,

. ..
., -d 9367

UNSOLICITED
;
A COMPUTER PROGRAMMiFOR 14ULTICOMPONENT,
HYDROCARBON PHASE

J
EQUILIBRIUMCALCULATIONSUSING THE PEW-ROBINSON OR A MOOIFIEO
—<
i SOAVE-REDLICH-IOIIONG
EQUATION OF STATE

.“

by R.S. McGlashan (Graduate)4andJ .M. Peden (Member)

Departmentof Petroleum Engineering,lierfot-Watt

—.

-.
.

-,
 . .,

“? 1

A computer programe for mu~ticomponent.hydrocarbonPhas@ equi~ib~fum

calculationsusing the Peng-Robinsonor a modified Soave-Redlich-Kwon~

“ equation of state

Abstract

A robust interactivecomputer model has been developed using

thermodynamicpropertiesevaluatad by the opt~mal selectionof Peng Robinson

(PR) or modified Soave-Redli

ch-Kwong (MSRK) equations of state for high

pressure multicomponent,hydrocarbonphase equi1ibrium calculations.

Evaluationof the thermodynamicproperties using the PR or MSRK equations

of state requires input data of critical pressure, critical temperature

and acentric factor for each component. A procedure is developed for

.
evaluating this data for petroleum fractions such that the equation of .
state matches experimentalreservoirfluid phase behaviour.

The developed phase behaviouralgorithm is tested on a wfde range

of reservoir fluids includinggas condensatesand volatile oils.

The phase behaviourmodel is successfullyapplied to predicting the

depletion performanceofa gas condensatereservoir.

Introduction

Proper descriptionof the Pressure-Volume-Temperature

relationshipand

phase propertiesof real hydrocarbonfluids is essential to enable accurate

predictionof reservoirperformance,liquid recoveriesand the performance

of surface processingequipment.

Recently two constant equations of state such as the Peng-Robinson(PR) ‘1)

or modtfied Soave-Redlich-Kwong(MSRK) ‘2) have been used ”to.pred~ctaccurate

(3) (4)
and consistent thermodynamicproperties for reservoir fluickmixtures ●
T

2
* ...
.$

This paper presents a robustiterativemethod for flash, bubble and

dewpoint calculationsusing eitherof these equations to calculate th@

thermodynamicproperties.

The compositionaland’lysisof a petroleum !’eservoirfluid fs generally

reported as a series of pure componentsfrom atethanethrough,hexane (or

nonane) with the remaining componentsgrouped together as heptanes plus

{or decarteplus). The heptane plus contains a mfxture of hydrocarbon

components heavter than hexane wfifc~typ~cally includes paraffins, aromatics

,
and naphthenes.

The MSRKor PR equation of state can then be used to predict the phase

behavtourof the petroleum reservoir flufd by treating itas a multicomponent

system of pure and pseudo-components.

Applicationof the Equatton of”State

The propertiesrequtred from the equation of state are the equilibri~m

ratios enthalpiesand liquid and vapour densities.

The detailed derivationsof the equations for obtaining these properties

are gfven in the AppendIx; the resultingequattons for the equilibrium ratio

are however summarisedin the followingpages.

The required thermodynamicpropertiesmay be evaluated as follows:

Equilibriumratio, ki, is dsfined as,

.

kf ~ ~i/xi ~ (~tL/xtp) / (~iv/Ytp) s ~tL]oiv (1)

The above equation may be representedas,

,, ,
In kf = (ln$ ‘- In$lv) (2)
 3’
. ..

‘i
-v
In $L and In $i are evaluated using the f’ollowlngthermodynamic

relationship,

1- R?’
ln~i =—l (ap/3ni)T,V,nj,j#~ - ~ dV-lna
Rrv+ [ 1 .,

When equation MSF&KlorPR is used tQ evaluate the right handsldeaf

equation (3), the followlng expression is obtained, :

*18 -...n ~a . . ..b
A
In?i
bi
=~(a-1)-ln~a-b)+—in —
C38 .[ a+c# H 2$:i. W-A
a b1

... (4)

al, bi~ a, b, A and B Involved in equation (4) are evaluated usfng equations

described in the next section.

