Professional Documents
Culture Documents
Datos Sismicos
Datos Sismicos
SecbtyofPetreteum Engineers
SPE 24752
Effects of Properties in Seismic Data on Reservoir Characterization and
Consequent Fluid-Flow Predictions When Integrated With Well Logs
lJ,G. Araktingi* and W.M, Bashore, Chevron (Xi Field Research Co.
SPE Memkrer
f-l
W
ABSTRACT
INTRODUCTION
In recent years, recognitionof the importentimpaot reservoir heterogeneity has on fluid dwplscementprocesseshoa led to the use of deti
geologicallyrealisti~ quantitativememoir models as input to fluidflow
simulators. Such large models require the assignmentof petrophysical
and flow properties to each location represented in the nuntcriealfluid
flow sinndstion. Unfortundy, for moat reservoir suoh dsta cre only
mmilableat few diacmte sample locadom, namely wells. Ropcrty
values csn be assigned to the remahdog grid cells by Mcrpoladng the
available well data or by generadng steehaadc realeach of
which is a possible representationof the reservoir given the available
data. Both of these methods have sexious &awbaoka, Smooth
intqidona
of properties UC kI10W21 to prOdU~ biased MCOVCIY
JXCdCti02Ui in fluid displacement dCtlhtiOrtS)2
(h the Oth
hisI@
@dtsatic
realization conditioned by my
few data tend to genctate
mcovi+rycurves that cover a vcxylarge rsnge of values, as will be seen
later in the results section., Such problems may be alleviated by
integrating information fmm 2D and 3D seismic dam whi& are
availableon a much deruw areal coveragethan we~ extendingthe use
of such data beyond its traditional application in m@ng Iarge-soale
subsurfacestructures. Seismic trem inversion to pseudo-logsprovi~
acousticdata which em be used as soft information Whi~ in tunL can
then bs used to infer the variation of perrophysical properdes in the
interwellregion.
Re_
913
,.
The original prosity model was generated using core and neutnm
poroaitiesfrom S6 wells. A baaemapof the area showingonly ten out
of the 56 wells is presentedin Figure 2, l%eaeten wells were used h
in the fluid flow simulations. A realization using the Squendd
Indicator Simulation techniqueconditioned by the 56 wells was used
to generatea 3D, one million cell (1OMOOX1OO)
quantitativereamoir
model of porosity. This stochastic iechniquo permits different @al
correlationsto be imposedon sapuate !hreshoklsof the porosity rartge
(0.00 - 0.37). Seven such thresholds were chosen (0.05, 0,10,0.13,
0,18,0.220,25, 0,30) with the respctive anisotropicvariogranw
SPE 24762
cutoff
Rx
1st
400 m
100 m
50
2nd
400 m
150 m
50
3rd
350 m
200 m
30
4th
300 m
250 m
20
5th
250 m
250 m
6th
400 m
150 m
7th
400 m
100 m
Poro81ty-Veloclty Transforms
9M
. .
~.
b.!
..
*,
. .
SPE 24752
915
Flow R@sults
SPE 24752
porosity grids. Tim grid dimensions are 35 ncdca in both the x and y
directions and 48 nodes irt the vertical directiomaddingup to a total of
58,800 cells. The average interval thiclomssof ~e area being studiqi
is 79 m resulting in cells that are approximately1,65 m thick Well-bg
and cmederived porosity from the ten wells was integrated with 143
seismic velocity traces. Of the many different me$hodsavailable to
integrate different data types, cdrxiging with the Merkov-Bayes
formalism was chosen. This approach respects the differences in
qualhy existing betweemdata types while providing the cepaMlityof
ass@ing separate spatial correlations to different ranges of the b
The identicalveriogrammodels chosen to generatethe originalporosity
model are re*
to model the spatial continuityof the well*
thus
eliminatingdtiercnccs in integrationtechniqueor spatial correlationas
potentialcxtusesfor changesin flowbehaviorbetweenresemoir modelk
Naturally, as the carelstion level bstween the seismic and weIl data
vtics, so do the spatial coffelations for the seismic and the crosscorrelationbetweenseismic and well-log data accmdingto the MarkovBayesapproximation.Conditionalsimulationsof porositygrids for each
syntheticseismicdata set were generatedusing the SequentialIndicamr
Simulationtechnique. These porosity grids were then transformedinto
permeability @da using a Iognonnal porosity-permeabilitylrmusf~
producingpermeabtity mapsthat are qualitativelysimilertotheporcsity
maw meaning a region of high porosity correspondsto a region of high
permeability. The permeabtity values ranged horn Omd to 1,800 md.
