SPE 10067 !

WE
Evaluating Constant Volume Depletion Data

by Curtis Hays WhHson, Roga/and Dktrlct Co//ege and Stein Barre Torp, Norwegian
Institute of Technology

@Copyright 1981, Society of Petroleum Englneere of /IIME

This paper was presented at the 56th Annual Fall Technicel Conference and Exhlbltlon of the Society of Petroleum Engineers of Al ME, held in
San Antonio, Texas, October 5-7,1981. The material Is subject to correction by the author. Permieslon to copy is restricted to en abstract of
not more than 300 words. Write! 6200 N. Central Expressway, Dallas, Texae 75206.

ABSTRACT Few engineersare aware of potentially-useful data

This paper presentsreeulta of analyzingconstant which can be derived from CVD data, some of the most
volume depletiondata obtainedfrom experimental
analysesof gas condensatesand volatileoils.
Theoreticaland practicaldevelopmentsare supported
by analysesof experimentaldata from two North Sea
condensatereservoirs.
The threemajor contributionsof this work are:
(1) presentationof materialbalance equationsused
to calculatefluid (particularlyliquid)properties
from measuredconetantvolume depletiondata,
(2) a simplemethod for calculating“b’lackoil”
formationvolume factorsand solutiongas-oilratios
for volatilesystemsusing materialbalanceresults
and a separatorflash program,and (3) investigation
of the Peng-Robinsonequationof state as a tool for
matchingmeasuredPVT data and studyingvapor-liquid
equilibriaphenomenaduring constantvolume depletion. PVT properties- i.e., formationvolume factorsand
The main examplepresentedis a rich ga~
condensatewhose measured,calculatedand simulated
phase behaviorare fully documentedin tablesand
figures. Completedescriptionof the heptanes-plus
fractionis also includedao that other engineerscan
check,modify and hopefullyimprovefluid character-
izationusing the Peng-Robinson(or.any other)
equationof state.
INTRODUCTION
Constantvolume depletion(CVD) experimentsare
performedon gas condensateand volatileoil fluids
to simulatereservoirdepletionperformanceand comp-
ositionalvariation. Resultingdata can be used in
a variety of reservoirengineeringcalculation,
among the most usefulbeing materialbalancecalcula-
tiOnsjgenerating“black oil” PVT propertiesand more
recex,tly,the tuningof empiricalequationeof state.
All of these applicationsare addreasedin the
presentwork.
Referencesand illustrationsat end of paper
importantbeing &q& composition(and therefrom
K-values),density,molecularmass (and specifically
C7+ molecularmass); WUpOit density (usingtwo indep-
endentmethods);and &L72& 4g4@n molecularmass.
No assumptionscirempiricalreIationsare used to
calculatethese data - only experimentalCVD data
and simlllematerialbalance equations.
A procedureoutliningthese calculationswas first
presentedby Reudelhuberand Hinds? Their descrip-
tion,however,is somewhatdifficultto follow and
not extensivelyknown or used by the industry. We
thereforedecidedto present the materialbalances
in equation-formusing currentSPE nomenclature.
Using the materialbalance-derivedproperties,
a method is proposedfor calculating“black oil”
solutiongas-oilratios used in two-phaseflow
equationsand reservoirmaterialbalances. The
method is not new in principle,as it was first
suggestedby Dodson, Goodwilland Mayer2 in 1953 for
solutiongas/crudeoil systems. Their method,
however,requiresexpensiveand time-consumingliquid
sampleremovalsand experimental flash separations.
The proposedmethod followsthe same procedurebut
uses experimentally-determined vapor compositions
and materialbalance-derivedliquid compositions
togetherwith a routinemulti-stageseparatorflaeh
program (usinglow pressureK-valuesindependentof
systemcomposition). PVT propertiescalc~latedusing
this method are comparedwith those calculatedusing
the Peng-Robinson3equationof state.
Though more complicated,empiricalequationsof
state can also be used to evaluateCVD data. Several
investigators5’1shave used the Peng-RobinsonEOS to
simulatePVT studiesof light gas condensatesand
crude oils, needleesto say avoidingsystemsoperat-
ing near the criticalpoint. Resultshave ranged
from excellentto poor, dependingon which properties
were compared. Conrad and Gravier16proposeda
method to improveliquiddensity estimationsby
adjuetingpropertiesof the heaviestplus fraction
(boilingpoint and Cl interactioncoefficient).
Firoozabadi,Hekim and Ketzs studiedanotherlean
gas condensateand found that by only adjuatingthe
methane-plusfractioninteractioncoefficient,the
Peng-RobinsonEOS overestimatedliquid drop-outby
nearly 100%. [As discussedlater in this work,
materialbalancecalculationsof CVD data for this
system indicatethat measuredliquidvolumes are
approximately100% low - i.e. calculatedliquid
densitieswere much too high.]
Over 30 constantvolume depletionstudies
parformedby commercialand privatelaboratories
were analyzedusing the materialbalanceapproach.
Three of these (NorthSea fluids)were chosen to
be analyzedusing the Peng-Robinsonequationof state.
Their choicewas based on (1) internalconsistency
of measuredCVD data, as indicatedby material
balancecalculations,and (2) availabilityof
extendedcompositionaldata for the heptanes-plus
fraction. All three fluidshave similarparaffin-
nephthene-aromsticcontent,with Watsonk character-
izationfactorsrangingfrom 11.95 to 12.05 for the
C7+ fraction.
The first fluid (NS-1) is a rich gas condensate
and was chosen to illustrateproposedtechniquesfor
analyzingconstantvolume depletiondata. Extensive
data for this sampleand its heptanes-plusfraction
have been includedin tabularform so that other
engineerscan duplicate,mdify and hopefullyimprove
our analyses.
The secondfluid (NS-2) is a lean gas condensate
similarto thesystemsanalyzedin References5 and 16.
Our discussionof NS-2 is limitedto behavioror
observationswhich differ from those presentedfor
the rich gas condensate. The last fluid (NS-3)is a
volatileoil operatingnear its criticalpoint. We
had not completedour analysisof NS-3 using the
Peng-RobinsonEOS when this paper was written;
convergenceproblemswere encounteredwhen tryingto
simulatethe CVD process. NS-3 is thereforeonly
mentionedwith regard to materialbalance calculations
and K-valuebehavior. More informationon any or all
of these fluidscan be obtainedfrom the authors’!
DESCRIPTIONOF THE CONSTANTVOLUNE DEPLETIONPROCESS
A constantvolume depletionexperimentis
conductedat reservoirtemperatureand begins at
saturation pressure. Cell volume,Vcell, or the
volume containedby the fluid, is initiallynoted
and used as a referencevolume.
Mercury is then withdrawnfrom the bottom of the
cell, therebyloweringthe pressureas fluid expands.
During this process,a second phase develops-
either retrogradeliquid (for gas condensates)or
solutiongas (for volatileoils). Mercurywithdrawl
is ceasedwhen a predeterminedpressureis reached.
Some laboratoriesmeasure liquidvolumesduring the
first pressurereduction,before any vapor has been
removed;thesevolumes,reportedrelativeto Vcell,
representconstantcompositiondepletion. They
closelyapproximate,however,volumeswhich would
have been meesuredif the processhad been constant
volume depletion. [Thiswas checkedusing the Peng-
RobinsonEOS simulatorfor lean and rich condensates. 1
* RogalandDistrictCollege,Ullandhaug,4000
Stavanger,Norway
Mercury is reinfectedinto the cell at constant
pressurewhile simultaneously withdrawingan equiv-
alent volume of vapor. When initialcell volume is
reached,mercury injectionis ceased. Withdrawn
vapor is analyzedusing gas chromatographyto deter-
mine composition, Yj. Moles of vapor producedare
calculatedusing the real gas law and are reportedas
a cumulativepercentof initialmoles, np. Compres-
sibilityfactor,Z, is also calculatedby noting
producedvapor surfacevolume and equivilantcell
volume (at pressureand temperature).From measured
vapor gravityand composition,heptanes-plus
wlecular mass is back-calculated.Liquid volume
is measuredvisuallyand reportedas a percentor
fractionof cell volume,which in essenceis a type
of hydrocarbonliquid saturation,SL.
The experimentalprocedureis repeatedseveral
times (6-7)until a low pressureis reached,say
700 psig (4828kPa). The remainingliquid is removed
separated(i.e. distilled)and analyzedusing gas
chromatography.Measuredliquidcompositionshould
checkwith materialbalance-derivedcomposition.
[Some mujokkbotititi &noo-thand’’adjwt” meauzed
vapo~ compo.6&on4 W &ema@f.iaZ Wanw chech.
lti pmoceduu.d tieouzagedingenti. I$h good
pzaticeto Lnquhewhtiti a.tabomtomjfiepoti
mtxw,tkedok “~moo.thd” data, and to toha.textent
mat- bathtce-dtived da.t atwe tied .& @wJ!CWQ
kepoti. 1
MATERIALBALANCEEQUATIONS
Liquid Compositionand K-valueCalculations
Perhaps the most useful applicationof constant
volume depletiondata is for calculatingliquid
compositionswhich, togetherwith measuredvapor
compositions,yield high pressureK-valueshaving
many importantreservoirand processengineering
applications. To arrive at the final expressionfor
liquidcompositionin terms of measuredCVD data,we
‘firststate molal a~:dcomponentmaterialbalances,
respectively,
. . . . . . . . . . . . . . * . . . . (1)
● ● ● ● ● ●
‘& = ‘Lb+ ‘vk
. ..*..***.*.. (2)
‘&” ‘jk = ‘Liz”
‘.jk+ ‘vk”gjk
whera nL = moles of liquidwith composition Xj$
~ = moles of vapor with compositionyj and nt =
totalmoles in the systemwith compositionz“, each
quantitybeing determinedat pressurestage i.
Subscriptj designatescomponentnumbersmaking up
each phase.
Eq. 1 states that totalmoles of the two phase
systemequals the sum of liquid and vapor moles, whil
Eq. 2 states that totalmoles of componentj in the
two phase system equalsmoles of j in the liquidplus
moles of j in the vapor. The only data measured
directlyand appearingin eitherof the equationsis
vapor composition. The remainingunknownscan be
determinedfrom reportedCVD data and modifiedforms
of the materialbalancerelations.
,Jrla 7Avuufl
cmn nnc7 t!.
- . H.
.. . WWTTSC)N
--------- AND
---- S.
-- B.
—- TORP
---— 3

