1. A Massive 3D VSP in Milne Point, Alaska
Claire Sullivan, Allan Ross, James Lemaux, Dennis Urban, Brian Hornby, Chris West, and John Garing,
BP Exploration Alaska
Bjorn Paulsson, Martin Karrenbach, and Paul Milligan, Paulsson Geophysical Services, Inc.

2. What we did: Why we did it:
Acquired a 3 million trace 3D VSP with 4 wells, each instrumented with component receivers recording simultaneously.
Why we did it:
High resolution data set for field development.
Determine variation in permafrost velocities.

3. Outline
Introduction
Planning the VSP
Acquisition
The Data
Conclusions

4. Milne Point Location S-pad B e a u f o r t S e a Milne Point5 10
Miles
B e a u f o r t S e a
Milne Point
Northstar
S-pad
Pt. McIntyre
Kuparuk
West
Dock
Niakuk
Prudhoe
Endicott
Lisburne
Bay
SDI
Pt. Thomson
Milne Point is located on the North Slope of Alaska, just to the west of Prudhoe Bay Field.
The S-pad development is located here and is considered part of Milne Point field.
PS1
Prudhoe Bay
Tarn
Badami
Sourdough
TAPS
ANWR
Not to scale

5. Schrader Bluff S-pad Highlights
53 mmbo recoverable
Reservoir depth = 4000 ft.
Viscous Oil – API 14-24, biodegraded, low temperature
Economic breakthrough with multi-lateral drilling design.
Success depends on accurate placement of laterals in the reservoir and elimination of drilling sidetracks.

6. Schrader Bluff Geology
The Geology at Schrader Bluff consists of 2widespread shallow marine sands, the OA and OB. These sands are currently being developed now.
Later development may target the N sand, a fluvial deltaic sequence above the O sands
shallow tvd low temp.
N sands are fluvial, deltaic
O sands are shallow marine.
close to permafrost, unconsolidated, biodegraded
viscous API
low productivity
Multiple, stacked sands
Main reservoirs are O & N at Milne Point

7. Motivation for the 3D VSP
Current seismic data was low fold (4-6).
Thin sands (20-30 ft) with sub-seismic faulting.
Need a high resolution data set to steer development drilling and avoid sidetracks.
Current surface seismic data is only 4-6 fold at Schrader Bluff interval at 4000 ft depth. The survey was originally shot to target the Ivishak formation at 9000 ft.
The sands are thin, ft thick with sub-seismic faults. 15 feet of throw on an unseen fault will throw the lateral out of the sand interval, resulting in a sidetrack. Our goal was to predict the small faults ahead of time to avoid sidetracks.

8. Feet from surface location
vertical exaggeration =10x
-3,850
ESE
SSE
-3,875
MPH-08A SIDETRACK BORE HOLE
PROFILE CROSS-SECTION
-3,900
-3,925
-3,950
-3,975
-4,000
Depth
First hole
-4,025
-4,050
-4,075
Second
hole
-4,100
-4,125
This what we call threading the needle. The challenge is to direct long 4000 ft multi-laterals in ft sands with offsetting 15 ft faults that are below seismic resolution. The goal is to eliminate costly sidetracks. The focused 3D VSP method was chosen as solution.
The red track represents the first attempt and missed the sands, the well was then sidetracked, shown in teal. Before the development of S-pad, this team had several examples of wells requiring 5 or more sidetracks. This had to be avoided.
Is a slotted liner really necessary?
Gross hole collapse should result in the fine grained Schrader sand spilling through the slots.
The only other use for slotted liner would be in the event of interbedded shales collapsing, or at fault rubble zones, where larger pieces of rubble might block the hole.
H-08a was completed barefoot to test this theory. Proving slotted liners aren’t necessary could save $500K per future multilateral well.
Early results show a wellbore blockage has occurred. Further study is ongoing to determine the cause.
'OA' Sands
-4,150
-4,175
-4,200
6,500
6,750
7,000
7,250
7,500
7,750
8,000
8,250
8,500
8,750
9,000
9,250
9,500
9,750
10,000
Feet from surface location

9. Planning the Survey

10. Image coverage modeling
P/GSI Illumination modeling.
Various scenarios were simulated:
single well and combined deviated wells
different shot spacing and coverage
varied placement of geophone arrays in borehole
designed trajectory of borehole
Complete 3D volume of hit counts computed.
Paulsson Geophysical did the illumination coverage modeling for us. This work was done in the summer of It had to be done early since we had no wells on the pad, and were planning to pre-drill injector wells for the VSP survey.
Various scenarios….etc

