We-01-05 Ekofisk PRM: A Look at Some of the Technology and How it Might Evolve H. Hoeber* (CGG), S. Buizard (CGG), A. Gresillaud (CGG), S. de Pierrepont (CGG), A. Bertrand (ConocoPhillips), P.G. Folstad (ConocoPhillips), A. Grandi (Total) & H. Nakstad (Optoplan)
SUMMARY In this paper, we take a closer look at some of the technology used and the research carried out over the first two years on the Ekofisk PRM project. This includes QC issues for the acquisition; signal processing issues (such as 4D robustness of rotations; interference and VZ noise methods; surface consistent amplitude corrections); sensor performance and solutions to problems found; optimal imaging solutions such as FWI. Furthermore we report on the evolution of the Ekofisk operations in the coming months and discuss opportunities and possibilities arising via enhanced acquisition technology, improved processing and imaging, data management, and the potential for other reservoir oriented methods such as passive seismic monitoring.
Second EAGE Workshop on Permanent Reservoir Monitoring 2013 – Current and Future Trends Stavanger, Norway, 2-5 July 2013
Introduction 4D seismic is an important tool for reservoir management at the Ekofisk field in the southern part of the Norwegian North Sea. As subtle 4D seismic changes related to production and injection develop rapidly, frequent and highly repeatable 4D seismic monitoring is required. In 2008 ConocoPhillips decided to install a permanent reservoir monitoring (PRM) system at Ekofisk. An Optowave fibre optic system, consisting of nearly 4000 mostly buried multi-component sensors was chosen as the best long term solution to support the intensive drilling program planned for the next 15 years and for fast delivery of high fidelity 4D seismic products (Folstad et al., 2011). The system was installed according to plan and was fully functional in October 2010. Four Life of Field Seismic (LoFS) surveys have been acquired so far and two surveys are planned in 2013, one of which is ongoing at the time of writing. As a brief reminder, the Ekofisk PRM system consists of 24 cables, separated by 300 m, giving roughly 200 km cable length and coverage of about 60 km2. The total array has 3996 4C receivers at 50m spacing, the majority of which are trenched 1-1.5 meter deep into the sea-floor. The acquisition is performed parallel to the cable direction with a 25m shot and 50m line spacing. Each acquisition has approximately 120,000 shots over an area of 143km2, and follows a fixed preplot pattern (Bertrand et al., 2013a). CGG is currently responsible for all the main elements of the seismic delivery chain for the Ekofisk LoFS project: equipment supply, acquisition and processing (Buizard et al., 2013). The involvement of a single contractor group enables good coordination between the different phases of the project. As a result, processing turnaround was reduced to less than 4 weeks from final shot. For the last two years, the Ekofisk Permanent Reservoir Monitoring Project has delivered high quality time-lapse data with clean 4D signals and NRMS of around 3-5%. In this paper we look at some aspects of research and development, focussing on the seismic processing and imaging, that has been carried out since the system was installed in 2010. Optimal 4D/PRM Processing The focus of the first two years of operations on Ekofisk has been on setting optimal procedures across all phases of the project. The seismic imaging team, in particular, was tasked with finding a robust processing sequence able to deliver high fidelity 4D seismic within a turnaround time of 4 weeks from the end of the acquisition. This needed to factor in some knowns, such as a method to remove the “tank noise” diffracting off the obstructions (Ekofisk’s large concrete tank), and of course many unknowns, such as the various marine noise sources. In addition, with this being the first large scale installation of a subsurface fibre optics installation, the processing team, in conjunction with the co-located acquisition QC staff, were tasked from the beginning to monitor closely the quality of the data during and after each of the acquisitions. Frequent repeat processing with severely limited turnaround throws up some challenges, which, of course, will be of interest to 4D processing in general. Unlike in conventional parallel 4D processing, each new vintage is processed with the same repeat sequence as the base. One of the key questions then, both for repeatability and for speed, is whether the same detailed processing parameters can also be applied in this manner. For example, some vintages will be contaminated by much stronger seismic interference (SI) than others; as an example, we find that the SI technique we use can be safely tuned for harsh noise and still be repeated in a 4D friendly manner, when SI is weaker. It was also clear, early on in the project, that processing would be split into two phases. Firstly, as the data are streamed in near real-time into the office in Stavanger, where acquisition QC staff and the processing team are co-located, we perform all acquisition QCs and all shot-domain processing. For the second part of the processing, which is more compute intensive, the data are transferred to CGG’s compute hub in the UK. In order to limit the size of the data early in the sequence, PZ summation is part of the shot-domain processing; this implies that necessary denoise steps to remove VZ (lowfrequency, low-velocity) noise on the Z components also have to be applied at this stage, rather than Second EAGE Workshop on Permanent Reservoir Monitoring 2013 – Current and Future Trends Stavanger, Norway, 2-5 July 2013
waiting for a full 3D aperture data configuration. We therefore designed a noise removal sequence on 2D gathers, in which the tank noise and VZ noise are removed in a combined processing step. Whilst the tank noise can be easily modelled for subtraction, the VZ noise is efficiently attacked on shot gathers, where the noise is incoherent, using FX projection filtering. We enhance signal preservation by a conditioning step in which we remove primary energy. Figure 1 shows this processing step and its impact on receiver gathers and stacks. Primaries are well preserved and noise is efficiently removed. The process is 4D friendly and it commutes well with the actual PZ summation operation, which further allows us to simplify the processing flow.