?irstwe note that total moles at stage k equals All unknownsinEq. 1 have now been defined in
initialmoles minus cumulativemoles of vapor terms of measuredCVD data except liquidcomposition,
?roduced. We assumea basis of one mole initial which when written in termsof the other variables,
Eluid,that is n&l = 1, yielding becomes

k - nvh.g.k
%’ ‘jk (9)
(3) ● ● ● ● o . . . . .* . . . . ● ●

%k= 1 ‘i:2&pi ““*””””””**”*’*””**”*”” ●

‘jk = (n~- nvk)

rhe same materialbalance can be appliedon a

componentbasis, resultingin Equilibriumconstantsor K-valuesare definedas the
ratio bf equilibriumvapor to liquid composition,or
k
%k”zjk=zjl - ~~ttipk”gjk ““b””””””-””‘4) (10)
‘jk = yjkf~jk ● *..*,.***.* ● , . . ...*.*.* .,

whereAnpl. = the incrementalmoles of vapor produced

from the cell during stage i, and zjl = the initial
fluid compositionat stage ? (saturatedconditions).
Moles of vapor remainingin the cell can be
calculatedusing a volumetricbalanceand the real
gas law (pV=nZRT). Recallingthe basis of one mole
initialfluid,cell volume can be calculatedfrom
initialfluid properties,which for gas condensatesis
Zd*R.T
V . . . ...0....
cau’— Pd
An effectivemeans of correlatingand checkingthe
consistencyof calculatedK-values (i.e. liquid
compositions)is to plot log Kp vs the component
characterization factorF$ as suggestedby Hoffman,
Crump and Hocott! It has been our experiencethat
such a plot, when exhibitinglinear trendsapproach-
ing a common convergencepoint, indicateshigh
qualityvapor compositiondata, and to a lesser
extent,goodnessof liquidvolumemeasurements.
A more completediscussionof this method is present-
. . . . . . . . . . . ,...0. ( Sa ) ed in a later section,

and for volatileoils (existingas a liquidat bubble PhysicalPropertyCalculations

point pressure),
Constantvolume depletiondatn can also be used
to calculatephysicalpropertiesof equilibriumvapol
Vca ‘ M@b ...... . * .+ ● ● ● ● ● ● ● ● ● .....*. . (5b ] ● ● and liquid. A mass balance is employedto carry out
the necessarycalculations,

where R = 8.3143J/mol-K for preferredS1 units, and I

R = 10.732 sia-ft3/mol-oRfor field units. [T(K), ‘mLk+muk ““**”’””””*””’””*”’**’”*’17]
f ‘Xk
p(kpa), V(m ) and T(”R), p(psia),V(ft3), respective-
lyl *
where mt = totalmass of the system,w = liquidmas:
At each depletionpressure,liquidvolume is and ~ = vapor mass at stage k. Anotherway of
measuredvisuallyand reportedas a fractionof the statingthe mass balance is that totalmass at stage
cell volume, SLk. Actual liquidV01UK02vLk Can then k equals initialmass minus cumulativevapor mass
be calculatedfrom producedfrom the cell. Recallingthe basis of one
mole initialfluid,and therebyequatinginitialmas!
with initial (saturation)nmlecular~ss~ Ms> gives
. ● ● ● ● . . .* ● ● ● ● . . , ,* ● ● ● , ....
● (6)
‘Lk ‘ ‘Lk”vcUf3
k
. . . * .* ● ● ● ● ● , . . (72)
●

‘a% = h - & ‘np.i”%i

and from a volume balance,vapor volume Vvk is

w~i~~e~~i is the vapor phase molecularmass at

Vvk= (~ ‘S~k)”VC~ . ....
● ● ● ● . .** * *
● ● ● ● ● ● (7) stage I.” Both Ms and Mv can be easilycalculated
using Kay’s pseudocriticalmixing rule and approprial
componentmolecularmasses.
Using the real gas law, the correspondingmoles of
vapor nvk are calculatedfrom Vapor mass can now be calculatedby noting that
mass equalsmolecularmass timesmoles, or
Pk”‘Vk
*******...*.*****.********* (81 ● ●

‘vk ‘~ (131
. . . . . * . . * $, . $. . . . . . . .
● ● ● ● ●

y
where pressurepk correspondsto vapor compressibility
factorZk.
fMb and pb are buble point molecularmass and
density,respectively. Zd and pd are dew point
compressibilityand pressure,respectively.
‘vk = ‘vk*”vk
* F = b(l/Tb - l/T) where b=[lOg(pc/pa)]/(l/Tb-l/Tc
and: Tb = boilingpoint at atmosphericpressurePas
Tc and PC are criticaltemperatureand pressure,
respectively,and T is the system temperature.
4 EVALUATINGCONSTAWTVOLUMEDEPLETIONDATA SPE 10067

Liquid mass is calculatedas the differencebetween “BLACKO

totalmass and vapor mass,
,... . ● ● .
‘lb = %2 - ‘vk
Having calculatedmasses and volumesof equilibrium
liquidand vapor, respective densitiescan be
calculateddirectlyfrom the ratio of m to V, i.e.
p = m/V (wherevolumescome from Eqs. 6 and 7).
An independentcheck of vapor densitycan be
used to check the consistencyof measuredZ factors.
The relationis deriveddirectlyfrom the real gas
law and can be statedas
‘Vk“P~
...*..**.(15)
.,**...........0.0..
‘vk ‘~
We can also calculatemolecularmass of the
equilibriumliquid,and specificallyits heptanes-
plus fraction. Rewritingthe mass balanceas
‘.tk‘ MLk”nLk ‘%k”nvk
IL” PVT PROPERTIES
Dodson, ti.d~ suggestedan experimental
procadurefor determiningso-calledblack oil PVT
● ● ● . ● . . ● ● . . .* ● . ● . + (14) propertiesused in two-phaseflow equationsand
solution-gasdrive materialbalancerelations.
Current laboratoryproceduresfor estimatingoil
formationvolume factorB. and solutiongas-oilratic
Rso only approximatethe Dodson, @d. method -
without flashingthe liquidphase at each stage of
the differentialvaporizationprocess, For medium
to low volatilecrudes this procedureappearsvalid
for most engineeringcalculations.The “vapor”
solutiongas-oilratio Rsg is also assumedequal to
infinity- i.e. liquid condensationis neglected.
Highly volatileoils and gas condensatefluids
cannot,however,be analyzedor describedby the same
differentialprocess, The basic problemposed by
thesemore volatilefluids is that during two phase
flow there exist both two phases and two components.
That is, flowingoil containssolutiongas which,
when undergoingpressurereduction,evolvesand mixef
with the existingvapor phase. Likewise,flowinggas
containsretrogradeliquidwhich also evolvesand
mixes with the existingliquidwhen pressuredeclines
This complex thermodynamicphenomenonis, for all
.0.. ..... * . ● ● ● ● ● ● . (~6) practicalpurposes,impossibleto simulatein the
laboratory.

