Professional Documents
Culture Documents
Abstract
Seismic provides critical information for drilling, such as 3D structural images showing geological targets or hazards and
formation properties relevant to drilling such as pore pressure. However, pre-drill estimates of formation pressures and
structural depth images typically have large uncertainties. These uncertainties present drilling risks and could increase the
cost of wells in the deepwater and other drilling environments. We present a new method that reduces the uncertainty up to
large distances ahead of the bit by optimally integrating existing seismic data with new information acquired in real-time
from the well being drilled.
Often the pre-drill seismic earth model remains mostly unchanged during drilling even though real-time LWD data contain
significant new information about the formations being drilled. Real-time checkshot measurements provide constraints for
the velocity model, real-time logs reveal formation tops, and other while-drilling information such as pressure measurements,
mud weights, tests, and drilling events can be used to calibrate the earth model.
Recent advances in acquisition, processing, and integrated earth model building technologies have made this type of
utilization of seismic and while-drilling well data to provide results in time for drilling decision making a reality. Results of
two field tests are presented. First field work shows the ability of the technique to image a fault accurately in 3D space. The
second field work demonstrates the ability to predict pore pressures up to 3000 ft ahead of the bit. Predicted pore pressures
were within 0.25 ppg of the actual measured formation pressures.
Motivation
Seismic is one of the key inputs to drilling planning and execution. Pre-drill planning of a well is made using the seismic
depth image and estimated properties important for drilling such as pore pressure, fracture gradient, geomechanical properties
etc. We will call this the earth model. The motivation for our work is that this model is not unique. There are multiple models
that will fit the same surface seismic data. There is no unique velocity model, pore pressure estimate, and structural image.
We call this the seismic uncertainty and this is the cause of many risks encountered during drilling. If we reduce this
uncertainty we also reduce the risks.
SPE 164786
Figure 1 shows an example from the Gulf of Mexico. These are the interpreted horizons from seismic images of the same
target horizon. Without any further information they are all equally probable. Which is the right one?
Figure 1. Example of uncertainty in target horizon. All interpretations are equally probable based on seismic data
alone.
Figure 2 shows the overlay of 500 map migrated realizations of a fault. We see that the uncertainty across the fault plane is
about 400 ft. For a vertical well this would create a 700+ ft uncertainty on where the well will cross the fault. If we wanted to
take a drilling action at the fault location, such as setting casing, this uncertainty will cause problems.
400 ft
SPE 164786
A mark of Schlumberger
(a)
SPE 164786
(b)
(c)
Figure 4. Method for reducing velocity, geological structure, and pore-pressure uncertainty ahead of the bit by using
checkshot and log measurements down to the bit depth.
We now rebuild the earth model everywhere including deeper sections ahead of the bit as shown in blue in Figure 4 (c). We
now have a velocity model and structural image that is consistent everywhere both with surface seismic data and with well
data. One consequence of this update is the possible change in the spatial positions or drilling targets (depicted by a blue bar
in the figure) and the change in the pore pressure estimates obtained from velocity predictions ahead of the bit.
The above example shows, in a simplified way, how we can reduce uncertainty ahead of the bit by using information behind
the bit. Other external information such as geological knowledge, rock physics models, and petrophysical analysis are also
used in the process to constrain the model. We will show examples of these in the next section.
Field Examples
First example shows how a fault location is moved with the new while-drilling updating process a large distance and imaged
at the right location. The primary challenge in this project was to set a casing below a secondary fault as shown in Figure 5.
This was necessary for the hole-size requirements in the final well completion. Locating both primary and secondary faults
accurately was deemed critical. Data from one offset well were used to build a local anisotropic model extending into the new
well location. However, offset well data were limited to the deeper sections and could not be used to build a reliable pre-drill
model in the shallower section in the drilling volume of interest. Due to this and likely lateral variations in moderately
complicated geology large uncertainties were expected in positioning of events with existing seismic data. It was important
to improve the velocity model and reimage while drilling using well information from the well being drilled to reduce the
positional uncertainty of the fault locations.
SPE 164786
Real-time LWD checkshot and log data were acquired all the way from to the mudline down to complement the offset well
for a good velocity model. Anisotropic velocity models were created in several stages by seismic tomography where the
vertical velocities were constrained by well data. The volume for velocity models included the offset well to ensure a proper
tie to that well, in addition to the new well.
Figure 6. The figure on the left is the legacy image that existed prior to the project. The figure on the right is the
seismic image after the final update. Red lines are the fault interpretation on the legacy image. Blune lines on the right
are the fault interpretation on the new image. One vertical grid is 1000 ft. There was a significant shift in the spatial
locations of the faults targeted for casing point.
For each updated velocity model, a new earth model and structural image was generated during drilling, enhancing fault
location accuracy in time to impact drilling and casing decisions. The desired casing location was accurately predicted within
+/- 50 ft (Figure 6). The prediction was performed using re-imaging performed at approximately 1500 ft above the planned
casing depth.
SPE 164786
The second example demonstrates pore-pressure prediction in the Gulf of Mexico. Pore pressures are estimated from a rockphysics model that relates the measured velocity of a formation to its pore pressure. Therefore, for accurate pore-pressure
prediction we need two things: (a) good estimates of velocities and (b) a rock-physics transform appropriate for the geology
at the drilling location.
Pore pressure predictions before drilling (pre-drill predictions) are based on estimated seismic velocities and an assumed
rock-physics transform; as described earlier neither of these is accurately known particularly in exploration drilling.
Consequently, pre-drill formation pressure estimates typically have large uncertainties.
