Log Editing

8/13/2019 Log Editing

1/124

DATA QUALITY, EDITING,DATA QUALITY, EDITING,

AND RECONSTRUCTIONAND RECONSTRUCTION


2/124

WIRE.GR_8

GAPI0 200

WIRE.CALI_2

MM150 400

WIRE.CALS_1MM150 400

2350

2400

2450

2300.0

DEPTH

METRES

WIRE.DT_1US/M700 100

WIRE.RHOB_1K/M31950 2950

WHY DO WE EDIT LOG DATA?WHY DO WE EDIT LOG DATA?





Sonic Density

DONT THE LOGS

MEASURE WHATTHEYRE DESIGNEDTO MEASURE?

2500

2550

2600

2650

2700

2750

2800

28502875.0

RCHD

TGLU

RCHD

TGLU

RCHD

TGLU

RCHD

TGLU

For best results, it requires some

knowledge experience

common sense, and ALL CURVES !


3/124

BASIC EDITING WORKFLOWBASIC EDITING WORKFLOW

LOAD ALL DATA

UNDERSTAND WHAT YOUVE GOT

MERGE LOGGING RUNS (CREATE COMPOSITE LOGS)

ENVIRONMENTAL CORRECTIONS (if necessary)

NORMALIZATION (if necessary)


NORMALIZATION (if necessary)


4/124

BASIC EDITING WORKFLOWBASIC EDITING WORKFLOW

IDENTIFY AND FLAG BAD DATA

DEPTH SHIFTS

ADD

ATA

GENERATE PSEUDO DATA

REPLACE BAD DATA WITH PSEUDO DATA

MULTIWELL TREND PLOTS AND HISTOGRAMS TO QC

IDENTIFY

ANDR

EPAIR

B


5/124

UNDERSTAND WHAT YOUVE GOT

SOME KEY QUESTIONS:

HOW MANY LOGGING RUNS?

UP-LOGS OR DOWN LOGS?

MUD LOGS AVAILABLE?

CORE DATA AVAILABLE?

CASING POINTS

AVAILABLE CURVES UNITS


6/124

Some common LAS naming practices:

MP, MAINMAIN PASS

UNDERSTAND WHAT YOUVE GOT LAS FILE NAMES

,

TVD, TV TOTAL VERTICAL DEPTHHR, H HIGH RESOLUTION DATAAIT, DT TOOL STRINGS OR LOGS

Load everything in measured depthand convert to TVD using the directionalsurvey data.

Stay away from files measured in TVD; often mislabeled in LAS

NOTE THE FOLLOWING:


7/124

Much of what we need to know for a petrophysicalanalysis can be found in the log header

information. Key pieces of information include KBelevation, mud type, Rmf, and casing points.

When working with neutron log data, we need to

know, prior to our analysis, the type of neutron tooland the matrix on which the data was recorded.Sometimes this is clearly indicated in themnemonic (e.g., NPLS, NCNPL), but more often

than not, we need to look in the LAS file to find thisinformation. Oftentimes we actually need to refer

to the hard copy of the log to find this information.

Density is also frequently presented as a porositycurve, in which case we need to know the matrixand fluid densities used to calculate the density

porosity.


8/124

LOAD ALL DATA

Load all data from all logging runs

Either organize the data in a coherentdatabase, or name the curves in such a

run

Load repeat data!

EXAMPLE


9/124

LOAD ALL DATA

Exercise!!!

CLASS PROJECT

EVALUATE LAS FILES

LOAD BOTH LOGGING RUNS TOPOWERLOG/HRS???


10/124

LOAD ALL DATA

DEPTHM

A_GRGAPI0 150

B_GR

GAPI0 150

B_C13

MM0 500

A_C13

MM0 500

A_RDOHMM0.2 200

B_M2R9

OHMM0.2 200

A_DT24QSUS/M500 0

B_DT24QS

US/M500 0

B_PORZSSPU40 -10

B_CNCSS

PU40 -10

100

200300

400

500

600

700

800

900

QUESTIONS:

1) WHERE ARE THE PROBLEM DATA?2) WHY ARE THERE NEGATIVE POROSITIES?

1100

1200

1300

1400

1500

1600

1700

1800

1900

2000

21002200

2300

2400

2500

2600

2700

2800

2900

3000

3100

3200

3300

3400


11/124

Merge log runs to create single curves to be used

In multiwell projects, compare curves well-to-wellto check for differences in logging contractors, tool

MERGE LOGGING RUNS (CREATE COMPOSIT LOGS)

ca rat on, too type v ntage, an ot ermeasurement differences.


12/124

RUN1.GR_1

GAPI0 200

RUN2.GR_1

GAPI0 200

SP

MV-160 40

RUN2SONIC.GR_1

GAPI0 200

RUN1.RHOB_1

K/M31950 2950

RUN2.RHOB_1

K/M31950 2950

RUN2.PE_1

B/E0 10

RUN1.PE_1

B/E0 10

SP

MV-160 40

RUN1.CAL2_2MM150 400

RUN2.CAL2_1MM150 400

RUN1.CALI_1MM150 400

RUN2.CALI_1

MM150 400

SP

MV-160 40

RUN1.DT_1

US/M500 100

RUN2.DT

US/M500 100

RUN2SONIC.DT_1

US/M500 100

SP

MV-160 40

3375

3350.0

DEPTHMETRES


3400

3425

3450

3475

3500.0

When merging curves, it is usually best to take theuppermost curve down as far as reasonable. Havingbeen logged earlier in time, it is less likely to beimpacted by deteriorating borehole conditions andadditional invasion.

Merge at a point where logs are reading same,whenever possible.

Casing point


13/124

EVALUATE AND DISCUSS CASING POINT (see nextslide)


MERGE LOGGING RUNS


14/124


DEPTH

M

A_GR

GAPI0 150

B_GRGAPI0 150

B_C13MM0 500

A_C13MM0 500

A_RD

OHMM0.2 200

B_M2R9OHMM0.2 200

A_DT24QS

US/M400 100

B_DT24QSUS/M400 100

B_PORZSS

PU40 -10

B_CNCSSPU40 -10

SPL_GR

API0 150

1525

1550

GAMMA RAY MERGE

- WHAT IS WRONG WITH THIS?

MERGE POINT

1575

1600

1625

1650

1675

IS THE DATA ACROSSTHIS ZONE USEFUL?WHY OR WHY NOT?


15/124

MERGE THE REMAINDER OF THE CURVES


As with anything in petrophysics, I like to be very consistent (if possible) with mynaming conventions. When I splice curves together, I usually prefix the newcurve with an SPL_*. If I do multiple splices during the editing process, I add a

version number to the splice. For instance, if I do further splice work on the GRcurve, the new curve will become SPL2_GR. If you are reasonably consistent, itwill be relatively easy to remember what you did at a later date.


16/124


CONTRACTOR SPECIFIC CORRECTIONS

CORRECT FOR BOREHOLE CONDITIONS

Most of us dont typically apply environmental corrections. There are numerous reasons for this, but ingeneral it is because the magnitude of the corrections are small, and the uncertainties in how they areapplied are large (most frequently, we dont even have the requisite data to apply the corrections).