we can solve for liquidmolecularmass ~k, An alternativemethod is suggestedwhich, by

making certain simplifyingassumptions,approximates
the true model describedabove. Basically,
‘Zk - ‘vk-nvk individualphase compositionsdeterminedfrom CVD
● *.**. *.. **., ****...,. (17]
‘Lk = analysis(measuredor calculated)are flashed througl
‘Lk
a multi-stageseparatorsimulatorrepresentingfield
Using Ksy’s mixing rule, the heptanes-plusmolecular conditions. Fig. 1 describesthe processdiagrsm-
mass can be back-calculatedto yield atically.
N-l Beforewe begin our discussionof the proposed
MLk - method, let us define the four basic PVT properties
~~1 ‘j ‘jki
M — .*. *,*,**. (18) used in two-phaseflow and reservoirmaterial
LkC7+ = balance equations:
‘kC?+
.Li.quLd
vo.tumeoffx.a.tkuwivo4L contioti
where Mj are molecularmasses of “pure” components. %0 =
~.todz.tunko.il volumefiuu&&g @om dttih od x
:
The averageCT+ molecularmass of the two phase
system shouldbe calculatedusing the relation
Ao.td g~ volumektiting @om tie ~ltih od x,
R=
‘Lk*XkC7+”~hLkC7+
+ ‘vk”ykC7+”MvkCT+ 40 htick &nk oil vo&nemw&tAg @om &!atJt od x
m (!7+‘ (?9)
‘Lk”xkC7+ + ‘vk*ykCT+
vapok vo&me od y, & atiehvoh cotiti
8=
9 .totig~ vo&unewwWng &om tie ~tih od g
i

Xotiga4 vohmemuting d~om tie f@h od y,

‘hg= ~tiektinko~ volwemwl$ing ~kom i~h od Y.

