You are on page 1of 10

. ,.

SPE 36105

EMPIRICAL PVT CORRELATIONS

FOR COLOMBIAN CRUDE OILS

F. Frashad, J. L. LeBlanc and J. D. Garber, SPE, University of Southwestern Louisiana and,


J. G. Osorio, University National de Colombia

cwmwfm.wd~

to their applicability to all types of @oil mixtures with API


gravities ranging between 18 to 44.9 (single stage separation),
14.3 to 29.0 (two stage separation) and 40.3 to 44.1 (three stage
separation).

EI@IW81Y

lhii PwwvMszforwcarWmI
titi
Fwth Lath AmaicMmmd CuL=aI
Ca#awca
W
in Pml-al-SIMh, TM
~ Tc&wJ, 2M6 A@
Polmhwn E~
19%.
Thiipapwwss
toMWfWfWSNIWM
LVUNSPE P~C4f?ImkINfCUUWhJ*OI
mfumak-mc=xtmedhmtib_
81@Mki@d
bytlwudtmf(s).
Cadlh
WWS9
Ervhnrsmdmu@ocl Io
pmsw-tcd, hmrKIIbe6n
rwwmdbYWNScd@tY
c4Pdrohm

oarncknbylhaqt)

Thlmluicl, up8utM9idanf

cw t-alhluy

INTRODUCITON

or its mambsm PWUS pmndd


al SPE
pOM.
c4m8S0ciay
c4P0smklnl Engmalings nsubjac410wblkAon
_bYlho
Ediil
CWIITIilha c411m SocWfol
P61rohum Enginws.
Pumiswon 10 COPYm rwdrka.d to m ~
C4nd IMM UWI 2#J
Th9ab8u-aa
IhOumcOnm@l cmaF4mls
Ibtmtwu
may rmi be cc@M
Wwds
dnAu@md
dW8t9NId Lvti
UNWWW@$rm-ntad WHo LSruiu!SPE.
P O SS2SSS,Rk+urdwn, TX 7SOl?3-SL?36
USA fan 01 -214-ss2-94sE

AIS acauate knowledge of Pressure-Volume-Tempemture


(PVT) properties is essential in reservoir and production
engineering calculations. Estimation of reserves, determiiation
of oil reservoir performance, recovery efficiency, production
optimization and design of production systems are some of the
areas which require precise determination of a fluids physical
properties at different conditions of pressure and ternpemture.
Ideally, the physical properties of the reservoir fluids are
determined experimentally in the laboratory. However, due to
economical anrVor technical reasons, quite otlen this
information cannot be obtained fkom laboratory measured
values. In this case, PVT propertks must be estimated iknn
empirically derived correlations.
Several correlations have been proposed for determining
the PVT properties of reservoir fluids. Some of the most widel
1
c4dhOlU1S,
Standings, LiW5k#,
used correlations are:
Trubesl*, Chew-Comallys5, Beals2, Glasos, VszquezPetroskyDokla-osman,
Beggs9, Be~Robinsons3,
Farshadll, and petrosky-Farshad12. These cmmekkms are M
on reservoir fluid samples from certain specific regions of the
world.
Because of the varying compositions of cnr& oils from
different region$ predktion of PVT properties horn empirical
correlations may not provide aatisfaetory results when they are
applied to hydmxbons behaving differently 6orn the fluid
samples on which the correlations were baaed. previous studies
have shown that extrapolation of empirical PVT correlations
should be umkrtaken with caution. Examples of such studies
are Oatermsumset al10 fix Alasksm crude Oik?and Sutton and

ABSTRACT
Empirical PVT correlations are presented for estimating bubblepoint pressure, solution gas-oil ratio, oil formation volume
factor, and isothermal compressibility. To develop the above
comelations, the data base consisted of ninety-eight PVT
laboratory analyses for Colombian crude oils. llse gas-oil
ratios, gas gravities, oil gravities, and formation volume factors
involved in the development of the correlations are the result of
one, two and three-stage flash separation as recorded tim PVT
samples analyzed in the laboratory.
Tle effect of separator conditions on the predktion of the
bubble-point pressure, solution gas+il ratio and oil formation
volume t%ctor is studied. A new correlation that corrects the
separator solution giss-oil ratio for separator conditions is
provided. Improved ccsmelationsfor estimating the bubble-point
pressure, hnsed on the corrected separator solution gas-oil ratio,
are developed. In addition, total solution gas-oil ratio and oil
formation volume factor correlations based on separator data are
presented. Since the stock-tank gas-oil ratio and stock-tank gas
gravity are not usually measured in the fiel~ these correlations
PVT properties.
represent a rwdiatic form of Mtmladng

Although the correlations presented are based on


Colombian crude oils and gases, consideration should be given

311

Famhada17for ~

of Mexico csude

tibothatdk

aahowtha
tthaema

Oik.

~C

madts ObtZid

ybeconsidemble errors
emprncal PVT Curelationa

inl@ve41 in.tryingto gemdize


Wti.?!$@l@*
:.
*X*
-;*
with p~

Gfnprnd Carelations

exdudWy bak!dm FVfpropdes of Colombian csude oils.