GR is about the only log I will sometimes correct, simply because the changes can be large andimportant. Note, however, that the corrections to the GR can often be too large!

HAVE SIMPLY DIGITIZED THE PUBLISHED

CHARTS


17/124

DEPTHFT

DCALIN-5 5

DCAL 0

GRGAPI0 150

CORE_GR

API0 150

GRCapi0 150

DSCOREGR

API0 150

10700

10710

10720

10730

ENVIRONMENTALCORRECTIONS TO GR

MINOR DEPTH SHIFT OFCORE DATA


10740

10750

10760

10770

10780

10790

10800

10810 RAW DATA CORRECTED GR

AND DEPTH-

SHIFTED CORE

DATA


18/124


19/124

DEPTH SHIFTS

DEPTH

M

B_GR

GAPI0 150

B_C13MM0 500

( )0 500

B_M2R9

OHMM0.2 200

B_DT24QS

US/M400 100

B_PORZSS

PU40 -10

B_CNCSSPU40 -10

2520

2530

2540

WHICH CURVE IS OFF DEPTH?HOW DO YOU KNOW?

2550

2560

2570

2580

2590

2600

2610


20/124

For effective depth shifting, pick a reference curve you believe (usuallyshallow resistivity or first-run GR: has good character and vertical

resolution)

Be careful of too many shifts up and down. Fewer is better. As withseismic data, you can get off a cycle if youre not careful.

DEPTH SHIFTS

NOTE: DEPTH-SHIFTING CANNOT CURRENTLY BE DONE IN H-R

CLASS EXCERSIZE: WE WILL SCROLL DOWN THROUGH THELOG, AND INTERACTIVELY DEPTH SHIFT THE SONIC LOG


21/124


BADD

ATA

IDENTIFICATION AND REPAIR OFIDENTIFICATION AND REPAIR OF

BAD DATABAD DATA


REPLACE BAD DATA WITH PSEUDO DATAIDENTIFY

ANDR

EP

AI


22/124

KEY POINT: tool readings do not reflect the properties of the formation

Although the terms bad data and badhole are not particularly scientific, nor

perhaps grammatically correct, they are the common terminology among log

analysts. Be aware when conversing with log analysts, however, as they will oftendifferentiate between truly bad data (i.e., tool failure), vs. valid tool readings in poor

borehole conditions. As seismic petrophysicists, we want to edit the log data suchthat it represents what we believe to be the true formation properties.

WHAT IS BAD DATA?WHAT IS BAD DATA?

We will look at some examples and discuss:

Causes some typical reasons for bad data- (obvious things you should always check first)

Recognition practices to identify poor quality data- (some ways to recognize and isolate bad data)

Edits and Reconstruction useful methods to repair curves

We will focus on the two slowness curves and density


23/124

12700

12800

12900

13000

13100

Type G sand

Logrun break clearly indicated by

bitsize change, caliper, straightlined curves

Resistivity and neutron-densityseparation suggest no sand here

THE EASIEST: CASING POINT PROBLEMSTHE EASIEST: CASING POINT PROBLEMS

13200

13300

13400

13500

13600

13700

13800

13900

ALWAYS REFER TO LOGHEADER INFORMATION!

BEWARE OF CASING POINT

PROSPECTS!


24/124

COMMON LOG DATA PROBLEMSCOMMON LOG DATA PROBLEMS

BS_1MM150 400

CALI_2

MM150 400

GR_2

GAPI0 200

SP_1

MV-160 40

3125

3115.0

DEPTHMETRES

DT_2

US/M500 100

NPHI_1

V/V0.45 -0.15

RHOB_2

K/M31950 2950

PEFZ_1

B/E0 10

SFL

OHMM0.2 2000

ILM

OHMM0.2 2000

AF90_1

OHMM0.2 2000

LLD

OHMM0.2 2000

Tool problems (dead)

Log run breaks

Hole conditions

Cycle skips

3150

3175

3200

3215.0

Tool pulls Casing

Mud and mud cake

Digitizing errors

Depth shifts


25/124

BSMM150 400

CALI_1MM150 400

GR_1

GAPI0 150

800

775.0

DEPTHMETRES

DT_1

US/M500 100

DTSM

US/M1200 200

WIRE.DEN_1

K/M31950 2950 .

.

.

Tool problems

Log run breaks

Hole conditions

Tails

Cycle skips

Tool pulls

Poor hole

conditions


825

850

875

900.0

as ng

Mud and mud cake

Digitizing errors

Depth shifts

Cycle Skips

In this example, we see poor holeconditions affecting the density logs. Thesame hole conditions are causing cycleskipping on this sonic data.

WHAT IMPORTANT QUESTIONSHOULD WE BE ASKING ABOUT THE

REASONS FOR THE BOREHOLEPROBLEMS?


26/124


27/124

BS_1MM150 400

CAL2_2

MM50 300

CALI_2

MM50 300

GR_2

GAPI0 200

VSH_GR_1V/V0 1

3665.0

DEPTHMETRES

NPHI_2V/V0.45 -0.15

RHOB_2

K/M31950 2950

DRHO_2

K/M3-400 100

PE_2

B/E0 10

Tool problems

Log run breaks

Hole conditions

Tails

Cycle skips


3700

CasingMud and mud cake

Digitizing errors

Depth shiftsBarite mud filling fractures (or rugosityalong the borehole wall) can result in

very high RHOB and PEF readings.


28/124

Recommendations for badhole identification:Recommendations for badhole identification:

Borehole and measurement quality indicators

Log header information for log run breaks Reasonable log values and frequency content

Comparison to other data using crossplots andhistograms

RECOGNIZING BAD DATARECOGNIZING BAD DATA

. .,

Trend plots and cross-plots Empirical curves or regression equations

Also look for consistency amongst specific lithologies(i.e. do the curves make sense collectively!)

Our ability to recognize poor quality log data will depend

on our understanding of the following:

local geology

tool responses, and the borehole environment


29/124

1600

300

400

500

600

700

800

900

UNDERSTANDING THE GEOLOGY WILL HELP TOUNDERSTAND QUALITY vs. BAD LOG DATA.

ALWAYS LOOK AT LOGS USING A VARIETY OF VERTICAL SCALES!ALWAYS LOOK AT LOGS USING A VARIETY OF VERTICAL SCALES!

1700

1000

1100

1200

1300

1400

1500

1600

1700

1800

How much of this spiky data is bad?

Closer examination of the data shows that much of thespikiness may be due to geology. Editing out thesespikes would be inappropriate!


30/124

DEPTH

FT

A_GR

GAPI0 200

A_HCAL

IN6 16

A_HCGR

GAPI0 200

A_RLA5

OHMM0.1 10000

A_RLA3

OHMM0.1 10000

A_RLA1

OHMM0.1 10000

RAW_RHOB

g/cc1.95 2.95

A_NPHI

CFCF0.45 -0.15

A_PEFZ

0 10

RAW_VP

ft/s10000 25000

500

1000

1500

2000

2500

VALID DATA?VALID DATA?