A black oil,simtdat.d was then used to simulatea watdood (mobility
ratio = 2) through each reservoir realization. Upscaling was not
performed on the property grids, and identical reladve permdWy
cumes were utilized for each fhdd flow displacuncn~ Approximately
60 such wateribods were conducted for each syntheticseismic data net
providiig dismbtion curves for different flow per~
m
probab@y .distributiona for cumulative field oil recovery and total
fractional flow of oil at 0.7 pore volumes injected wexe used to
dutinguish among the flow characteristics of the different reservoir
models.
As discussed earlier, the effects on flow behavior of varying three
properties within the seismic data were investigated correlation with
well information waveformphase estimatirm and frequencybaruiwidth
In addition, anotherset of stochasticrealizationsof the raervoir model
wm generatedconditionedby the informationfrom the ten wells alorm
This case was needed to determine whether acquiringseismb data and
combhdngwith the well data provides a benefitover ustig rkdy the data
from the wells. Randomly sampled flow results,for this sparse~
are prssented in Figure 8 in terms of cumulativetotal oil recovery end
total ffectional flow of oil as a function of pore volumes injectd ht
additi~ the results for the original reservoir description used to
generate the syntheticseismic daw in other wcrda b rsal reservoir,
and randomly sampled flow results through a model ob@ned by
integrating an average seismic data set (R=O.6, frequency, bandwidth=10-125HZ zero phase estimation error) are also showm The
flow predictionsobtaild with the ten wells abne clearly ov~
for all realizations the real rccovexy behavior. The range in flow
behavior is also quite a bit larger than the range exhibitedby the fluid
displacementsthrough,the reservoir model obtained by including the
916
.,:,
,.
. .
SPE 24752
estimation error cmnrnitted. Also, these results are very similar to the
case when no estimationerror is made (R=0,6 in Figure 5),
DISOUSBION
917
3.
Alabq I? a, stochastic
Imaging
of spatia) Distributions
using Hard and Soft Informati% M. SC,ThX Stanford
University,CA 1987, 197 ~.
4.
5.
Koef~
&ophysicd
7.
8.
9.
10.
Bemberger,A. et al., Inversionof Normal IncidenceSeismograms; (hphysich v. 47, 1982 pp. 757-770.
11.
S., Recovery of
~kkmburg, D. W., Scheuer, T., and by,
the Acoustic Impedance from Reflection Seismograms,
Geophysics,V. 48, 1983, pp. 1318-1337.
12
13.
14.
6.
CONCLUSIONS
1.
SPE 24752
ACKNOWLEDGMENTS
REFEIWNCE~
1,
2,
15.
918
Arsktingi
I
i
1
~
~
1
...
. . ,.,>.,,.
. .
LJ@&@ii@~&q
.m.a@qEhJ&,;, .&#/!4g~..
t
7ransfonn Ptmvdty
toVdoclty
Cbudktmfonnfrcm
flddpww&mdc
bgm(%mmtn study)
I
S@nt&3DDSsim8iD&
*w@b
rdiKMy mum
conv0h9VldmlswawfcMITH
(WdJtbfl 8EJdy)
FtillbsWQtsQn(we8slmtbn Wm!y)
Riiil
Resarwir
Figure 2. Basemap
Porosity Model
SF%
720(3 ~
_..
.--...
62.00
r--
[1
11WQIIS
752 m of section
N = 4939
R = 0.s9
I
4
I
._.
r
k.n2Ea-Tnnn.nLJ
. .
0.0
0.1
0.2
0.3
Pormmy
Figure 3, Porosity-velocity
vdationship
W20
*,U
SF%
Fbroslty
-_
24752
@yer 24)
VelmMy (R=w3)
Vw.wlty
(R RO.6)
Voltwty .
