Professional Documents
Culture Documents
JCPTb5 -0 / - 07
Correlating and Averaging
Connate Water Saturation Data
By R. S. BUCKLES"
To be presented at the -
(16th Annual TechHical 3Ieeting. P.&N.G. DilJision, C.l.AI.. Calgary. !lIay, 19(5)
The use of simple hyperbolic relationships in correlat- may be jointly related to saturation in a ~ingle COL'-
jng connate water saturation data is shown to ha\-e wide relation. However, this may only reflect a mutual
3()plication in a ,'ariet:,.. of resen'oirs. This can be of con- relation::;hip between porosity, permeability and some
siderable practical significance to the reser"oir engineer other more basic rock property, while contributing
because "olumetrically weighted a\-crage saturations for
any combination of core samples, resen'oir strata or lease nothing to the accuracy of derived .-;aturations.
areas can be obtained directly from such correlations with Undoubtedly, the irreducible connate water satura-
a knowledge of a\-erage Ilorosit:r alone. Also, in systems tion in a rock is mo!:.:t directly related to propertieH
represented b:r a simille equilateral hYI)erbola the po-
rosity-saturation product is constant and hydroC'arbon other than the porosities and permeaLilities normal-
volumes are conveniently derind from the difference be- ly measured in routine core anal,}'ses. Intuitivel,}', it
tween the pore ,"'olume and a constant fraction of the would be expected that the number and type of grain
bulk volume. Potential applications and limitations of the contacts or the exposed area of the mineral 1:iurface.
technique ha ...'e been explored for a number of sand and
carbonate resen'oirs, and se ....eral examples are describcd_ aH determined by the size and distribution of the
pores, would hm'e a controlling influence. However.
INTRODUCTION until methods for determining ::iuch factors (5) can
IrCURVE B"
I--l-j-JI Sw· 250 d l -8
28 ~/CURVE A
~ Jomes A. LeWIS Correlalion From "Engineering Anoly~is 01 Methods FOr Increased
Oil Reserves, Cardium ReserVOIr", Prepared For Corll~dion Seoboord Oil Compony
I _-,h~/_·k-":M;::':::":h~I:::9::S8;:,,,:8::':;":':,,:0:;":,,,,::19::;6:,,:5::,::;m:::,':.:,:,'.:.F:.:"::jm:..:8::....:;0::il-..:8:.:;';:":.C::'~":..::A"~':::'Y:.:
2."11- ~
;:.:'_r--j-_+--+_+--+_+--+_+---1_+--1_+--1
..
c \ \' :
•
,. ..•" 20 "
. •,1* "'
,.
J
~,t"
;To~'
.. ,"." :
. .
16 "
!:: " " ~
"'
.. .".~ ~ :{ ..
0
0: ."
0 2 "
...
• .;"~r--.. ::;~h.. '" : "'~.''''' ! '.
4
.. . •
0
so
0
" 2'
CONNATE WATER SATURATION
" - Percent
55 60 70 7S
Figure l.-Po1"osity 'U.s. Connate Water Relati01'lShips for the Pembina. Cfu·di1l-m Pool - 562 points from
analyses ot 19 oil-base co-res.
"
Even the best oil-core data may be unreliable where tionship exists j often this relationship does exist, or
porosities are low. Although water content may be is approximated for all practical purposes) even though
measured with normal precision, this volume is ex- we may fail to recognize it. If it does not exist, the:
pressed as a percentage of a pore volume having about value obtained from the saturation correlation at the ,.