3000

3500

4000

4500

5000

5500

6000

6500

7000

NOTE:

Washout on caliperAnomalously low density valuesVelocity drop


31/124


32/124


33/124

NOTICE THAT DENSITY IS NOTNECESSARILY BAD EVERYWHERETHERE IS WASHOUT OR LARGE

DENSITY CORRECTIONS. BE

BE CAREFUL WHEN AUTOMATINGTHE PROCESS OF CREATING A

BADHOLE FLAG!

CALIPERDRHO

HOLE QUALITYHOLE QUALITY

Be careful when Drho > 100 kg/m3

(+) correction when mud and large holesize cause density to read too low(common)

() correction when heavy mud or smallborehole diameter

Sonic appears tobe largely

unaffected by thewashouts


34/124


35/124

DEPTH

FT

GR

GAPI0 150

DCAL

IN-5 5

AT90

OHMM0.2 2000

AT30

OHMM0.2 2000

AT10OHMM0.2 2000

RAW_RHOB

g/cc1.95 2.95

NPHI

V/V0.45 -0.15

PEFZ----0 20

HDRA

G/C3-1 1

RAW_VP

km/s2 7

1000

1500

2000

2500

3000

3500

4000

4500

5000

BADHOLE from automated process

Hand-picked BADHOLE


36/124


37/124

COMMON PROBLEMS

Digitizing errors

Differences in tool type / contractor / vintage

REPAIR TECHNIQUES

OTHER SOURCES OF BAD DATAOTHER SOURCES OF BAD DATA

IF POSSIBLE, USE MULTIPLE WELLS IN AN AREA TO DETERMINE WHETHER OR NOTIF POSSIBLE, USE MULTIPLE WELLS IN AN AREA TO DETERMINE WHETHER OR NOTYOUR LOG CURVES ARE RECORDING REASONABLE VALUES.YOUR LOG CURVES ARE RECORDING REASONABLE VALUES.

Re-digitize

Normalization

Rescaling


38/124

2450

2500

2550

2600

2650

2700

2750

2400.0

TGLU

DT LOOKS NORMAL

100

160

220

280

340

400

460

520

580

640

700

0 0

500 500

1000 1000

1500 1500

2000 2000

2500 2500

DEPTH vs. WIRE_INT5.DT_SMT CrossplotWell: 7 Wells

Range: All of WellFilter:

(METRES)

4354

4354

0

0

0 0

Beware can be hard to catch.DIGITIZING ERRORSDIGITIZING ERRORS

2800

2850

2900

2950

3000

3050

3100

3150

3200

3250

3312.5

100

160

220

280

340

400

460

520

580

640

700

3000 3000

3500 3500

4000 4000

4500 4500

5000 5000

DE

PTH

WIRE_INT5.DT_SMT (US/M)

Well Legend:300C426930134450 300D556930134450300F436930134450 300G336930134450300H066930135000 300H546930134450300P036930135000

Functions:

ri_ne_opress : No description given.

xtrend_ri_ne : No description given.

Sometimes log data looks normal in terms of frequency content and correlation

to other logs but can be much different when compared to offset well data.

Digitizing errors can be difficult to identify using only the log display.

Always go back to the hardcopy logs for verification !

LOOK FOR SCALE CHANGES THAT THE DIGITIZER MAY HAVE

MISSED!


39/124

500

1500

2500

3500

4500

5500

6500

7500

500 500

1000 1000

1500 1500

2000 2000

DEPTH vs. 1000000/ESSO_INT.DT CrossplotWell: 7 Wells

Range: All of WellFilter:

S)

7001

7277

275

0

0 1 These erroneous logs can hang around fora long time particularly with the multipledata sources available to geoscientists

Heres a number of velocity logs still beingused today although they were drilled in

VELOCITYVELOCITY--DEPTH CALIBRATIONDEPTH CALIBRATION

500

1500

2500

3500

4500

5500

6500

7500

2500 2500

3000 3000

3500 3500

4000 4000

4500 4500

5000 5000

DEPTH

(M

ETR

1000000/ESSO_INT.DT (US/M)

Well Legend:300C426930134450 300D436930134450300D556930134450 300G336930134450300H066930135000 300H546930134450300P036930135000

t e m s nto t e . c ag u e

in the Mackenzie Delta

SHOULD DATA BE HUNG

STRUCTURALLY ORSTRATIGRAPHICALLY?


40/124

As we have seen, many logs measure

similar properties of the formation (e.g.matrix, porosity), and we expect there tobe reasonable correlation between them.Log analysts use the collective behavior

CALI_RES_1MM100 350

GR_2

GAPI0 200

SP_5MV-25 75

780

785

790

775772.0

DEPTHMETRES

NPHI_2

V/V0.6 0

RHOB_2

K/M31650 2650

PE_3B/E0 10

DT_2

US/M500 100

DTSM_1

US/M2400 200

SFL_1

OHMM0.2 200

ILM_1

OHMM0.2 200

ILD_1OHMM0.2 200

PATTERN MATCHINGPATTERN MATCHING

evaluation as will we. Qualitativeinterpretations help to identify log datathat just doesnt fit our expectations.

795

805

810

815

820

830

835

800

825

839.5


41/124


BADD

ATA

IDENTIFICATION AND REPAIR OF BAD DATAIDENTIFICATION AND REPAIR OF BAD DATA


REPLACE BAD DATA WITH PSEUDO DATAIDENTIFY

ANDR

E

PAI


42/124

GENERATING PSEUDO DATAGENERATING PSEUDO DATA

When repairing bad data, we often times use simple hand-patches. However, this approach becomesunwieldy when trying to process thousands of feet of data. Therefore, it is frequently useful to generatepseudo data. This is generally done via application of an empirical transform, or via application of multilinearregression or neural network analysis. Empirical relationships typically use one log to compute an unknown.For example, Gardners famous relationship uses P-wave velocities to estimate density. Although the

empirical transforms are often very useful, I prefer to use multilinear regression when generating pseudo

data. This allows the user to include more than one curve type when estimating an unknown.

In this section we will focus on practical (meaning time-efficient) methods of editing sonic and density data.We will cover the empirical relationships, but also application of multilinear regression and neural net analysis.

We will also explore the various options that Hampson-Russell gives the user for generating pseudo data.


43/124

EMPIRICAL RELATIONSHIPS

Compressional Velocity

Faust: Dt = 513.3 / (Depth*Rt)**(1/6)

Density

Gardner: Rhob = 0.23 * Vp**(1/4) (Vp in ft/s; rhob in g/cc)Gardner-Castagna: Rhob = fcn(Vp,Vsh)

Shear VelocityGreenburg-Castagna coefficients


44/124

affected by local stresses and formation properties

also affected by invasion and near-wellbore alteration

these issues will be discussed later today and tomorrow morning

USED TO ACQUIRE FORMATION VELOCITIES

SONIC LOGSSONIC LOGS

Schematic of Schlumbergers long

spaced sonic tool. Taken from

Bassiouni.

Note that sonic tools are run centered

(ostensibly) in the wellbore.

MONOPOLE (primarily P-wave velocities) ANDDIPOLE SOURCES (P- and S-wave velocities)

NOTE: sonic logs can be thought of as refraction devices. Thus, you need to have

a critical angle to receive a signal (we will discuss critical angle later in the class).