11. Modeled illumination at 4000 feet
with 320 receivers placed in 4 wells.
10,000-30,000
30, ,000
# of hits
Receiver
locations
By pre-drilling the 4 injectors wells, the team was able to use these wells as appraisal wells for the field, as production drilling did not start until 3-4 months later.
2X3 miles of coverage
One mile
Contour interval = 200 hits

12. Placement of geophone arrays
Each array has
80 levels spaced
50 ft apart.
Length of array is 4000 ft.
Total of 320 multi-component geophones
Receivers in deviated wells are placed below permafrost.
Permafrost
1000 ft
This slide illustrates the placement of the receiver arrays in a 3D view.
The 4 colored lines represent the receiver positions in the borehole, each well has 80 receivers for a total of 320 receivers in the ground at one time.
The receivers in the deviated wells were placed beneath the permafrost. This was done for 2 reasons, one to avoid 2 way travel through the permafrost and to move out the receivers laterally away from the pad so to maximize coverage. We could not move them out too far though, because we would lose image aperture by placing our receivers too deep or too close to the reservoir. This image is has vertical exaggeration, the receivers are much more spread out than what appears here.
2000 ft

13. Acquisition of the Survey

14. This is the work platform that was used to deploy the geophone arrays
This is the work platform that was used to deploy the geophone arrays. You can see most of the pad in this picture, Shown in the foreground is the S-21 well which was also used in the VSP survey.
The Klondike platform was equipped with a crane, but was not available for use.
The crane was used to lift the tubing, cables and geophone pod housings onto the work floor and also during the deployment process.

15. Geophone cable array, pod housings, and tubing on rig floor ready for installation

16. Placing geophone cable into
pod housing and securing before
deploying into wellbore. Bladder
style clamping mechanism is seen
in picture on the right.

17. Here is a good view of the 3C geophone, with the bladder behind it, cables and geophone pod housing all in place. When the bladder is pressurized at the surface, this geophone will be forced against the borehole, providing the coupling.

18. Here is one of the well heads with geophone array in place
Here is one of the well heads with geophone array in place. Bops required heaters and generators. The pressure gauge for the bladder system is here

19. Source point interval 250 ft
8 second sweeps, 1 vib, Hz, 6 sec listening, sec tapers, vib, Hz, “ “
2 vibes, 2 sweep pattern. 2nd pair of vibes operates during move-up. Buggy travels along to pull-out stuck vibes.

20. VP coverage and active receiver wells map
1 mile
The is the a map of the source covarage. We recorded over 3000 Vibe points into the 4 instrumented wells at one time. We also recorded 4 walk-above lines directly over the receiver positions at a 50 ft spacing. The original plan was to shoot a circular pattern around the wells, with an 8000 ft offset to the farthest phones. We were losing signal as we progressed to the NE and decided to stop recording. Time was also a factor as we had 3 – 4 week window to complete the acquisition. It took 10 days to deploy the equipment, 6 days to test and shoot, and another 7 to remove the equipment. The rig arrived on the pad one day after we had left.
3232 VPs into 4 wells, 80 three-component geophone levels, 960 channels, 3 million traces.
4 walk-above lines at 50 ft source interval to aid in 3C geophone orientation.

21. A Look at the Data

22. Raw records: near offset (500 - 670 ft)
MPS MPS MPS-31
Direct P wave
Direct S wave
Upgoing
reflections
500
1000
Time
This is an example of the raw records at a near offset (500 ft) to 3 of the wells. Time is the vertical scale and depth or receiver position in the borehole is the horizontal axis. The direct P wave, and direct S waves are clearly visible. Here are the upgoing P wave reflections.
Receivers or Depth down the wellbore (ft)

23. Raw records: far offsets (5000 - 6200 ft)
MPS MPS MPS-31
Direct S wave
Upgoing
reflections
Direct P wave
500
1000
Time
This is an example from the same wellbores only at a far offset 5000 ft. The data quality is somewhat diminished but still fairly strong.
Receivers or Depth down the wellbore (ft)

24. Direct wave spectral analysis Near offset – 376 ft 10-170Hz bandpass
200
150
Frequency (Hz)
signal
100
This is a frequency spectrum of one of the near offset records. The frequency range is from Hz. 50
-0.01
+0.01
Wavenumber (cycles/ft)