The values of constants Cl, C2 and C3 depend won the equa~ion ofs~*
●

used in the computations.

i:
For PR equation the constants are defined as follows:o

=1+42 (4a)
c1
c~ =1-#2 (4b)

c~ = c2-c~ (4C)

For MSRKequat~on the constants are defined as follows:-

CI = 1.0 (5a)

C* = 0.0 (5b)

C3 =c*-cl (5C)

The procedure for obtaining the above mentioned thermoc@amic properties

using either of the equations is the same~ and hence only the Peng-Robinson

equation of state will be discussed here in detail. The PRaquation of

state may be representedas, .

 a ●
t b .“ . .
i
,.. .. ..
* . .

“ , ..
.: “

, .“
.4
~.,.,,
‘~
m “ a(T] ..
pa—. (6] “’
v-b v(i+b) +b(v-b) :

where ‘a’ and ‘b’ for pure componentsare defined ast

.*

al(T) = act al(T) -. (7).,.

,..
.
.,
0.45724 R2Tct21E& .
act =
ai*5(T) a 1 + Ml (1’-“(T/Tc~)a5] . .
. .
2
ml = 0,37464 +lS4226Ut - ,2699uf
. .
= 0.0778 RTci/pci (8) “ :“”
bf

Pengand Robinsmamivedat the relat~onshipofa;5~T) using the

vapour pressure data from the normal boiling point temperatureto the critical

temperature.

The mixing rules to be used for mu~ticomponentmixtures with the PR

equation of state are: .,:?

b= ~Xi bi
“i
a=
f f ‘i ‘j aij
* = (1 - dij) 0.5 aj
ai 0.5
aij
where 6ij are binary interactioncoefficientsobtained from experimental

binary VLE data.

If the compressibilityfactor ~ = PV/RT is introduced,the following

cubic equation is obtained:

23-(1 -B) s2+(A-3B2- 2Bjs-(AB- 82-63)=0 (9]

a? bP
where A == andB=—
RT RT

-. The equation of state is used to evaluats the compresstbllityfactor.

—
—
The MSRKQr PR equation is solved analyticallyto produce 3 roots. For

obtaining the liquid phase compressibilityfactore liquid composition Is

used and the smallest root is selected. Forobtain~ng thevapour phase

compressibilityfactor, viipourcompositionts used and the largest root is

selected.

The Pseudo-ComponentMethod

In this method, the C7+ fraction is arbitrarilydivided into narrow cuts

representingpseudo-components,each identifiedby its mean boiling point and

Spec,ificgravity. To obtain the pseudo-components,theT8P cuweofthe C7+

fraction may be divided into equal size cuts, or if theTBP curve is irregular

in shape, the cuts may be taken at irre9ular int@~va~s~ Th@ numb@@ QfPs@ud@

components into which the fraction is broken depends on the accuracy desired
.
and the type of phase equilibriumcomputationto be made. For example, in
.
dew point calculations,the higher boiling portion of the fraction has the
(5);
greatest effect on phase behaviour and therefore the cuts are taken at

closer intervals~near the high temperatureend of the curve. When making

bubble point calculations,the lower boiling portion of the fraction is mare

important.

Froma knowledgeof the boiling po~nts and specific gravities of the

pseudo-components,many of their other properties such as molecular *ights*

critical pressures,critical temperaturesand acentric factors can be estimated

by different correlations.

If a “completeTBP analysis”’is not ~vailable and the C,+ fraction is

thereforeconsider’ed
as one pseudo-component,the simplest characterisation

method is to use the propertiesof an n-paraffin having approximatelythe

same molecular weight as the C7+ fraction.

However, the fact that the densitiesof the heptanes plus are usually
 ..
“i

higher than those of normal paraffins having the same molecular weights,

Indicates that they contain aromatic and naphthene compounds. To characterise

the heavy fraction using only n-parafftnparameters could lead to erroneous

results (4). A better approach is to specify the average boiling point,

molecular weight and specific gravity Of t$e C7+ fraGtiOns and us@ the

covelations to estimate the propertiesof this one pseudo-component.