What does this mean for the lower limit of velocities we can detect with sonic tools?

transmitters emit a pulse, which is recorded by an array of receivers

records transit time between receivers


45/124

TWO METHODS FOR MEASURING SONIC TRANSIT TIMES:

First arrival picking (older tools like BHC, LSS)

Full waveform capture entire waveform from which various arrivals are picked.

Compressional wavestravel parallel to the direction of particle motion. They readthe rock matrix, fluid in the rock, and elements of pore structure.

Shear wavestravel perpendicular to the direction of particle motion. They measure


the rigidity of the formation and are influenced by the rock matrix and elements of

pore structure.

Tube waves(or Stonely waves) travel along the wellbore

FIRST ARRIVAL PICKING


46/124


From Tang and Cheng, 2004

ADVANTAGES OF ARRAY PROCESSING:

- different moveouts for different modes

- better processing and results

FULL WAVEFORM ARRAY PROCESSING


47/124

f

=

= wavelength

= formation velocityf= frequency

DEPTH OF INVESTIGATIONIS RELATED WAVELENGTH, WHICH IS

RELATED TO THE FREQUENCY OF THE MEASUREMENT AND

THE FORMATION VELOCITY


2000.0 4000.0 6000.0 8000.0 10000.0 15000.0 20000.0

6000.0 3.0 1.5 1.0 0.8 0.6 0.4 0.3

8000.0 4.0 2.0 1.3 1.0 0.8 0.5 0.4

10000.0 5.0 2.5 1.7 1.3 1.0 0.7 0.5

12000.0 6.0 3.0 2.0 1.5 1.2 0.8 0.614000.0 7.0 3.5 2.3 1.8 1.4 0.9 0.7VEL

OCITY

(ft/s)

FREQUENCY (Hz)

Commonly encountered ranges for the DWGOM

If low frequency data can be acquired, the sonic data will be less affected by invasion.

While sonic data is less susceptible to well-bore washout than density data,


48/124

it still frequently requires considerable editing. Editing of sonic data usuallytakes the form of correction of cycle skips, and de-spiking noisy data.SONIC LOGSSONIC LOGS

WHAT DOES BAD SONIC DATA LOOK LIKE? abundant spikes (fast and slow) very noisy (abnormally high frequency)

too fast for the area

CYCLE SKIPS (common in older logs) acoustic wave is attenuated below the threshold of the receive abnormally long travel times (records Vp that is too slow)

WHERE ARE CYCLE SKIPS LIKELY TO OCCUR? thin beds with large velocity contrasts gas sands gas-cut mud

poorly consolidated formations fractured formations


49/124


FAST SPIKES (noise) receivers record early noise arrivals results in apparent fast rock

EDITING CYCLE SKIPS AND NOISE simple approach is to simply filter the data generate pseudo-sonic data

neural networks and multilinear regression hand edits

WHEN ARE SPIKES NOT CYCLE SKIPS OR NOISE? must be careful to not edit lithology

coal

hard streaks (e.g., carbonate interbeds) look for lithology indicators compare behavior of resistivity, neutron, and density to sonic data.

Do they move in the same directions?


50/124

Cycle skipsNoise

Note the noise in the resistivity and the density data. Some of the

spikes in the sonic data may be reflecting true lithology effects

across this zone.


51/124

SIMPLE SMOOTHING

RAW DATARAW DATA


52/124

DEPTHM

TVDM

GRGAPI0 150

HCALMM100 300

LCALMM100 300

AHT90OHMM0.2 2000

AHT30OHMM0.2 2000

AHT20OHMM0.2 2000

VP_RAWft/s10000 20000

DPHI_RAWv/v0.4 -0.1

NPHI_SSv/v0.4 -0.1

2750

2760

2770

2780

2710

2720

2730

RAW DATARAW DATA

SIDERITE

My general approach is to generate

pseudo-sonic data for editing purposes.

Usually I do this via multilinear

regression. The best answers I usually

get involve resistivity, neutron, depth, and

some lithology indicator.

CADOTTE

2790

2800

2810

2820

2830

2840

2850

2860

2870

2880

2740

2750

2760

2770

2780

2790

2800

2810

2820



53/124

DEPTHM

TVDM

GRGAPI0 150

HCALMM100 300

LCALMM100 300

AHT90OHMM0.2 2000

AHT30OHMM0.2 2000

AHT20OHMM0.2 2000

VP_RAWft/s10000 20000

VP_PSE10000 20000

1400

1500

1600

1700

1400

1500

1600

1700

Multilinear regression

Typical data used for Vp

estimation:

oResistivity

oNeutron porosity


DUNVEGAN

SHAFTSBURY

CADOTTE

1800

1900

2000

2100

2200

2300

2400

2500

2600

2700

2800

1800

1900

2000

2100

2200

2300

2400

2500

2600

2700

o N-D difference

oVshale

SONIC EDITSSONIC EDITS


54/124

19000

00

16000

170

00

18000

1.2

0.8

1

SONIC EDITSSONIC EDITS

BAD SONIC DATA

VP_PSE2

9000 1900010000 11000 12000 13000 14000 15000 16000 17000 18000

VP_

RA

W

9000

10000

11000

12000

13000

14000

15

VCLGR

0

0.2

0.4

0.6

R2 = 0.88

EDITED SONIC DATAEDITED SONIC DATA


55/124

EDITED SONIC DATAEDITED SONIC DATA

DEPTH

M

TVD

M

GR

GAPI0 150

RDEEP

ohmm0.2 2000

RMEDohmm0.2 2000

RSHAL

ohmm0.2 2000

VP_RAW

ft/s10000 20000

VP_FINALft/s10000 20000

2780

2790

2800

2740

2750

There are multiple reasons why the sonicdata in the Cadotte might not be valid in thiswell. However, from multiwell analysis, weknow that the recorded velocity across theCadotte is invalid. The velocity estimatedusing multilinear regression is within the

expected range of values, as determinedfrom analysis of the other wells in the field.

2810

2820

2830

2840

2850

2860

2870

2880

2760

2770

2780

2790

2800

2810


56/124

NOTE BAD DATA AT CASING POINT.

DO THE OTHER LOGS SUPPORT THIS VELOCITY?

LOOK FOR CONSISTENCY WITH OTHER CURVESLOOK FOR CONSISTENCY WITH OTHER CURVES

SONIC LOGSSONIC LOGS EMPIRICAL TRANSFORMSEMPIRICAL TRANSFORMS


57/124

SONIC LOGSSONIC LOGS -- EMPIRICAL TRANSFORMSEMPIRICAL TRANSFORMS

Besides transforms such as Faust, various relationships exist that relate

velocities to porosity. The most famous, of course, is Wileys time-average equation, which treats the velocity as a volume weightedaverage of its components. All of these empirical porosity-velocitytransforms have some uses, but by and large they do a poor job ofpredicting velocities.

As we will see when we discuss seismic rock properties, the reasonsthese transforms typically dont work well are numerous. Nonetheless,we present them here, as they locally may help you create pseudo sonicdata.

