You are on page 1of 50

Why do we need an error model?

n All measurements have errors.


n Errors that are not corrected for result in an uncertainty applicable to
the measurement.
n The uncertainty may exceed the allowable tolerance for a particular
application.
n Therefore there is a need for a valid quantification of the uncertainty.
n For a downhole survey tool, the uncertainty is calculated using an
“error model”.

© 1998 Baker Hughes Incorporated


All rights reserved.

1
Types of error.

n Mistakes !
n Systematic.
n Random.

No error model will account for mistakes.

Current (‘98) error model does not account for random errors.

© 1998 Baker Hughes Incorporated


All rights reserved.

2
Mapping systems

n Wellhead and target will be defined in a global co-ordinate


system.
n Wellbore surveys are usually reported in local co-ordinates.
n Failure to ensure correct conversion and/or correlation
between mapping systems constitutes a gross error, and can
cause the target to be missed completely.

© 1998 Baker Hughes Incorporated


All rights reserved.

For drilling operations position is usually reported in local co-


ordinates, but at some time early in the planning phase it must be
ensured that the local wellbore co-ordinate system is related correctly
to the wider world. Similarly it must be ensured that all locations of
interest are referenced to a common mapping system.
Failure to get this right constitutes a gross error, that is an error that
is not included in planning presumptions, or in estimates of position
uncertainty.

Opportunities for error ocurr when data is transfered within the


operating company and between the operating company and
contractors such as seismic and directional.

The survey company is normally not a user or recipient of the


mapping data (other than convergence). It is normally the directional
contractor who must be aware, in order to correctly initiate its
database.

3
The effect of randomization.

A 10 per 100 error will grow as:


Depth Systematic Random
0 0 0
100 10 10
200 20 14
300 30 17
400 40 23
500 50 28
↓ ↓ ↓
1,000 100 32

© 1998 Baker Hughes Incorporated


All rights reserved.

4
Other positional errors

Our task is to drill from wellhead to target, but...

n The wellhead position will have some error.

n The target location will have some error.

© 1998 Baker Hughes Incorporated


All rights reserved.

How and when these uncertainties are be included depends on what


we are evaluating.
Both can affect the probability of hitting the target, but surface
location uncertainty much less so. It is more often associated with
anit-collision monitoring.
Many wells are drilled without taking either into account.

5
Benefits of More Accurate Surveys

n Closer drilling possible, less shut-ins


n Provides larger drilling target, cheaper drilling
n Greater chance of hitting target
n Better reservoir definition
n Easier relief well drilling

© 1998 Baker Hughes Incorporated


All rights reserved.

One approach to defining the target for the driller, is to shrink the
original geological target according to its uncertainty, then to further
shrink it based on wellbore survey uncertainty. This produces a
“driller’s target”. If the driller lands the well within it, we are confident
that the well has hit the actual target (with the specified degree of
confidence). This exercise demonstrates the benefits of accurate
surveys, giving a bigger target for the driller to hit, and therefore
reducing drilling time.

The design of a survey program that is “fit for purpose” in a strictly


limited sense can be a false economy. All factors must be
considered, including drillability and possible future requirements.

Better reservoir definition is not usually a consideration, but there is


surely an advantage in both improved validity of uncertainty
estimation, and in improved accuracy.
ie it is possible that position uncertainty may allow the target to be
intersected with sufficient confidence, but may be too large to allow
improved interpretation of the reservoir,
or, if the uncertainty is reported as being smaller than it actually is,
false interpretation of the reservoir model may ensue.

6
For directional survey data to be of real
use it must:

n be free from gross error.

n include a valid estimate of position uncertainty.

© 1998 Baker Hughes Incorporated


All rights reserved.

Unfortunately gross errors are not infrequent, and error models are
not always as valid as we would lke them to be.
The industry is working on improvements in both areas.

7
Development of an Error Model

• Develop model based on theory.

• Test for all conditions.

• Modify the model based on results.

• Retest.

• Enforce lab and field procedures so that


presumptions made in model apply to
operations.

© 1998 Baker Hughes Incorporated


All rights reserved.