the same magnitude as the probable experimental er- average porosity will not be equivalent to a volumetric
ror. Therefore, the water· saturations reported for average. The unique relationship mentioned here is
cores having porosities of only 1 or 2 per cent can discussed in more detail under the heading UAverage
be considered, at best, as rough approximations. Pos- Saturations from Hyperbolic Correlations." ,
"
sible errors of similar or greater magnitude must be Two examples are given below to illusti"ate the dif~ ,: -
recognized in the other methods of saturation deter~ ference between the average saturation and the satu-
mination. Indirect laboratory measurements maJor suf- ration corresponding to the average porosity. The
fer from wettability and interfacial tension uncer- first example uses data from the Pembina Cardium
tainties) and induction log calculations, for instance, pool. In this case·, eighteen wells were selected so as
may tend toward increasing errors ,vhere saturations to COver a wide range of average porositfes. An avail-
are low or where invasion is significant. In any case, able plot of porosity versus connate water saturation
if We are to avoid compounding errors) it is desirable (Figu7'e; 1, Curve A) was accepted per se, and satura- r.'-?('
to use a true volumetric mean value for whatever we
believe to be the most I'eliable data available. This is
tions were assigned to each core sample from the
eighteen wells on the basis of individual porosity val-
" -
the only number which, when multiplied by total pore ues, By totalling the thiclrness-porosity-saturation :.
volume, gives the correct total water content of the product for all plugs in each weIi, an appropriate av-
reservoir, erage saturation was obtained in each case. At the
The process of volumetrically weighting saturation same time, each well's average porosity was deter-
data can be exceptionally tedious. In a pool with 100 mined and corresponding saturations from the cor-
core samples for each of 100 wells) we would find, for relations were noted. From the ratio of the satura-
'.0:-,
example, that 10,000 calculations of </> x h x S would tions obtained by the two methods, Figure 2 was con-
bave to be made after picking 10,000 values from an structed and used in estimating volumetrically E.'~ :..
appropriate correlation and assigning them to 10,000 weighted saturations for about a hundred wells in a
individual core samples. It is not surprising, there- particular segment of the field, In this instance, the
fore, that short-cuts, although not necessarily valid, maximum difference betw"een Utrue" (volumetrically
are frequently used as a "first estimate" in prefer- weighted) average saturations and "apparent" aver-
ence to the mOre rigorous approach. ages, as shown by FigU1"e 2. amounts to 15 per cent.
Where saturations correlate with porosity, one of In terms of hydrocarbon volume this is not particu-
the most common estimating practices is to com- larly significant, as average oil saturations will be
pute the average pool porosity and then interpret the close to 90 per cent in any case_
corresponding connate water saturation from the es- Altbough tbe above example has limited quantita-
tablished correlation. There is nothing fundamentally tive significance, it does illustrate one approach to-
wrong with this procedure if a unique geometric rela- ward estimating true volumetrically weighted aver-
"rr-------------, ,,~--------------,
0
,f
- "o
~
-
0
~~O f- ;'; I-
~
!I~
0
~
i
w w
~
u u
~ I ~ ,
~ 0
o
1:::±=:J'l:::::::::J ~ :---L~':::::::±=:±':::::::::J
L_L----o-!= O.O~ 004
0
o
I
( -'
" In processing empirical data, the foregoing deriva-
V / tion, through simplified analogies, suggests that there "
V may be some basis for expecting hyperbolic correla-
.V / / tions between connate water saturations and porosi-
V c ./
0
ties. The forms which might he expected are:
. / V- • / I Type I - For origin at (a, b) (0.0)
S - Cf<l> or<l>5 ~ C
I 1/ • Type II - For origin at (a. 0)
(S±a)=C/<I> or<l>S ~ C±<I>a
I/' J Type III - For origin at (0, b)
5 = Cf(<I> ± b) or <IS ~ C ± Sb
I
Y 1/ Type IV - For origin at (a, b) .-
, '
,..
"n
The linearity of these inverse relationships further
..;~,i'~~~
shown in FigUTC' 8-C must result. of the oil core distillation results was verified by
The equation of the first curve (FiguTe 8-A) is selecting thirty preserved samples at random and
easily derived from the assumed geometry of the processing these at a different laboratory. _,'-;: ,
--- .
~ -,
idealized system composed of spherical particles. It is In the course of studying these data, it was dis- ,.
