Integration of Processing and Reservoir Geophysics

1. Avo Inversion and Processing: Dedication and Integration Marcel Zwaan, Yvan Charreyron, Dave Bateman Shell E&P Europe

3. Signal processing Psdm initial velocity model Psdm Prio cube Velocity model updating Final Volume migration AI/EI/ AVO Inversion Discover Need for Further processing A project is a series of “throwing over the fence exercises” petrophysics The “Problem”

5. Pre-processing Psdm initial velocity model Psdm Prio cube Velocity model updating Final Volume migration AI/EI/ AVO Inversion Inversion Feasibility Identify Post-processing steps Project integration is enabled By a feasibility step in parallel to the depth migration cycle Petrophysics (QC) The “Solution”

8. Discovered 1974 - Sub Sea tie-back to Brent-C - 65 km Production Flow-line Penguins Cluster Brent Charlie The Penguin Field Some facts

10. Get the expectations right: Are Fluid and Litho discriminations possible?

11. SST versus Shale discrimination possible Vshale logs From top Brent To TD Vshale coloured – Versus P-Impedance and S-Impedance Vshale logs From top Brent Top Dunlin Shale Volume X-Plots IP IS IP IS

12. Logs with synthetics (Aki-Richards) Synthetic: Left brine – Right gas condensate Brine and Hydrocarbon synthetics brine Gas condensate Offsets: 300 – 3000 m P-Sonic Density S-Sonic Poisson’s P-Son_chk

14. Wavelet estimation and near far stack alignment Aim: check the alignment of the data

15. Wavelets estimated from the Near and Far seismic respectively Near wavelet Far wavelet Wavelets

16. Left: Near Right: far (Near and far wavelets estimated) Near match good – far match: misalignment – spectral balancing needed near stack far stack Seismic to well tie – and Frequency contents Blue: synthetic Red: trace at well location P-Sonic – S-Sonic - Density P-Sonic – S-Sonic - Density

17. Conclusion: Near (mid) and far stacks need to be aligned And the spectra need to be balanced to a common reference (mid stack)

18. Angle versus offsets How to compute the angle versus offsets QC of angle versus offsets with angle versus offset in the well

19. Colour: Angle – every colour is an angle slot of 3 degrees. Angle versus offset functions at well location compared with angle vs offset relation from P-Sonic log Angles behave linear with offset offset time Angle values Colour 6 12 18 24 25 36 Angle versus offset Gathers

20. Relation between angle and offset Compute angles versus offset by the Walden approximation: (A.T. Walden: Geophysical Prospecting 39, pp. 915 942 1991) with and Snell’s law: X Ttwt, Vrms Vint

22. Pre- and Post stack diagnostics

23. The aim of these diagnostics is to assess whether the data – be it pre stack data or near mid and far sub-stacks – satisfy the two term AVO behaviour: Where L is the intercept and M is the Gradient. Does the data obey the classical Aki and Richards Amplitude versus offset trend? (cf. “Making AVO Sections More Robust” by Andrew Walden, BP, 52 nd EAGE Meeting Copenhagen, 1990 ). Two term AVO equation - Diagnostics

24. Diagnostics on pre-stack data amplitude Fix a zero offset time – Fit the Aki and Richards equation Output: L and M Compute the amplitude from The Aki and Richards equation from the given L and M Subtract this amplitude from the data amplitude and square it Sum (stack) this error data over The different sub-stack ranges – e.g. near mid and far ranges Output: Error cubes Red=positive impedance Offset or angle offset Amplitude of real data Compute error Per sample error Compute the squared error amplitude

25. Normalised AVO ERROR – Near - Mid - Far The relative high values of these maps are considered to correspond to large errors in the data AVO behaviour Error on Near – Mid – Far sub-stacks

26. Large residual move-out visible at top Brent (may become better after model update) Red=positive impedance Error on Depth Migrated Gathers X-unconformity Top Brent Top Dunlin

27. PSDM Gather at locations of large error Interference/Noise Or residual move-out? Red=positive impedance Error on Depth Migrated Gathers X-unconformity Top Brent Top Dunlin

28. PSDM Gather at locations of small error Reasonably flat top brent Red=positive impedance Error on Depth Migrated Gathers X-unconformity Top Brent Top Dunlin

30. Left: original – Right: pre-stack multiple removal (TX deconvolution) and post-stack multiple removal Red=positive impedance Processing after Migration

32. M estimates. Left: from near and mid – Right: from near and far The Mid map shows larger M values as compared to far M (same colour-scale) M estimated on Near – Mid – Far sub-stacks

33. M Error maps. Signed error – negative means a flip in the sign of M Orange: sign flip The Sign Flip Diagnostics

34. Inspection of the stacks at the location of a sign-flip according to the error map Yellow Horizon: Top Brent Relation of Sign Flip with sub-stacks weak stronger weakest Red=positive impedance

35. Stacks at low error location – Near - Mid - Far Yellow Horizon: Top Brent Relation of Sign Flip with sub-stacks weak stronger strongest Red=positive impedance

37. After the inversion (Jason Rocktrace AVA inversion)

38. Vshale Review of computed line with lithology flag curve Only Brent interval Wells 13-6, 14-4RE, 141S2, 13-2, 14_3 Ip vs Is and computed distance from Vsh line V-shale cube computation Vsh estimate: Good sands in Brent Characterised by: Vsh est < .8 (< .75) Vshale logs Vshale distance from line

39. Review V-sh seismic vs wells – well 13-2 – good Comparisons at full frequency and seismic frequency scale (not shown) QC of V-shale cube Lows and highs do match

40. V-shale QC - wells: 13-2, 13-6. V-shale from well log and seismic band-pass filtered (8 – 45 Hz) Excellent tie of 13-2 well. Good tie of 13-6 well over Brent sands, less convincing over Dunlin shales QC of V-shale cube – filtered to seismic scale

41. V-shale QC - wells: 14-3 – 14-4RE V-shale from well log and seismic band-pass filtered (8 – 45 Hz) Good ties over Brent sands. Reasonable tie over Dunlin. QC of V-shale cube – filtered to seismic scale

42. Porosity column from the reservoir model (left) compared with the one of the inverted cube (right). The inversion result sows more detail at several locations. Porosity Reservoir model vs Inversion result

43. NtG : Left reservoir model – Right Jason inversion – same colour-scale Note that the NtG from the reservoir model is more homogeneous than the Jason NtG Cross-section through the field NtG Reservoir model vs Inversion result High NtG High NtG

44. The horizontal C2 well (not indicated) encountered an up-thrown shale block in the reservoir section. The vertical 211/13-2 well (indicated) shows a very thin Kimmeridge section of approximately 30 ft. The V-shale cube from the inversion ties the well log very well over the reservoir section. A similar story holds for the D1 well. “ Blind” Well Results Vshale Blue: low Red: high

45. “ Blind” Well Results The horizontal C2 well (not indicated) encountered an up-thrown shale block in the reservoir section. The vertical 211/13-2 well (indicated) shows a very thin Kimmeridge section of approximately 30 ft. The V-shale cube from the inversion ties the well log very well over the reservoir section. A similar story holds for the D1 well. Shale: red – yellow Sand: blue