8
Why an ellipse?

Error, or uncertainty, in one dimension.

© 1998 Baker Hughes Incorporated


All rights reserved.

9
Why an ellipse?

Error, or uncertainty, in a second dimension.

- +
X

© 1998 Baker Hughes Incorporated


All rights reserved.

10
Why an ellipse?

Two dimensional uncertainty?


_ +

© 1998 Baker Hughes Incorporated


All rights reserved.

11
Normal Distribution

-2σ -1σ 0 +1σ +2σ

© 1998 Baker Hughes Incorporated


All rights reserved.

12
Why an ellipse?

_
+

© 1998 Baker Hughes Incorporated


All rights reserved.

If the two eror sources are uncorrelated, it is unlikely that an extreme


error on new will coincide with an extreme error in the other.

13
Why an ellipse?

© 1998 Baker Hughes Incorporated


All rights reserved.

14
Why an ellipse?

© 1998 Baker Hughes Incorporated


All rights reserved.

15
The ellipse is a “contour”
drawn at a particular confidence level

© 1998 Baker Hughes Incorporated


All rights reserved.

16
The ellipse is not an area of even
probability

X X X O

© 1998 Baker Hughes Incorporated


All rights reserved.

There is a natural tendency to think that the well has equal probability
of being at any of the positions marked X, and zero probability of the
well being at point O, but this is not correct .

17
The ellipse is a “contour”
drawn at a particular confidence level

X X X O

© 1998 Baker Hughes Incorporated


All rights reserved.

18
Confidence levels.

1-D 2-D 3-D

1σ 68 39 20

2σ 95 86 74

3σ 99.7 98.9 97

© 1998 Baker Hughes Incorporated


All rights reserved.

19
Three Basic Directional
Measurements

n Measured depth errors affect vertical and radial uncertainty.

n Inclination errors affect radial and vertical uncertainty.

n Azimuth errors affect lateral uncertainty.

© 1998 Baker Hughes Incorporated


All rights reserved.

20
Position Uncertainty (2 sigma, half axes)
LAT 60, KO 800’, BUR 3/100, Hold 60, TD 13,900’md

MWD type hole direct’n 20 ° hole direct’n 70 °

lateral vertical lateral vertical

MTC 243’ 73’ 278’ 73’

MTC + sag 243’ 37’ 278’ 37’

MTC + sag 261’ 37’ 593’ 37’


+ mag
© 1998 Baker Hughes Incorporated
All rights reserved.

21
Position Uncertainty (2 sigma, half axes)
LAT 60, KO 800’, BUR 3/100, Hold 60, TD 13,900’md

TOOL Azimuth 20 ° Azimuth 70 °


Lateral / Vertical Lateral / Vertical
“Good Mag” 279 82 281 82

“Poor mag” 606 168 1495 168

NaviTrak 243’ 73’ 278’ 73’

NaviT + sag 243’ 37’ 278’ 37’

NaviT + sag + 261’ 37’ 593’ 37’


mag

© 1998 Baker Hughes Incorporated


All rights reserved.

22
Drillstring interference

The effect of axial interference increases,


generally, with:

n higher latitude
n higher inclination

n higher azimuth (towards east/west)

© 1998 Baker Hughes Incorporated


All rights reserved.

Axial interference is the magnetic interference caused by the


magnetic field that exists in the steel drillstring components.
Magnetic survey tools are positioned in non magnetic drill collars to
reduce this effect. Increasing non magnetic spacing to reduce the
effect increases cost and moves sensors further back from the bit
which makes steering the well more difficult.

23
Drillstring interference corrections

The uncertainty associated with such corrections tends to


increase with:

n higher latitude
n higher inclination

n higher azimuth (towards east/west)

© 1998 Baker Hughes Incorporated


All rights reserved.

It is possible to correct for drillstring interference. This allows


reduced NMDC length, and allows the sensor to be placed closer to
the bit, but the algorithms used have error (or uncertainty) inherent in
them. This error is a function of tools accuracy and of the accuracy
with which we know the local magnetic field.