::! '-.; -
simply A = 53_3/d, for "d" in inches and "A" in covered that a minor adjustment in the curve previ- ,'.
r·' .
acres of surface area per acre foot (7)_ Similarly, if ously drawn through the connate water - porosity
particle diameter "d" can be replaced by the linear data \\'ould lead to a very convenient simplification
function a.p± b' (considering the previous discussion applicable to all subsequent volumetric calculations.
of Figure 1,.), then the equation for the curve in Fig- The adjusted curve, as shown in FiguTe 5, can be
w-e 8-B will be of the form A = c' / (rjJ ± b), where seen to provide an excellent fit with the available
c' and b are constants. The third curve, being pro- data. It is, in fact, an equilateral hyperbola with an
portional to the second curve under the conditions pos- equation of the form xy = constant, or, more specifi-
tulated, must then be expressed by S = Cf(</> ± b), cally, cPS = 85. The bulk volume water saturation in
where C is a different constant- When b = 0, the Leduc (cPS), \vhich was reported as early as 1953 as
constant C is equal to the product cpS, which is sim- being "nearly constant," can be treated, therefore, as
ply the irreducible water saturation expressed as a being precisely constant for all practical purposes. Ac-
fraction of bulk .volume instead of pore volume. cordingly, the hydrocarbon porosity (or oil satura-
The relationship S = C/ cp can be described as an tion as a fraction of bulk volume), which is given by
equilateral hyperbola.* Alternate forms, such as (.p) (l-S), can be reduced simply to (</>-0.0085), where
S = Cf(</> ± b), as derived above, or (S ± a) = CN, both cP and S have been expressed as fractions rather
will of course represent the same function, but with than percentages_
the ~ and S axis shifted by fib" units or "au units, By establishing that cPS can be considered constant
respectively" for Leduc D-2, the variable connate water factor is
eliminated, and all volumetric calculations can be
made solely on the basis of area, thickness and po-
*The equation (y - ar- (x - b)'
= 1 is the form com-
k' h'
monly ~een for a hy}:erbola with transver~e axis parallel to the y
rosity data. In engineering studies of this reservoir,
conversion from total effective porosit}r to hydro-
..
axis. When k = h, this reduces to the rectangular or "equilateral'" carbon porosity was made by providing, in the core
to' -
form: (y - a)2 - (x - b)::! = h2 _ By rotating the axis through 45 analysis computer program, an arrangement where-
r ~.:- :: '
degrees and establishinE',; the center at (a. b) = (0,0), this is trans- ',- '
lated too y ~ (h'f2) ('fx), or xy ~ constant. by 0.85 per cent could be automatically subtracted
from all measured porosities as reported in the 210 conclusive indications in other cases complicated by
available core analyses. It \',ras unnecessary to as~ign greater reservoir heterogeneity, possible transition
individual pore volume saturations to each core sam- zones and generally less-reliable data.
ple, because, by \lirtue of the correlation's geometric In 1957, the Texaco Exploration Company filed all
form, a saturation corl'e~ponding to anJ.c average ef- I\:IPR submission (9) with the Albel'ta Oil and Gas
fective porosity in Figure 5 is :t true weighted aVC1"- Conservation Board which induded oil-base core data
age :mturation regardless of the number or tJrpe of hom a well in the 'Vizard Lal{e D-.3 reef pool. A pre-
. samples contributing to the avenlge porosity. This "ioL1s submission for the neighbouring Bonnie Glen
was determined to be a general property of Type I D-3 pool (10) provided oil-base core data showing a
and Type II hyperbolas, and is mathematically veri- similal' porosity-saturation trend. Texaco proposed a
fied under the heading "Average SatLlration~ from logarithmic equation of the form Sw = aeplo to repre-
H)"perbolic. Correlations." sent each set of data. The optimum l"ocfficients ob-
'Vem:el', in other investigatiuns, has made a paral- tained by the method of least square,g (presumaLJly
lel ubservation regarding the occurrence of constant gh'ing equal weight to each data point) were reported
cpS relationE>hipti. For Archie's rock classification, to be as follows:
Type I BCD, he noted (8) that it is only the size and
frequency of the connected secondary "',lugs (and
hence the porosity) that contrubi saturations as. long ~nnnie Glen D-3 53 (i5 -O.9R4
a::-; the dense matrix retains essentially the same char- \\. izard Lake D-3. 78.5 -1.1
1m.. L!duc N~ 509 (0,1 CD'!) Imp Led~c Nc. 6~c. (0,1 Ce,el
, I I I J
1m.. L!duc No 595 (0,1 Ce,!) Check S~mp[e. 1011 C~,"l
'" ,.,.--,---,----,-----,---,------,---, I I 1 1 I I I I I
I 1 I I I I 1 1 I
"K'--+---+---+---I---j---+----1
D
I
I I
I
I H'
I
TEll Wll.o.no LAKE CRO .... N 0-1
,'-i-i'H--t-1:r--Tt·IfTrT"'
I I I . TEll DONNIE ~LEN CR~~ tH.