AlgoritMa~or the Solution of the EquilibriumCalculations

The calculationof the bubblepoint, dewpoint and the proportionsof

vapour and liquid in equilibriumwhen two phases coexist fs based on the

principlesof mass conservationand thermodynamicequili riumat the given

temperatureand pressure.

For examples consider the solution of an equilik .m flash calculation

●
F
when the overall feed stream composition,~i~ enthalpy!ih! and system

pressure are specified. The equilibriumflash ca..ulatlonbecomes the so~ution

of three simultaneousalgebraic equattons:

n =i (1-~f)
f, (G,T,a) = Z Molar Balance (lo)
j=l 1 +a(kl - 1)
~F V
- ah - (1-a) hL
f2 (a,T,~) = Enthalpy. Balance (11)
hF

f3 (a,T,z) = 23- (1-B)a2 + (A-3B2-2B)z- (AB-B2-83)

for the PRE.O.S. (12)

or f3(a,T,z) = z3- Z2 + (A-B-B2)~- AB

for the MSRl(E.O.S.

In orderto solve the nonlinearalgebra~c equations used to model flash,

bubble and dew point calculationsa method must be found for evaluating
 t

,’ ?’
*

component equilibriumratios or k-valu~s~ kio The k-value of a @Wo~ent

In a mlxtwe is a function of the temperature,pressure and compositionof

the mixture. The program uses a cubic equation of state to evaluate

component Iwalues. At a *gtven temperature,pressure and 1tquid or vapour

co~oslt!on the equation of state can be salved to give the liquid and vapour

phase compressibilityfactors which arethen used to evaluate the liquid and . L

vapour cmponent fugaclty coefficientsand hence the component k-values.

S4nce the equation of state is cubfc it can be solved exactly for

.
every fteration in the implicit w.rfables~ SO for the %ul~ibti~ flash

calculationonly equation (10) and (11) are solved simultaneouslyby a

quasi-ffeuton
method ‘6).

As an example consider the solution of an equi1ibrium flksh calculation

when the overall feed stream compositio~,enthalpy and system pressure are

specified. First an estimate of the temperatureT, is made using a m-arranged

..
Antotne equation. Then in order to evaluate initial vapour and liquid

compositionsa compositionindependentequation for calcula’ ng initial

k-values,ki is useds
(7)
~ = exp [5”37 (~*#l-lflrjjl
. (13)
i
Pti

where:

‘1 = acentric factor of component i.

Tri . reduced t~perature of co~onent i.

Ppi = reduced pressure of component i.

The initial liquid compositionis obtained by assuminga, the fraction

of feed vaporised is 0.5 and using the composition independentk-values,$

in equation (15). The inf~ial vapour compositionIS obtained by applylng the ..

-
 8

equ~libriumrelationship:

Y~ =

These compositionsa?e used to calculate compositiondependent k-values

by solvtng the chosen equation ofstateortce for the Iiquld and once for’the

vapoup phase. The equation ofstateylelds liquid and vapour phase

o compressibilityfactors which are used to calculat@the Ilquid fugacftY

coefficients -
$iL ‘v
and vapour fugacity coefficients,@l for each companent

and hence the k-values:

~L
TV
kl=(bxipm ‘/ ytP ) (14)

The liquid and vapourenthalpies areobta’inedby combining the ideal gas

enthalpy with the Isothermaldeparture in enthalpy for each phase calculated

from the appropriateequation of state. v

The equilibriumflash equations (10) and (11) are then evaluated. If

the values of these functions do not satisfy a defined tolerance then the

implicit variablesaand Tare updated by a quasi-newtonmethod and the

compositiondependent k-values are calculatedat the new temperature.