where x- are liquid compositionsdeterminedfrom

“1 balance equations.
msterla are vapor composition
Yj

measured experimentally.
SPE 10067
First, liquidcompositionx“ is flashedusing a
set of appropriateK-valuesand ~asic vapor-liquid
equilibriaequations. [GlasOand Whitsonghave
documentedthat Standings‘ 1° low pressure“black oil”
K-valuesare quite accuratefor flash calculationsof
black oils. We have since found that they are also
accuratefor flash calculationsof medium to highly
volatilegas condensates- i.e. systemswith gas-oil
ratios less than about 50 000 SCF/STB (9000 Sm3/Sm3).]
The sum of surfacegas volumesdividedby stock
tank oil volume is definedas the “liquid”gas-oil
ratio Rso.
Oil formationvolume factor B. is calculated
from the relation
btigti
“z.m . + m~To
g4
B* = L— ● * .
● ● * *
● ● * , , (20)
● ●
v Severalalternativenumericalmethodswere tried
STO“‘L
where VSTO is stock tank oil volume (e.g. 1 STB) and
where mg and mSTO are masses of total surfacegases
and stock tank oil, respectively. Liquid density,PL, system (NS-3)which, from all indications,lies very
can be determinedfrom eithermaterialbalance
calculations(~ from Eq. 14 and VL from Eq. 6), or
from one of severalcompositionaldensitycorrelations free over a wide range of temperatures.
available6Jusing materialbalance-derivedliquid
compositions. We cautionthe use of PL calculated
from materialbalance equationssince only a slight
error in retrogradeliquidvolume can result in a
substantialerror in liquiddensity- and therefore
Bo. The same errorwill .notaffect liquid composition consideredfor the CG fraction. Heptanesand
to the same degree.
At the same depletionsta~e k, vapor phase with
composition is separatedthroughthe flash
Yj
simulatorusing identicalK-values. The resulting
surfacegas volumesdividedby stock tank oil volume
definesthe “vapor! solutiongas-oilratio Rs .
Gas formationvolume factor,on the other han%, can
be accuratelyestimatedfrom the CVD compressibility
factorZ using the real gas law,
pbe*Z*T
‘s70 p
B =—*(1-— .*. *,**.. *,..** (21)
9 P“TAe n@d
where nSTO = 33301eS of stock tank oil resultingfrom
the flash of nfeedmoles of reservoirvapor.
The me.jorassumptionimplicitin the proposed
method,assumingmeasuredCVD data are “good: is that
liquidand vapor compositionsare solelydependent
on pressure. That is, the composition-pressure
relationis unique and not alteredby physicalflow.
Examplesof PVT propertiesfor the rich gas
condensate(NS-1)are presentedin Figs. 2-5.
Here we have comparedpropertiescalculatedusing
materialbalanceresultswith those calculatedusing
Peng-RobinsonEOS simulateddata. Identicallow
pressureK-valueswere used for both sets of data.
“Vapor” solutiongas-oilratio is nearly the
same for both methodsof calculation. “Liquid”
solutiongas-oilratio and oil formationvolume
factor are both low for materialbalance-derived
properties. The differenceis clearlya result of
the lower liquiddensitiesestimatedby the Peng-
Robinsonequation.
C. H. WHITSONAND S. B. TORP 5
.
APPLICATIONOF THE PENG-ROBINSONEQUATIONOF STATE
MeasuredCVD data and materialbalance-derived
propertieswere controlledusing a fluid properties
packagebased on the Peng-Robinsonequationof state
and developedby RogalandResearchInstitute. A
completedescriptionof the computerprogramscan be
obtainedfrom the authors. The PVT packagenot only
includesgeneralvapor-liquidequilibriumoptions,
but it also includestwo optionsfor characterizing
the heptan>s-plusfraction- me~hodspresentedby
Whitson” or Robinsonand Peng.
The numericalsolutiontechniqueused includes
a pre-iterativesucessivesubstitutionsmethod
followedby Newton’smethod using analytical
derivatives. Convergenceproblemswere encountered
for the NS-1 fluid at temperaturesapproachingthe
● criticalpoint - i.e. below reservoirtemperature.
● ● ● ● ● ●
(Powell’smethod and a newly-developedaccelerated
sucessivesubstitutionmethod” ) withoutsuccess
Similarproblemswere noted with the volatileoil
near the criticalpoint at reservoirtemperature.
The lean gas condensate(NS-2)was solved problem-
Pure componentproperties(criticalpressure,
criticaltemperature,acentricfactor and molecular
mass) were used for non-hydrocarbons and hydrocarbon
from methane to n-pentane, Only n-hexanewas
heavierpropertieswere estimatedusing the procedure
and equationssuggestedby Whitson~lwith several
modificationsgiven in AppendixB.
To manipulatethe retrogradeliquidvolume curve
the Watson characterization factorof the heaviest
componentwas adjusted,making sure that adjusted
propertieswere physicallyrealistic,
Binary interactioncoefficientswere set equal
to zero except*3: Nz- Nz = -0.02,COZ - hydrocarbons
=0.15, N2 - hydrocarbons= 0.12, and Cl - ~n,
n = 6~7*...*which ware esti~ted using a llnear
fit of the Katz and Firoozabadi13
data (theirTable 2)
= o*14*yn - 0.0668 ● .***,.....***. [22]
~cl-cn
“Theinteractioncoefficientbetweenmethane and the
heaviestcomponentwas then adjusteduntil a match
of the measureddew point pressurewas obtained.
Fluid Description:Rich Gas CondensateNS-1
NS-1 is a rich gas condensatefirst testedat a
gas-oilratio of 5500 SCF/STB (980 Sm3/Sm3)from an
initialreservoirpressureof 7300 psia (50340kPa)
and temperatureof 280 oF (138 oC). Stock tank oil
gravitywas 44 oAPI (0,8055gin/cc).Separatorsamplf
were takenwhile flowingthe well at 16.3 MMSCF/D
(460*103Sm3) and a flowingbottomltole pressureof
7260 psia (50070kPa).
6 EVALUATINGCONSTANTVOLUME DEPLETIONDATA
Heptanes-PlusCharacterization
Extendedcompositionaldata of the C7+ fraction
was not availablefor the NS-1 fluid,only molecular
mass and specificgravity. Completetrue boiling
point (TBP)data were, however,availablefrom an
offsettingwell, NS-lb. These data were adapted to
the NS-1 fluid using the method presentedin Ref. 11,
slightlymodifiedas discussedin AppendixB,
Molal distribution(mole fractionvs molecular
mess) of the NS-lb fluidwas fit using the ganma
distributionparameteralpha and variableupper
boundrymolecularmasses. The optimalalpha was 0.712
for eta (minimummolecularmass in the C7+ fraction)
of 92. Table 2 gives resultsof the match,
Molal distributionof the NS-1 C7+ fractionwas
then calculatedusing a = 0.712, q = 92 and l&7+ =
184 (as comparedto 177 for the NS-lb fluid). We also
chose to hold upper boundrymolecularmasses constant
(equivalentto paraffinvalues),giving the rasults
presentedin Table 3.
Propertiesof the single carbonnumber (SCN)
groupswere estimeted*~by definingthe Kuop factors
from NS-lb SCN molecularmasses and specificgravities a physicallyunrealisticsystem. Adjustedphysical
using the relational
K =4. S579.M0*]5J7g.#”g4573 ........ (23)
Uop
Eq. 23 was then invertedand combinedwith NS-1 SCN
molecularmasses to yield SCN specificgravitiesand
normal boilingpoints,Tb, for the NS-1 fluid,
Y =6.010770K-1”1$24
Uop
● M0”17q47 . . . . . . . . . . . (24] (1) extendingthe CT+ split to Cqo+ such that the
where Tb = (y*~ )3, per definition. SCN data for
NS-1 calculated;~ing Eq. 24 are given in Table 3,
togetherwith criticalpropertiesestimatedusing the
Riazi-Daubert15 correlations(exceptfor Tb > 850 oF,
when modifiedcorrelationswere usedll).
Tuning the Peng-RobinsonEquationof State
Single carbon number groupswere combinedinto
five multiplecarbon number (MCN) groups- CVC9,
C10-C13,C14-C17,C18-C24and C2S+ - as suggestedin
Ref. 11. Group propertieswere calculatedusing
Kay’s pseudocriticalmixing rule, except for specific
gravitieswhich used a volume-weighted mixing rule.
Methane interactioncoefficientswere estimatedusing
* The TBP analysiswas performedaccordingto the
procedureoutlined inRef. 14 and discussed”byKatz,
ti.d.~s The laboratoryonly reported,however,
single carbonnumber molecularmasses,mole fractions
and specificgravities. Cumulativevolume percents
were then calculatedby noting that incremental
volume (per mole) = mole fractionx molecularmess +
specificgravity. Little curvaturewas exhibitedby
the TBP curve and normal boilingpointswere, there-
fore,merelyaveragesof the boilingpoint range for
a given SCN group.
SPE 10067
MCN specificgravitiesand Eq. 22, as was the C1-CG
coefficient. Using thesedata in the Peng-Robinson
EOS yieldeda dew point pressuremuch lower than
measured. The C1-CZS+interactioncoefficientwas
then increaseduntil dew point pressurematched.
The CVD simulatorwas run using the MCN
properties,as given in Table 3. The overallmatch
was good to excellent,except for liquidvolumes
which were much too high (32% simulatedmximum vs
22% measuredmaximum). To check if measuredvolumes
were low we comparedmaterialbalance liquid densitie
with Alani-Kennedy7densities(usingmaterialbalance
compositionsand molecularmasses). Table 4 shows
resultsof the comparison,indicatingthat measured
volumes are consistentexcept for perhapssmell
errors in the first two volumemeasurements.
Based on these results,it was decided to lower
the Peng-Robinsonliquidvolumesby adjustingthe
characterization factorof the CZ5+ fraction.
By loweringthe factor from 12.42 to 11.80 resulted
in a decreaseof the liquidvolumes- 8% for the
maximum drop-out(from 32% to 26%). The adjustment
had littleor no effect on other estimateddata. To
have loweredthe %Qp factorrmre would have created
propertiesfor the CZE.+fractionare found in
Table 3, as is the methane interactioncoefficient
used to adjust dew point pressure. Completeresults
of the CVD simulationare presentedin Table 5.
Peng-Robinsonliquiddensitiesare comparedwith
Alani-Kennedyestimatesin Table 4.
Over twentyother adjustmentsof the C7+ char-
acterizationprocedurewere attemptedfor improving
liquidvolume predictions. None of thesewere part-
icularlyhelpful, thoughsome are worth mentioning:
last componentwas very heavy, (2) increasingthe
numberof MCN groupsused to nine, CZ5+ inclusive,
(3) splittingthe C7+ fractioninto eight SCN groups
and a Cls+ fraction,(4) using TBP p factors**
insteadof those estimatedfrom Eq.%’3
(5) using the Lee-Kesslerzpropertyco~r~~tions.~**
** ActuallyKuo is definedas T~13 /Y and could,
therefore,have Eeen calculateddirectlyusing normal
boilingpoints determinedfromTBP analysis. Using
Kuop estimatedfrom Eq. 23 and measuredmolecular
masses and specificgravities,estimatednormal
boiling pointswere calculatedfrom Tb = (y*Kuo)3
and are presentedin Table 2. Some of thesevafues
were higher than upper boilingpoint boundriesdefine
for the specificSCN group. mo possibleexplanation
are provided:(1) due to distillationunder vacuum it
was not possibleto duplicateexact boilir.g point
boundriesas defined in Ref. 14, or (2) measured
uolecularmasses of the heavier fractionswere in
error. It was found,however,that using estimated
%op factorsfromEq, 23- when used for critical
propertyestimationsin the Peng-RobinsonEOS - gave
better match of measuredconstantvolume depletion
data; the differencewas only minor.
*** Info~tion on these or other simulationruns can
be obtainedfrom the authors. We would also apprec-
iate suggestionsas to how one might improvethe
liquidvolume prediction.
SPE 10067 C. H. WHITSONAND S. B. TORP
RESULTSANO DISCUSSION
Most resultspresentedin this paper are taken
from the threeNorth Sea systemsNS-1, NS-2 and NS-3.
It has been our experience,however,that some
featuresof CUD analysisare common to all systems.
We have tried to differentiatebetweenobsemations
which are specificto a given syetem,and thosewhich
are more general in nature. We note in particular
that the K-value correlationdevelopedin AppendixA
was developedfrom our analysisof many fluids,
rangingfrom volatileoils to light gas condensates,
Fig. 6 comparesmeasured (or more correctly,
“emoothed”)vapor compositionswith those simulated
using the Peng-RobinsonEOS. The match is excellent,
showingonly slight deviationfor the C7+ and C6
components. Deviationof the hexane componentis
probablydue to its incorrectcharacterization as
n-hexane.
Fig. 7 presentsheptanes-plusmolecularmassesof
liquidand vapor phases and the total system.
Simulatedand materialbalance-derived values match
well. Our experiencehas ahown that a good match of
C7+ molecularmaas using the Peng-RobinsonEOS is
usuallydifficult,and very dependenton proper
characterizationof the plus fraction.
Calculatedequilibriumconstantswere correlated
using the Hoffman,ti.t@.8method. Three main reasons
are given for this choice: (1) the @ Kp vs T plot
providesa simplemeans of definingthe approximate
pressure-and temperature-dependence of K-values,
(2) materialbalance-derived K-valuescanbe evaluated
for consistencyby checkingthat ~Og Kp vs F plots are
linearand converge,more or less, to a single point
and (3) an approximateestimateof con~ergence
pressurecan be determinedby extrapolatingthe slope
(of ~og Kp vs F plots)vs pressurecurve to zero,
which can in turn be used to improveinitialK-value
estimatesfor the Peng-Robinson(or any other)
equationof state. See AppendixA.
Fig. 8 presentsNS-1 K-valuescalculatedusing
materialbalancerelations. The hg Kp vs F plots
are linearand appear to approacha commn point.
As discussedin AppendixA, the convergencepoint can
give an estimateof the apparentconvergencepressure.
Actually,the most accuratevalue is obtainedby
extrapolatingthe elope vs pressurecurve to zero,
as done in Fig. 11. The resultingestimateof pK#
8000 psia (55170IcPa).
Fig. 9 presentsNS-1 K-valuescalculatedfrom
the Peng-RobinsonEOS. Once again linearplots of
tog Kp vs F convergeto a point. From the extrapola-
tion of slope to zero in Fig. 11, ~*7500 psia
(51720kpa). Experiencewith the Peng-RobinsonEOS
and materialbalanceevaluationof CUD data has shown
that rich gas condensatesand volatileoils exhibit
a more well-definedconvergencepoint than leaner
systems.
Another interestingfeatureshown in Fig. 9 is
that heavy componentsare better correlatedusing the
log Kp vs F method at higher pressurea. This ~~
suggestthat methane interactioncoefficientsof the
plus fractionshave most influenceon K-valuesat
low pressures.
7
Temperatureeffectson the &g Kp vs F plots of
NS-1 were investigatedby runningthe Peng-Robinson
EOS airnulatorat 340 ‘F (171.1oc], some 60 ‘F higher
than reservoirtemperature. Fig. 10 presentsthe
results,indicatingthat temperatureinfluenceis
(1) largestfor heavy componentsat large pressures,
(2) negligibleat low pressures(as was found in
Ref. 10), (3) relativelysmall comparedto the
influenceof pressure,and (4) not significantin
changingthe apparentconvergencepressureof the
system. These observationsare also illustratedin
Fig. 11.
For lightereyatemssuch aa NS-2, there does not
alwaysappear such a unique convergencepoint for
@ Kp VS F plots. We thoughtthat this perhaps
resultedfrom a change in the totalcompositionof
the system,or from alterationin the heptanes-plus
properties. We investigatedthese possibilitiesby
runninga constantcompositionsimulationof NS-1
(at 280 and 340 oF) andNS-2 (at 241 ‘F). Resulting
K-valueswere comparedwith CUD K-valuesand are
presentedas bg Kvs &lg p plots in Figs. 12, 13 and
14. All three systemsclearly indicatethat compsi-
tionalchange during constantvolume depletionis ?zo.4
significantenough to influenceK-valuesor conver-
gence pressure,if in fact there existsa true
convergenceof K-valuesto unity. As seen in Fig. 14
the lean gas condensate(NS-2)does not appear to
have a convergencepressurefor componentsheavier
than hexane.
CONCLUSIONS
1. Measuredconstantvolume depletiondata for two
gas condensatesand a volatileoil were analyzedusin
simplematerialbalancesand the Peng-Robinson
equationof state (EOS).
2. A simplemethod is proposedfor calculating
‘tblackoil” pw properties(formationvolume factors
and solutiongaa-oilratios)of gas condensatesand
volatileoils.
3. Materialbalance-derived K-valuescan be correlate
to yield an estimateof the apparentconvergence
pressurewhich, when used in a newly-developedK-valt
correlation,helps calculatehigh pressureK-values
used as initialeatimateain equationsof state.
4. Simulatedconstantcompositionand constantvolume
depletionstudiesof lean and rich gas condensates
using the Peng-RobinsonEOS indicatethat K-values
are &depf2nde& of the depletionprocess.
5. Temperatureeffectson the Hoffman,&&c&.8 K-valu
correlatingtechnique(log Kp vs F) were studied
using the Peng-RobinsonEOS.
6. The Peng-Robinsonequationof atate usuallyover-
estimatedliquiddrop-outfor gas condensatesduring
constantvolume depletion. The problemwas normally
correctedor improvedby reducingthe Watson”
characterizationfactorof the heaviestcomponent.
NOMENCLATURE K = convergence
A or A(p) = slope of @ Kp vs F plot L = liquidphase
b = elope of the strai~htline connectingthe n = carbonnumber
criticalpoint and atmosphericboiling o = oil
point on a @ vapor pressurevs l/T
plot, cycle-oR; cycle-K P = produced
B = formationvolume factor,Bbl/STB ; m3/Sm3 R = reduced
CT+ = heptanes-pluscomponent s = saturated(bubbleor dew point)