The error model includes a term for drillstring interference, or for the
residual error from correction. At high angle, approaching east/west it
can be the dominant azimuth error term.

24
MAGCORR uncertainty

Azimuth uncertainty, North Sea Region


90

Azimuth Error (degrees) - 2 Sigma


10
80
5
70
Azimuth (degrees from N-

60
2
50
S)

40
1
30

20 0.5

10 0.25
0.1
0
0 10 20 30 40 50 60 70 80 90
© 1998 Baker Hughes Incorporated
All rights reserved. Inclination (degrees from vertical)

This chart indicates the azimuth uncertainty contribution from


applying our drillstring interference correction algorithm. Second
generation algorithms are now being deployed which might improve
on this performance.

The big advantage of such corrections is that they allow the sensor to
be placed much closer to the bit than would normally be the case.
They do not necessarily improve accuracy.

The residual error after correction can be significantly reduces if the


magnetic field parameters are obtained from a more accurate source
than a global model, eg aeromagnetic survey, site survey,
interpolated in-field reference, etc.

25
Separation Factor

r1 r2

C
SF =
r + r2
1

© 1998 Baker Hughes Incorporated


All rights reserved.

26
Separation Factor

c
SF =
c−e
© 1998 Baker Hughes Incorporated
All rights reserved.

27
Separation Factor

c
SF =
c−e

© 1998 Baker Hughes Incorporated


All rights reserved.

28
PROXIMITY CALCULATIONS

© 1998 Baker Hughes Incorporated


All rights reserved.
Maersk TS - 10/98 - 29

A combination of “while drilling” surveys (used to monitor and correct


drilling progress), and “multishot” surveys (used to update or confirm
the while drilling surveys).

“Geometric” used here to differentiate from geologic positioning.

29
PROXIMITY CALCULATION

n The customer must quantify risk (outcome).

n Based on risk, customer will specify acceptable


probability of intersection.

n Contractor plans and steers well accordingly.

© 1998 Baker Hughes Incorporated


All rights reserved.

30
PROXIMITY CALCULATION

n In the beginning there was centre to centre distance (c).


n Then we included position uncertainty and calculated
“ellipse to ellipse” separation (e).
n Then we introduced Separationn Factor (SF), a ratio of c and
e.
n The final goal is calculation of “probability of intersection”.

© 1998 Baker Hughes Incorporated


All rights reserved.

31
Separation Factor

r1 r2

C
SF =
r + r2
1

© 1998 Baker Hughes Incorporated


All rights reserved.

32
Separation Factor

c
SF =
c−e
© 1998 Baker Hughes Incorporated
All rights reserved.

33
Separation Factor

c
SF =
c−e

© 1998 Baker Hughes Incorporated


All rights reserved.

34
PROXIMITY CALCULATION

“C” 50 100

50 100

© 1998 Baker Hughes Incorporated


All rights reserved.

35
PROXIMITY CALCULATION

“E” 50 100

10 30

© 1998 Baker Hughes Incorporated


All rights reserved.

36
PROXIMITY CALCULATION

“SF” 50 100

1.25 1.43

© 1998 Baker Hughes Incorporated


All rights reserved.

37
Proximity Calculations.

Value Well 1 versus 2 Well 2 versus 3


C 50 100
E 10 30
SF 1.25 1.43

© 1998 Baker Hughes Incorporated


All rights reserved.
Maersk TS - 10/98 - 38

38
PROXIMITY CALCULATION

n BP pioneered the use of probability of intersection - QRA.

n They have now backed of its use in critical situations


because:

– “Mistakes” are too common for uncertainty distributions to be


sufficiently valid.

– Differences between the actual and calculated unceratinty


distributions give rise to large dimensional differences at small
probabilities.

© 1998 Baker Hughes Incorporated


All rights reserved.

39
Target versus survey uncertainty

Survey program A Survey program B

Target Target

© 1998 Baker Hughes Incorporated


All rights reserved.

Survey program B may cost several thousand $ more than program


A, and since program A appears to be fit for purpose, there is no
need to incur the additional cost of specifying program B.

40
Driller’s target

Survey program A Survey program B

Target

Target

© 1998 Baker Hughes Incorporated


All rights reserved.