Equations (10) and (11) are then re-evaluated. This constant

composition inner loop is repeated until the functions satisfy the defined

tolerance. The new compositionsare evaluated using the temt~al k-values,

[i anda.

at
(15)
“i = —
1 + a(kl-l)

(16)

These new composit’ions

are then compar+i to the hjtial ampositions~
 “f

and if the square of their difference Is not within a defined tolerance the

toner loop is resolved using the new composttlons. A f~owsheet for the

algorithm ofan equilibriumflash calculationis given In Fig. (1). A

similar algorithm is used-to solve the other models. All the algorithms

have been formulatedtnto a computer programne called FLA!M4AN. The

prograrmneis written in FORTRAN IV and implementedas a conversational

timeshare package usfnga visual display unit lfnkedto alloneywell 66/6000

computer. The programne contains a 60 component data bank and the following

calculationopttons are available for use with either the PRor MSRK

equation of state.

1. equilibriumflash calculation

2. isothermalflash calculation

3. dew pointor bubble point pre~sureor temperaturecalculation

4. component equilibr’fum
values and total phase enthalpy calculation

5. optimisationof binary interactioncoefficient

6. fitting the equation of state to saturationpressure and densfty

of a multicomponenthydrocarbonmixture

Evaluationof the Prowamne

A classificationof natural hydrocarbonfluids is given in Table (1).

The-delineationbetween fluid types is not precise so onlya detailed study

of the particularhydrocarbonsystem beingconsideredcan produce meaningful

results. Although the progranme has been tested on synthetic hydrocarbon

mixtures for the purposes of this paper only real hydrocarbonmixtures will

be described. ..

IlryGas System

Erlmr ‘5) reports a dewpoint curve fora dry gas sytem for which the
* .

10
. ..
‘?

C6+fractlon has been split into ten pseudo-componentsof known boiling

point, molecular weight and density. The FLASHMAN program was used to

generatea series of dewpoint curves for different characterisationsOf

th@ C6+. The results are-presentedin Fig. (2) for the PR equatton.

CharacterisfngtheC6+ fractton as n-nonane or one pseudo-c~ponent

gave a poor overall fit to the expev~mentaldew point curve. The d~ point

curves genera$ed by treatirgtheC6+ as

. ~~n pseudo-compotients
and then

modifying the calculatedcritical temperatureof each pswiomxnponent

produced the best overall fit to the experimentaldew point curve.

Volatile Ofl

Fig. (3) gives the compositionofa volatile oil reservoir fluid

together with the bubble point pressure.anddensity at a reservoir

temperatureof395 K. The FLASHMAN program was runkdth theopt?on to .0-,

fit the equation of state to the experimentalbubble point pressure and

densityat 395 Kby adjusting the boillng point and critical temperatureof

tbe Clo+ pseudo-component. The resulting percentagedifferences are given

In Fig. (3).

The fitted equation of state was then used to performs series of

isothermalequilibriumflash calculationson the volatile 0+1 at pressures

ranging between 29.52 MN m‘2 and 4.08 w m-2@ A comparison between

experimentaland calculatedmixture densities Is given in Figs. (4).

.2’
The PR equation gives a maximum differenceof-2.4% at 1S.89 MN m .

The fittad equation was used to perform an equilibrium flash

-2
calculationat 290 Kand 101.35 kN M on the nservoi$’ fluid composition

given in Fig. (3).

The resultingphase compos’itions

anddens~t~es are given in Table 2for

the PR equation. Theequatlon gave excellent agrement w!ththeexperimental

 ,
%* I*

il
, ,.
I
. .
r
values. However, the experimentalerror for mole fractions in the range
‘(8) ‘“
0.001 to 0.0001 is estimated to be between 30 - 100 per cent .

The comparison between mixture densitiespredicted &y laboratoryPVT

measurementsand the FLASHMAN program are given fn terms of a percentage

dlf~erencebecause the errors associated in the laboratorytechniqueswere

not avaflable.

