Improved AVO Fluid Detection and Lithology Discrimination Using Lamé Petrophysical Parameters _ CSEG Recorder Online

5/13/2015 Improved AVO fluid detection and lithology discrimination using Lam petrophysical parameters | CSEG Recorder Online

http://csegrecorder.com/articles/view/improved-avo-fluid-detection-and-lithology-discrimination-using-lame-petrop 1/6

Improved AVO fluid detection and lithologydiscrimination using Lam petrophysicalparameters: "p", "p", and "/ fluid stack": fromP and S inversions

Bill Goodway, T. Chen and J. Downton

PARADIGM GEOPHYSICAL

CSEG RECORDER | SEP 1997 | VOL. 22 NO. 07

Introduction

Traditional AVO and petrophysical analysis extract and exploit anomalous variations betweenseismic compressional wave velocity (Vp) and shear wave velocity (Vs) to indicate changes primarilyin pore fluid, as well as lithologic properties (Gassmann 1951, Pickett 1963, Tatham 1982, Castagna1993a). Other analysis methods using seismic measurements derive Poisson's ratio formulated

from (Vp/VS)2 (Ostrander 1984) or P and S reflectivities, ie. impedance contrast (Gidlow et aI. 1993,Fatti et al. 1994 Wallace & Young 1996). The underlying emphasis on seismic velocity and densityarises from the Knott-Zoeppritz equations for continuity of displacement (u) and stress ( (u))across an interface between different rock properties for a propagating seismic wave.Displacement and stress are usually derived from a plane wave solution of the acoustic wave

equation; u = Aeil(t-x/V) However the underlying physics in the wave equation; d2u/dx2= /M(d2u/dt2) does not involve seismic velocities, but instead the ratio of density () to modulus (M). Soconverting velocity measurements to Lame's modulii parameters of rigidity () andincompressibility () offers new insight into the original governing rock property factor /M. It willbe shown that an improved identification of reservoir zones is possible by the enhanced sensitivityto pore fluids from pure compressibility, as well as lithologic variations represented byfundamental changes in rigidity, incompressibility, and density parameters as opposed to mixedparameters of seismic velocities.
http://csegrecorder.com/editions/issue/1997-09



Theory, method and log analysis motivation

Standard analysis methods given above, though appearing different, rely fundamentally on Vp, Vsand density variations, thereby masking the original modulus parameterization as mentioned.Some authors point out the need for a more physical insight afforded by rigidity (Wright 1984,Thomsen 1990, Castagna et aI. 1993b) in the above equations. Castagna also indicates that the linkbetween velocity and rock properties for pore fluid detection, is through the bulk modulus that isembedded in Vp. However both and more so Vp have the most sensitive pore fluid indicator Adiluted by varying factors of the rock matrix indicator (ie. nonpore fluid). This can be seen in the

following relationships; Vp2 = (A + 2)/ = (+ (4/3))/ and VS2= /.

Recent AVO inversion schemes incorporate an explicit density term (Stewart 1995, Smith 1996) topotentially extract modulii, but as the number of unknowns increase and exceed the measuredquantities (intercept and offset gradient amplitude) so these complex equations are less robust andthe extracted values more inaccurate. From these observations the standard approaches may beconsidered either too insensitive or unnecessarily complex as rock property indicators.

The proposal here is to use modulii/density relationships to velocities V or impedances I, given as;

Ip2 =(Vp. ) = ( + 2) and Is2 = (Vs. )2 = , in order to extract the orthogonal Lame parameters and , from logs with measured density , or and , from seismic without known density. Thesimple derivations are; =Vp2. - 2Vs2., =VS2., and = Ip2- 2 Is2, = Is2.
http://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-fig01.jpghttp://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-fig02.jpg



Note, Poisson's ratio analysis being related to (Vp/VS)2, comes closes to measuring the most "rock

property sensitive" modulii ratio of / (i.e. (VpVS)2 = (/) + 2). Unfortunately the constant 2 in theequation is of the same order as the / ratio thereby reducing the full impact of the relativechange of this ratio between different rock properties. Incompressibility is however non physicalunlike rigidity , but the extraction method can be seen as a form of stripping off the sensitiverock matrix to reveal the most sensitive pore fluid indicator .