The propdea preamted are bubble-point presam, the solution
gas oil ratio, oil fotmatioO volume t%ctor,and the isothermal oil
several
widely
known
Furthermore,
compresd%ility.
corrections am evaluated to detelmine their applicalilii for
predicting the PVT pperdes of Colombian crude oils.
Dit%rent approaches were used for developing the
Condationa. -fherefm several Co17elationsare presented for
some PVT properties. The number of correlations provided per
PVT propesty depends on the number of stages used for the
surfhce sepamtion of oil and gas, the variables correlated, and
the range of variation of the database used.
Most of the comelations am developed by applying
multiple nonliiear regression techniques. Linear regression
~w~mw~edinaf~iowhichthe
mathematical models utilized could & expmsed in linear form.
Ihe data base used consisted of ninety+ight results of
PVT laboratory anrdyam. lYIis data covers over thirty-two
reservoirs tiom diffixent Colombkm oil producing regions.
They are the results of one, two and three-stage flash separation.
REVIEW OF CORRELATIONS
EmprncaI PVT correlations have been a subject of study since
tie begiiing of reservoir and production engineering. Several
correlations for determining propaties of reservoir fluids were
publish~ among others, by Standrngi, Calhoun, TNLdg,
hsate?, &tat, and (kw and -.
For SeVGld y4XWSy
these conditions
were the only source of information available
to estimate the physical properdes of reservoir fluids from field
)M13, however, there has been an increasing
data. h recent
interest indeveloping new empirical PVT correlations for other
geognsphical areas. Some of the latter studies have been
perfbnned by Glaso, A1-Marhoun, and V~uez and Beggs19,
Sutton and Farshad*7, Dolda and Osman, Wtrosky and
Farshadil, and Petrosky and Farshad*2.
In 1947 standing presented empirical PVT Cor7ektions
for determining bubble-point pmasuma and oil fbnnation
of solution gas-oil ratio, gas gmvity,
volume fhctm as fimctions
used 105
oil gravity and tempemture.
standing

experimentallydetermined data pints on 22 di&rent cnldeoilhatuml-gas mixtures from C4difomia to develop his
cmdations.
The PVT data used were the result of a 2-stage
flash sepamtion at a cxmstant temperature of 100F. The first
stage pressure varied between 250 psi imd 450 psi and the
second stage pressure was maintained constant at atmospheric
~.
~e-~t~~=m-w~
*of

312

nhrogen~dhydmgen
atdfk%carbon dioxide was~at
cmccntratmns less than one mole pcmcat. Wading
pointed
out that the condunder which his cadatkm
were
developed are considered to approximate the avemge Califmia
Opemting clmiitions.
In 1957 Trubel* presated a graphii
cxmelation fm
determining compressibility of umk@uaM
hydmabon
reservoir fluids. The pmdwdud
CnnpreKdil&, (Mined
as the oil compressibility times the paeudoaitid
pmsaurG was
cOrmh@ graphically, as function of psewbmduced prcsure
and tempwature. Trubels did not presmt stadadd ~
concerning the data used fm developing this codadon.
In 1958 bsaterg presented a caelation of the bubblopoint presswe as function of thegasgmvity, tempaatureand
gas mole tiaction. Since the gas mole &actkM is a fimction of
the oil mokcular weighg Lasate? also presemted a cormkkion
for determining the molecular weight of tank oil fkom the API
gravity so that the bubble-pointpressure cxmdation could be
applied when only field data is available. His correlations wem
based on 158 PVT sampks fiorn 137 independent systems
produced in Cam+
Western and Mid-C0ntin4mtal United
States, and South America
In 1973 Cronquist6 published a set of dimensionkss
graphical empiricat PVT correlations for Gulf Coast reservoir
oils. In the development of these cmdati~
Cronquist
defined dimensionless oressure. dimensionless cumulative
gas evolution, and dimensionless shrinkage, which allows
one to determine the bubble-point pressure, the solution gas-oil
ratio, and the oil fommtion vohune fhctor, respectively. Ihe
correlations were based on 80 oil samplesfkom31 fields.
In 1980 Glasos presented comelations for -ming
the
bubble-point
pressure and oil fmation
volume fhc$or as a
function of solution gas-oil ratio, total gas gravity, reservoir
Mnpemtm andtank-oil gravity. The PVTdatausedwerethe
result ofa2-stage tlaahsepamtion ata constant temperatumof
125F. he wpamtor pressure was held constant at 414.7 psia
fix the tlrst stage, and 14.7 psia fm the second stage. GIMo*
used data tl-om 45 oil sampl~ most of which came fi-om the
Nwth Sea re on.
F
GISSOS CQ~hthS
were developed for oil with U(3P
&mtmhtion
fbctom of 11.9 (oik with pamtlicinities
equivaknt to No* Sea Oik). A comection to the API ~ity
Wassuggeated wheathe correlations are applied tocrudeoik
Uhrmnntely, this
with parafficinities other than 11.9.
corrccdonrcqubes osletousethegravity ofthe Ktidualoil6wm
adiffknntial
sepamdqinfimnation
which ismtsdi.ly
avaikbk fknn field data
Gkso* provided a method for cormctm
gtheprdcted
bubblepoint pressure fm the pmaence of carbon nitrogen and hydrogen sultlde. The cmedioo
factors area
function of the mole fractions of the non-hydmcadmn

components. fhii limits the appliiilhy

indicated thatauniveraal eomelation doansxexist andthatd@a


f?omlocalfegioms sbouldbeusedtodevekp
local cwrektk3ui.
In 1993, PeUosky and Famhad presented ~phiC5d PVT
COSTChti031Sfor estimating bubbk point ~,
aoiutioo gasoil ratio, bubble point fmation
volume &tar
and
undmmmted
isothermal and compraaibil~
fm Gulf of
Wtrosky and Farshad utihzed 81
Mexico crude oils.
laboratory PVT sample analyses to develop there mrrelations.
Fluid samples were obtained tim resmvoirs located offshore
Texas and Louisiana Petrosky and Farshad12 ako presented
empirical viscoshy congelations for Gulf of Mexico crude oils.
Petrosky and Farshar+z pmented ViSCOSii congelations for
estimating dead oil, satwted oilj and undmatmmd
oil
ViSCOSitiC&A totfd of 126 hbomtmy PVT analyaa W~
used to
develop the oil viscosity correlations. The data base utilii
was constructed horn diffkrentird liberadon data and two stage
laboratory aepamtor tests conducted 0ssbubble-point oil.

