Wytch Farm - A reprocessing case study...Wytch Farm – A Reprocessing Case Study 2 Figure 1 Left: representative shot gathers on input from the 1999 survey (1.8kms maximum offset)

1

Wytch Farm – A Reprocessing Case Study Tom Massip1, Alain-Christophe Bon2, Phil Smith1, Henna Sherazi-Selby1, Gordon Roberts3 1 ION GX Technology, 2 Perenco UK, 3 Perenco Holdings Abstract Multi survey vintage data acquired over the Wytch Farm field in Dorset (off the south coast of England) were previously processed in 2008 resulting in a Kirchhoff depth image. However, due to noise associated with shallow water and variable data quality, especially for some older 1992 and 1999 3D data, the results were sub-optimal. Recent enhancements in shallow water demultiple techniques and noise removal methods permitted a detailed re-working of the pre-processing sequence, resulting in cleaner gathers with useful information at their longest offsets. These enhanced gathers were input to iterative velocity model update, using non-parametric moveout picking in conjunction with hi-resolution tomographic inversion. Subsequent Kirchhoff and beam migration produced an improved depth image from the four multi-survey input data volumes.

Introduction

Wytch Farm is one of Western Europe’s largest onshore oil fields, located in Dorset, England. The field was discovered in the 1970’s in the Bridport Jurassic sandstones and the Sherwood Triassic sandstones at a depth of about 1500m. After a production plateau of 110,000 bbl/day in the 1990’s, the field has entered its mature phase. Operating efforts are now focused on optimizing the recovery of the remaining oil. A better geophysical imaging of the structure was deemed necessary to understand the production pattern and to identify remaining reserves. In particular, the poor signal-to-noise ratio at the reservoir level and the poor definition of faults in some areas motivated the decision to reprocess the available 3D seismic data. Multi survey vintage data acquired over Wytch Farm were previously processed in 2008 resulting in a Kirchhoff depth image. However, due to noise associated with shallow water (~20m depth) and variable data quality, especially for some older 1992 and 1999 3D data, the results were sub-optimal.

Recent enhancements in shallow water demultiple techniques and dispersive noise removal methods permitted a detailed re-working of the pre-processing sequence, resulting in cleaner gathers with useful information at their longest offsets (this long-offset information was not preserved in the 2008 processing). These enhanced gathers were input to iterative velocity model update, using non-parametric moveout picking in conjunction with hi-resolution tomographic inversion. Subsequent Kirchhoff and beam migration produced an improved depth image from the four multi-survey input data volumes. The data under consideration comprised three vintages with mixed azimuths: 1992 (900 azimuth), 1999 (00 azimuth), 2003 (00 azimuth), and 2003 (600 azimuth), with maximum offset ranges 1.2km, 1.5km, and 1.8km respectively. Data Quality Data of variable quality and limited offset distribution, as can be expected when mixing vintages from the 1990’s, pose a serious challenge for noise suppression, due to the under-sampling of the noise trends. Contemporary interpolation techniques combined with various de-noise methods were able to significantly improve on the previous (2008) processing of these same data. Preserving more usable offset information, by removing noise at those offsets, was a benefit to the subsequent velocity analysis, and especially for the autopicking involved in the tomographic update of the velocity model for preSDM. Figure 1 shows three representative shot gathers from the input data and then after several passes of noise suppression. For such shallow water, we often have problems with both guided waves and mud roll, and if we are unable to adequately suppress these noise trains at the far offsets, then we seriously limit our velocity resolution.

Following these comparisons through to a brute stack, gives the images shown in figures 2 & 3, for the input brute stack and pre-processed stack (respectively)

Page 4289SEG Denver 2014 Annual MeetingDOI http://dx.doi.org/10.1190/segam2014-0048.1© 2014 SEG

Main Menu

T

Wytch Farm – A Reprocessing Case Study

2

Figure 1 Left: representative shot gathers on input from the 1999 survey (1.8kms maximum offset). Right: shots after swell noise removal, linear and guided wave removal, shot and receiver tau-p deconvolution, shallow water multiple suppression, SRME, diffracted noise suppression

Figure 2: representative brute stack. Short-period multiple contamination is evident

Figure 3: stack after swell noise removal, linear and guided wave removal, shot and receiver tau-p deconvolution, shallow water multiple suppression, SRME, diffracted noise suppression


Main Menu

T


3

Velocity model building

Five iterations of hybrid gridded tomography (Jones, 2010) were employed in the velocity model update, using non-parametric picking of residual moveout after each iteration of Kirchhoff 3D preSDM.

Figure 5: well logs showing good correspondence between the sonic (grey) or its 50m smoothed average (blue) and the hi-resolution tomographic result (orange and brown)

Calibration to several well markers was also performed to assist in updating the anisotropy parameter fields (epsilon was taken as being equal to twice delta). Figure 4 shows a cross line where the high velocity chalk layer is evident at about 700m depth, below which is a velocity inversion. The interval velocity structure derived via tomographic inversion is confirmed by well control. Figure 5 shows several sonic logs overlain with the interval velocity obtained from tomography: for the most part, there is good correspondence at the scale length of the seismic wavelengths.

Figure 4: 3D Kirchhoff preSDM with interval velocity model overlay after the 4th iteration

Figure 6 shows a comparison of the 2008 vintage processing and the final 2013 preSDM image, showing improved imaging especially at the sea bed (where we can see sub-cropping dipping reflectors) and below improved reflector continuity below the chalk. Faulting at the chalk level is also better resolved.


Main Menu

T


4

Figure 6: comparison of vintage and final preSDM image (converted to time), showing improved imaging especially at the sea bed (dipping reflectors) and below the chalk.

Conclusions

It is well known that reprocessing seismic data after a few years can often result in uplift in the data quality. In addition, if the uplift in the pre-stack data quality is significant, then subsequent velocity model building can be enhanced resulting in greatly improved imaging. This has been the case here.

Acknowledgements

We thank Perenco UK, its partners on the Wytch Farm field, Premier Oil, Maersk Oil, Talisman Energy and Summit Petroleum, and ION Geophysical for permission to present this work and to Ian Jones for help in preparing the material.


Main Menu

T

http://dx.doi.org/10.1190/segam2014-0048.1 EDITED REFERENCES Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2014 SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web. REFERENCE

Jones, I. F., 2010, An introduction to velocity model building: EAGE.


Main Menu

T

Documents

Wytch Farm - A reprocessing case study...Wytch Farm – A Reprocessing Case Study 2 Figure 1 Left: representative shot gathers on input from the 1999 survey (1.8kms maximum offset)