VELOCITY SYSTEMATICSVELOCITY SYSTEMATICS


58/124

Wyllie (1956, 1958, 1963)( )11

+=

The most commonly used relationships are either

heuristic or empirical, and are often times based

on limited data sets. Note that these models are

essentially a linear mass balance of the velocities

of each constituent in the rock.


matrixVpfluidVpVp ,,

Raymer et al., 1980

fluidVpmatrixVp1Vp 2 ,,)( += - for porosities < 37%



59/124

Han, 1986

068086Vp .. = 286064Vs .. =

- clean sandstones at 40 MPa (~5800 psi)

- shaley sandstones at 40 MPa (~5800 psi)

NOTE: velocities are in km/s

and porosities and clay content


...

C891914523Vs ... =

Eberhart-Phillips, 1989

( )eP716e 01P4460C731946775Vp.

.....

+=

are fractional values

NOTE: velocities are in km/s

and porosities and clay content

are fractional values. Pressure

is in kilobars

( )eP716

e 01P3610C571944703Vs

.

.....

+=



60/124

Castagna et al., 1985

C212429815Vp ... =

- shaley sandstones of the Frio Formation (log-based)


C042077893Vs ... =

NOTE: other Vp-porosity relationships have also been published. In addition, Vp-porosity relationships can also be derived

from theoretical models. Model-based results, however, often predict velocities which are much faster than the measureddata.



61/124

20000

25000

Vshale = 0

Vshale = 0.2Eberhart-Phillips

/s)

Porosity decrease due to cementation


0

5000

10000

0 0.05 0.1 0.15 0.2 0.25 0.3

POROSITY

VELOCITY(

ft

Porosity decrease due to increasing clay content



62/124

20000

25000

Vshale = 0

Vshale = 0.25

Vshale = .5

MARLIN A5 PAYCastagna, Vsh = 0

Castagna, Vsh = 0.5

s)


Note that E-H and Castagna are similar in form.

However, each model yields different results, especially

when porosities are greater than approximately 10%. This

highlights the need to locally calibrate whenever possible!

Note the difference in porosity values for a

velocit of 10 000 ft/s. These t es of

0

5000

10000

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

POROSITY

VELOCIT

Y(

ft

SHALE

uncertainties are one reason why empirical

porosity-velocity models are difficult to use.

What is the dominant control on porosity reduction?



63/124

Invasion-corrected P-wave velocity


Sonic data can be affected by

invasion, especially when thereservoir is charged withhydrocarbons. Unfortunately, theeffects of invasion on sonic dataare difficult to detect and quantify.Correcting sonic data for invasion is

also complex, and loaded withassumptions. We will discuss sonicinvasion corrections in ourdiscussion on fluid substitutionpitfalls.


64/124


65/124

DENSITY LOGSDENSITY LOGS


66/124

PAD TYPE DEVICE strongly affected by well-bore washout also affected by invasion

EDITS ARE ALMOST ALWAYS REQUIRED! fortunately, density is easier to edit and estimate than sonic data ALWAYS closely evaluate the density you are using for rock

proper y wor pay close attention to the caliper

NOTE: Not all wellbore wash-out will result in bad density data is dependent upon the rugosity of the well-bore. if the washout is large, yet the well-bore is smooth, it is possible to

get good density readings.


67/124

EDITING DENSITY LOGSEDITING DENSITY LOGS


68/124

Typical approaches include the following:

- Gardners relationship (Gardner et al., 1974)- Castagnas modification of Gardner (Castagna et al., 1993)

- neural networks- multilinear regression- cross-plotting of local data

,

Neither method works particularly well in poorly consolidated rocks

25.23.0 VPb =

Gardner et al., 1974

VP is in feet/second

25

b VP7411..= VP is in km/second



69/124

LITHOLOGY a b c

Shale -0.0261 0.373 1.458

Sandstone -0.0115 0.261 1.515

Coefficients for the equation rhob = aVP^2 + bVP + c

CASTAGNA et al., 1993:

- . . .

Dolomite -0.0235 0.390 1.242

Anhydrite -0.0203 0.321 1.732

LITHOLOGY d f

Shale 1.75 0.265

Sandstone 1.66 0.261

Limestone 1.500 0.225

Dolomite 1.74 0.252

Anhydrite 2.19 0.160

Coefficients for the equation rhob = dVP^f

NOTE: it is my experience that the sand and

shale coefficients generally do a good job in

more consolidated clastic basins.

VELOCITYVELOCITY--DENSITYDENSITY

It is important to note that there is no unique relationshipbetween velocity and density. The are various reasons forthis, many of which we will cover during the course of thisclass. You should always be aware, however, that no singlerelationship between velocity and density may be applicable


70/124

relationship between velocity and density may be applicable.

DEPTHFT

VSHALEv/v0 1

PHIE_FINv/v1 0

( CORE_POR )

100 0

RDEEPohmm0.2 20

FIN_RHOBg/cc1.65 2.65

NPHI_SS0.6 0

FIN_RHOB NPHI_SS

FINAL_VPkm/s2 6

FINAL_VSkm/s1 3

FINAL_SWv/v0 2

PHIE_FINv/v0.4 0

BVWv/v0.4 0

14200

14300

14400

GUERRA SANDGUERRA SAND

Guerra Sd base

14600

14700

14800

14900

15000

15100

15200

15300

15400

VELOCITYVELOCITY--DENSITYDENSITY

It is important to note that there is no unique relationshipbetween velocity and density. The are various reasons forthis, many of which we will cover during the course of thisclass. You should always be aware, however, that no singlerelationship between velocity and density may be applicable


71/124

2.8

2.5

2.6

2.7

1

0.6

0.8

GUERRA 20GUERRA 72GUERRA 21GUERRA 36GUERRA 104GUERRA_109GUERRA 20

GUERRA 72GUERRA 21GUERRA 36GUERRA 104GUERRA_109

relationship between velocity and density may be applicable.

9 Jan 2007 @ 7:42

VP (km/s)

2 52.5 3 3.5 4 4.5

RHO

B(g/cc)

2

2.1

2.2

2.3

2.4

VSHALE

0

0.2

0.4

Density calculatedusing both Gardnerand Castagna.


72/124

In the absence ofother wells to use forcalibration it can bedifficult to know what

density values areappropriate for theshale.


73/124



74/124

CROSS-PLOTTING- cross-plot local data and do regressions on the valid data points

R^2 = 0.68

CAUTION!

Velocity-density (or porosity)relationships are extremely complex.Indeed, it is often difficult to find ameaningful relationship between the

VP (ft/s)

RHOB(g

/cc)

R^2 = 0.87

.

cracks in the rock matrix, grainboundary effects, pore-geometryeffects, complex mineralogy, fluiddistributions, and other microscopicflaws in the rock matrix.

1) regress valid sand and

shale data points

2) linearly mix the two

equations using Vsh (or

some other lithology

curve)

Note that the localdensity estimationworks best.


75/124


As with sonic data, multilinearregression is often an effective tool forediting density data. Key inputs tend tobe velocity and Vshale (or GR)


76/124

DEPTH

FT

GR_2

API0 200

CALIN6 16

RAW_RHOB

g/cc1.8 2.8

FIN_RHOBg/cc1.8 2.8

RAW_VP

km/s2 7

FINAL_VPkm/s2 7

RAW_VS

km/s0 4

FINAL_VSkm/s0 4

RAW_PR

v/v0 0.5

PRv/v0 0.5

7500

8000

8500

9000

NOTE: LARGE UNCERTAINTY WITH REGARDS TO DENSITY EDITS.