of the method when


field data is the only rnformabon
- avaikbk.
In 1980 Vaquez and Beggal presented cmekrtiom for
determining the solution ~il
ratio and the oil formation
volume factor as a function of the bubble-point pressure, tankoil gravity, ternpcmtum and gas gravity. A correlation was
provided for comecting the gas gmvity to sepamtor condtions of
100 psig. In order to improve the acu3racy of the correlations,
Vazquez and Beggs19divided the measured data into two groups
based on the oil API gravity. The divisiin was made at 30
API.
Vazquez and Beggs19 used more than 6,000
measurements of solution gas-oil ratios and oil formation
volume factors iiom fields all over the world.
The correlation for comecting the gas g3avity to a separator
pressure of 100 psig was based on 124 data points tim 27
different reservoir fluids.
Vazquez and Begga19 also presented a correlation for
detenninirrg undemtumted
oil compressibilities. The
correlation
for
determining
the
compressibility
of
undersaturated crude oils was developed as a function of the
solution gas-oil ratio, temperature, tank-oil API gravity, gas
gravity and the undersaturated oil pressure. A total of 4486 data
points were used for the development of this correlation.
In 1988 A1-Marhounl derived empirical equations for
calculating the bubble-point pressure and oil formation volume
factor for Middle East crude oils. Both correlations were
developed as functions of the gas-oil ratio, temperature, gas
gravity and tank-oil gravity. A total of 160 data points from 69
oil reservoirs were used for this study.
In 1983 Ostermann, et alio utilized laboratory PVT sample
arralyscs obtained from 4 fields in Alaska to analyze the
accuracy of existing correlations for bubble-point pressure, oil
formation volume factor at the bubble-point ptessure, dead oil
viscosity, and saturated oil viscosity. They concluded that is
necessary to evaluate the applicability of existing PVT
correlations before using them with confidence.
In 1990 Sutton and Famhad7evaluated sevend wellknownempiricalPVT correladons for application in the Gulf of
Mexico. The fluid properties examined were bubble-point
pressure, solution gas-oil ratio, formation volume factor,
isothermal compressibility, dead oil viscosity, gas saturated oil
viscosity and undersaturated oil viscosity. ~ey concluded that
even the PVT correladons yieldrng the best results for thk
region may present large errors. Sutton and Farshad17 also
showed that there are significant discrepancies in the magnitude
of the error obtained by applying different correlations fbr
calculating PVT properties of crude oils from the Gulf of
Mexico.
In 1992, D&la and 0sman7 published a act of correlations
for UAE erudea. They presented corrdation equations for
bubble point
pressure
and oil formation vohune factor. A total
of51 bottomhole samples tiom UAE reservoirs were used. He

DESCRIPTION OF PVT DATA


Results of PVT Iabomtory analyses for Colombian crude oils
constituted the bases of information needed for thii research. A
total of 98 reservoir tluid samples from more than 32 reservoirs
were made available for this study.
Most of the correlations are based on a di!%rent number of
experimentally determined data points. The factors determining
the data base of each correlation are the number of sepmtor
stages and the simges of variation of the data for which the
correlation is valid.
In order to study the effect of the number of sepamtor
stages on the amuracy of the correlations, the data has been
divided into three groups. The first group consists of 43 data
points obtained from single-stage sepmation. The second group
includes a total of 146 data points based on two-stage
separation. The third group is formed by 15 data points
obtained horn three-stage sepamtion. lle nmgcs of values
covered by these data are presented in Tables 1,2 and 3.
Only 49 of the 146 two-stage sepamtion data poiuts
include information on the stock-tank gas gravity. lhe total gas
gravity comeaponding to these data points has been calculated.
Most of the statistical models used m this study are
intrinsically nonlinear models. Consequently, most of the
comelations are developed by applying mukipk nonlrnear
regression techniques. Linear regression procdms were only
applied m a few cases in which the mathematical models could
1% expressed in linear form.
For nonlrnear models, the
Nonlinear Regression (NUN) program fiorn the SAS13
atatisdcal package was used. The RSQUARE and STEPWISE
regression program were used for the ~mearmorkls.
several modellrng altemadves were considered for some
oftbe PVTprop@ka.
033eexampk oftbeaeakmativa
concerns the bubble-point pressure. All of the corrections

313

indic.ateatbatrnomrealiadc models can be formulated i~instead


of total solution gas-oil ratio and total gas gravity, sepamtm gasoil ratioa and gas gravitiez are correlated.
lhe corrected sepamtm gas gravity cmelation assumes
the following general relationship].
(1)
Yac= flYupYmT#sp)
Numeamrs liiear and nonliiear models were eonaidared as
~~i~
eq@kmz. lle model yielding beat results is a8