One advantage ofuslng a computer IWWWII such as FLASHMAN to calculate

PVT’relationshipsts that once the program has been “calibrated”to some

experimentaldata the program’s results are consisten&-whereasthe

experimentalPVT analysis has a random error associatedwith the experimental

Discussionof the Results

,
The FL.ASHMANprogram has been successfullyapplled to predict the

phase behaviourof various reservoir fluid mfxtures. No trouble was

experiencedwith the algorithms for the solutionof the equilibrium

calculations. To represent the phase behavlourof a gas condensateor dry

gas system extended analysis of the C7+ fraction is needed. The program can

then be “fine tunqxt”to the experimentaldew points by varying the calculated

critical temperatureof the pseudo-components. The two-phasemixture density

of a crude o?l system can be reproducedby representingth@ C7+ fraction as

one pseudo-componentof known specific gravity and molecular weight. The

pseudo-componentboiling point is the variable used to match the experimental

and calculated bubble point pressure and the critical temperatureis the

variable used to match the experimentaland calculated saturationdensity.

This is an ’iterativeprocess since the correlationsfor the pseudo-component

critical properties described in the Appendix are a function of the

pseudo-compunentboiling point.

The equations of state have to befitted for each reservoir fluid.

u *

t ..
12
. ..
.
“’l
Since a semi-empiricalequation of state is befng used to represent the

PVT properties of real reservoirfluids the vindicationof any matching

process is in the results obtained. The results obtained justify the

matching techniquesused with ths FLASHMAN progmm.

“ petroleum EngineeringApplicationfor the Proqram : ~ ‘

Since Petroleum Engineeringis essentiallyconcerned wfth the flow of

a complex hydrocarbonfluid from the reservoir through the mechanical

equipment such as tubfng and Initial separationttathe downstream processing .

activities, it would seem reasonableto suppose that compositionalmodelling ‘

offers great scope in fts applicationto Petroleum Engineering.

Potentfal applicat~ons for such a compositionalpackagewe as follows:-

(A) Reservoir Studies \

1) Reservoir fluid depletion studies with declfning pressure. ~

2) The applicationof compositionalmodel1ing to material balance

studies on the reservoir.

*
3) Applicationto reservoir simulation.

4) Estimationof well productivitiesby compensateng for compositional

effects.

5) Mi,s@bility studies in enhanced oil r=over% e*9t C02mis@bility

with hydrocarbonsystem.

(B) Well Studies

1) Vertical 11ft performancein production tubing‘by calculationof

the two phase pressure drop. ~

2) The design of gas 1ift systems.

*
13 . .. ,.
. t
“#

(C) Anctllary Areas

1) Surface separator design.

2) Calculationofhydrocarhon density variationwith pressure

for use in well ’controlcalculations.

?%Wofth eapplieationsar eat present being developed hut in viewef

the limited space only two of thenarediscussed below.

Reservoir Depletion Calculationsfor Gas CondensateSystems

Amass balance subroutine,DEPLH, was incorporatedinto the FIJWMAH

program in order to handle stepwise depletion calculationsfor gas condensates.

Firstly, the dew point pressure is determinedat reservoir temperature. The

program then performs an isothermalflash calculationat a chosen pressure

less than the dew point pressure, follo~ed by the removal of the excess

gas created by the expansion. The phases are then recombined to fom a “

new composite reservoir fluid which is isothmnally flashed to a new pressure

which is defined by the user.

This procedure is repeated for a series of incrementalpressure

reductions. These calculationsgive the produced phase compositionsand

the volumetric behaviourof the gas condensateas well as the vesidual

reservoir fluid compositionat each depletion pressure.

Data on depletionmeasurementsand phase separationscf a gas condensate

fluid from an Iranian gas well, Pazanan 10, has been published by Firoozabadi ‘9).

He used this data to investigatethe effect that t~~,binary interactionparameter,

61j, between methane and the heavy fractions has on the predicted phase
.
behaviour calculatedby the PR equation of state. He concluded that ntodfying

did not significantlyimprove the results even with extended analysis of

6ij
(10)
the heavy fractions to C25; Recently, peneloux reported a new
s b

14
.,,-
.,;

characterisationmet~ad for the C7+ fra~tiono The C7+ fraction is considered

to be a pseudo-binarycomposed of a “llght” n-alkane and a ‘heavy” n-alkyl-

benzene In such proportionsthat the molecularweight of their mixture is

equal to that of themeas~red C7+ fraction. Peneloux used this characterisation

method and a non-cubic equation of state similar to the Redltch-Kwongtype

to.perfoma constant volume pressure depletionan the Pazanan 10 gas

.
.,..
condensate,

Peneloux concluded that hfs method was as accurataas the extended

4 z. . .
analysis approach of Ffroozahadiand also easier to apply. ..
.
The FLASHMAN program was usedto perform the pressure depletion