I r·:-1-
l±~B'~'j'.'jll~1'1'~t~'~i~i'i±t~ji=-I
"H,----j---j---j---j---j---j-----1 I , I I I 1 I' --!--l~+--'f-+++--I
! I I I I I I I I
o ie' CU'V! For EQuilot~rol Hyperbole Wllh Elluellon "3 = 85
, I' I III i I I' T'
0. Frocllen"J Hydr~corbon Pc,o",l~ • 0 -0 C:005 I II I I ,
" f--h--v---r---'-----,---..---,----+----1 ,: I I 1 I I j--I--------L~·T.----L ~J
I I
I III .I !I . -r ,- ~ f-! -.. - '- +-
~
, I J
I I I I
I
I
I
,--,.,- ,
---,-~'~_ ---,-- __
-- - L
_
__
--'----
---.-
-,
•
~ " !·II·UI'lS~6E L I 1, ~ __ .
H,-7,+I?'!-n--,--, - 1 'T - T---.l -_. - - ,
..
•• " I:: I~' I1+'+ 1::-1 ~i ~:~ J=+=-;
•"
"•
.. ! 1 I 1 ,- -
II I I :"::1- -
I I I
I
I,'
}---1-~-! ~
1 _~ l-t~J
'I.t~---I
<0
.. ; : " f------]-- - L,--r-,- ~Y-;--I
Ii: t~~~~';t;~
,. r::q=i=~~= - ~_ 8:L :; l ;lLI - ..
i f : . ~l~;lrF~I
1 1· -, I
~ 1 1 L§-r:~;~
d:-':
,_p::r::L:~CC:LJ_'--f'""
i 11.. _ i [I:___.-; :,
'-":' 1 1 ~.. I
djr'+Ht~,:,,;·,.,!~:-L," - -!.
~H+F+tFFi-l =:;i1~_
J~1ttl I, I I I I ,
,'I_IT: -'-I'
C l.O 2'
POROSITY P.r ......
(T
J ••,
ship of some sOrG exists. As illustrated in the Leduc
D-2 example, and others to followl it is the consistency
and convenience afforded by thfl hyperbolic relation-
D-3 pool, for example, shows a slightly different po-
rosity-saturation trend for different rock classifica-
tions (Fig,,,"e 7). Data for samples described as skele-
;-1 ship in processing and averaging data that is of prin- tal and cemented skeletal have heen correlated using
,.' cipal value to the reservoir engineer_ I a hyperbolic function identical to that described ear-
.J
Tra.nsitional Sutu.l-ations and Va;riable Lithology
lier for Bonnie Glen and Wizard Lake (opS = 66).
However, Figu1'e 7 shows that for conglomerate as 7
One observation which must be made is that a sin- well as algal and reef framework material, a better
gle equilateral hyperbola can hardly be expected to correlation is given by the hyperbola opS = 80.
fit data from reservoirs having thick transition zones.
In several such cases, cpS has been found to show a Statistical Methods
changing trend with depth. Accordingly, any applica- The saturation data generally available may some-
tion of the principle under these conditions requires, times have a frequenc:~! distribution of porosity rang-
.f~~~' ':::'/'~ "~.-
'.