Studkdwere devdopedunda
difkatumditiorw
depding
onthermmbero ffleklw pamtoratagea
artdkwthevariabka
conelated. Baeause of thelwge number ofmodcls uzedas
q~
~04
@
tboze models yielding the beat
comelations m pmaellted here.
CORRECTING
CORRELATIONS
FOR
SEPARATOR GAS GRAVITY AND SOLUTION

THE
GAS-

OIL RATIO

Yp= 7- + 15.5727(YJ-QIog@Jl 14.7)


Wbctw
YF=ww*(+O~would-~~@~14.7*
Y*=? IiWdtiy(-1)
obtained atcmditions of

Charlges inprmaure
and ~~-fhe
produchofre-servoir
fluid scauzmsohltior lgastobelii
fium the oil p&se. llw volume and gravity of the gas evolved
atwhiiaurhce
wpamdonis
-~~~
perfbrmed.
Theamountofgas
liiiaminimii
attheoptimum

(2)

yO= oil!~ifi~gravity (water=l)


P*=actual wpamtorpreaaum, psia
Tq=actual aepam@r~OF
lllisequation is based on%data
pointaobmined mm
analyses of 27 difft
Colombian lWMVOir fluii. For this

-~~.
rfti~--ubbw
the optimum wpamtor pressure, the volmne and gravity of the
premureincmses.
gas liberated decrmsm asthesepammr
Camrsety,
fix wpamtor pressurea above the optimum
y~,tivolumedmoftigmfiti
Creaseaastbe sepmtorpmasure incremm.
The e!lkclofthe wpamtortempemture and preasum on the
gas gravity and solution gas-oil ratio suggests that their
inclusion in the development of empirical PVT comdations
should result in an improvement of the pmdkted values. A
method was proposed by Vasquez and Beggs19 accomplishing
irr part this objective. It consists of cmeding the gravity of the
gas liberated ibm a sepammr opemting at a specific pressure to
the value that would have resulted if the sepamtor presme were
114.7 psia. Ihe comected gravity is then used as an independent
variable in the PVT comelationa.
Two correlations were developed to detamine the effect
of separator conditions on the prdcdon of PVT popertiea
Introduced intbisstudy isanaddidonal caredlo n fm solution
gas-oil ratio. These correlations correct the solution gas-oiI ratio
andthegas gravity msultingfkorn asepamtor opemting ata
specific presawe to tbe values that would have resulted if the
~oetiamfmu~.
Itwasfoundrhatin
most of the PVT analyses aveilabk for tbia research the
optimum aepamtor pmaure was in the range of 54.7 and 164.7
paia (these reauka C&l&k with Wk4t hquez and -9
found). Therefore, it was deckkd to select a value of 114.7 paia
asthereferlmce presaum.
Itwaspoatulated tbattheecambon should be performed
not Udy on the gas gravity, as suggested by Viizquez and
Beggs9,but also on the solution~il
ratio for the following
reasons. F~asstatede
arlicr,b othpamrnetemd ependonthe
condkions at whkb the aepamtion is made. Secondly, in routine
Iieldpmcticeq thegaslii
fiorntbe wpamtorto thestock
tank k vented to the atmosphere, lltemfore, the stock-tank gas
gravity and adution gas-oil ratio is seldom measured. fhis

equadomtbeaverage relativeerror was4M4%wit


baatan&rd
deviadonof 3.93%. The correlation coefficient was 0.951. The
crossplot forthiicorrelation isshownin F~
of the data used are m followx

landthemnges

Gas gravity at aepmtor


conditions of 114.7 psia (a& 1)

0.599 to 1.329

Gas gravity at sepmtor


condkions other than 114.7 psia
(a*l)

0.573to 1.337

Solution GOR at sepamtor conditions


of 114.7 Pa@ sct7sTB

67 to 1~00

SolutionGORat separatorcondii
otherthan 114.7psia!scfWI-13

66 to 1230

Oil specific gravity (water=l )

0.827 to 0.931

-r

km-,

68 to

p-m

OF

Pk

100

34.7 to 514.7

Using the same data base, linear mgmsakmanalysiswaa


applied todevelop
tbemrectedaqamtor
giwoilmtio
eorleladon. Tbe following equadon Wm obtaimk
IQ= ~[1+27.6417x~#xlog(F#l
14.7)]
(3)
where:
%-solutiongaaoilmtiotllatwould
result fkom aepamtor conditions of 114.7 ps~
Scmm

314

&= %pamtorgas-oil
ratio obtained at corsditions

T = tesnpemture, F
lle avemge relative error for this eurrelation is 13.32??with a
standarddeviation of 37,02V0. Fi~ 3 shows the crossplot for
this correlation. The ranges of the data covered m the
Comelationare presented in Table 1.

of T= and Pv SCVSTB
llw average reladve error for this correlation is 0.57V0
with a standard deviation of 6. 17%. The torielationcoefficient
is 0.997. The crossplot for this correlation is shown in Figure 2.

p,

33.22[

p18281
~.

o(o000037T-00
42Ap)
An improved correlation was obtained by correlating the
corrected separator gas gravity, Yw,and the corrected solution
gas-oil ratio, ~. These values were estimated fiwnr equations 2
and 3, respectively.
The regression analysis yielded the
following equation:
l%is equation showed an average relative error of -3.49% with a
standard deviation of 14.61?40.The crossplot for this correlation
is shown in Figure 4.