calculationson the Pazanan 10 gas condensate. The reservoir fluid

compositionis given in Fig. (5). The reservoir flutd is characterised

by nine pure components and ten pseudo-$omponentsrepresentingthe

(It) ‘
boiling point ranges C6 tO C15 as giv@n by Benman Q :
A comparisonof the results of Firoozabadiand this study are given

in Table 3 and show good agreement except that FLASHMAN under-predicts

the exiJ@rimental
volume percent liquid at lower pressurws compared to

Firoozabadi’smodel which overpredictsthe experimentalvolume percent

liquid. Both models underpredictthe compressibilityfactor for the produced

gas.

The method of characterizingthe C7+ fractfon ~fa 9a~ condensat@ us@d

in the FIASHMAN program gives encouraging resultsn predicting the phase

compositionsand volumetricbehaviour of a gas condensate undergoing a

constant volume pressure cfepletton.

‘.
“,
 .
● ‘v
* b
* 15 ., , . .!,..
.‘
;0
v
. .

‘
‘t

Pressure Drop Predictionin Oilwell ProductionTubin~

The high flawratecondftions in North Sea Oilwells are such that the

main contributionto the pressure gradient is due to the gravitational

Wponent. Because the gravitationalcomp~nent ts directly proportional

to the in-situ fluid density, an accurate calculationof the amount of gas

tttatwill come outof solutio nata given pressure and temperature is

extremely important for this calculation.

.
Uhen awell is producing at a steadyflow rate the total mass flow

rate of oil and gas ip the wellbore is constant, The assumptionof no-slip

or homogemaousflow means that the process through which a barrel of

reservoir otl undergoes in flowing up the wel1bore from the producing

formation to the wellhead can be consideredto be constant overal1

compositionseparationor an equilibriumflash liberation {12)-

.

The in-situ mixture density at any ‘point in the we?1 can therefore

be defined in terns of the input volumetricflow rates of each phase under

the flowing pressure and temperatureconditions.

The FLASHMAN program was used to calcutate the mixture density as a

function of pressure for a constant temperature.

So that these data can be used to calculate the density in the wellbore,

sfakameans of compensatingfor the temperaturechange from bottom to surface

conditions is required. The simplest way of doing this is to interpolate

betwen the density curves at bottom-holeand wellhead temperatures. The

value of ~ used is the value of the fluid denstty at the mean of the two

pressure points being considered.

The pressure gradient equation used in this study for calculating the

pressuredropAP/A% between points one and two in the well was developed by
(13)
Cornish .
. . . . . . ,.

“.. . . ;..

‘
* b
. .
*
●
,.

.$ ,..
16
, * ...
2
AP ;gcose’ Gf G2 Ii (P1/P2).
144— ~ — +— +— (17)
AZ 9~ 2P gc di ~ gC A*

where:

i = average mixture density between points 1 and 2

P2 = mixture density at point 2

P1 = mixture density at point 1 “

G = total mass flow nte per unlt are~

f = Moody frtct~on factor

,“
dj = internal diameterof tubing

The three terms on the right hand side of equation {17) represent the

gravitational,frictionaland accelerationcomponentsof the total pressure

gradient.

Fer flow in a deviated wel1 the gravitationalterm is corrected for

the true vertical depth. This permits ~riction and accelerationeffects
?
to be taken over the total tubing depth while the gravitationaleffect

is only taken over the vertical length.

Figure (6) gives a comparisonof pressure profiles fa~ different

methods of calculatingthe flowing mixture density. Model I has

empirical correlationsex~ressed in terms of measured field parameters

used as produced gas gravity, tank 0$1 gravity and initial 9as-oi~ ratio

while Model II used the compositionalapproazh. The reservoir fluid

compositionis given in Fig. (7).

