Third EAGE Workshop on Rock Physics

15-18 November 2015 Istanbul, Turkey

Improved Seismic Inversion and Facies Using Regional Rock Physics Trends: Case Study from Central North Sea

A.V. Somoza*, K. Waters & M. Kemper (Ikon Science Ltd)

Introduction

Kemper and Gunning (2014) introduced a new seismic inversion system, where key inputs are per-facies low frequency background models (instead of using only one low frequency background as used in simultaneous inversion). These per-facies low frequency background models are typically derived by fitting a simple compaction model to the corresponding well elastic logs, limited to the facies in question, and referenced to an appropriate datum. As such the wells used to make this compaction model need not come from wells within the boundaries of the seismic datasets to be inverted. In this paper we will demonstrate that deriving these trends elsewhere in the same basin results in accurate facies and impedance predictions at the wells within the seismic datasets being inverted, without these wells being used at all. Case study

Upper Palaeocene log data from 30 wells located in the Central North Sea, covering an area of 35,000 sqkm, were analysed using clustering analysis to determine what elastic log responses could be differentiated and to what geological facies they relate. Based on this analysis, the log responses were classified into soft shale, hard shale, brine and oil bearing reservoir. These facies categories, in conjunction with the corresponding elastic logs (Vp, Vs and Rhob), were used to generate per-facies compaction trends, referenced to the mudline (Figure 1).

These per-facies trends were subsequently used to build per-facies low frequency background models for two seismic datasets, to the north of the 30 wells mentioned. This is achieved, per facies, by ‘hanging’ the relevant trend from the seabed horizon at each trace location in the 3D survey area.

For both seismic surveys statistical wavelets were generated per partial angle stack. In other words, the wells within the seismic surveys were neither used to derive the per-facies trends nor to derive the wavelets – they are not used at all, other than for display.

The objective of the two inversions was to image both the Palaeocene sands and hydrocarbon distributions within the Forties Formation in the Forties field (Rose et al., 2011) and within Upper Balmoral Member of the Montrose Group in the Brenda field (Jones et al., 2004).

Results

Inverted impedance and facies results were compared with impedance and facies logs of Forties and Brenda field wells (again, this is a blind test because data from these wells were not used at all). Figure 2 shows this comparison. Overall, there is a good agreement between the inverted facies and facies logs. Also a good match in terms of AI and Vp/vs is observed at well A (Forties field). However, this match is poorer for Well E (Brenda field); this can be explained by the poor quality of the seismic data, the gathers needed further conditioning but this was outside the scope of work.

Third EAGE Workshop on Rock Physics

15-18 November 2015 Istanbul, Turkey

Figure 3 shows the impedances (AI and Vp/Vs) and facies sections inverted from the Brenda and Forties seismic datasets. In both cases the inverted facies images accurately depict the oil-water contacts (transition from green (oil-bearing sand) to yellow (water-bearing sand) going downwards), and also deliver accurate estimates of the absolute elastic rock properties, AI and Vp/Vs. Due to the nature of the inversion, the AI and Vp/Vs transition between layers is more discrete/blocky.

Net sand and net pay attribute maps were calculated from the Forties facies volume over the reservoir interval (Figure 4). It can be seen in detail channel sands with hydrocarbon accumulations where good sandbodies are located (sands with an average porosity of 27% (Rose et al., 2011)).

Conclusions

Thirty wells from the Central North Sea were used to generate compaction trends to build per-facies low frequency background models for two seismic datasets to the north of these wells. The two case studies prove that trends from a neighbouring area can be used to invert seismic data with limited or no well control and obtain, with high level of accuracy, absolute impedances (from which, per facies, rock properties can be derived using facies-specific rock physics models). In this case study wells were used within the same basin; work is ongoing to assess to what degree wells in similar basins (conjugate margins?) can be used/borrowed to achieve the same high quality result.

Figure 1 Regional compaction trends referenced to the mudline, per facies, generated using log data from 30 wells located in Central North Sea.

Third EAGE Workshop on Rock Physics

15-18 November 2015 Istanbul, Turkey

Figure 2 Comparison of inversion outputs and insitu logs at the wells location.

Figure 3 AI, Vp/Vs and Facies from facies-impedance inversion. The impedances and facies are compared with up-scaled log data at the wells location. Facies classification are: soft shale in purple, hard shale in brown, brine sand in yellow and hydrocarbon sand in green. AI scale varies from 5000 g/cc_m/s (blue) to 9000g-cc_m/s (red). Vp/Vs scale varies from 1.5 (blue) to 2.8 (red).

Third EAGE Workshop on Rock Physics

15-18 November 2015 Istanbul, Turkey

Figure 4 Net sand (left) and net oil (right) attribute maps in ms (TWT) from facies-impedance inversion. Scale for both maps is from 0 to 40ms. Forties field.

References

Jones, I. F., R. Christensen, J., Haynes, J., Faragher, I., Novianti, H. Morris, and G. Pickering [2004] The Brenda field development: A multi-disciplinary approach: First Break, 22, 85–91. Kemper, M. and Gunning, J., [2014], Joint Impedance and Facies Inversion – Seismic inversion redefined, First Break, 32, p89-95. Rose, P., Barker, G., Koster, Klaas and Pyle, J. [2011] Forties Infill Drilling Eight Years On; Continued Success Through the Application of Thorough Development Geoscience Driven by 4D Seismic, SPE 145433.