CVD E constantvolume depletion Sc = standardcondition

exp(x) = ex ; e = 2.71828... STO = stock tank oil
EOS = equationof state t = total (two phase]

F or F(T) = componentcharacterization
factor,cycle v = vapor phase
K = equilibriumconstant(K-value)
K = Wataon characterization
factor ‘GreekLetters
Uop
&t = naturallogarithmto base e a = parameterin gamma distribution
log = logarithmto base 10 @ = parameterin gamma distribution
m = UISSS, lbm ; kg r = gamma :Eunction
M = moleuclarmass, ibm/lb-mole; kg/kg-mole Y = specificgravityrelativeto air or
E = molecularmass of the total system water (60/60)
n = (1) moles, ibm-mole; kg-mole A = incremental

(2) exponentin K-value correlation (s = interactioncoefficient

NS = North Sea sample n = parameterin ganma distribution

(minimummolecularmass)
P = pressure,psia ; kpa
P = density,lbm/ft3;kg/m3 (gin/cc)
p(x) = probabilitydensityfunction
w = acentricfactor
Pr = cumulativeprobabilityfunction
R = universalgas constant,10.732psia-ft3/
mole-oR ; 8.3143J/mole-K
ACKNOWLEDGMENTS
R = “vapor”solutiongas-oilratio, SCF/STB ;
Sg
Sm3/Sm3
The authorswish to thankH. Xorvik,H. Asheim,
R = “liquid”solutiongas-oilratio (same) D. Murphy,V. Dalen and G. Nielsen for useful comment
so
s concerning thie paper. We also acknowledgecomputer
= saturation,fractionor percent
time and facilitiesdonatedby RogalandDistrict
T = absolutetemperature,‘R;K College,NorwegianInstituteof Technology(NTH)and
ContinentalShelf Institute(IKU). PhillipsPetro-
v =volume, ft3; m3
leum Norway and Statoilshouldbe thankedfor contri-
X = liquid composition,fractionor percent buting well-neededfluid data to the petroleumliter-
ature. Economicalsupportfrom fryingpan publica-
Y = vapor composition,fractionor percent
tions,Inc. is, as usual, appreciated.
Y or Y(p) = interceptof .bg Kp vs F plot
z = total system composition
REFERENCES
z = vapor compressibility
factor
1. Reudelhuber,F.O. and Hinds, R.F.: “A Composition
al MaterialBalanceMethod for Predictionof
Subscripts
Recoveryfrom VolatileOil DepletionDrive Reser-
a voirs,”TMW.,AIME(1957)2?0,19-26
= atmospheric
b = bubble point (pb) or boiling (Tb) 2. Dodson~C.R., Goodwill,D. and Mayer, E.H.:
“Applicationof LaboratoryPVT Data to Reservoir
c = critical
EngineeringProblems;’T&zM.,AIMB(1953)198,
cell = cell, pertainingto PVT cell volume 287-298
d = dew point
3. Peng, D.-Y. and Robinson,D.B.: “A New ‘ho-
g = gas ConstantEquationof State,” lnd.Eng,Ckm.Fund.
(1976)IS,N0.1,59-64
i = index for sumatian
j = componentidentifier
k = depletionstage
 SPE 10067 C. H. WHITSONAND S. B. TORP 9

4, Watson,K.M., Nelson,E.F. and Murphy,G.B.: 20. Standing,M.B.: Vo&una.t&icand Ptie 8ehavio& o{

“Characterizationof PetroleumFractions,”Znd.
Eng.Chn.(1935)27,1460-1464
5. Firoozabadi,A., Hekim, Y. and Katz, D.L.:
“ReservoirDepletionCalculationsfor Gas Conden-
satesUsing ExtendedAnalyses in the Peng-
RobinsonEquationof State,”Can.J.Chsm.Eng.
(Oct.,1978)56,
610-615
6. Standing,M.B. and Katz, D.L.: “Vapor-Liquid
Equilibriaof NaturalGas-CrudeOil Systems,”
Th.uti.
,AIME(1944)755,232
Oil F.i&d ffyhoumbon Sy&tem, 8th printing,
Societyof PetroleumEngineersof AIME, Dallas
(1977)
21. Robinson,D.B. and Peng, D.-Y.: “The Characteri-
zationof the Heptanesand HeavierFractions,”
ResearchReport 28, GPA Tulsa,Oklahoma(1978)
22. Kessler,M.G. and Lee, B.I.: “ImprovePrediction
of Enthalapyof Fractions,”~@RO.phOC. (March,
1978)153-158

7, Alani, G.H. and Kennedy,H.T.: “Volumesof Liquid

Hydrocarbonsat High Temperaturesand Pressures,”
Tkuti.,,AIME(1960)219,288-292 I APPENDIXA

8. Hoffman,A.E., Crump, J.S. and Hocott,C.R.:

“EquilibriumConstantsfor a Gas-Condensate
I IMPROVEDK-VALUEESTIMATIONAT HIGH PRESSURES

Solutionof the Peng-Robinson(or any other)

System,”Tfiati.,AIME(1953)198,1-10
9* Glas@,0. and Whitson,C.H.: “TheAccuracy of PVT
ParametersCalculatedfrom ComputerFlash Separa-
tion at PressuresLess Than 1000 psia,” SPE Paper
8033 (1979)
10* Standing,M.B.: “A Set of Equationsfor Computing
EquilibriumRatios of a Crude Oil/NaturalGas
Systemat PressuresBelow 1,000 psia,” J.Pti.
Tech.(Sept.
,1979)1193-1195
cubic equationof state requiresinitialestimates
of K-values. At higher pressures(> 500 psia) and
particularlynear phase boundriesor the critical
point, these estimatesare very importantfor deter-
mining the “correct”sol Itionto the equation.
AccurateK-value estimatescan also reducenumerical
divergencewhen searchingfor the solution.
WilsonD proposedthe followingthermodynamic
relationfor estimatingK-valueswhich should,in
pra~tice,only be used at low pressures,

110 Whitson,C.H.: “CharacterizingHydrocarbonPlus

Fractions,”EUR Paper 183 Presentedat the EUROPEc ~j=~xp 5*37(~+~j)(~-J/TRj) /pRj ..*** (A-f)
Meetingheld in London, England,Oct. 21-24, 1980 { }

12* Risnes,R,, Dalen,V. and Jensen,J.I.: “Phase where

EquilibriumCalculationsin the Near-Critical
Region,”Paper Presentedat the 1981 European E T/T
‘Rj cj
Symposiumon EnhancedOil Recovery,Sept. 21-23,
1981

13* Katz, D.L. and Firoozabadi,A.: “PredictingPhase

Behaviorof Condensate/Crude-Oil SystemsUsing
I pRj = plp~j

Tb ./Tc .
Methane InteractionCoefficients,”T~ti.,AIME j+~.=? 10g (pcj/pa)
(1978)f6~,1649-1655 J 71- Tbj/__fcj

14. “SelectedValuesof Propertiesof Hydrocarbonsand

RelatedCompounds,”API Project44, Texas A&M 5.37 = $.W1O)
Univ.,College Station (1969)