! ..
in effect, a famil~r of saturation versus porosity es which is not typical of the entire reservoir. In
curves. If each curve is drawn to represent a certain curve fitting, therefore, statistical analyses of porosity
depth interval and if the relationships are hyper- distribution for the reservoir or for its component
bolic, the method will still simplifl' volumetric cal- rock types can be most usefuL Mathematical curve-
culations to almost as great a degree as when the en- fitting techniques, such as the method of least
cantly different rock type. The Golden Spike South be readily checked by sketching the straight-line func-
tion on reciprocal graph paper. The constant "a", for
*The significance of a transition zone depends more on example, in S = (IN) ± a can be obtained directly
its size relative to the total oil zone than its absolute from the intercept on the saturation axis at "infinite"
thickness (see Case I under "Further Examples.'l porosity, and (C ± a) is read where the straight line ; ~
0
cent porosity. This last expression (Type II) has the
0
" 0]'\
'" 0 Average Satu1·ations frorn Hyperbolic C01'relations
0
f',
Analytically, the equivalence of volumetric average
" 0 j-..o 0 0
0 saturations and saturations read from a ·Type I or
", 0
Type II hyperbolic correlation at average porosity can
I be demonstrated as follows:
•
'"
POReSITY - "
·Poreon!
" " " let Sv be the volumetrically weighted average connate water
saturation;
Figure 7.-Connate liVater Saturation as Related to let Sc be the correlation's indicated saturation corresponding
Porosity and Rock T'ype (Golden Spike South D-8). to the average porosity. ".
Technology, JDnuary~March, 1965. Montreal 47
POOR IMAGE DUE TO ORIGINAL DOCUMENT QUALITY
''----'1----'--1--'-----1'----'--'---111---' ~, : +I1f'~::
o
o I 1 I-j----j--jf-+I_I _ ""- : 'I 11_. _
"
f-+I--:I~i-I---:---I---+----:lc-+-II l -"l..-'·-.'*<llli:-
r-=--I. 1,> 1 I I II ~_-= "
.
,
~
o
,I\
I I I I!
I .1..-" .!L,' -
----¥'-- ', 1
r--"I-
I 'j~" '"'" ,s- i;;, II
1 I'
:
~_'
'
2
I
I
L !"::-LUl.L+-,
r l'h II ...Li
I I ' .i I "
r--i -.J J I ~ -r
±
~ 1 1\\ I 1 1 U ~--!--r l ! ~ .", L'\-.
~ 0 I ~ ! 1 ! LI .,----,----,-',.'_',-----,---,-"
~ o-~---~ U __ I_LL l~ LLL ! _
~ I I ' I I I I .;. som~'e~L
~
~ -r' --e,f 'I-:-rj -[I I T c- -
r i
r
,----,-- I
r-
~,
i : L
t-:-I',.I~.- i
I
1 I I -i~- T 1-;- I
0o~..L---j,r---"c-.!...----i,--'--------+.--'--,';;m--'-,,",_L-,,:,.-----;,:.,---;",-'---,:"
POJ'iOSITl' - Percen. Of Rock Volum.
S = C(l/¢) +-
a. where "a" mayor may not be zero,
then:
~hC + ~h~la
:'.:cjlh
C ~ h +a
~o?h
..' t\:J h \
C(1/;j;j +a ~ Sc (q.e.d.)
'0
\
\
Note that lIsing an average porosity to obtain an
"average" saturation from non-hyperbolic correlations
and hyperbolas of Type III and Type IV does not give '0
1\
\
E\', the true weighted avemge saturation. In certain z
~
in~tances the difference will be small, such as in the ~
~
\Vizard Lake - Bonnie Glen example, where both
S = 78.5¢-1.L and S = 53.65¢_n.9'~ are roughly
~
~
!:i 30
\
~irni]ar to the hyperbolic form, S = GG¢-I.O. Another ~
~
point to be noted is that the "gross" average porosity w
~
common!.}' used in some pools for volumetric calcu- ~
\
u
:-,"
I "S~66 n •• lobh.~.d Iram: ~_ ~I---I---Hi\--I--1---1---1-+-+-++++++-+-+-+-+-+-f-H
I :----;. Wl10,d La~. (14~ .a",p',~ -FiC.6) ~.S'Zl~
,I
J,ddl';"nol Palnla PlolI.d I•• ,
.o.ch".n 1~-34
Lod.c 60<: (32 lampl.. - •
Rlmhf I-Z3(2D •• mplll -01
tlD,ampl.. _.c.)