BUBBLE-POINT PRESSURE, Pn, CORRELATIONS


Standing reported the following general relation between the
bubble-point pressure of so oil and gas mixture with its fluid
and reservoir proprties.
(4)
pb = XK,Y*,AP1,T9
where:
~
= total solution gas-oil ratio
&J = average gas gravity
API = Stock-tankoil API gravity
= temperature
T
Several mode;s were tried as regression equations to
obtain a general bubble-point correlation. It was not possible to
develop an accurate correlation for all ranges of the data
available for this study. Therefore, the data was divided by
groups. The oil API gravity, the solution gas-oil ratio, the gas
gravity and the number of separator stages are the variables used
as criteria for this division. In each case, it was observed that
splitting the data into at least two groups and developing
correlations for each of them improved the accuracy of the
predicted values. The best results were obtained when the data
were divided based on the number of separator stages.
Consequently, correlations were developed for one, two and
three-separator stages.
Bubble-point Pressure Correlation
Separation Data

Bubble-Point
Separation

Pressure

Correlations

Based on Two-Stage

Nonlinear multiple regression was performed to obtain a


correlation having the following general form:
Ph = qIQqT,@
(7)
The following correlation was develo d:
4ti510<0.~TW.0153~APl)
P~= 30.21 ~w &
(8)
The relative average error of this equation is 3.323% with
a standard deviation of 16.26V0. The crossplot for this
correlation is shown in Figure 5. The ranges of the parameters
covered are presented irrTable 2.
An improved correlation was obtained by correcting not
only the separator gas gravity, but also the separator gas-oil
ratio. Equations (2) and (3) were used for these corrections.
The bubble-point pressure was then correlated as a timction of
the corrected separator gas gravity, the CO*
separator gasoil mtio, the reservoir tempemture and the oil API gravity. llre
following equation was obtained
)
p~ = 10[1.~l+l~~SxMAW.31WX[~A)]
(9)
where:
A = ~m yvnfl~% Ap~%2
The average relative error is -0.95% with a standard
deviation of 13.08%. The crossplot fbr this correlation is
presented in Figure 6. The data base used is the same as in
Equation (8).

Based on Single-Stage

Nonlinear multiple regression analysis was used to develop a


comelation of bubble-point pressure as a function of field data.
Two different cases were considered. First, the bubble-point
pressure was correlated as a function of the gas gravity, solution
gas-oil ratio, oil API suavity
and reservoir temperature. The
-model best fittin the data ielded the following correlation:
)
p,= ~fO.~W+l.Wl?MAN.M.MA)]
(5)
where
~@@xT4,~&I)
A = ~g-13~X&l.~3X

Bubbl&Point Pressure
Three-Stage Separation

Correlation

Based on Two and

where
~ = total solution gas-oil ratio, scfZSTB
Y*= average gas gravity (air=l)
API = stock-tank oil API gravity, API

Because of the small number of data points, it was not possible


to develop a correlation based only on three-@ge separadon
PVT data. However, a correlation based on data coming fkom

316

two and -stage


sepamtionwas obtained. The equationfor
bubble-pointpressure,P is:
.,,,k,o(o.m35xT4.mxm-7

~8,81

P~= 64.14 ~0w3 yg


where
y = total average gas gmvity

with a standard deviation of 16.85/0.


correlation is shown in Figure 9.

The

crossplot for this

~,o)

solution Gas-oil Ratio Correlation Based on Two and


Thre&3tage Separation Data

Equation(10) showed an averagerelative error of 1.91%


with a standard deviation of 9.80%0. The crossplot for this
correlation is shown in Figure 7. The ranges of the data used
are presented in Tables 2 and 3.

The bubble-point correlation presented in Equation (10) is based


on two and three-stage separation data.
The following
correlation is obtained by mathematical rearrangement of t.hk

SOLUTION GAS-OIL RATIO, Its, CORRELATIONS


This correlation show~ an average relative error of3.56% with a standard deviation of 16.85%. The crossplot for
this correlation is shown in Figure 10. The ranges of the data
used for its development are the same as that used to develop
Equation (10).

Three correlations for estimating the total solution gas-oil ratio


are presented in this paper. The fnt two correlations are based
on single and two-stage separation daa respectively. The third
correlation is based on two and three-stage sepamtion data. The
correlations are obtained by mathematically rearranging the
bubble-point pressure correlations given by Equations (6), (9),
and (10).

Solution Gas-Oil Ratio Correlation Based


Separation Data

on

Oil Formation Volume Factor, B., Correlations


The following general relation between bubble point pressure of
an oil and gas mixture with its fluid and reservoir properties is
assumed ~:
BO= flFQs,~,T)
(14)
where:
Be= oil formation volume factor, bbl/STB
Numerous models were tried as regression equations to
develop a correlation for the oil formation volume factor.
Several approaches were used. The best three correlations arc
presented in this paper.
The fmt correlation is based on one, two and three-stage
separation data. The correlating variables are the average gas
gravity, the total solution gas-oil ratio, the reservoir temperature

SingI&Stage

An improved bubble-point correlation based on single-stage


separation data has been presented in Equation (6). The
correlation is a fimction of the corrected gas gravity, corrected
solution gas-oil ratio, the reservoir temperature and the stocktank oil gravi~. The corrected solution gas-oil ratio is estimated
from Equation (3). Substitution of Equation (3) into Equation
(6) and a rean-angement of the resulting equation yields the
following correlation for determining the solution gas-oil ratio:

(11)

and the stock-tankoil gravity. This correlation was fud to be


very accurate. Therefore, the data was not divided according to
the number of separator stages, as it was done for the bubblepoint pressure correlations.
Two correlations based on two-stage sepamtion data are
also presented. The fmt correlates the oil formation vohrme
factor as a fhnction of the coneded sepamtor gas gravity, the
total solution gas-oil ratio, the reservoir tempemture and the
stwk-tank oil gravity. llc second con-elation utilizxs tire
corrected wparator gas-oil ratio, instead of the total solution
gas-oil ratio. The other correlating parameters are the same as
in the fmt case.