●

Conclusions

computer program, FLASWIAN. has-been written to predict

An interactive

mixtur~s using
the phase behaviourof mult’icomponent either the Peng-Robinson

or modified Soave-Redlich-Kwong.
equation of state. Mhen modelllng.reservoir
..
fluids, it is necessary to represent .he heavier hydrocarbonsby means of

pseudo-components. The pseudo-componentpropertiesare derived from mpirical

correlations. These properties have a significanteffect on predictionsof

phase behaviourat high ptiessuresand have to be tunedto match the

experimentalbehaviour. Proper tuningof the pseudo-componentproperties

has allowed the applicationof the phase behaviourmodel to predicting the

depletion performanceof a gas condensate resewoir and providing density

data for pressure drop prediction in oil well productiontubing.

Acknowledgements

One of the authors (RS McGlashan) wishes to thank Shell U.K.

Explorationand Production forproviding him with the financial

dssistartcefor the work.

—..
 ,
.,
9 *

18 . ..
.,
“r!

APPENOIX

Calculationof Enthalpy and Density using the Equations of State

b

The liquid or vapourenthalpy is obtained by combining ideal gas

enthalpy and the isothermaldeparture In enthalpy from the Ideal gas as

follows,

(u%)

~L=n
Z xi HiO + HOEP (19)
1=1

The ideal gas enthalpy is evaluated at the

\
system temperatureusing the six

coefficientequation described by l%ssut and Danner (14).. The reference

?
state for their equation, stated as follows, is selected as OOR.

HIO = (Ai + BIT + C*T2 + D1T3 + E1T4 + FfT5) * MWi (20)

where MWi is the molecular weight of component i.

The isothermaldeparture fn enthalpyis calculated using the following

thermodynamicrelationship,

P
HDEP = -f (v- 7#)) dP (21)
o aT p

Use of the equation of state into the above equation gives the following

result,

awl
B
[1
(a’(T) T - a)
HDEp = RT(a- 1) + In — (22)
c3b Z*2B

The equation $ora’(T) is obtained as,

 .,
,
. Ii

. O** (23)

The Iiquidor vapour density is obtained

(24)

(25)

where :

aL is the liquid phase compressibilityfactor

*V *S the vapour phase compressibilityfactor

R is the universal gas constant.’

Correlationsfor the Predictionof Pseudo-ComponentProperties

Consider that only the boiling point, specific gravity and molecular

weight of the pseudo-componentis known. There are ag least six correlations

available to predict the critical propertiesand acentric factor of the pseudo-

cotilponent
using its boiling point and specific gravity. These are the Caveti

Method ‘15)S “MwM hod” ‘16), Lin and Hopke Method ‘17), the method in

GPA RR13 (18), Robinson’sMethod (74), and Bergman’s Method (17), .

All the methods except that of Robinson are basedon correlationswhich

attempt using various parameters such as boiling point, critical properties,

gravity, etc., of the lumped fraction to attempt to characterise it based on

the Paraffin Napthane Aromatic (P.N.A.)distribution. Robinson’smethod

however offers a slight departure from theothers in that a series of

* ●

20
● “’..
“t

slmultanaousequations is setup and solved for the P.N.A. distribution

usfng stmilar measured parametersof the lumped fractfon as suggested

*
above.

For this wark the method proposed by Cavett has been used and Is

discussed in deta~l below.

‘..

Cavett Method .

In 1962, Cavett ’15) proposed correlationsfor calculatingthe pseucfo-

component properties. His correlationsrequire the mean average, cubic

average and molal average boiling point temperatures. Although these

temperaturesntaybe evaluated from the rtonnalaverage boiling point

temperatureusing graphical correlation from “theAPI data book ’20},

all the average boiling point temperaturesare approximatedto the normal

.
average boiling point temperature. (This is exact if the C,+ is characte~sed

by one pseudo-component). The correlationsforobtatning the critical

properties are as follows:

.