Reducing the geological target by the dimensions of the survey


uncertainty generates a “Driller’s target”. This is what the driller must
hit to ensure that the geological target is hit with the required degree
of confidence.
The additional cost of program B may be deemed worthwhile when
the drilling cost (predominately rig time) of hitting the smaller driller’s
target generated by program A is taken into account.

41
Survey Tool Performance
can be Sensitive to:

n Location
n Inclination
n Azimuth
n Dogleg
n Time
n Duration
n Tempeature
n Etc.

© 1998 Baker Hughes Incorporated


All rights reserved.

42
Depth Control

lGyro survey.
Measured depth error 0.5/1000.
Cumulative undertainty at 5000 = 2.5

lMWD.
Measured depth error 2.0/1000.
Cumulative uncertainty 5000 to 10000 = 10.0

© 1998 Baker Hughes Incorporated


All rights reserved.

43
Depth Control

Cumulative uncertianty at 10000:

• Random summation (EC*TRAK) = 10.3

• Systematic summation = 12.5

• Actual uncertainty in real time = 20.0

© 1998 Baker Hughes Incorporated


All rights reserved.

44
Wellbore Uncertainty Estimates

Clearance Factor
n Purely geometrical computation

n Relies entirely upon ‘100%’ assumption

n Untenable with ‘100%’ ellipses

n Cannot be used for QRA

n Require alternative to CF for QRA

© 1998 Baker Hughes Incorporated


All rights reserved.

45
Definition of EC*Trak error term

n Type
n Mean asumed to be zero
n Variance
n Station to station correlation asumed to be 1
(ie systematic)

© 1998 Baker Hughes Incorporated


All rights reserved.

46
Definition of an error term

n Type
n Mean
n Variance
n Correlation station to station
n Correlation survey to survey
n Correlation well to well

© 1998 Baker Hughes Incorporated


All rights reserved.

47
Example horizontal well.

Program 4000m MD 6000m MD


Lateral / Vertical Lateral / Vertical
MWD 69 22 105 36
MWD sag 69 11 104 17
MWD sag/spaced 80 11 163 17
MWD sag/mag 76 11 619 17
Seeker+MWD 42 6 142 11
1
RIGS+MWD 36 5 141 10
2
RIGS+MWD 32 5 135 10
RIGS+MWD3 32 5 132 10

© 1998 Baker Hughes Incorporated


All rights reserved.

Well is at latitude 61N, profile has large azimuth turn out of 13 3/8”
shoe, horizontal section is almost due east.
4000m is approx heel of the horizontal, 6000m is the toe.
Casing shoe at 3000m is taken as deepest gyro survey.
Uncertainty values are at a 2 sigma confidence level.

MWD1 = MWD with correct spacing input and sag corrected


MWD2 = MWD magnetic interference and sag corrected
MWD3 = MWD with correct spacing input, sag corrected, and
declination improved by eg, aeromagneic survey.
MWD4= MWD with correct spacing input, sag corrected, and
declination improved by eg, in-field referencing.

The uncertainty values, relative to each other, are not intuitive. this is
typical of more complex wells. A different profile may indicate much
greater benefit from switching Seeker for RIGS, or may indicate that
mag correcting the MWD is a viable option.

48
Geometric versus Geologic position.

n Both estimates of position are of value.

n A combination of both should be used to best define the position of


the wellbore.

n Uncertainty relative to wellhead and to geological markers are both of


interest.

n We may provide more than one estimate of wellbore position, tailored


to the end use.

© 1998 Baker Hughes Incorporated


All rights reserved.

Geometric data can be tied to geological markers to reduce


uncertainty relative to the target.
At present we tend not to evaluate both sets of data together. We
need to get better at understanding how to combine the two. Better
communication between the Survey/Directional company and G&G
departments is required.

49
The Future?

n Industry Steering Committee on Wellbore Survey Accuracy.

n BP, INTEQ, STATOIL, Sysdrill consortium.

n Sysdrill “Director”.

© 1998 Baker Hughes Incorporated


All rights reserved.

50

You might also like