The average relative error of this correlation is -7.9?! with


a standard deviation of 22.7/0. The crossplot for this correlation
is shown in Figure 8. The ranges of the data used are presented
in Table 1.
Solution Gas-Oil Ratio Correlation Based on Twr#Nage
Separation Data
Equation (9) is based on 146 experimentally determined data
points obtained from two--e
separation.
Solving this
equation for ~ yielda the followin relation
x&o.wklo,o.m37xT@.olnl.m
~12)
&= 0.01936 xPbis74
The ranges of the data on which this cormkttion is based
w provided in Table 2. The average reladve error is 0.79?!

Oil Formation Vohne Faclor Correfatbns Based on he,


Two Orrd+hree-stagelikparadon Data

316

A total of 107experimentallydetermineddata pointswere used


to developthe oil formationvolume factorcodation basedon
one,two and three-stageseparation. The correlatingpammeters
are the total sohitionges*il ratio, the averagegas gravity,the
reservoirtemperature,and the stock-tank oil gravity. Nonlinear

This correlation showed an average tdative esmr of 6.85% with a standard deviation of 32.5%. The cmsplot fm
this condation is shown in Figure 14. The ranges of the
parameterscoveredin the correlationare presentedin Table 4.

regression analysis was performed to obtain the following

COMPARISON OF CORRELATIONS

reiation:

BO= , + , ~1-2.W14.SSKwAW.3331

HA)]

Calculationsof various PVT properties were made with several


widely used empirical PVT correlations. In this paper, several
correlations have been developed for the same PVT properly.
The correlation achieving the highest acaracy
is selected fw
this comparison. llre equations or charts of the following
comclations were used
1) For the corrected sepamtor gas gravity, VazquezBeggs9
2) For bubble-point pressure, Standrng4, ViwxpEZBeggs9
3) GIuoa, and A1-Marhoun, Dokla-0smrm7, and
Petrosky-Farshad.
4) For solution gas-oil ratio, Standing, VazquezBeggs9,
5) Glasos and A1-Marhoun, and Petrosky-Fsrshad.
6) For oil formation volume factor, Standing,
Vwquez7) Beg#, Glaso and A1-Marhoun. Dokla-0srnan7,
and Petrosky- FarShed.
8) For isothermal compressibility, Vazquez-Beggs9 and
Petrosky-Farshad.

(15)

where:
+ ()<09T6xT
A = R,0sgMX&an@X&-132n
The averagerelativeerror of the correlationis 0.00028%
with a standarddeviation of 0.03380A. The crossplot for this

correlation is shown in Figure 1I. The ranges of the data used


are presented in Tables 1, 2, and 3.
Volume Factor Correlations Based on Tw*
Stage Separation Data

Oil Formation

The experimental data for 146 bubble-point liquids were used to


obtain a correlation having the general form of Equation (14).
Nonliiear multiple regression analysis was used to obtain the
following equation:
BO = ] + 10(4.74~+2.l WMMAW.
1223x[Ios(A) }
(16)
where:
+ 0.33 127xT
A =&o.7534x&41036.%-1017
The average relative emor of this correlation is 0.427%
with a standard deviation of 3.26/0. The crossplot for this
correlation is shown in Figure 12. The range of the parameters
covered in this correlation arc presented in Table 2.
As stated earlier, the stock-tank gas-oil ratio is seldom
measured. For this reason, it was decided to develop an oil
formation volume factor correlation based on the corrected
Gther correlating
separator gas-oil ratio (Equation 3).
parameters are the corrected separator gas gravity, the reservoir
temperature and the stock-tank oil gravity. The data base used
is the same as in Equation (16). Tle nonlinear multiple
regression analysis yielded the following equation:
BO= 1 + 3x10-3xiQ3~7 + A
(17)
where
A = (0.000292+OOOO0459XI&0"
4sw)X(T-60)0M7XAP10
"nXy_o"2's
This con-elation showed an average relative e;or of
0.073% with a standard deviation of 3.63%. The crossplot for
this correlation is shown in Figure 13.