J
I I ,
~
I
I
.. I
I
I I
""R~.IT,T - """'.'
"
I I ~ B.A. CORRELATION be said that 10 to 15 per cent of the reservoir's bulle
"olume is water, 80 to 85 per cent is rock solids and
I I the remainder, less than 10 pel' cent, is hydrocarbons.
los ~ 120:-..\ The large surface area of the sand evidently holds
"
I I mOlit of the water in an immobile state, as a third of
,. I the welh, in the pool consistently produce clean oil.
I \
I I I I
" I I
\\ I
"
·
·
~,
,
" .....
I
I
"'- -p
I I
."
" ..'" '~11.
"~I·-
,,~, _ .:--~
l
'" . >1
I f'1([ :1· or
."1-
'"
.-'-;:'l~~T-
, 6·Z9· . . ·~ '*l [
· "'-
-
~ ,,- f [l~i :_~'" .
I.'" J ""."
. _
~~ "iJ-
! :: i
,
. . . .
:,. U ['!
I i "'"..
':~=+j fl ~l - ~T:
0
,.
0
" "
CONNATE. WATEA - P.'.8nl
"
.,
': Another example using data from a sandstone reser- In this example, the oil core data have been chissi-
voir is presented as Figure 14. The plotted points rep- fied according to whether the samples were from
resent oil-base c01:e anal)rsis results for four wells in elevations above or below the oil-water contact_ As ex-
the Provost Viking gas field of east-central Alberta_ pected, the values plotted for samples from the water
The hyperbola is of the form referred to as Type IV zone fall above the curve presented in F'iguTe 15, al-
under "Hype!~bolic Relationships," with both porosity though it is probable that much, of the free water
and saturation axes displaced from the origin (8 = was flushed from these samples during coring opera-
265/(</> ':-14) + 14)_ As such, it doegnot offer the tions.
advantage of a constant cpS product, nor does it pro-
vide any added convenience in a,'eraging saturation >,'.
; ,.
data. The fact that the correlation is hJrperbolic is, in
this illstance, primarily of academic interest. A cur- 0
?,~ .
~.,
"
\ 0
,
0 z
o
~, "
. "\ • Imp. S leelmon 9-B-4-5W2
• Imp, Sleeiman 13-26-"l-5 W2
. .
0
0
\
0
•
0
., I
I ••
"
f\ ~ , I
\
0 0
: 1\ , I :
1
. .
0
,, , "
. . '~K
0
• 0
"
.... . I
,
~ -...
>
.
0
, I I
, " ----.1 ,
•...... +l'l
0 ~s. l:~'l
...... I- . ---:-- >
"
,
"
\ I I .
, "
.
,..
,.
" " " "
PClRDSITT -
,.
Poreanl
" " ~
" " . " "
POROSITY - "
Percenl " " "
Figu1'e 14.-Connate TYater 'VS, Porosity Provost Fig'lt1'e l6.-Connate Water Saturation '118. Porosity.
sand, P"ouost Viking gas field. (Steelman A1'ea - Charles Reservoi1")
I
I--+-+--+-+--+-+-+-+---+-+__jf-+-f-+: :~~:: ~~~E!ONE_I ~ __
"I----'H--+-+-+-+-+-+-+-I-I--+-I-+-+-l--_-+11--+-+--l
I-+-+-++-+-+--+-If-~'- -- -II-I--!-
"f-+-+-.-h-j:\-+-+-+-+-+--+-+--+-+-I--+-+--+-+---1-!---1
~ I~f-+,,,,,,,,I-" -,-,,"Cl-,,""",~~X·_+-+-_+-+-_+~+--+-+--l-+__j-+__jf__l_-f__i
"5 ~ ~20
, .
.
~o l;o
WliTER S.l.TURATION - P.".~I