Downloaded 09/16/16 to Redistribution subject to SEG license or copyright; see Terms of Use at

Transcription

1 Ehsan Zabihi Naeini*, Ikon Science & Russell Exley, Summit Exploration & Production Ltd Summary Quantitative interpretation (QI) is an important part of successful Central North Sea exploration, appraisal and development activities. Accurate determination of hydrocarbon facies is particularly vital as the oil and gas industry currently faces low oil prices and fewer subsurface opportunities. This paper presents an integrated workflow on a recent North Sea discovery using broadband seismic data and a new joint-impedance and facies based inversion. The focus was, in particular, on analyzing the best and worst case scenarios for the distribution of facies to help optimize future appraisal and development decisions. Introduction De-risking, via QI, is an essential part of successful Central North Sea exploration and appraisal where new discoveries tend to be close to economic limits. This is currently particularly important as the oil and gas industry faces a prolonged period of low oil prices with a subsequent decline in exploration activities. Additionally, the North Sea is a mature basin with numerous undeveloped discoveries which could be economically viable if key uncertainties are reduced. To do so, it is essential to use the latest, state-of-the-art technologies. An example is shown here utilizing a broadband long offset seismic dataset, broadband well tie estimation, followed by a newly developed facies based seismic inversion. The case study shown in this paper centers on a Paleocene discovery, known as Avalon, in block 21/6b of the UK Central North Sea located at the north-western edge of the Central Graben just south of the Buchan Field. The discovery was initially made using conventional simultaneous pre-stack inversion followed by a discovery well that successfully drilled an 85 ft column of oil in good quality sands. The reservoir sands lie within the proximal part of the prolific northwest to southeast late Paleocene Forties and Cromarty depositional trend. This fairway includes the giant Forties Field. Locally, Cromarty sands directly overlie and down-cut into the underlying Forties sands and Lower Sele shales along the Dornoch shelf edge. The Balder and Upper Sele shale intervals typically act as the regional seal to Cromarty and Forties hydrocarbon accumulations. Generally, Cromarty and Forties reservoirs have high porosities, high net-to-gross and a high degree of lateral and vertical connectivity. As a result these sand fairways act as important conduits for the lateral migration of hydrocarbons and make these reservoirs particularly suitable for AVO based inversion techniques. Method This paper demonstrates a workflow using a novel facies based Bayesian seismic inversion technique to analyze the distribution of reservoir bodies through a range of facies based sensitivities. Facies based seismic inversion was introduced by Kemper and Gunning (2014) in which the low frequency model is a product of the inversion process itself, constrained by per-facies input trends, the resultant facies distribution and the match to the seismic. So the inversion benefits from a rock physics model (and therefore a low frequency model) per-facies to optimize the inversion. This new Bayesian inversion system simultaneously inverts for facies and elastic properties. In this study the input seismic consisted of conventionally acquired but broadband processed data with two important processing steps as follows. Firstly, a pre-imaging deghosting technique, for broadening the bandwidth of the conventionally acquired towed streamer data, was used to remove the frequency notches caused by ghost wavelet interference. Secondly, the processing workflow included a multi-layer, non-linear, slope tomography to derive the velocity model for imaging and Kirchhoff pre-stack depth migration. The advantages of using such broadband seismic data have previously been demonstrated in the literature (e.g. Zabihi Naeini et al., 2015) providing increases in both the low and high frequency signal thereby enhancing resolution. The presence of seismic signal at low frequencies however is more important in the context of seismic inversion as it specifically helps reduce the dependency on the initial low frequency information. QI workflows often consist of rock physics analysis, fluid substitution, synthetic modeling, followed by well tying and subsequent inversion to elastic properties and facies. Zabihi Naeini et al. (2016a) demonstrated an example of the importance of an accurate well tie (and therefore accurate wavelet estimation) for inversion, specifically when using broadband seismic data. They concluded that one has to use broadband wavelets when inverting broadband seismic to fully benefit from the broad signal bandwidth. The problem of wavelet estimation for broadband seismic data, however, arises during the well tie process when the length (in time) of the well-logs is often seriously inadequate to provide sufficient constraints on the low frequency content of the resulting wavelet. Zabihi Page 2906