In most drilling projects, the conventional approach is to update this prediction during drilling using Logging While Drilling
(LWD) logs, mud weights, cuttings, and drilling events such as kicks and mud losses. These pore-pressure updates, however,
use logs from already drilled section of the well and cannot accurately predict ahead of the bit. No actual measurement
ahead of the bit is involved in these updates. Our new approach goes beyond this by using actual seismic velocity
measurements ahead of the bit, integrated with well data, to update the pore-pressure predictions.
Figure 7 shows the Gulf of Mexico Green Canyon area where the well is located. Geology is moderately complex with
changing dips and structures nearby. The well is vertical with a total depth of about 11700 ft. The field study is conducted to
assess the ability of the described method to estimate the pore pressures ahead of the bit. For this, reason an area with known
pore-pressure issues was chosen. At each step, 3D velocity, structural image, and pore-pressure cubes around the well
location were obtained. For clarity, in the following we will show the estimated and measured velocity functions and pore
pressures along the well trajectory only.
(a)
(b)
Figure 7. (a) Well location in deep water Gulf of Mexico. (b) Seismic cross section.
First 3D pre-drill velocity and pore-pressure estimates were obtained without well data using conventional methods as a
reference. Figure 8 (a) shows the obtained velocity function along the well trajectory. We see that velocities increase down to
about 9000 ft and then there is a velocity reversal. This reversal is associated with an increase in pore pressures.
When the well is drilled to 8000 ft, well data (checkshot and wells logs) down to this depth become available for use in our
new method to provide look-ahead estimates of velocities and pore pressures ahead of the bit. Figure Figure 8 (b) shows in
red, the velocities obtained from checkshot measurements down to 8000 ft. The checkshot velocities increase more or less
monotonically and do not yet show any velocity reversals. The measured velocities, however, do not agree with the pre-drill
model indicating the model is inaccurate in this shallower section, and it might be inaccurate as well below the bit depth.
From a drilling perspective, the challenge now is to be able to look ahead hundreds to a thousand meters and predict the
location and magnitude of the over pressures.
SPE 164786
(a)
(b)
Figure 8. (a) Pre-drill velocity estimates along well trajectory. (b) Pre-drill estimates compared with checkshot
velocities (red) measured down to 8000 ft.
Figure 9 (a) shows in blue the velocity prediction ahead of the bit down to 11500 ft with the new technique. The prediction
shows a velocity trend increasing down to about 9000 ft, followed by a sharp reversal. The amount of velocity reversal given
by this updated prediction is larger than the pre-drill estimate in black, indicating higher pore pressures ahead of the bit than
were predicted from the conventional pre-drill model. The blue curve in Figure 9 (b) shows the pore pressures predicted
using the look-ahead velocities and a velocity vs. effective stress relation calibrated with the LWD checkshot velocities and
the shallow pore-pressure estimates to invert for the parameters in the transform as described by Sayers (2010). The new
pore-pressure estimates in the 8000 11500 ft are significantly higher than the conventional pre-drill estimates.
(a)
(b)
Figure 9. (a) Look-aehad velocity prediction from the new method below 8000 ft. (b) Velocity predictions (blue)
compared with measured velocites in the deeper zone (dotted red) confirming predictions.
SPE 164786
(a)
(b)
Figure 10. Look-ahead estimates made at 8000 ft compared with actual measurements in the 8000 - 11500 ft section of
the well.
Finally, these estimates were compared with actual measurements from the well in the deeper, 8000 to 11500 ft section.
Figure 10 (a) shows in dotted red the actual velocities measured from 8000 to 11500 ft. These velocities agree much better
with the look-ahead predictions from the new method (blue) compared with the pre-drill velocities (black). In Figure 10 (b)
red dots are the actual formation pressure measurements made after this section is drilled. The agreement between the lookahead predicted (blue curve) and measured (red dots) pore pressures is seen to be excellent; whereas the pre-drill predictions
underestimate the required mud weight by as much as 2.1 ppg (0.25 g/cc) in this zone. We see that in the look-ahead
prediction zone of over 3000 ft the estimates are mostly within 0.25 ppg of the actual measurements.
Conclusions
Uncertainties in the locations of the geological features (markers, targets, faults, etc.) and the estimated formation pressures
present challenges and risks in drilling projects. We have presented a new technique where we reduce these uncertainties
ahead of the bit as the well is being drilled. This new process involve integration of surface seismic data, well data (LWD
or wireline), rock-physics information, logistics and planning, and fast turnaround times for in-time results.
We have shown with a field example that structural uncertainty in the location of geological features such as faults, drilling
targets, or markers can be significantly reduced. A second field example demonstrated that uncertainty in the pre-drill pres
pressure estimates can be significantly reduced and pore pressures up to 3000 ft ahead of the bit can be estimated within 0.25
ppg.
References
Esmersoy, C., Hawthorn, A., Durrand, C. and Armstrong, P., 2005, Seismic MWD: Drilling in time, on time, its about time:
The Leading Edge, 24, 56-62.
Esmersoy, C., Kania, A., Kashikar, S., Ramirez, A., Hannan, A., Lu, L., Teebenny, S. and Duan, L., 2011, Optimum Use of
Seismic Data to Reduce Drilling Risk and Improve Well Placement: 73rd EAGE Conference & Exhibition, Extended
Abstracts, C046.
Osypov, K., Nichols. D., Woodward, M., Zdraveva, O., Qiao, F., Yarman, C.E., Vyas, M., Yang, Y., Liu, Y., and Ivanova,
N., 2011, From quantifying seismic uncertainty to assessing E&P risks and the value of information: 81st SEG Annual
International Meeting, Expanded Abstracts, 30, 3683.
Sayers C.M., 2010, Geophysics under stress: Geomechanical applications of seismic and borehole acoustic waves. SEG
Distinguished Instructor Series, No. 13. ISBN 978-1-56080-210-5.