15, Riazi,M.R, and Daubert,T.E.: “SimplifyProperty ‘T and p define the system’stemperatureand pressure,
Predictions,”ff@o.Pkoc.(March,1980)l15-116 Tb is the atmosphericboilingpoint at pa, and Tc and
pc are criticaltemperatureand pressure,respectively
16. Conrad,P.G. and Gravier,J.F.: “Peng-Robinson Actually,Eq. A-1 appearedin the petroleumliterature
Equationof StateChecksValidityof PVT Exper- some 10 yeara before Wilson proposedhis relation.
iments,”0ti4GaAJ.(April 21,1980)77-86 Hoffman,ti,ti.apreeentedEq. A-1 graphically(their
Fig. 4) and suggestedthe followinggeneralization
17. Yarborough,L.: “Applicationof a Generalized (albeitgraphically),
Equationof State to PetroleumReservoirFluids,”
from EqtutttotiO{ $&W by Chao and Robinson
log KjP = A(~)* Fj(T) + y(p) .. .... ........ (A-2)
180 Wilson,G.M.: “A ModifiedRedlich-KwongEquation
of State,Applicationto GeneralPhysicalData
Calculations,”paper presentedat the 65th
NationalAIChE Meeting,Cleveland$1969 I where 1/Tb . - l/T
19. Brinkman,F.H. and Sickling,J.N.: “Equilibrium Fj(T) E log (P~j/PU) . . . . . . . (A-3)
Ratios for ReservoirStudies,”Tuw.,AIME(1960) - 1lTcj
219,313-319 7‘Tbj
10
A(p) = ~a~tie-dependetihtope
It is easily shown that Eqs. A-l&A-2 are identical
for A(p=pa) = 1 and Y(p=pa) =.&?g pa - i.e. at
atmosphericpressure.
Engineersfamiliarwith Eq. A-1 are aware that
its accuracyis usually limitedto low pressures.
An investigationof the pressure-dependentslope in
Eq. A-2 showed,however,that Eq. A-1 could be
reformulatedto yield equally-accurateresultsat
higher pressures.
The quasi-thermodynamical model used to extend
Eq. A-1 is based on the suggestionby Brinkmenand
Sicklinglgthat plots of tog Kp vs F(T) at several
pressuresintersectat a common point definingthe
apparentconvergencepressureof the system. Their
relation,after droppingpressure-and temperature-
dependentnotation,is
PK A(Fj - FK]
K.=— 10 .,.*● ● .,.*...... O. (A-3)
Jp
where ~ = convergencepressureand FK = the F-value
correspondingto pK at the comnon point of intersec-
tion. Eq. A-3 is es{]ilyderivedby noting that Kj=l
at convergencepressurepK. We can also show from
Eq. A-1 that FK = ~og(pK/pa),or in terms of the
interceptin Eq. A-2,
Y = Y(pl =@PK-&hg(PK/Pa)
The pressure-dependence of slopeA was investi-
gated and found to have the followingtwo properties:
(1) it ranged from O~A~l for pK<p<Pa and
(2) its form varied from sligtlyconcaveup to
linear (see Fig. 11). From these observations,a
generalequetion-formwas found to fit A(p),
P - Pa,n
A=A(p)=l -(— . . . . . . . . . . . . . . . . . (A-5)
pK- Pa
where exponentn varied from 0.5 to 0.8 for various
condensateand volatileoil systems. We found,
however,that n = 0.6 gave reasonableK-value
estimatesfor all fluid systemsatmost pressuresand
temperatures. In practice,exponentn and pK for a
particularsystemcanbe determinedexactly.by
plotting@(l-A) vs ~Og(p- ~) where n is the slope
and the interceptequalsn.1Og(pK-pa). SlopeA is
usuallydeterminedby drawinga best straightline or
using linearregressionthroughdata for components
ethane throughhexane,where our experiencehas shown
thatmethaneand carbon dioxidealso lie on the line.
For more generaluse of Eq. A-5 (for example,
initialK-valueestimatesfor an equationof state),
we have found that a good estimateof apparent
convergencepressurecan be obtainedfrom the
heptanes-plusmolecularmassm
EVALUATINGCONSTANTVOLUME DEPLETIONDATA SPE 10067
pK (P4ti]=60*MC7+ - 4200 ............. (A-6)
I
and that exponentn = 0.6 for all systems.
We can now reformulateEq. 1 (the Wilson-type
equation)to give K-valuesat all pressuresand
temperatures,
Pc” A-j
Ki. (d) .exp 5,37* At(?-ui)(l-l/TRj) /pR
PK { }<
. . . . . . . (A-7)
Note that A=l for p=pa, reducingEq. A-7 to Eq. A-1.
For p=pK, K.=1, thus making Eq. A-7 consistentat
both pressuie boundries. Using Eqs. A-5 (n=O.6),
A-6 and A-7, K-valuescan be estimateddirectlyfrom
input data used by equationsof state.
●
APPENDIXB
MODIFIEDC7+ SPLITTINGMETHOD
Methods presentedin Ref. 11 were used to
determinethe molal distributionof the C7+ fraction
with the followingmodifications:
......... (A-d)
I (1) Minimummolecularmass (eta)was set equal to 92
(2) The error ftinction
used to determinethe optimal
alpha ($) was
E(Q) =2(2-2)2 .,0..,,,., . . . . . . . . . . . . . . (B-l)
and was minimizedusing the half-intervalmethod.
(3) The upper molecularmass boundrywas allowedto
vary between 14*i - 12 and 14.i + 12, except for C7
which had a range of 93 to 14*i + 12. Measuredmole
fractionwas matchedby varying the upper boundry.
Only two iterationsare requiredsince the relation
betweenboundrymolecularmasses and mole fractions
is nearly linear.~That is, if upper boundryMi yield
zl, and M! yields 22, and z = measuredmole fraction,
then the optimalupper molecularmess boundry is very
nearly
M; -M;
M:pk = M;+ (z-~2)~ ......... (B-2)
22 - $
In some cases,Eq. 2 was used twice in order to
increasethe accuracyof the estimatedmole fraction
SPE 10067 C. H. WHITSONAND S. B. TORP 11

(4) In the originalmethod presentedinRef. 11,

sin Ie carbonnumber molecularmasses corresponding
R
to z. with molecularmass boundriesM: and M!+l w~s
mereiy the arithmeticaverage. Dale Embry (~hllllps
PetroleumCompany,Bartlesville)noted in a
personalcommunicationthat this approximationwas
not necessaryand that the exact analyticalexpres-
sion is given by

where ● ** (%-3)
I
i3=——
‘%,+- Q
● .**** o*@** .*e** *e***..* (B-4)
a

Pk(X:x,a,b,c) = e-g ; [gU+j / r(u+j+l ) 1 (B-5)

j=o

g= (x-c)/b ● *+********. *.*.**,*,*,.,, (B-6)

●

and, as previouslymentioned,q = 92. Concerning

Eq, B-4, @ is the same whethera or a+l is used in
the cumulative.probability functionPr.
TABLE 1 - MEASURSDCONSTANTVOLUMSDEPLETXONDATA FOR THE NS-1 FLUID AT 280 %+

Coolpos i tions
Equilibrium Equilibrium
Vauor Liauid
Pressure - Psia Exp , Calc .
CO~ponent 6764.7 5514.7 4314.7 3114.7 2114,7’ 1214.7 714.7 714.7 714.7
—— —. —— .—
Carbon Dioxide 2.37 2.40 2,45 2.50 2.53 2.57 2,60 0.59 0.535
Nitrogen 0.31 0.32 0.33 0.34 t3.34 0.34 0.33 0.02 0.017
Methane 73.19 75.56 77.89 79.33 79.62 78.90 77.80 12.42 10.704
Ethane 7.80 7.s3 7.87 7.92 8.04 E.ho 8.70 3.36 3.220
Propane 3.55 3.47 3.40 3.41 3.53 ~.,
74 3,91 2.92 2.S96
iso-Butane 0.71 0.67 0.65 0.64 0.66 0.72 0.78 0,91 0.916
normel-Butane 1.45 1.37 1.31 1.30 1.33 1.44 1.56 2.09 2.103
iao-PentaOe 0.64 0.59 0.55 0.53 0,54 0.59 0.64 1,40 1.417
normel-Pentana 0.68 0.62 0.58 0.56 0.57 0.61 0.66 1,60 1,624
Hexenea 1.09 0,97 0.S8 0.83 0,82 0.s5 0.90 3.68 3,755
Heptanes-plus 8.21 6,20 4,09 2,64 2.02 1.84 2,12 71,01 72, S15
—— —— ,——. .

Totala 100,00 100.00 100,00 100,00 100.00100.00100.00100,00100,000.