9500

10000

10500

11000

11500

12000

12500

13000

DEPTH

FT

GR

GAPI0 150

SPMV50 250

AT90

OHMM0.2 2000

AT30OHMM0.2 2000

RAW_RHOB

g/cc2 3

PRED_RBg/cc2 3

RAW_VP

km/s2 7

PRED_VPkm/s2 7

DENSITY EDITS: WHAT IS THE GROUND TRUTH?


77/124

DCAL

IN-5 5

AT10

OHMM0.2 2000

PRED3_RB

g/cc2 3

4100

4200

4300

4400

4500

DENSITY:

GREEN = RAW DATARED = 1st-PASS REGRESSIONBLUE = REFINED REGRESSION

MBFL

UBRNTSH

ORD_UNC

4600

4700

4800

4900

5000

5100

5200

5300

VELOCITY:

BLUE = RAW VP

RED = REGRESSION



78/124

RECOMMENDATIONS:

- generate pseudo-density for all wells- locally calibrate if possible- top to bottom of data

- use pseudo-data where raw data is bad

If no density data is available for calibration, I recommend usingCastagnas modification of Gardners relationship.


79/124

EDITING DENSITYEDITING DENSITY


80/124

Creating Pseudo Densities

Gardner coefficients (eLog)Emerge (H-R)

Editing Density based on new logs andgeologic understanding

EDITING DENSITY LOGSEDITING DENSITY LOGS -- INVASIONINVASION


81/124

porosity fluid density

During drilling, the near-wellbore environment is invaded to some degree with mudfiltrate. Thus, in reality the bulk density is responding to the rock matrix and thedensity of the mud filtrate. While this may not be particularly problematic for brine-filled sands, it can cause problems for gas sands. We hope to demonstrate this withthe following examples.

n

flgB += )1(

Measured bulk density

grain density

ra n ens y,

=i

iig

1

=

=n

i

iifl x1

Fluid density,

PROBLEM SETPROBLEM SET


82/124

PROBLEM 1a:

A gas sand has a bulk density of 2.05 g/cc and a grain density of 2.65 g/cc.

- calculate the porosity assuming a fluid density of 1.0 g/cc- ca cu ate t e poros ty assum ng xo = . ; assume a gas ens ty o .

g/cc and a brine density of 1.0 g/cc

- correct the bulk density for the effects of invasion; assume an Sw of 0.18 (18%)- calculate the percent difference between the measured and corrected bulk

density.

PROBLEM SETPROBLEM SET


83/124

BULK DENSITY 2.05 Fluid density Fluid density

Grain density 2.65 Gas 0.25 Gas 0.25

Filtrate 1 Brine 1

Sxo 0.55 Sw 0.18

PART 1 FLUID DENSITY AT Sxo 0.6625 Fluid density at Sw 0.385

porosity with fluid density = 1 0.363636

PART 2

porosity with 55% invasion 0.301887

PART 3

Corrected bulk density 1.966226

% differen 4.086516

SHEAR EDITS AND ESTIMATIONSHEAR EDITS AND ESTIMATION


84/124

As seismic petrophysicists, most of our work isdependent upon reliable shear measurements orestimates

Dipole measurements not available prior to about 1990

Need to frequently estimate shear velocities, especially

for older wells

It is also desirable to estimate shear velocity as a guidefor shear log editing

It is ALWAYS good practice to purchase and evaluate theraw wave-forms and evaluate the vendors shear picks!

DIPOLE SONICDIPOLE SONIC


85/124

Needed for Vp and Vs

Modern dipole tools utilize amonopole source and twoorthogonally-arranged dipole

sources Two orthogonally-arranged

receiver arrays

Diagram of Schlumberger's DSI tool (visit theirWEB page for additional information)

Why dipole?

Older monopole tools could record shear

waves, as long as the velocity of thewellbore fluid was slower than the shearwave velocity of the formation (that is,fast rocks). Dipole technology can

avoid this limitation.

and slow shear waves when

run in crossed-dipole mode Ideally the tool is centered in

the wellbore

Less susceptible to wash-outs than density


86/124

SHEAR ESTIMATIONSHEAR ESTIMATION -- comparisoncomparison


87/124

10000

12000

14000

SHALE

ft/s)

0

2000

4000

6000

5000 7000 9000 11000 13000 15000 17000 19000

G-C

WILLIAMS

VP (ft/s)

VS


88/124

SHEAR ESTIMATIONSHEAR ESTIMATION -- comparisoncomparison


89/124

6000

8000

10000

12000

14000

SHALE - G-C

SAND - G-C

14000

0

2000

4000

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

0

2000

4000

6000

8000

10000

12000

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

SHALE - WILLIAMS

SAND - WILLIAMS

WHAT ARE THE IMPORANTDIFFERENCES?

WHY MIGHT THIS BE IMPORTANT IN

ROCK PROPERTIES WORK?

SHEAR ESTIMATIONSHEAR ESTIMATION


90/124

)

6000

9000

8000

7000

1

.6

0.8

Many wells in the DWGOM have lowvelocity contrast between sands andshales. If we were to apply the G-C

coefficients to these wells, the shearvelocities would be too fast (note,however, that both Williams and G-Cwould predict similar shear velocities forshale). Later in the class we will discussthe importance of this for the AVOresponse.

20 Oct 2005 @ 11:11

VS (ft/s)

0 8000800 1600 2400 3200 4000 4800 5600 6400 7200

DEPTH(BML;feet

14000

13000

12000

11000

10

000

Vs

hale

0

0

.2

0.4

0

7 1

Note that sands are much faster than theshales in this well. This is an indicatorthat G-C coefficients might be moreappropriate for shear estimation. Use of

the Williams estimator may result in Vp/Vsratios that are too high


91/124

P(km

/s)

5

6

Vs

ha

le

0.6

0.8

ratios that are too high.

Please note that this is a rule of thumb,and may not be globally applicable. Localcalibration is ALWAYS desirable!

11 Apr 2007 @ 9:32

DEPTH (feet)

500 55001000 1500 2000 2500 3000 3500 4000 4500 5000

2

3

4

0

0.2

0.4

BARNETTSHALE

THE IMPORTANCE OF COMPOSITIONTHE IMPORTANCE OF COMPOSITION

DEPTHM

GRGAPI0 200

RTOHMM0 2 200

RHOBG/C32 3

VS_RAWft/5000 10000

PR_RAW/0 0 5


92/124

M GAPI0 200

DCALin-5 5

0 DCAL

OHMM0.2 200

RXOOHMM0.2 200

MSFLOHMM0.2 200

G/C32 3

VP_FINALft/s9000 19000

PERC_DIF%-100 100

ft/s5000 10000

VS_CASTft/s5000 10000

v/v0 0.5

PR_CASTv/v0 0.5

4400

Vp

13,000 ft/s (3.96 km/s)

Vs

7,500 ft/s (2.29 km/s)

Note the positivecontrast in PRbetween shale and

sand

4500

4600

DIFFERENCE BETWEEN PREDICTED AND

MEASURED VS LESS THAN 10%

COMPOSITIONCOMPOSITION


93/124

IMPORTANT DIFFERENCES IN

RESULTANT AVO RESPONSE

CLASS IV vs. CLASS III

IMPLICATIONS FORPROSPECTING?