CORRELATION
FOR
COMPRESSIBILITYOF OIL

THE

CORRECTED SEPARATOR GAS GRAVIW


The values predicted by this study for gas gravities at separator
conditions of 114.7 sia were compared to the values predicted
by Vazquez-13eggsE correlation.
This studys comelation
showed lower avera e relative emors and standard deviations
Com,ation
~m vwuez.BeJ9
BUBBLE-POINT PRESSURE
The values predicted by this study for bubble-point pressures
were compared to the values predkted by Standings4,Glasos
and A1-Marhouns and Vazquez-Beggs 9, Dokla-Osmans7 and
Petrosky-Farshads correlations. The correlations presented in
Equation (6), which is based on single-stage separation da@ and
Equation (10), which is based on two and three-stage sepamtion
&@ were selected for this comparison.
The correlations for the bubble-point Pure
of this study
achieved the lowest relative errors and stmdard deviions and
WSS seconded by Petrosky-Farshadsllcamdations. standings

ISOTHERMAL

Nonlinear multiple regression analysis was performed to obtain


the following relation
. aOMOWJ035XA )
,.(-$4511+0.000303xA
(18)
co=

where:

~=

~o.,m

f3.66w

and Glasos

correlations

performed

Very pooriy

at km bubble-

pressures.
For intermediate and high bubble-point
pmsures, the results obtained !ium Standings correlations
point

~c4210sm11.011y4.1616

317

were more awcptable. A1-Marhounsl correlation underpredicts


low bubble-pointpmssmes and overpredictshigh bubble-point

Vazquez-Beggslv, Petrosky-Famhadsl1, and this St@%


comelation of isothermal compressibility of oil were compared.
This studys correlation achieved higher accumcy than both
Vazquez-Beggs19and Petrosky-Farahadsl 1correlations.

This stud)% correlation obtained higher accumcy than


VazqueAeggalv correlation. It was noted that VazquezCOrrdatkm
OV@Ctd
bubble
Beggs9 and kkkkMU1S7
point premuw%

CONCLUSIONS
I. Empirical PVT correlations for estimating bubble-point
pressure, solution gas-oil ratio and oil formation volume
fiactor have been developed W
on sirdar work by
Standmg.Equations (6), (9), (10),(1 1), (12), (13),(15),(16),
(17) and (18) form the basis for calculating the bubble-point
pressures, solution gas-oil ratios and oil formation volume
factors and undemmmkd -oil compressibility. TIIese
correlations are based on results of PVT laboratory analyses
of Colombian crude oils.
2. A new correlation that corrects the separator solution gas-oil
ratio to a ref-~
sepator pressure has been developed.
Equation (3) forms the basis for calculating the corrected
separator solution gas-oil ratio. Estimated and measured
conected separator solution gas-oil ratios were compared.
This comparison as well as the stdistical analysis of this
correlation showed that the correlation can be applied with a
high degree of accuracy.
3. Improved correlations for estimating the bubble-point
pressure have been developed. The improvement of these
correlations was achieved by introducing the new correction
factor on the sepamtor solution gas-oil ratio.
4. Bubble-point pressure, oil formation volume factor, and
solution gas-oil ratio correladom based on corrected
separator data have been developed. Since the stock-tank
gas gravity and stock-tank solution gas-oil ratio is seldom
measured in the field these correlations represent a more
realistic form of estimating PVT properties thao PVT
correlations based on totrd solution gas-oil ratio and average
gas gravity.
5. The numtm?rof surface sepiuator stagea was used as criterion
to develop dfierent correlations for the bubble-point
pesswe, the solution gas-oil ratio and the oil formation
volume f-r.
6. Deviations fkom experimentally determined &@ indicated
as average relative errora, standard deviations and croaaplots
were lower for thii study than for estimations baaed on other
published empirical PVT correlations.

SOLUTION GAS-OIL RATIO

standings,Glasos, fi]-hfalhomls, WZqUeZ-BeggS9,


Petrosky-Famhadsl 1, and this studys correlations were used to
estimate the solution gas-oil ratios.
Equations (11) and
Equation (13) of this study were used for thii comparison.
This studys solution gas-oil ratio correlations achieved the
10WW errors and standard deviations. Petmsky-Farahadsll,
Standings and A1-Marhouns COITektiOllS stood amn~ thii
and fourth in accuracy, respedvely. In general, PetroskyFarahads 11showed good amrracy while Glasos$ correlation
showed poor accuracy.
standings and AhMarhounsl
correlation tends to overpredii low solution gas-oil rati~ while
Cilasos*correlation underpmdicts solution gas-oil ratios.
Vazquez-Beggs19 correlation and Equation (12) of this
study were used to predict the solution gas-oil ratio. Both
correlations utilize the comeeted sepamtor gas gravity, instead of
the average gas gravity. This studys correladon achieved
higher accuracy than Vazquez-Beggs19.
OIL FORMATION VOLUME FACTOR
The experimentally detmnined oil formation volume
factors were compared to the values predcted by Standings,
GIssos*, A1-Marhounsl, Dokla-Gsrnan7, Petrosky-Farahad,
and Vazquedkggsv and this StUdyS correkttionso @MtiOnS

(15) and (17) of this study were used for this evaluation.
standings, Glasos:, A1-Marhounsl, Dokla-osnlans7,
Petrosky-Farshads1, and this studys results am expressed as a
tlmction of the total solution gas-oil ratio, the average gas
gravity, the stock-ta@ oil gravity and the reservoir temperature.
~is atud~s, Standmgs*4,and Petmsky-Farahadsll correlations
yielded approximately equal maults. In general, the four
emrdations showed very low relative emors, lhii confirms that
the oil formation volume factor cxmelations are more general
thanthe otbercomhtionsconsid
eredinthis
study and can
safely be used for @mates on a wide variety of crude oils, as
WW pointed out by Standing].
Vazquez-Begga19 correlation and Equation (17) of tlm
study were used to *
the oil formation volume factor.
Both correlations again showed approximately same accuracy.
ISOTHERMAL

NOMENCLATURE
API
BO
B&

COMPRESSIBLIJTY

318

= oil API gravity, API


= oil formation volume fhetor, bbl/STB
= differential bubble-point oil formation
vohune tbctor, bbVSTB

= flashbubble-pointoil formationvolume
factor, sct7sTB
= isothermaloil compressibility,PsK1
= presum, psia
= bubble-pointpressure,psia
= averageprmsure,psia
= actualaepamtorpressure,psia
= total solutiongas-oil ratio, sct7STB

5. Chew, J. and C.A. Connally, Jr., 1959,


Correlation

(1959)216,23-25.
6. Cronqui~ C., 1973, Dirnensionkw PVT Behavior of Gulf
Coast Reservoir Oils, ~

1973) 538-542.