Table 1, shows the justification and power of the method in petrophysical analysis. Actual Vp, Vsand values from a shallow gas well (not in the study area discussed later) have been combined togive various rock property values and average % change i.e. contrast at the interface fordetectability. The unusual behaviour of a very limited VpVp compared to VsVs for this thick, goodquality gas sand zone requires some explanation, as most standard measurements concentrate onthis non-responsive Vp change. The answer lies in comparing columns 4 to 7, where it is apparentthat because of Vp's dependence on both and , the effect of decreasing A (incompressibility) as adirect response of the gas porosity, is almost completely offset by an increase in (rigidity) in goingfrom capping shale to gas sand. However by breaking out A from Vp and combing it into a / ratio,the average changes of 70% for A and 110% for / are by far the most sensitive to variations inrock properties going from shale to gas sand. The best conventional measurements from Poisson's

ratio () of 45% and (VpVs)2 of 55% are far less sensitive. This , analysis works well for log data,however for surface seismic without an independent measurement of density, the extraction of and is not possible with any certainty.

For seismic data, a common starting point for AVO analysis is the simpler linearized approximationto the Knott-Zoeppritz equations given by Aki and Richards (1979 p. 153). This equation can bereformulated in terms of modulii and density as;

Though this equation has some merit in revealing the AVO variation for modulii and a totallyindependent term, it is not practical for AVO analysis. A modified equation to extract P and Sreflectivities or impedance contrasts, given by Gidlow et al. 1992, Fatti et al. 1994, is used instead,
http://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-fig03.jpghttp://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-eq01.gif



where the last term in only, cancels for most VpVs ratios around 2 and small angles;

Having extracted the P, S reflectivity sections, the next step is to obtain Ip and Is through inversion.Finally by using the (modulii x density) to impedance relationships given above, extraction of and is possible. Results of this method on synthetic, prestack AVO models derived from subtlechanges in carbonate rock properties, show significant improvement in DID detectability over theconventional analysis based on VpVs or Ip, Is, variations.

Case study log analysis and seismic AVO

Fig. 1, shows gas logs in depth from the study area where a conventional Ip v's Is analysis (1a) haslimitations in clearly discriminating between all the various lithologic and gas sand zones (eg. gas Awith VpVs of 1.5 is just discernible). Because Ip and Is share both rigidity and density the log curvestend to track each other and never crossover, with only the lowest impedance shale or highestimpedance carbonate zones being clearly distinguishable. By contrast the and curves (1b)have similar value ranges, that do crossover with < for gas zones (note, gas zone A has abetter poro-perm than zone B as shown by the wider separation of the curves in zone A), while > is an excellent indicator of thin, tight shale breaks as seen separating gas zone A from B, as wellas the capping shale above gas zone B.

AVO analysis requires crossplotting Ip, Is or in this new approach , for a "fluid factor" (Smith &Gidlow 1987) or threshold type stack that isolates only the anomalous gas zones from backgroundrelationship. Figs. 2a, b, compare the Ip, Is to , crossplots, with , showing a significantadvantage in isolating both lithologic properties, ego sand, shale, and carbonate facies as well asgas zones. The Ip, Is points in Fig. 2a, cluster in a close linear relationship (hence Castagna's mud-rock line) with shale having the lowest values on both axes. By contrast in Fig. 2b, the lowest (incompressibility x density) point has the best gas sand values along with (rigidity x density)values higher than shales. Simply put, for the Ip, Is plot all rock types plot to the upper rightdirection from the lowest shale values, while for the , plot the anomalous gas sands are in theupper left hand quadrant from the lowest shales while other more competent pure lithologies(silts, cemented sands) plot in the opposite upper right quadrant relative to the shales. The reason
http://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-eq012.gifhttp://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-fig04.jpg



for the separation improvement in 2b, compared to 2a, is that the v's axes are orthogonalwith regard to modulii, unlike Ip v's Is, thereby making the crossplot more discriminating. Thisdifference is better seen between figs. 3a and 3b, showing the equivalent crossplotted results froma seismic line through the gas well. By being able to clearly identify the upper left quadrant of the, crossplot as good gas sands unlike the tighter Ip, Is crossplot scatter, the resulting "fluidfactor" stacks in Figs. 4a v's 4b, show the improved discriminating power of the new , approach. Fig. 4b, has the gas zone only shown plotted in green on wiggle traces of P impedance,while the Ip, Is "fluid factor" stack has numerous ambiguous gas zones that are false. Finally a pure (incompressibility x density) stack in Fig. 5, shows clear isolation of the gas zone in blue,sandwiched between higher zones both for the overlying shale and underlying carbonates aspredicted by logs.