FLUIDSUB TO GAS USING

IN-SITU SHEAR LOG

FLUIDSUB TO GAS USING

G-C COEFFICIENTS

UPPER SAND

-0.1

-0.05

0

0.05

0.1

0 10 20 30 40 50 60

Incident angle

Rpp

()

AVO response calculated

using the measured

properties

AVO response calculated

using G-C Coefficients

Measured = positive AVO gradient

Predicted = negative AVO gradient

Note the clay rich matrix and that the

COMPOSITIONCOMPOSITION


94/124

33.6

11.8

41.1

Detrital Quartz

Total Feldspar

Total Lithics

Total Diagenetics

Note the clay-rich matrix, and that the

quartz and feldspar grains are not in

contact with each other.

K = 1.24 md; = 6.4%

13.5

0.01.6 4.0

5.6

85.6

0.0

0.0

0.8

2.4 0.0

Clay coatings

TOTAL SMECTITE

TOTAL CHLORITE

TOTAL QUARTZ AND FELDSPAR

TOTAL CARBONTATE (CC, DOLO, SID)

HALITE

ILLITE

TOTAL TiO2TOTAL PYRITE

Bitumen



95/124

KEY INDICATORS OF BAD SHEAR DATA:

1) anomalous (or unexpected) Vp/Vs ratios

2) negative (or anomalously low) Poissons ratios

3) shear values that are significantly different from estimated values



96/124

NEGATIVE POISSON RATIOS

It is not uncommon to encounter negative Poisson Ratios when doing rock property work.There are a couple of reasons for this, but it is necessary to correct this problem.

, ,to calculate PR and look for negative, or anomalously low, values.

NEGATIVE POISSON RATIOS: BED



97/124

NEGATIVE POISSON RATIOS: BEDBOUNDARY EFFECTS

- Mostly a problem in Class I and Class II rocks- Solutions:

1) stretch/squeeze log data (Vp and Vs),2) estimate shear and splice in pseudo shear

across the bad data zone

appears to affect

NEGATIVE POISSON RATIOS: P-wave ATTENUATION? CRACKS?


98/124

appears to affectfairly tight rock

data can beproperlyprocessed, andstill yield thisresult.

if shear data

looks reliable,generatepseudo-VP

NOTE: note that the shear data mimics the geometry

of the sand better than the P-wave velocity. This is an

indicator that we should probably correct the VP, and

not the shear.



99/124

1) Simple, but reasonably accurate

2) All of these are derived for shale and brine-filled sands

- a lication of an em irical shear estimator to a as sand will

EMPIRICAL SHEAR ESTIMATORS

result in a shear velocity that is too slow.

3) Need to locally calibrate to known dipole and/or coremeasurements

4) In the absence of local calibration, I use contrasts in P-wavevelocity as a guide for selecting the appropriate shear velocityestimator.

NOTE:

- estimated shear velocity accurately

predicts measured shear velocity

- shear velocity for gas sand is not

predicted

- however, note that cross-plottingestimated vs measured shear may


100/124

Gas Sandestimated vs measured shear may

provide a means of identifying gas

sands!

R^2 = 0.91

*Data are from East Anstey (MC607)

1) ll f th l d h ti t f h l d b i fill d



101/124

1) all of the commonly used shear estimators are for shale and brine-filled

sands.

2) estimation of shear velocities in pay is problematic

- Greenburg and Castagna, 1992; an iterative approach

- P-wave modulus approximation

3) Use the P-wave modulus (VP2) to take the gas P-wave velocity back tobrine (Mavko et al., 1998). Then we apply one of the empirical shear

.

00

2

0

1

1

M

M

MM

MM

MMdry

fl

dry

drysat

+

+=

Msat = VP2 ; the saturated P-wave modulus (measured from the log data)Mdry = the P-wave modulus of the dry frame

M0 = the P-wave modulus of the mineral matrix

Mfl = the P-wave modulus of the pore-filling fluid

= porosity

Since we are calculating the P-wave modulus for two fluids, we can algebraically eliminate the



102/124

)20(

2

20

2

)10(

1

10

1

MflM

Mfl

MsatM

Msat

MflM

Mfl

MsatM

Msat

=

Since we are calculating the P wave modulus for two fluids, we can algebraically eliminate the

need to calculate Mdry:

We can solve forMsat2 and calculate the compressional velocity for the brine case:

2

2

satMVP=

Note that the bulk density must also be fluid substituted back to brine!



103/124

( )( )2

222

22

/ matrixfluidmatrix

fluidsituin

situin

VSVPVP

VPVPVS

=

Kriefs model for the rock frame (we will discuss this later in the class) can becombined with Gassmanns equation to determine the shear velocity across a pay

sand. This technique is an easier approach than via the P-wave modulus

approximation.

VSHv/v0 1

DEPTHFT

RDEEPohmm0 2 200

VSft/s0 10000

SWv/v0 2



104/124

v/v0 1

PHIEv/v1 0

0 VSH

FT ohmm0.2 200 ft/s0 10000

VS_DWGOMft/s0 10000

XVS_INSIft/s0 10000

v/v0 2

PHIEv/v0.5 0

BVWv/v0.5 0

12200

Shear estimate without

Measured shear velocity

12300

12400

12500

gas correct on

Shear estimate with

gas correction



105/124

0.05

0.1

0.15

0.2

oefficient

-0.2

-0.15

-0.1

-0.05

0 10 20 30 40 50 60

INSITU RESPONSE

GAS CORRECTION

NO GAS CORRECTION

Refle

ction

Offset (degrees)

- note phase reversal and brightening at the far offsets for the insitu gas case

- these characteristics are lost if shear is not properly estimated.


106/124

P-wave MODEL

COMPARISON STUDYCOMPARISON STUDY


107/124

DEPTH

FT

VCL_FIN

v/v0 1

PHIE_FINv/v1 0

0 VCL_FIN

AO90

OHMM0.2 200

AO30OHMM0.2 200

AO10

OHMM0.2 200

RHOB_FIN

g/cc1.65 2.65

NPHI_SSv/v0.6 0

RHOB_FIN NPHI_SS

VS_RAW

ft/s3000 8000

VS_PWAVEft/s3000 8000

PR_RAW

v/v0 0.5

PR_PWAVEv/v0 0.5

10300

10400

10500

10600Use of the P-wave shear estimator is based on Gassmann. It

also requires that the user select one of the empirical shear

estimators. In this case, I used the G-C coefficients for sand

and shale. Note the quality of the match in the shales and

brine sand. Whether using Krief or P-wave, it is useful to

calibrate the estimated shear to either shales and/or brine

sands.