25. Common Source Gather Depth 1787 3219 Depth 1787 3219 Depth 1787 3219
Time (us)
1000
500
100
Offset from well head (ft)
365
3806
Raw axial data
Data rotated towards source.
Data rotated towards reflectors
Data after wavefield separation – upgoing P wave
As preparation for the pre-stack depth migration, all data must undergo rotations and a 3C wavefield separation. I have several slides showing different rotations. Each slide is a Common source gather from a source about 2400 ft SW of the bottom geophone in the well. It is fairly typical of the data set. The first slide is the axial data, with no rotations. The second example is after the data has been rotated towards the source. Notice that the first break energy dominates the record. This is where the first break picking is done. Next is an example rotated towards the reflectors, emphasizing the upgoing P wave reflections. The final slide is after the wavefield separation and this is what is input into the per-stack depth migration. The upgoing P wave reflectors are prominent, but note there is still some residual shear energy. The will get stacked out in the migration process, as the raytracing and velocities will be incorrect for the shear.
Source point is 2400 ft
SW of lowest receiver

26. North – South example line through center of S-pad
-4350
-4150
-4050
Depth
-3950
Next I will show a comparison between the surface seismic and the VSP. It is a N-S line through the center of the pad and covers the central, near vertical MPS15 well. The structure map shows a low, faulted graben in blue in the center of the field. The producing wells are the curved laterals and the injectors are the straight tracks. The 15 well was the first well in the field, a pre-drill injector for the VSP.
¼ mile

27. Reprocessed Surface Seismic Preliminary VSP
1/4 mile
1/4 mile
We did some in-house reprocessing to the surface data this spring and that is shown on the left. The equivalent VSP data is on the right.
The depth range is from 3000 to 6000 feet. As you can see, the frequency content of the VSP is much higher than the surface seismic. A lot of time was spent handpicking the velocities in the reprocessed surface and it is about as good as it can get. The VSP still needs work on the velocity model, once that is correct, the image should improve even more.

28. Massive 3D VSP / W-E Profile

29. Reprocessed Surface SeismicHz
-dB
10 – 70 Hz
Reprocessed Surface Seismic
25 – 125 Hz Hz
-dB
Preliminary 3D VSP
more than 10 Hz to the high end of the data. The VSP has a bandwidth from Hz. There has been a low frequency attenuation applied to this data to help remove some residual shear energy.

30. Well Planning Impact
The MPS13 injection well was moved 400 ft to the SW avoiding a faulted zone.
MPS-13
S-Pad
This is an example of the impact the 3D VSP had on the well planning. This is a coherency plot at the reservoir level, the OA sand. The injection well originally was planned to set in this fault zone here, marked by the red dot. The fault zone was picked up by the VSP and the well was moved 400 ft to the SW.
Cost of well 0.9 million
Original OA OA
Coherency plot of VSP data

31. Velocity issues

32. Correlation between sonic data, offset VSP’s and 1D migration velocity
Velocity (ft/s)
This graph shows the correlation between sonic data, offset VSP’s and the iD migration velocity profile.
The vertical axis is velocity in ft/s and the horizontal axis is depth in ft. The velocity ranges from 6000 to over 14,000 ft/s. The permafrost zone is here, and you can see the strong variation in velocity. The variation is not only vertical but lateral too, as shown by the difference between the green and red tracks, representing different VSP offsets. Note that the VSP matches the sonic log quite well.
Depth (ft)
Shallow permafrost velocity variations

33. First Break Tomography
Direct arrivals from surface to deviated wellbores are used to determine local shallow velocity variations
source
surface
In order to get a handle on the near surface permafrsot velocity variations, first break tomography has been done. The direct arrivals from each source to the deviated and straight wellbores are used in the tomography. The advantage of the deviated wells is that you end up with a lot of near vertical raypaths. Next I will show some preliminary results of the tomography.
Permafrost boundary
wellbore
receivers

34. North – South example line through center of S-pad
-4350
-4150
-4050
Depth
-3950
Next I will show a comparison between the surface seismic and the VSP. It is a N-S line through the center of the pad and covers the central, near vertical MPS15 well. The structure map shows a low, faulted graben in blue in the center of the field. The producing wells are the curved laterals and the injectors are the straight tracks. The 15 well was the first well in the field, a pre-drill injector for the VSP.
¼ mile

