Wellbore Survey Uncertainty and Its Application to
Di outside diameter of ith drillstring component, m Collision and Target Intersection Probability Analysis” SPE
di inside diameter of ith drillstring component, m 36863, Europec, Milan, 1996
DREF depth reference uncertainty, m
DSF depth scale factor uncertainty 10. Winters, W. and Warren, T.M. “Determining the True Weight-on-
DST depth stretch uncertainty, m-1 Bit for Diamond Bits” SPE 11950, SPE Annual Technical
Conference and Exhibition, 1983
E Young’s modulus, Pa
G geothermal gradient, °C/m SI Metric Conversion Factors
g acceleration due to gravity, m/s2 lbf × 4.448 222 E + 00 = N
K coefficient of thermal expansion, °C-1
Li length of ith drillstring component, m
MD measured depth, m
n number of drillstring components, numbered from
top down
TMi mean TVD of ith drillstring component, m
TVDi TVD at bottom of ith drillstring component, m
W weight on bit, N
∆L change in drillstring length, m
∆P mean differential pressure across drillstring, Pa
ν Poisson’s ratio
ρm density of drilling fluid, kg/m3
ρs density of steel, kg/m3
σ uncertainty (standard deviation)

Acknowledgements
The authors wish to thank the management of BP and Baker
Hughes for permission to publish this paper.
SPE 89899 7

Appendix A

Simple models were developed as follows:

For mechanical stretch in a drillstring with n components,

(ρ s − ρ m ) ⋅ g (ρ s − ρ m ) ⋅ g ⎡ ⎤
) ∑ {(D )( )}
n n −1 n

∑ [L ⋅ (TVD − TM i )] + ∑ ⎢⎢ (D
Li 2
∆L = ⋅ i i ⋅ 2
⋅ j − d j 2 ⋅ TVD j − TVD j −1 ⎥ . . . . . . . . . . . (A - 1)
E i =1
E i =1 ⎣ i − di 2 j =i +1 ⎥⎦

This model assumes that the drillstring is rotating, and the rotation eliminates effects of axial drag.

For thermal expansion,

n
∆L = K ⋅ G ⋅ ∑ [L ⋅ TM
i =1
i i ]
. . . . . . . . . . . . . . . . . . . . . . . . (A - 2)
This model assumes that the local geothermal gradient normalized by dividing it by the measured depth and the TVD
provides a satisfactory approximation to the circulating of the appropriate wellpath. This expressed the stretch in the
temperature profile. form of a coefficient analogous to the DST term in the
ISCWSA error model. The mean value over the 522 cases
For hydrostatic pressure, modelled was 4.3×10-7 m-1, with a standard deviation of
ρ m ⋅ g ⋅ (1 − 2 ⋅ν ) n 0.9×10-7 m-1. This mean value is slightly smaller than the bias
∆L = −
E
⋅ ∑ [L ⋅ TM
i =1
i i ] . . . . . . . . . . . . . (A - 3) estimate of 4.4×10-7 m-1 suggested by Williamson9, and the
residual uncertainty should be assigned a value somewhat
greater than 0.9×10-7 m-1 in order to account for unmodelled
For differential pressure and ballooning, sources of error and sampling errors in the cases selected for
∆P ⋅ (1 − 2 ⋅ν )
n ⎡
L ⋅d 2 ⎤ study. It is therefore proposed that 4.3×10-7 m-1 is a reasonable
∆L = − ⋅ ∑(
⎢ 2i i 2 ⎥
) estimate of the bias, with an associated residual uncertainty of
E ⎣ Di − d i ⎥⎦ . . . . . . . . . . . . (A - 4)
i =1 ⎢ 1.3×10-7 m-1 at 1 standard deviation. This approach provides
This assumes that most of the differential pressure is more valid estimates of depth and uncertainty than either of
developed close to the bit, and an average differential pressure the ISCWSA models.
is applied to the entire drillstring. The term (1 - 2⋅ν) accounts The adjustment for weight on bit was also estimated for the
for ballooning. The effects due to bit nozzle thrust, hydraulic various cases studied using a simple model described by
pump-off, and annular viscous drag are assumed to be second Equation A-5, and results were normalized by dividing the
order and are neglected in this analysis. depth adjustment by measured depth. The average result was a
correction of - 2.0×10-4 per unit measured depth, for a weight
For weight on bit, on bit of 25000 lbf. The estimate was made for a weight value
n ⎡ ⎤ typical of smaller holes, since the correction becomes more
W 4
∑(
L
∆L = − ⋅ ⋅
)
⎢ 2 i 2 ⎥ significant at greater depths where the hole size is reduced. A
E π i =1 ⎢⎣ Di − d i ⎥⎦ slightly larger correction may be required when drilling with
. . . . . . . . . . . . . . . . . . (A - 5)
drag bits, as additional WOB is required to overcome the
hydraulic pump-off effect10.
In order to obtain an estimate of the magnitude of this bias and
its residual uncertainty, these simple models were populated
with 522 cases representing a wide variety of wellpaths,
depths, drillstring designs and sizes, geothermal gradients, and
mud densities. The 58 wells included straight holes (both
vertical and inclined), J-shapes including slant and horizontal,
and S-shapes including slant and horizontal. Depths of 2000m,
5000m, and 8000m were modelled. Drillstrings were either 5”
drillpipe with 8” collars, or 4.5” drillpipe with 6” collars.
Some models used heavy weight or compressive service
drillpipe where appropriate, and some included tapered
drillstrings with additional 3.5” or 2.875” drillpipe. Each of
the 58 wells was modelled with geothermal gradients of 0.2,
0.3 and 0.4°C/m, and with mud densities of 1000, 1500, and
2000 kg/m3, for a total of 522 different cases.
The effects due to mechanical stretch, thermal expansion,
hydrostatic pressure, and differential pressure were summed
for each case. Each computed stretch value was then
8 SPE 89899