P-wave MODELKRIEF MODEL



108/124

8000

6000

6500

7000

7500

1.2

0.8

1

R2 = 0.87

8000

6000

6500

7000

7500

1.2

0.8

1

R2 = 0.82

WELL: East Cameron 131 #1ZONE: 5410.500 - 10819.000 FTDATE: 14 Oct 2005 @ 9:10 PHIE_FIN > .04

VS_KRIEF

3000 80003500 4000 4500 5000 5500 6000 6500 7000 7500

VS_

RAW

3000

3500

4000

4500

5000

5500

VSHALE

0

0.2

0.4

0.6

WELL: East Cameron 131 #1ZONE: 5410.500 - 10819.000 FTDATE: 14 Oct 2005 @ 9:09 PHIE_FIN > .04

VS_PWAVE

3000 80003500 4000 4500 5000 5500 6000 6500 7000 7500

VS_

RAW

3000

3500

4000

4500

5000

5500

VSHALE

0

0.2

0.4

0.6



109/124

CAUTION!

If you are using estimated shear velocities that have not been gas-corrected, the AVO response you get will be wrong. Even in relativelyti ht and fast rock incorrect shear estimation for a can make asubtle difference on the calculated AVO response.

It is not uncommon to see people do this!

VCL_FIN

v/v0 1

PHIE_FIN

v/v1 0

GR

API0 150

DEPTH

FT

RDEEP

ohmm0.2 2000

RMED

ohmm0.2 2000

RHOB_FIN

g/cc1.65 2.65

NPHI_SS

v/v0.6 0

VS_EST

ft/s4000 10000

VS_FINAL

ft/s4000 10000

PR

v/v0 0.5

PR_MEAS

v/v0 0.5

SW_FINAL

v/v0 2

PHIE_FIN

v/v0.5 0

A BLIND TESTA BLIND TEST


110/124

0 VCL_FIN

RSHAL

ohmm0.2 2000 RHOB_FINNPHI_SS

BVW

v/v0.5 0

7600

7700

LIMESTONE LAYER

7900

8000

8100

8200

8300

8400

11000

100

00

1

0.8

1: VS_EST-VS_FINAL-VCL_FIN

Correlation Coefficient:

r = 0.9074 r-square = 0.8234


111/124

URED

VS(ft/s

)

8000

9000

Vs

ha

le

0.6

25 Jan 2004 @ 19:55

ESTIMATED VS (ft/s)

4000 110005000 6000 7000 8000 9000 10000

MEAS

4000

5000

6000

7000

0

0.2

0.4

NOTE: the estimated shear velocity

predicts values which typically are slightly

faster than the measured values. However,

the estimated shear velocity in this case

was calculated using Castagnas

coefficients for sand and shale. If local

coefficients were developed, a better fitwould probably be observed. You will

note on the following page, however, that

these small differences make no

appreciable difference in the AVO models.

0.1



112/124

0

0.05

0 10 20 30 40 50 60

-0.2

-0.15

-0.1

-0.05

RED = estimated shear

BLUE = measured shear

CASE NAME: DIPOLE CASE NAME: ESTIMATED SHEAR

SHALE SAND SHALE SAND

VP 11772 13491.26 VP 11772 13491.26

VS 6069 8488.129 TRUE VS 6231 8731.205 TRUE

RHOB 2.64 2.318 RHOB 2.64 2.318

Display Results Display Results



113/124

SOME OBSERVATIONS:

If velocities are greater than approximately 9,000 ft/s, Castagnas coefficients will result inrealistic frame properties. Shear estimators designed for soft rock will result in Vp/Vsratios which are too high.

, ,frame properties which are too stiff.

As always, locally calibrate when possible, and build shear coefficients which areappropriate for your area.

PREDICT SHEAR VELOCITIES (USING G-C) AS A QC TOOL



114/124

Look for large discrepancies (>15%) Possible data quality issues Smaller differences (< 10%?) may be due to unexpected

composition

INCORRECT SHEAR VELOCITIES CAN ESTABLISH THE

AVO models

Geomechanical analysis Registering of multicomponent datasets

COLLECT RAW WAVEFORMS

stacked semblance is not enough! evaluate STC plots at key depth intervals (especially across the

shales)

IF NECESSARY, REPROCESS

TWO FINAL EDITING STRATEGIESTWO FINAL EDITING STRATEGIES


115/124

PATCHES

RESCALING OTHER CURVES

EXAMPLEEXAMPLEDEPTH

M

A_GR

GAPI0 150

B_GRGAPI0 150

B_C13

A_RD

OHMM0.2 200

B_M2R9OHMM0.2 200

A_DT24QS

US/M400 100

B_DT24QSUS/M400 100

B_PORZSS

PU40 -10

B_CNCSSPU40 -10

SPL_GR

API0 150

Lets repair the data across the casing point.Remember, we want to generate final curves that more-or-less represent what we think is in the earth.

Since the deep resistivity penetrates deepest into thereservoir, we will start by making a few minor edits toour merged RT


116/124

MM0 500

A_C13MM0 500

1525

1550

1575

1600

1625

1650

1675

SIMPLE HANDSIMPLE HAND--PATCH ON RTPATCH ON RT


117/124

ARGUMENTS FOR OR AGAINST MY PATCH?

RESCALED SONICRESCALED SONIC


118/124

The correlation between the sonic and resistivityacross the casing point indicates that theSPL_RT may not require extensive editing.

RESCALED SONICRESCALED SONIC


119/124

Rescaled sonic data to mimic RT. Notice thatthis more closely mimics the resistivity than mysimple hand-patch.

What does the resistivity profile suggest aboutlithology?

What are the potential pitfalls in our model?

WHAT ABOUT GR?

RESCALED GR?RESCALED GR?


120/124

COMPARE RAW vs. EDITED DATACOMPARE RAW vs. EDITED DATA

DEPTHFT

RAW_RHOBg/cc2 3

FIN_RHOBg/cc2 3

QUAL RB

RAW_VPkm/s2 7

FINAL_VPkm/s2 7

QUAL VP

RAW_VSkm/s1 4

FINAL_VSkm/s1 4

QUAL VS

RAW_PRv/v0 0.5

PRv/v0 0.5


121/124

QUAL_RBflag0 20

QUAL_VPflag0 20

QUAL_VSflag0 20 RAW_PR PR

1500

2000

2500

MBFLUBRNTSHLBRNT

ORD_UNC

3000

3500

4000

4500

5000

5500

6000

6500

SUMMARY: HANDLING BADHOLE DATASUMMARY: HANDLING BADHOLE DATA

Recognition of bad data


122/124

g

Look for consistency between various log measurements!

Log plots Histograms Trend plot Cross-plots

Look for deviations from expected behavior

Look for badhole indicators such as caliper (washouts) anddensity correction

Various tools available for reconstructing bad data(regressions, empirical equations, patches, rescaling)

EDITING STRATEGYEDITING STRATEGY


123/124

Load all data check it carefully

Display the raw data Identify all possible rhob, dt, dts curves; compare and select Merge, depthshift, clip tails, compare to nearby wells Spend some time looking at log display at variety of scales, cross-

plotting curves, histograms, trend plots

Study the well. Look at the curve patterns in light of the geology

Always create a display illustrating the raw data and theedited data


124/124

Log Editing

Documents

Transcript of Log Editing