35. Original 1D velocity modelW E
2000
1000
14,000
11,000
8000
Velocity
(ft/s)
6400
Permafrost boundary
This is the 1D starting model, based on the zero offset VSP. A E-W line through the center of the pad is shown. The permafrost boundary is here at 1700 ft.
Depth (ft)

36. What does the permafrost really look like?
Thaw bulb
2000
1000
Depth (ft)
We know the permafrost is heterogeneous. This is a cartoon of what we think the permafrost might look like. The North Slope is covered with lakes, and these lakes do not always freeze through solid. This can cause warming of the underlying soil and thaw bulbs can develop, Also lenses of slower, non-frozen material are most likely present. In addition the permafrost base is not likely to be flat, but probably rugose as shown here. The next slide is and image from the preliminary tomographic velocity model.
11,000 – 15,000 ft/s
6000 – 9000 ft/s
7500 ft/s

37. 3D velocity model from FB tomographyW E
2000
1000
Depth (ft)
14,000
11,000
8000
Velocity
(ft/s)
6400
Permafrost boundary
Note the variations that are developing in the permafrost layer. They resemble some of the thaw bulbs or the rugose permafrost base. To check the validity of this model, the calculated travel times are compared to the picked travel times and also the areas of velocity changes checked to see if they match up with the lakes or other features.

38. Conclusions VSP data supported first phase of field development.
No sidetracks were required.
Production started Sept. 1, 2002 with 8000 bopd.
VSP data provides twice the frequency of surface seismic.
First break tomography identifies permafrost velocity variations.
Processing is ongoing and survey will be used for reservoir management and identification of deeper Kuparak targets.

39. Acknowledgements
Thanks to the S-pad development team, BP Exploration Alaska, for the opportunity to work on and present this project.
Thanks to Grunde Ronholt, READ Well Services, Sue Raikes, BP Sunbury, and Jonathon Woolley, a summer intern from the Colorado School of Mines.
Trace Geophysical recorded the data.
Equipment used for this survey was funded under a DOE grant,
# DE-FC26-01NT41234.

40. Back-up slides

41. Data processing sequence (1):
Raw VSP data
Assign & check source-receiver geometry,
pass only valid records on for processing.
Survey data
Determine 3C orientation
Initial QC, 3C orientation
and Velocity Model building
Rotate towards source
Convert to zero phase
and band pass filter
Invert for source statics
Pick first break times
Invert for zero offset 1D velocity model
Tomographic inversion for 3D model

42. Data processing sequence (2):
Create source wavelet inversion filters
Common Source
Gathers (CSGs)
Convert to zero phase
and band pass filter
Rotate towards source
Align on first breaks & mix
200 ms window
wavelet inversion
Source wave filters

43. Data processing sequence (3):
3C upgoing wavefield separation
Common Source
Gathers (CSGs)
Convert to zero phase
and band pass filter
Rotate to true x,y,z
Rotate H1 towards source
Sum: H1 - V
Source wave filters
Convolve
Top mute FBs
AGC
Gapped decon
Band pass
Upgoing P-waves
Radon filter

44. Data processing sequence (4):
Prestack Depth Migration & Stacking
Upgoing P-waves
3D Velocity model
Time shift
PSDM
Trim Statics
Source statics
Stack
3D reflectivity image
in depth

45. Multi component summing:
1) Rotate H1 (azimuth only)
towards source
2) Reverse V polarity
3) Sum:
-V + H1
source V
source V sv + - p Q4 Q1 + + H1 H1 - - Q3 Q2 p sv - +
reflection
Results:
Downgoing p-waves are attenuated with a dipole response in the Q1 quadrant.
Downgoing s-waves are amplified with an isotropic response in the Q1 quadrant.
Upgoing p-waves are amplified with an isotropic response in the Q2 quadrant.
Upgoing s-waves are attenuated with a dipole response in the Q2 quadrant.

46. Reprocessed Surface Seismic
Preliminary VSP

47. N-S line from MPS15 volume: preliminary 3D velocity model
2000 ft

48. North – South example line through center of S-pad

49. Frequency spectrum VSP Surface Seismic 20-125 Hz (-20dB)

50. Velocity model 11,000 ft/s 6500 ft/s 7500 ft/s Depth (ft) 9500 ft/s
Distance (ft)

51. Schrader Bluff Structure Map
The central well in the VSP program is located in a narrow graben. It is the only well in the area with log data.
The velocities are anomalous at this well.
Original 1D velocity model was based on zero offset VSP at this well.
This
-4300
-4000
-3700
1 mile