UAE Crudes< SPE Formation Evaluation (March 1992), p.


41-45.
8. Glaso, O., 1980, Generalimd
Pressure-Volume Temperature Correlations, J.Pet.Tech. (May 1980) 785795.
9. Lasater, J.A., 1958, Bubble Point Pressure correlation;
Trans. AIME (1958) 213,379-381.
10. Ostermanrr,
C.A.,
C.A.
Ehlig-Economides,
and
O.P.Owolabi, 1983, Correlations for the Resewoir Fluid
Propertiesof AlaskanCmdes, Paper SPE 11703presented
at the 1983 California Regional Meeting, Sot. of Pti.Eng.
of AIME, Ventu~ CA (March 23-25, 1983).

actualseparatortemperature,F
gas gravity(air= 1)
averagegas gravity(air= 1)
correctedseparatorgas gravity(air= 1)

II. Petrosky-Farshad, 1993, Pressure-Volume-Tempemture


Correlationsfor Gulf of Mexico Crude Oils, SPE Paper

oil gravity (water= 1)


= undersatmated-oil viscosity, cp
= saturate&oil viscosity, cp
= dead-oil viscosity, cp

12.

b
c
d
d
i
o
g

-y

7. Dokla-Osman, 1992, Correlations of PVT Properties for

= cmected separator solution gas-oil ratio,


scf7STB
= differential solution gtas-oilratio, scffSTB
= initial differential solution gas-oil ratio,
scnm
= initial flash solution gas-oil ratio, scf/STB
= separator solution gas-oil ratio, sct7STB
= tempemture, F

=
=
=
=
=

A Viscosity
for Gas-SaturaW Cmde Oils, Trans. AIME

= bubble-point
=comcted
= dead oil
= di&ereotial
initial
= oil
P

13.
14.

15.

REFERENCES

26644 presented at 1993 68th Annual Technical


Conference and Exhibition, Houston, TX (Oct. 3-6, 1993),
p. 39S-406.
12.Petrosky-Farsh~ 1995, Viscosity Comelation for Gulf
of Mexico Crude Oils, SPE Paper 29468 presented at the
1994 Production Symposium, Oklahoma City, OK (Apr. 24, 1995), p.249-258.
SAS Users Guide Statistics, 1985, SAS Institute, Inc.,
Cary, NC (1985).
Standing M.B., 1947, A Pressure-Vohune-Tempemture
Correlation for Mixtures of California Oils and Gases,
Drill. md Prod.prac,, API (1947)275-287.
Standin~ M.B., 1981,Volumetric and Phase Behavior-of
011 Field Hydrocarbon Systems, 9th printing Society of
Petroleum Engineersof AIME,Dallas,TX (1981).

16. Sutton, RP., 1993, An Evahration of Using


Compositionally
DerivedPVT Parametersin Pressure
GradientCalculatio~ MS -nlesis, University of
SouthwesternLouisian&Lafayette,LA (1983).
17. 17.Sutton, R.P. and F.F. Farsh@ 1990, Evahmtionof
EmpiricallyDerived PVT Properties fix Gulf of Mexico
Crude Oils, SPE Reservoir Engineering Journal (Feb.
1990),79-86.
18. TNtw, A.S., 1957, Compressibility of Undersaturated
HydrocarbonReservoirFluids,Trans. AIME (1957) 210,
341-344<
19. VazqueAM. and H.D. Beggs, 1980, Carelatkms for Fhrid

1. A1-MarhouLM.A., 1988, PVT Correlations for Middle


EastCrude Otis; J. Pet. Tech. @kly 1988) 650-666.
2. ~
C., 1970, The Viscosi~ of Air, Water, Natural Gas,
Cmde 011 and Its Associated Gases at Oil Field
Tempemtums and
Prwsures, SPE Reprint Series No.
3, Oil and Gas Property Evaluation and Reseme
EstimaSociety of Petroleum Engineers of AIME,
Dal~ TX (1970) 114-127.
3. Begga, H.D. and J.R Robinson, 1975, Estimating the
of Crude Oil Systems, J.Pet.Tech. (Sept.
Viscosity
1975)1140-1141.
4. Cdhoq
J.C.~r., 1947, Fundamentals of Reservoir
Engineer-in&University of Wahoma Press, NormarLOK
(1947).

Physical Property Predction:


968-970.

319

J.Pet.Tech, (June 1980),

S1Mclric Ccmvcmion Factors


API 141.5/(131 .5+API)
E-01
bbl X 1.589873
Cp x 1.0
E-03
F
(W-32)/l .8
psi x 6.894757
E+OO
R R/l .8
sct?bbl X 1.801175 E-01
Conversion factor is exact.

= gkm
= m3
= ~x~
= c
Id%
-K
= std m3/m3

You might also like