Академический Документы
Профессиональный Документы
Культура Документы
Disc Contents
PA98
Contents
- 1 - 052
Search
ABSTRACT
Extended offset horizontal wells (out to a kilometer or
more) are becoming common in field developments.
In areas where overburden velocity varies laterally, it
becomes difficult to know where the well is relative
to the top of the reservoir away from the top reservoir
penetration point. This is important to know for depth
conversion and volumetrics and to know what part of
the reservoir is sampled by the well to build a good
simulation model for reservoir characterization.
We examine the use of the conventional offset VSP
method but it can only image reflectors below the
reservoirs in the well making receivers in the
horizontal section only good for imaging the base
reservoir. The top reservoir over the horizontal
section could only be imaged by a very far offset
source recorded by receivers in the vertical section
giving raypaths with too much moveout, poor signal
quality and little control on overburden velocity.
Instead, we have applied a checkshot solution to the
problem where we can use the seismic time and the
checkshot time to estimate borehole location relative
to the top reservoir and'the overburden velocity for
depth conversion.
Examples from onshore North Sumatra illustrate the
ideal and practical methods and the accuracy of our
estimates. The marine applications are very accurate
(2 20 feet in the estimate of well position relative to
the top reservoir) due to more homogeneous near
surface and the ability to easily locate the source
directly over the receiver. The land applications are
less accurate but still more accurate than no
information (260 feet), due to more variable near
INTRODUCTION
There are many reasons stated for not acquiring
borehole seismic data in development wells. They
usually consider that enough velocity information is
available in the field from wildcat and appraisal
control, velocity surveys, and development well
penetration points to make a reasonably accurate
depth conversion. Extended offset horizontal wells
have added a new complexity to the question that has
made us consider the value of additional data from the
development wells.
Mobil Oil Indonesia has operated the drilling of 20
highly deviated development wells in 1997 to develop
four Peutu limestone fields in North Sumatra. Seven
of these wells, in the SLS-A' & D fields, have offsets
of approximately 2,625 feet. After drilling each well,
we know the position of the whole well bore in depth
but we only know the depth of the top of the reservoir
where the well penetrates it. What we don't know is
the depth to the top of the reservoir over the
horizontal well bore. We only know that surface in
time from the surface seismic. If this were an area of
well behaved overburden velocity, it would be
sufficient to depth convert the top reservoir surface
with the velocity from the well penetration .points and
the wildcat velocity surveys. Unfortunately, this area
is an onshore area in the foothills of the Barisan
Uplift topography. Both statics and laterally varying
velocities are a problem. We would like to acquire
borehole seismic data that allow us to better know the
top reservoir surface in depth, the position of the
horizontal well bore relative to it, and which layers in
the reservoir have been sampled by the well bore (for
118
the simulator model). Figure 1 illustrates the time to
depth problem for a general case.
SOLUTIONS
The first solution we tried to model was whether a
119
OPERATIONAL CONSIDERATIONS
The marine case is optimal for doing checkshots over
horizontal wells (Figure 7). Airguns can be placed
directly over the receiver location (vertical travel path,
direct comparison to seismic stack trace) and very
close to the sea level datum (small correction from
source position to datum). The land case usually
presents more constraints. Topography, surface
conditions , ,physical, cultural, and environmental
obstacles often conspire to limit the possible source
locations. If dynamite is used, shot holes must be
drilled. If airguns are used, large pits must be
excavated. At SLS, we were only able to use one
airgun pit for each well. Due to the mountainous and
forested nature of the area, we always attempted to
dig the pit on or near to a road. We tried to find a
location as close as possible to the well TD position
to minimize the nonvertical ray path angle correction
error and as far away from the well penetration and
wildcat known control points as possible. The
nonvertical angle correction increases as the receiver
is moved away from the source position (Figure 8).
ACCURACY
The marine case has many less potential errors than
the land case. In marine acquisition, there is no angle
correction as the shot is vertically over the receiver.