184,0 160,0 142.0 127,0 119.0 115,0 114.0 213,0 207.9
%,+
%,+ 0.816 0.799 0.783 0.770 0,762 0.758 0,757 0,833 0,843
z 1,238 1,089 0,972 0.913 0,914 0.937 0.960
n -% 0.000 9.024 21,744 3S.674 55.686 72.146 81.301
P
-z 0.0 14.1 19.7 21.6 21,3 20,2 19.3
‘L

TABLE 2 - COMPOSITIONAL AND PROPERTIES DATA OF FLUID NS-lB SAYPLED FROM

A WSLL OFFSETTING NS-1 COMFARSDWITH CALCULATEDDATA
GENERATEDUSING THE METHODPFU%SENTEDIN RSF . 11

Measured Calculated

Single Boiling Upper

K
Carbon Mole Molal Specific Uop Point Mole Molel Molal
Number Percent }Iaea Gravity Factor (%) Parcant Mass Maes
—— .— .— —.
0.94 95 0,715s 12,05 641.7 0.935 95.1 99.6
: 0.s4 104 0.7365 11.93 678.3 0.83S 105,4 111.9
9 0.74 118 0.7757 11,90 727.3 0.739 11s .9 126.8
10 0,60 132 0,7639 11,99 768.4 0.600 134,2 142.4
11 0.41 144 0.7723 12.04 S04 ,0 0.410 148,7 155.5
12 0.34 154 0.7s14 12.04 832,7 0,340 161,7 168,4
13 0.31 167 0.7939 12.03 871,2 00310 175.0 182.2
14 0,26 180 0,8053 12.02 907,0 0.260 188,7 195.7
15 0.22 197 0.8096 12.13 947.1 0.220 202,1 209.1
16 0.19 212 0,8152 12,30 1008,1 0.190 215,4 222.4
17 0,17 226 0.8255 12.19 1019.0 0.170 228.9 236,1
18 0.15 234 0.8303 12.20 1039.4 0.150 242.8 250.1
19 0.13 250 0,8341 12.26 1069,4 0.130 256,8 264.1
20 0.11 262 0.8400 12,28 1097.6 0.110 270,6 277.8
21 0.08 277 0,8477 12,29 1130.8 0.080 283.0 289.0
22 0.07 292 0,8531 12.32 1161,0 0.070 294,1 299,9
23 0.06 308 0,8577 12.36 1191.4 0.060 304,8 310.3
24 0,06 329 0,8666 12,38 1234,9 0.070 316,7 324.0
25+ 0.51 471 0.8826 12.87 1465.6 0.508 439,1 .
—— ——
6,19 177 0,8061 12.25 6.190 177,0
(12.02)*

* The higher average hop valua was calculated using a weisht-average

mixing rule, wharaas the lower value was estimated using the Whitson
correlation.
* The gamma distribution (Raf. 11) was usad where an optimal alpha of
0,712 wae found for eta (minimum molecular maas in the C7+ fraction)
of 92. Upper molecular maaaea wera found by fittins the meaeured
compositions,
 TABLE 3- PHYSICAL PROPERTIES OF THE C?+ SINCLE AND lNILTIPLE CARMN NUf4BSR
~ .OUPS USED ~ THE PENG-ROBINSON EQIJAT~N oF STATE ~ DEP,c21BE
L6SERV01R FLUID SEHAVIOR OF THE NS-1 FLUID

Critical
Single Boiling Methene
Carbon Mole Molel Specific Point Temp. Prese, Acentric Interaction
Number Parcant Meae Gravity (OR) (OR) (peie) Factor Coefficient
—— —— —— .——.
1,2136 95,3 0,7177 646.8 971.6 457,5 0.2742
; 1,1730 106.5 0.7409 690,6 1021.4 423,4 0,3056
9 0,8600 120.7 0.7599 739.4 1073,2 383,2 0,3454
10 0,6872 134,7 0.76S2 781,4 1112.9 345,9 0,3861
11 0.5681 14s.7 0.7781 822.2 1152,1 316,8 0,4251
12 0.4783 162.7 0.7908 863.1 1192.3 294,0 0,4622
13 0,4074 176.7 0.8034 902.8 1231,2 2:4,9 0.4984
14 0.349s 190.7 0.8153 941.2 1268,3 258,4 0,5342
15 0,3021 204.7 0,8168 972.6 1294,2 240.4 0,5740
16 0.2621 218.7 0.8210 1029.8 1321.5 225.7 0.6127
17 0.2282 232,7 0.8310 1039,5 1354.0 214,6 0.6486
18 0.1992 246.6 0.8389 1072,0 1383,6 204,3 0.6855
19 0s1744 260,6 0,8432 1104.7 1408,7 194.0 0,7246
20 0.1529 274.6 0.8486 1131.6 1434.4 185,1 0,7634
21 0.1343 288.6 0.8554 1161.9 1461,0 177,4 0.8016
22 0.1182 302,6 0,8602 1190.2 1484.9 170,0 0.8416
23 0.1041 316.5 0.8639 1217.4 1507,0 162,9 0,8830
24 0.0918 330.5 0.8690 1245,1 1530,2 156.8 0.9241
2s+ 0.7054 462,3 0,9192 1.488.0 1734,6 91.4 1,0590
—— —
S,21OO 184.0 0.8160

MULTIPLE CARBONNUMBERPROPERTIES USED IN THE FINAL CVO SIMULATION

7- 9 3.2466 106.1 0.7385 688,3 1016.5 425.5 0.3044 0 .03659 ***
10-13 2,1410 152.7 0.7837 835,0 1163.6 313.1 0.4348 0,04292***
14-17 1.1421 209.2 0.8205 984,4 1304.5 237.4 0.5S56 0.04807***
18-24 0.9749 281,5 0.8524 1146.9 1446.0 182.7 0,7832 0 .05254 ***
25+ 0,7054 462.3 0.9192 1276.1* 1584,2* 168,8* 0.8819* 0.18400**
—. —
S,21OO 184.0 0.8160