Table 1. The most significant sources of measured depth error and uncertainty, with estimates of their
magnitudes.
Typical values, depth 5000 m
Reference point Bias, m Unc (1σ), m Weight Propagation Comments
Reference to survey datum None 0.1 1 Systematic
Wrong reference None gross error 1 Systematic
Rig Jacking None gross error 1 Systematic

Floating installations
Waves None 1 1 Random
Weather-pressure None 0.5 1 Random Could be systematic for wireline
Tides None 0.5 1 Random Could be systematic for wireline
Weather-wind None 0.3 1 Random Could be systematic for wireline

Wireline
Sheave movement None gross error 1 Systematic
Winch movement None gross error 1 Systematic

Drillpipe
Pipe stick-up None 0.2 1 Random Directional surveys only.

Depth measurement Bias, m Unc (1σ), m Weight Propagation

Wireline
Wheel wear -3 1 MD Systematic
Wheel slippage 2 1 MD Systematic
Wheel buildup 2 1 MD Systematic
Wheel thermal None 0.5 MD Systematic
Marking temp None 0.5 MD Systematic
Marking accuracy None 0.5 MD Systematic
Drillpipe
Measure vertical 0.5 0.2 MD Systematic
Measure temp None <0.5 MD Systematic
Wind on block height line None <0.1 1 Systematic

After measurement Bias, m Unc (1σ), m Weight Propagation

Wireline
Inelastic stretch 5 2 MDxTVD Systematic Only for new wireline
Temperature -4 1 MDxTVD Systematic
Pressure -2 0.5 MDxTVD Systematic
Torsion -1.5 1 MDxTVD Systematic
Cable sag -0.1 0.05 1 Systematic
Elastic stretch None 1 MDxTVD Systematic

Drillpipe
Elastic stretch 4 1 MDxTVD Systematic
Temperature 4 1 MDxTVD Systematic
WOB -1 0.3 MD Systematic
Pump-off -0.6 0.2 MD Systematic Only for diamond and PDC bits
Differential pressure 0.5 0.1 MD Systematic
Hydrostatic -0.4 0.1 MDxTVD Systematic
Annulus viscous drag -0.3 0.1 MD Systematic
Nozzle thrust -0.1 <0.1 MD Systematic
Torque -ve <0.2 MD Systematic Greater if buckling occurs
SPE 89899 9

Table 2, Uncorrected LWD MD

MD of logged feature (m) *MD uncertainty (m), 1σ *Weighted averages

Scenario (TVD) LWD w/l diff LWD w/l combined MD Uncertainty

1 2000 3554 3556 2 1.8 1.0 2.1 3555.5 0.9
2 1500 1859 1856 -3 0.8 0.5 1.0 invalid invalid
3 4200 6545 6556 11 6.3 3.7 7.3 6553.2 3.2

Resulting actions
Common practice Alternative practice
Scenario
1 Accept w/l depth of 3,556 m as A relative uncertainty of 2.1 m, (1 σ) means that there is a 34%
definitive, but with no explicit chance that the two MD’s would differ by 2m or more.
quantification of associated uncertainty. Considered an acceptable comparison.
No reason to question either depth.
Accept w/l depth as definitive, with uncertainty of 1.0 m (1 σ).

2 Accept w/l depth of 1,856 m as A relative uncertainty of 1.0 m, (1 σ) means that there is about a
definitive, but with no explicit 0.2% chance that the two MD’s would differ by 3m or more.
quantification of associated uncertainty. Considered an unacceptable comparison.
Investigate quality of both depths.
Discount depth that is found to be in error.
Accept remaining depth as definitive.
Assign appropriate uncertainty.

3 Accept w/l depth of 6,556 m as A relative uncertainty of 7.3 m, (1 σ) means that there was a 13%
definitive, but with no explicit chance that the two MD’s would differ by 11m or more
quantification of associated uncertainty. Although agreement is not particularly good, it is considered within
Complain to LWD supplier about depth specification.
and its negative effect on while drilling No reason to question either depth, but improved accuracy
decisions. possible via weighted average.
Possibly initiate failure investigation. Accept weighted average depth of 6,553.2 m as definitive, and
assign weighted average uncertainty of 3.2 m (1 sigma).

* Not considered in current common practice.

Table 3, Corrected LWD MD

MD of logged feature (m) MD uncertainty (m), 1σ Weighted averages

Scenario (TVD) LWD w/l diff LWD w/l combined MD Uncertainty

1 2000 3556.3 3556 -0.3 1.3 1.0 1.7 3556.1 0.8
2 1500 1859.8 1856 -3.8 0.7 0.5 0.8 invalid invalid
3 4200 6555.5 6556 0.5 3.9 3.7 5.4 6555.8 2.7
10 SPE 89899

Current practice. Stage 1. Stage 2. Stage3. Stage 4..

Choose one and Uncertainties included. Quality control and Improvement in Weighted average
reject the other. identification of accuracy. depth and combined
unacceptable or gross uncertainty.
error.

Figure 1. Stages in the alternative approach to improving depth accuracy.

Fig 2. Well X showing the difference between the apparent well and regional oil-water contacts.
SPE 89899 11

Fig 3. Well X showing the difference between the TLC and FEWD data at TD.