2 Naeini et al. (2016b) discussed this problem in detail and proposed three different solutions to overcome this issue. In this study one of their proposed wavelet estimation techniques was implemented, namely the parametric constant phase method to tie the seismic to the well and consequently use the wavelet for inversion. North Sea Case Study Figure 1 shows the well tie panel and the estimated wavelet for the mid-angle stack. The constant phase assumption helps reduce the degrees of freedom for wavelet estimation and results in a more stable wavelet for short log sequences. One can also observe reasonable low frequency decay on the amplitude spectrum obtained inherently as part of this technique by using multi-taper spectral smoothing and averaging over many traces around the well. A good quality well tie can be observed with a crosscorrelation coefficient of 0.78 and a phase error of approximately 10 degrees. Similar quality well ties were also achieved for the other angle stacks. Initial rock physics and forward modelling studies revealed the Avalon discovery to exhibit a text-book Class 3 AVO (Rutherford and Williams, 1989) anomaly from the top reservoir reflector. Figure 2 shows the RMS amplitude map from around the Avalon discovery for both the near and far partial angle stacks. The main reservoir anomaly is evident around Well 2. The first and most critical step for the joint impedance and facies based inversion technique was to derive impedance depth trends for each facies. From these per-facies depth trends equivalent low frequency models are generated, an essential input to the algorithm. The depth trends are shown in Figure 3 where five facies are classified: Overburden hard shale, overburden soft shale, intra-reservoir shale, oil sand and brine sand. The presence of soft shale can also be observed in Figure 1 just above the reservoir. Separating the various shales into different facies types was a critical factor to improve the inversion accuracy. Subsequent to running the inversion to derive facies and elastic properties, QC was performed. Figure 4 shows a resulting facies section on an arbitrary line crossing both available wells in this study, which shows an optimized facies match at both wells. After careful QC, the inversion was run in 3D with optimized parameters. A key input of the inversion to facies and elastic properties are the prior facies proportions which were estimated from the discovery well, but there was of course some uncertainty in these proportions away from the wells. In Figure 5 (left) we show the oil sand time thickness maps (readily constructed by summing the oil sand facies samples over the inversion window) for two end member scenarios, to investigate the sensitivity of the prior facies proportions. Also, one could further analyze the overall connectivity of the oil-sand facies and potential satellite anomalies in 3D (Figure 5, right). Figure 1: Panels of petrophysical and elastic properties including the brine (blue), oil (green) and gas (red) saturated cases from the discovery well (Well 2). Petrophysically derived facies before and after up-scaling are also shown in 6th and 7th panel which were used to QC the inverted facies. Well tie panel is the last panel along with the estimated wavelet for the mid-angle stack. Page 2907

3 The final optimized inversion results (prior oil sand proportion of 3%) demonstrated a very accurate correlation between measured (in the wells) and modelled (from the seismic inversion) acoustic & elastic impedances and resulting facies (Figure 4). The inversion facies output provided a good result not only matching the oil column thickness but also the brine filled sands and shales as encountered in the calibration wells. The inversion also successfully delineated a thin shale layer below the oil column observed in the well (previously unobservable using conventional simultaneous inversion) that had significant impact on the understanding of potential water drive during production. Additionally, the output of this novel inversion technique provided the ideal framework to quickly and efficiently generate static and dynamic reservoir models with the facies based output being very similar to a geo-cellular format. Also, of key importance was that the facies output was generated without the need for qualitative and potentially biased interpretation of conventional impedance products. Conclusions Figure 2: Reservoir RMS amplitude maps on near and far angle stacks. Facies based seismic inversion has been demonstrated, via a North Sea working case study, to provide significant advantages over more conventional impedance inversion techniques. When facies based inversion is combined with broadband data and appropriate broadband well tie techniques the resulting classified facies output provides a result ideally suited for geological interpretation and the generation of static and dynamic reservoir models. The joint impedance facies inversion technique successfully: Provides correlation wells. a better facies with calibration Inverts for an optimum low frequency model thereby removing one of the most significant sources of error in more conventional simultaneous inversion techniques, where a low frequency model is an input, not an output. Reduces interpretation burden by producing facies based output akin to a geo-cellular model. Figure 3: Depth trends for each facies. Allows a full range of potential sensitivities to be explored (Figure 5) therefore exploring the implications of inversion error. Page 2908

4 Figure 4: Inverted facies section shows a good match at wells (prior oil sand proportion is 3%). Figure 5: Left figures show the hydrocarbon time thickness map (in ms) and the right figures show the oil-sand facies in 3D obtained using facies based inversion in two different scenarios. Acknowledgements The authors would like to thank Friso Brouwer, Kester Waters, Denis Alexeenko, Michael Kemper, Richard Saxby and Andrew Howard for their contributions. Enquest Plc, Summit s partners in the 21/6b Block, are also thanked for their technical input and permission to publish. Finally, CGG are thanked for permission to publish results generated from their CornerStone seismic dataset and in particular Steve Bowman. Summit Exploration & Production Ltd is a wholly owned subsidiary of Sumitomo Corporation, Japan. Page 2909

5 EDITED REFERENCES Note: This reference list is a copyedited version of the reference list submitted by the author. Reference lists for the 2016 SEG Technical Program Expanded Abstracts have been copyedited 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. REFERENCES Kemper, M., and J. Gunning, 2014, Joint impedance and facies inversion Seismic inversion redefined: First Break, 32, Rutherford, S. R., and R. H. Williams, 1989, Amplitude-versus-offset variations in gas sands: Geophysics, 54, , Zabihi Naeini, E., N. Huntbatch, A. Kielius, B. Hannam, and G. Williams, 2015, Mind the gap Broadband seismic helps to fill the low frequency deficiency: 77th Annual International Conference and Exhibition, EAGE, Extended Abstracts, Zabihi Naeini, E., M. Sams, and K. Waters, 2016a, The impact of broadband wavelets on thin bed reservoir characterisation: 78th International Conference and Exhibition, EAGE, Extended Abstracts, WS01 B02. Zabihi Naeini, E., J. Gunning, R. White, and P. Spaans, 2016b, Wavelet estimation for broadband seismic data, 78thInternational Conference and Exhibition, EAGE, Extended Abstracts, Tu SRS3 06. Page 2910