(R=
O.4]
.
Vdtwmy (R=(M)
.
Pcmmy
(R=O.8)
(R=(M)
Fwosny
FmQwy
Iiz
O%
(f&
M 0.7 Pvl
0!0
6,3
Framml
25
2G
20
20
16
15
10
10
0:3
0.1
Figure 5. mm ofpmmityvc?k)city
0%
0,0
o
0:2
(h
Fvow
26
,;
o
0,1
Porosity (R=O.4)
o,~
0,3
0
0.1
921
0.2
0.3
- .__.
Wrxlty
(vweh Elm(i)
. .
I%mxlty (1OR2SO
Hz]
WIJCOwry fat0s7 WI
Hi
20
It
10
G
0,1
. .. . ..... .-.,,,,+,.
SW,,.
.,,,,.,.
...
.k m,..
k m,..-,
>.
0:2
():)
o
0!1
0,2
0,3
SI?E
. ...
VQhlty
Porosity
(-20 (leg)
..-...
01
(40 *g)
0:2
0::1
. ,. ,.,,,,,, ,.,..,,,,
,.,=,,,,.4
..,...._
. .. ..,,..,.
,___
.-...-.
___
24752@
,._.,
_.
/
............,. .,,,,...,. .............
.i
I ~
*M
...)....
... .
[:
~
Q
M) ---l. -
,y- ~~~
+::~~
~~
.+.
~~~~~~~~
~~
l~l!,88
.
-.*...
~~
.
,,,
...
..,.:.
..+....,
,,.
-+-.
.
...
:3
I I Iitg
h
-1
/ ,.... .,.
] i
+{
\a(!
\....#.......}.
.......... @
F..r
.......
tl\l
i/
..................~....
{
-- --
y-
Ild
.................... ................-
....+
.........r
.
...r..... ...rvo
j!
._..~_-p,...,...,,,
AJM03W WJ
%
..............\.....
f
............ . ........... .....1...
~,,.,
~,
.,..
,.
~ ................&L
!@
\
,,,...,,, ,.,,.,,.,,,,,
,,,
(
i
.,. .........../..
..,,. -/ ............
~
........ ./.......... .}..... .....
........ :
i
g
..._+...
....... . ...
u,,
--i--
-..............1.
.....................
~~~~~~~
. -.
+----tf%l
,,,
,,.,,,,...
,,.,+..,.
......
.... ..}.
hf!
M:H:%.
j,
~...,,
.......+
......... ../....
....
Q24
wt
.,
24?52
.,
. .
. .................... ....................................... .
>/
: J
ill
8
,.$
&
/2
#
. .. ...... . .: .:,... ,,
., :.:,,
..:; ... .. ...
,-
.. ..
/+
,/
,/
..,....
>...
-..m,,
./
.
. . .. .
..
..
...
.,.
... . .
....,
.:.
##r-
7
./ap
,.-$
?[
&#
/
4
After 730 days
I
F@re
. .
. .. .
.
.,
:.
...
.: ; .,:,..:.:$;:,;::.:.
. .. . .. ..,.. y,,.
~.
\
figure
10.Fluid-flow
simulation
SE modelhorn wellsonly
?:, a
/.
,,
,.
SPE
24752
..
........
.! ,:.....,::. ..:.:.:..
: : ;:.. : :.::...::: :.,..:.::,:::::,::..:?;.
:,,.:?
.2.: .,..,. ,,
. . ... .. . , .:.>
. .. ..-,.
. .. ..
. .. ..... . .
.. .. .
:, ..:.i:j ,,,
.: ,.
, :.,.,:;.
>.,
.,..
>7>
. . . . .. .
~,.>
.. .
> .. .
. . ... . . ,
,. ,. , ;.;:
,..1
:::,. .
:.
>..
.
. . . . . .. . .
,:.::,
. ...>;.
,:, . .; .: :
,. .<:.
.. . .....
,.
,......
,.~
Figure 11. Fluid-flow simulations with velocity model R = 0,6; 10-125 Hz
.,
I,-
926
..
,. ....