Both the seismic data and the checkshot data have the
source airgun near the datum in the known velocity
water layer. Therefore there should be no difference
in statics or datum corrections. Measurement and
calculation errors are all small including :
1. Seismic time error due to inaccurate migration
velocity if top reservoir is not nearly flat,
2. Error in picking checkshot first break up to 2 ms
= 13 feet,
3 . Error in reservoir velocity
and
2 500
ft/sec
6 feet,
120
CASE HISTORY
Checkshot data in some wells showed that the
overburden velocity was not highly variable and that
the data was not needed while others showed that the
velocity was quite variable and that the data added
value. The problem with this conclusion is that we
would not know this unless we had the data. One of
the wells that gave significant results was the SLSA10 well onshore 'B' Block North Sumatra. The
reflector at the top of the reef is a prominent peak on
this 3D data set (black, Figure 10) and is easily
interpreted over the well horizontal section and
appears to be almost flat. If the seismic top reservoir
surface had been depth converted with the velocity
from the well penetration, it would be nearly flat
(Figure 11). However, when the checkshot and
seismic data are used to calculate the depth of the
well bore below the top reservoir (delta D) and the
velocity of the overburden (by the method outline in
Figure 6, calculations Table 1) our depth estimates
would have been over 100 feet off (compare columns
K and L, Table 1 and graphically Figure 11). The
velocity decreases over the middle well section raising
the top reservoir in depth and then increases toward
the well TD depressing the top reservoir in depth.
Clearly, this information is valuable for depth
conversion, volumetrics and creation of a simulation
model.
We estimate that our top reservoir depths are accurate
to within approximately 2 50 feet. The datum
correction error has been minimized by using a
velocity (4,893 ft/sec) which makes the seismic time
(zero crossing, peak, minimum phase) and the
checkshot time equal at the well penetration point.
Physically, this is a very reasonable velocity for the
shallow overburden and it matches well with the
correction velocities calculated by a similar method in
CONCLUSIONS
VSP and checkshot methods have been examined for
application to highly deviated and horizontal wells.
VSP data does not meet the objectives of providing
control on the velocity and depth to the top reservoir
surface. There is, however, utility in the checkshot
method. If a good top reservoir reflection exists, the
seismic time and checkshot time can be compared to
estimate the depth that the top reservoir surface is
above the well bore. If a good top reservoir reflector
does not exist, the overburden velocity must be
assumed constant and only changes in depth across
faults and the like can be observed in the data.
The method we employ should be accurate to within
f 20 feet for marine data and f 60 feet for the land
data case. Results from seven horizontal development
wells show that significant volumetric error would
have resulted if we had not acquired the checkshot
data.
REFERENCES
Smidt, J.M., 1996, A limitation of well velocity
surveys in highly deviated wells drilled parallel to
bedding, Geophysics, 61, 627-630.
= (144 ft) / (4,893 ftlsec) = 29.4 ms; where 144 ft is the difference between
TABLE 1
PROBLEM
Vertical WC Control
FIGURE 1 - Schematic cross section of the knowns and unknowns for a horizontal well and a layered reservoir.
FIGURE 2
Schematic cross section of single offset source VSP raypaths Eor a horizontal well to image the
Top Peutu reservoir.
123
FIGURE 3
Schematic cross section of single ofiCset source VSP raypaths to image the Base Peutu reservoir
at a horizontal well.
Increasing Depth
-b Horizontal Section
Time
FIGURE 4
124
FIGURE 5
Schematic cross section of vertical checkshot raypaths to a horizontal well illustrating the
calculations when overburden and Peutu velocities are assumed constant.
GL
SL
Tchecknhot
i
Known in Depth and Time
Known in llme
born SL (seismic)
Welt path
known in depth
Vertical WC Control
FIGURE 6
Schematic cross section of a vertical checksbqt raypath illustrating the calculations when the
seismic time is also used.
12s
Operations Marine
SL
Time Vertical
E:
I
-(
TVDreoevla
! 8,000ft
nriowii iii
tiiiw
from SL (seismic)
Horizontal Offset
weii pain
known in depth
Vertical WC Control
FIGURE 8 - Schematic cross section of the desired source and receiver locations for land operations when only
one source location is used.
126
Accuracy Land
FIGURE 9 - Diagram of the single source to Ceceiver raypath for the VSP experiment and the CDP set of traces
used to create a stacked trace. The average GL and Vdatum of the CDP set may be different than
the VSP encountered.
128