Figure 1 VZ and tank noise removal. Receiver gathers (top) and stacks (bottom) before (left) and after (middle) application of the combined VZ and tank noise removal, and the noise removed (right most column). P and Z components are calibrated and summed. All aspects of repeatability, that of the system, the acquisitions, and the processing, are clearly of paramount importance in a PRM project. As the larger part of the 3996 receivers are trenched, one can reasonably expect receiver repeatability to be less of an issue than that of the shots. A repeatability analysis of the rotation angles between the first two acquisitions is shown in figure 2 and confirms this to be the case. Rotations were calculated with a scanning technique (Li et al., 2004) which we find to be a little more robust than an inversion method. All angles have median 4D difference values of near perfect 0 and the histogram has a standard deviation of less than 1 degree. However, at the end of the LoFS1 acquisition several sensors started displaying a variation in recorded pressure, and hence amplitude dimming, between primary and reshoot lines; this behaviour persists on subsequent surveys. However, we are able to remove this effect in processing, as we have sufficient redundant data to use matching operators to calibrate to unaffected data (see Nakstad et al., 2013). In order to validate the optimal type of correction, done early on in the sequence, we fast- tracked several alternative solutions through final imaging and time-shift analysis on provisional stacks and picked the best method. Figure 3 shows a subset of receivers with our optimized solution; on input, three shots are clearly impacted by this effect. After corrections, we have successfully removed the variations across gathers as shown on the 3D and 4D images. Solving this kind of problem benefits substantially from close collaboration between all parties, across contractor and operator, as well as within the CGG divisions executing the project components.
Second EAGE Workshop on Permanent Reservoir Monitoring 2013 – Current and Future Trends Stavanger, Norway, 2-5 July 2013
Figure 2 Repeatability analysis between LoFS2 and LoFS4 of one of the rotation angles. The angles have near perfect repeatability, so that the rotations from the first survey can be applied to subsequent vintages.
Figure 3 Receiver gather before (top left) and after (top right) pressure sensitivity corrections; bottom row shows the impact on the 4D difference (left before, right with correction). The 4D difference after corrections shows good continuity across all receivers. Future Directions & Conclusions Significant effort has gone into designing and optimizing a simultaneously fast and robust PRM processing sequence. Sensor QC and calibration represented a substantial part of this effort. Final NRMS values of 3-5%, and excellent interpretational results (Grandi et al., 2013, Lyngnes et al., 2013) validate the processing effort to date. Re-processing with an updated sequence, incorporating some of the lessons learnt, is ongoing at the time of writing. Other processing steps still under review are: Surface consistent corrections, where we are also considering a 4D correction step. Water velocity corrections: although small, at 75-80 meter water depth, these can clearly impact 4D timeshifts. We can explore the use of regular full water column measurements that have been acquired, to see if this provides an uplift over applying daily averages. Wavelet variations: We can investigate the use of continuous near-field hydrophone recordings to control wavelet stability. Imaging: To date, Ekofisk PRM 4D processing applies Kirchhoff time-domain imaging with the same velocity model. Time-shifts and time-strains are then calculated from the different vintages. Recently, Bertrand et al. (2013b) showed a velocity building workflow using Full Waveform Inversion with both well-tie and PP-PS tomography. For this application the data have been processed in COV volumes (different to
Second EAGE Workshop on Permanent Reservoir Monitoring 2013 – Current and Future Trends Stavanger, Norway, 2-5 July 2013
the PRM project) and with additional azimuthal RMO. There is significant uplift in the Ekofisk depth image with this high-graded velocity model. We may explore the advantages of this model in 4D. In addition, and notwithstanding that the FWI model refracted data only, we may ask if the improved velocity model can beneficially be employed as a starting model to search for differences relative to subsequent vintages. It is of course also our intention to continue with the 4D data analysis in stack and pre-stack domain, particularly also in order to further validate the processing sequence. Acknowledgements We thank the PL018 Partnership (ConocoPhillips Skandinavia AS, Total E&P Norge AS, ENI Norge AS, Statoil Petroleum AS and Petoro AS) for permission to publish this work. We thank all colleagues across ConocoPhillips, CGG and Optoplan who have contributed to the Ekofisk PRM project to date. References Bertrand, A. et al. [2013a] The Ekofisk Life of Field System: experiences and results after two years in operation; to be presented at the 75th EAGE Conference, London. Bertrand, A. et al. [2013b] WAZ PP/PS depth imaging at Ekofisk LoFS using FWI; to be presented at the 75th EAGE Conference, London. Buizard, S. et al. [2013] Ekofisk Life of Field Seismic: 4D Processing; to be presented at the 75th EAGE Conference, London. Folstad, P.G. et al., [2011] Ekofisk PRM – The technical case for this brand new installation. EAGE Workshop on Permanent Reservoir Monitoring (PRM) – Using Seismic Data, Expanded Abstracts. Grandi. A., Lyngnes, B. and Haller, N. [2013] Reservoir Management through Frequent Seismic Monitoring at Ekofisk Field; to be presented at the 75th EAGE Conference, London. Li, J., Jin, S. and Ronen, S. [2004] Data-driven tilt angle estimation of multi-component receivers; SEG Technical Program Expanded Abstracts 921-924. Lyngnes, B. et al. [2013] Life of Field Seismic at Ekofisk: Utilizing 4D seismic for evaluating well target; to be presented at the 75th EAGE Conference, London. Nakstad, H., Eriksrud M. and Valla, S. [2013] Pressure Recordings at Ekofisk - Experience and Route Forward; EAGE Workshop on Permanent Reservoir Monitoring (PRM) – Using Seismic Data, Expanded Abstracts.
Second EAGE Workshop on Permanent Reservoir Monitoring 2013 – Current and Future Trends Stavanger, Norway, 2-5 July 2013