* Adjusted valuea representing e Kuop factor of 11,80, The original veluee

correspond to a Kuop factor of 12,42 and ere given above (e.g. 1488 .0) ,
** Adjuetad valua ueed to match the maaeured daw POint Preeeur@.
*** Calculetcd ueing the Katz and Firoozabadi correlation, curve-fit tO yield
= o,14. y - 0.0668. Though not ehown in thie
~~~l&l~~i#~t~~~~n~Cl~n interect!on coefficient wae also calculated using
thie relation,

TABLE 4 - CALCULATEDLIQUID DENSITIES AS A FUNCTION OF PRSSSURE FOR NS-l

Calculated Liquid Deneitiee (gin/cc)

Measured CVD Data Simulated CVD Data

Alani-l(ennedy Alani-Kennedy
Material Daneity Peng- Denaity
Pressure Balanca (M. B. Liquid Robinson (P-R Liquid
(pela) Deneity PrOpertiee)* Dcne i ty** Properties)*

5514.7 0.670 0, 60B 0,541 0.570

4314.7 0,680 0,649 0.554 0,596
3114,7 0.688 0.670 0,580 0.632
2114.7 0.700 0,682 0.608 0,664
1214.7 0.711 0.700 0,636 0,692
714,7 0,722 0,711 0,653 0,707

* The Alani-Kanncdy density cm’ralat ion requires 1 iquid compoai t ions,

total liquid molecular mane, hcptanea-plue molecular mass ttnd specific
Sravity (ae well as preasurc and tempcraturo), Thase data were available
from either material balance calculations or P-R elmulntion reeulte,
** Tha Pcng-Robinson simulation uaad propertied giVenin Table 3 with an
adjuetcd KUOP ● il,8 for the C25 + fraction, Using the original KUO factor
of 12.42 gava evan lower liquid dcnnitlcs than thaac &ivcn ahovc, w! th a
larser deviation from tho Alani-Kenady valuca,
TABLE 5 - SIMULATEDCONSTANTVOLUME DEPLETIONDATA FOR THE NS-1 FLUID AT 280 OF USING THE PENG-ROBINSON
EQUATIONOF STATE
Compositions
Equilibrium Equilibrium
Vapor L<quid
Pressure- psia Exp. Calc.
Component 6764.7 5514.7 4314.7 3114.7 2114.7 1214.7 714●7 714.7 714.7

Carbon Dioxide 2.370 2.403 2.447 2.497 2.541 2.576 2.583 0.590 0.595
Nitrogen 0.310 0.323 0.336 0.344 0.343 0.334 0.321 0.020 0,029
Methane 73.190 75.549 77.644 79.135 79.712 79.242 77.772 12.420 11,939
Ethane 7.800 7.779 7.793 7,380 8.057 8.372 8.711 3.360 3.623
Propane 3.550 3.474 3.405 3.383 3.’444 3.660 3.989 2.920 3.133
iso-Butane 0.710 0.686 0.660 0.644 0.647 0.691 0.778 0.910 0.967
normal-Butane 1.450 1.390 1.326 1.281 1.282 1.375 1.567 2.090 2.314
iso-Pentane 0.640 0.604 0.564 0.530 0.516 0.548 0.638 1.400 1.509
normal-Pentane 0.680 0.639 0.592 0.550 0.532 0.563 0.659 1.600 1.770
Hexanes 1.090 0.996 0.889 0.789 0.727 0.744 0.877 3.680 4.223
Haptanes-Plus 8.210 6.157 4.343 2.969 2.198 1.895 2.105 71.010 69.897
—— —. —. —.
Totals 100.000 100.000 100,000 100.000 100.000 100.000100.000 100.000 100.000

Mc,+ 184.0 161.0 142.7 129.1 121.2 116.4 114.5 213.0 209.1
c,+
0.816 0.799 0.783 0.770 0.762 0.758 0.757 0.833 0.843
z 1.203 1.037 0.937 0.890 0.886 0.911 0.936
-z 0.000 9.637 22.581 39.492 56.196 72.413 81.535
‘P
0.00 19.55 26.11 26.65 25.11 23.00 21.58
‘L-%
 IDENTICAL MULTI-STAGE SEPARATION

“gi ~92 ‘g3

’03 r’”
p*c.z.T ‘ST”

Lo
.— mgl + %2 + %3 +
8 B. =
9 Tbc.p VAFOR Vg” 0 PL

‘j
.[l-—%“ ,-1
‘deed LIQUID
x.

PVT CELL

Fig. 1-Schematic description of the procedure for calculating “black

oil” PVT properties.

L
u
fA
‘“030
~’’’’’’’’’’’’’’ NS-I (’4
\ 280° F
‘.
Y-
+ PENG-ROBINSON Z-FACTORS
0.025 -
\ + MEASURED Z-FACTORS

0.020 -

ii!
3

z 0.015 -
0
H
1-
<
=
Ot
0
u. 0.010 -

0.005 -

I I I , , I , , I * 1 1 I , , I {
0.000
o 1000 2000 3000 4000 5000 6000 7000
PRESSURE, psia

2- Gaeformation volume factor vs pressure

Fig. for NS.1 at280 “F.
rnsoooo~~ 11111111111111
PENG-ROBINSON RESULTS
+fIIATERIAL RESULTS
BALANCE
USING MEASURED DATA
NS-I
280”F

L
SEPARATOR CONDITIONS*
10000 - PRESSURE TEMPERATURE
(psia) ● (“F)
1014.7 155
264.7 * 5TANDING LOU PRESSURE
14.7 !% ‘(BLACK OIL” K-VALUE
CORRELATION USED

O. ’’’’’’’’’’’’’’l’’” I“’’l ’’’’l’’”

1000 2000 3000 4000 5000 6000 7000
PRESSURE, psia
Fig. 3- “Vapor” solution gas.oil ratio vs pressure for NS.1 at 280 “F.

rn2.0~, l,,, , , 1 I I I r ,, I , 1 t 1 I , , I , I 1 I I t I I t 1 #
G +PENG-ROBINSON RESULTS NS-I

I
\
-A-MATERIAL BALANCE RESULTS 280°F
g
~MATERIAL BALANCE RESULTS
USING P-R LIQUID VOLUMES (SL)
~ 1.8 /

1.6

,.Qo~AJ
4000 5000 6000 7000
PRESSURE, psia

Fig. 4-011 formation volume faotor vs pressure for NS.1 at 280 “F.
 -1 , 1I I I I I I I I 1I I I I I I i I I , 1, I 1I 1, I 11, t
+ PENG-ROBINSON RESULTS - NS-I
-A- MATERIAL BALANCE RESULTS 280 ‘F
-a- MATERIAL BALANCE RESULTS
USING P-R LIQUID VOLUMES ($L) p

SEPARATOR CONDITIONS*
PRESSURE TEMPERATURE
(psia) ( ‘F)

1014.7 155 ) A
264.7 80
14.7 60

* STANDING LOW PRESSURE

“BLACK OIL” K-VALUE
CORRELATION USECI

J t 1 1 I 1 I 1 , I 1 , t 1 I I 1 I I I , 1 1 I I 1 , I # I 1 1#I 1
1I
I
1000 2000 3000 4000 5000 6000 7000
PRESSURE, psia

6- ‘J1-lquld”solution
Fig. gaa..oil ratio vs pressure for NS-I at 280 “F,

K 240, 1 I I I I 1 , 1 I i 1 1 I I 1 1 , I I I 1 1 I I , 1 I I , I 1
w
I
to +PENG-ROBINSON MATCH
W
e aMATERIAL BALANCE RESULTS
= [ USING MEASURED DATA
k
LIQUID

-i

~ +MEASURED (+) 280° F

I# , , 1 , t , I
loo~ ’’’’’’’’’’’’’”” “
1000 2000 3000 4000 5000 6000 7000
PRESSURE, psia

Fig. 6-Calculated and measursd vapor compositions vs preesure for NS.1 at 280 “F.
 102 I I I I , , 1I I I I , I n , 8 I , 1 1 ,
c1 o 4

— PENG-ROBINSON MATCH
o MEASURED (ACTUALLY THE DATA HAVE BEEN
“SMOOTHED’” AND ADJUSTED ACCORDING
TO CORE LABORATORIES PROCEDURE)

10
C2

P“=
N2 o

NS-I
280° F
11 1 , I t I tt I 1 , 1 I
10-’ ) , I I 1 ,
0 1000 2000 3000 4000 5000 6000 7000
PRESSURE, psia
Fig. 7-Calculated heptanes.plus molecular masses w pressure for NS.1 at 280 “F.

w
‘L”PESi I
‘XTR’PWNS”F
APPARENT CONVERGENCE
d,

F
PRESSURE s 7500 psia

CO*
II
~
Z
ICI-2
k
n lo’~
2.0
‘ ; 1
2.5
1
3.0
1
3.5 4.0
1
4.5
“’:
C25+ *ADJUSTED C7+F VIAL,UE,S
10-3 1 I i
-4 -2 0 2 4 6
COMPONENT CHARACTERIZATION FACTOR, F=b(l/Tb - I/T>

Fig. 8- NS.1 K.vak!es at 280 “F calculated using the material balance approach.
 I I , T I , , t I , * I , , , J

!
PRESSURE O MATERIAL BALANCE RESULTS I NS-I
(psia> USING MEASURED DATA .2 280*F
1-5514.7 —LEAST SQUARES LINEAR
2-4314.7 REGRESSION 4
3-3114.7 (N2ANDCF+ DATA EXCLUDED)
4-2114.7 /: ./~
-5-1214.7 A
6- 714.7

COMPONENT CHARACTERIZATION FACTOR, F=b(l/Tb - I/T)

Fig. 9- NS.1 K.values at 280 “F calculated using ths PengRoblnson EOS,

fo 105 I I I I I 1
.1-
In ’280 ‘F 1

PRESSURE
X 7500 psia-

NS-I
280 ‘F
d“’”
1 1 I 1 [ I
-2 -1 0 1 2 3 4 5
COMPONENT CHARACTERIZATION FACTOR, F=b (1/Tb - I/T)

~lg. 10-Temperature
effect on the Hoffman, et al. K.value corralatlon for NS.1.
 ‘PENG-ROBINSON MATCH (280”F)
‘-- PENG-ROBINSON MATCH (340”F)
o MATERIAL BALANCE RESULTS

u)
> 1-
=j
& 2.0
/ INTERCEPT Y (p)

g
~
1.0 g
w
>
z
$:
+W
~g
%$
$:
<o.
.NS-I

0.0
0 2000 3000 6000 8000
PRESSURE, psia

Fig. 11- Slopes and Intercepts of log Kp vs F plots vs pressure for NS.1.

102 , 1 1 , , , , I , I * 1
< —COSTANT CO!lPOSITION NS-I
> 280 “F
----CONSTANT VOLUME DEPLETION .?
II

= 10

+’
z
$1
z
o
u

e
m
H

10-3 r

10-4 r

10-5 F C25+ ?.
.?
~PRESSUREl
(psia) 714.7 I , , , 1 1 ,
10-6
3X102 103 104
PRESSURE, psia
Fig. 12- Peng-Robinson K-values for NS.1 at 280 ‘F representing two
depletlon processee.
 — —.
102 v I I , t * I , , , , ,

---- CONSTANT VOLUME DEPLETION NS-I

—CONSTANT COMPOSITION 340°F 1

10

10-’

1 o-i
C14-C17

10-2
cI&cz4

10-’

10-5
3X102 103 104
PRESSURE, psia
Fig. 13- PengRoblnson K.values for NS-1 at 340 “F representing two
depletion processes.

x lo2k 1 1 , , , t I 1 1 1 t 1 t 1 1 -i

<
r
E— CONSTANT COMPOSITION NS-2
x ---- CONSTANT VOLUME DEPLETION z410F
1 1

cl,+..--”
10-41 , I , 1 1 I

3X102 103 104

PRESSURE, psia
Fig, 14-Peng.Robinson Kwalues for NS.2 at 241 ‘F representing two
depletion processes.