Conclusions

1) Improved petrophysical discrimination of rock properties using , over conventional Vp, Vsanalysis.

2) Greater physical insight by isolating reservoir rock properties for pore fluid and lithology, into themodulin or Lam parameter terms of their seismic responses

3) Easier AVO crossplot thresholding for a more sensitive ", fluid factor" type stack.

4) A new pure (incompressibility x density) stack showing gas zones without interpretivethresholding or fluid factor choices.

About the Author(s)

Bill Goodway graduated from the University College of London in 1977 with a B.Sc. in Geology andis currently working towards his Masters at the University of Calgary. Bill worked until 1981 in both
http://csegrecorder.com/assets/images/articles/archive/1997-09-avo-fluid-fig05.jpg



Geology and Geophysics in the U.K. and transferred to Calgary from London with a seismiccontracting company in 1982. He has worked at a number of local Canadian and U.S. seismiccontractors and for the last 12 years he has been employed at PanCanadian Petroleum in theExploration Geophysics Department.

Bill has been a member of the CSEG since 1982, serving twice as a Convention Technical SessionChairman. He has also presented and coauthored technical papers both at CSEG conventions andthe SEG Development and Production Forum, and is a Professional Geophysicist with APEGGA.

References

Aki K, and Richards P.G., 1979, Quantitative Seismology, W.H. Freeman & Co.

Castagna J.P., 1993a "Petrophysical imaging using AVO" TLE Mar.

Castagna J.P., Batzle M.L.. Kan T.K., 1993b "Rock Physics The link between rock properties andAVO response" SEG Investigations in Geophysics #8 "Offset-dependent reflectivity theory andpractice of AVO analysis".

Fatti J.L., Smith G.C., Vail PJ.. Strauss P.L Levitt P.R., 1994 "Detection of gas in sandstone reservoirsusing AVO analysis: A 3D seismic case history using the Geostack technique" Geophysics 59, 1362-1376.

Gassmann F., 195 I "Elastic waves through a packing of spheres" Geophysics 16,673-685.

Gidlow P.M., Smith G.C., Vail P.J., 1992 "Hydrocarbon detection using fluid factor traces; A casehistory" SEG/EAEG Workshop.

Ostrander WJ., 1984 "Planewave reflection coefficients for gas sands at non normal angles ofincidence" Geoph. 49,1637-1648.

Pickett G.R., 1963 "Acoustic character logs and their application in formation evaluation" J. PetroTech., 659-667.

Smith G.C., Gidlow P.M., 1987 "Weighted stacking for rock property estimation and detection ofgas" Geo. Prosp. 35, 993-1014.

Smith G.C., 1996 "3-parameter Geostack" Ann. Int. Mtg., SEG. Expanded Abstracts, 1747-1750.

Stewart R.R., Zhang Q., Guthoff F., 1995 "Relationships among elastic-wave values; Rpp, Rps. Rss,Vp, Vs, , , " CREWES Report #7.

Tatham R.H., 1982 "Vp/Vs and lithology" Geophysics Mar., p 336-344.

Thomsen L., J990 "Poisson was not a geophysicist" TLE Dec.. 27-29.

Wallace R. Young R., 1996 "Pre-stack Inversion: Evolving the Science of Inversion" CSEG Recorder,December.

Wright J., 1984 "The effects of anisotropy on reflectivity offset" 54th Ann. Mtg., SEG ExpandedAbstracts, 670-672.

Documents

Improved AVO Fluid Detection and Lithology Discrimination Using Lamé Petrophysical Parameters _ CSEG Recorder Online