Tc = ao+al (Bp) +a2 (BP)2+a3 (API) (lF)+a4 (BP3) +a~ (API) (BP)2

+a6 (API)* (6p)2 ● ** (26)

logPc = b. + $ (BP] + b2 (fitP)*

+ b3 (BP)3 + b4 (API) (BP)

+b5 (API) (BP)2+b6 (API)* (BP) +b7 (API)* (8P)2

Q*Q (27)

141.5
where API = - 131.5
specific gravity

Tc = critical temperature(OR)

PC = critical pressure (psia) “ . .

API = API gravity

a II *

●
*

●
21
. ..

where the constantsa. tO a6 and b. to b7 are tabulated in Table 4.

The acentric factor, w is calculated using the Edmisterequat~on (28)

as follows:

@ u 3/7 -1.0 (29)

. rc
— “ 1.0
n.
. .

TB = normal boillng point (“R).

Equations (26) to (29) are recmnendad for pseudo-componentswith ‘API

greater than zero.

Cavett’s method is applied to calculate the critical temperature,

critical pressure and acentric factor (equations(26) tg (29)) for use in

the Peng Robtnson equation of %tate. l%eminimum informationrequired to

evaluate these properties is my two of the three pseudo-componentproperties -

normal boiltng point, specif!c gravity anti/ormolecular weight. Recently

Bergman (11) has prcoosed

, empiritally fitted equations whfch can be used to

evaluate the unknown density and molecular weight of a pseudo-component

from its boiling point.

MW = 54,389 + .17566 (BP)+ 1.2102 (BP/100)2+ .22885 (BP/100)3

● ** (30)
D = MW (8.5!37X 10-3 + 5.647 X 10‘6 (BP) - ~.303 x 10-8 (BP)*

+ 1.513 x 10“0 (BP]3 -8.742 X 10-14 (BP)4) . . , (31)

MW = molecular weight

D = density (g/cm3 at 60°F)

BP = normal boiling point (°F)

 .
, ●

.
22
* ...
* .
“P

frtation factor

fugacityofcmponent i In liquid phase “

din!enstionless
conversionconstant

total mass flowrate per unft area

enthalpy

idea? gas state enthalpy of component i

isotkennalenthalpy departure

equilibriumratioG, ki =YiiXi

number of moles

pressure

critical pressure of component \

b
reduced pressure, pR = p/pc

universalgas constant

temperature

critical temperatureof component i

reduced ts~perature,TR = T/Tc

molar volume

mole fraction of component i in liquid phase

mole fraction of component f in vaPour phase

compressibilityfactor

overall mole fraction of component i

Subscripts

c critical

i,j,k component number

 .
● ☛ P
●
8

Superscripts

F feed .

L liquid .b

v vapour

— ... Greek —..—.

a mole fraction of feed vaporised

interactioncoefficientbetween componentsi and j

‘ij

9 density
L
$i fugacity coefficientof component i in liquid phase

(0 acentric factor
 .
● * ●
9 6

24
●
..
i

References

1. Peng, D.Y. and l?obi~son,11.f3.,

1976, Ind. Eng. Chem. FiRO, 15, 1, : 59.

2. ;pb~k:,4~j and Daubert, T.E., 1978, Ind. Eng. Ch~. P.D.D.*

9 ●

3. Conrard, P.G. and Gravier, J.F., 1980, 011 and Gas J.,i78, 16, : 77.

4. Peng, D.Y., Robinson, 11.Biand Elshn@i, P.R., 1976, Pmt. 9thMwld

Petrol Congo, Tokyo : 377.

5. Erbar, J.Ii.,Ililson,A. and Maddox, R.N., 1978, O*I and Gas J.* 76* 34, : 76-

6. 8royden, C.G., 1967, Math. CAMP., 21, : 368.

7. Fussell, 0.0. and Yamsik, J.K., June 1978, J. Pet- Tech* : ~~30

8. Yarborough,L., 1972, J. Chem. !%g, Oata. 17, 2, : 129.

9. Firoozabadi,A. and Katz, D.L., 1978, Can. J. Chem. Eng., 56s 10~ : 610.

10. Peneloux,A., 1979, SPE Paper 8287 presented at 54th Annual Fall Conf.
at Las Vegas (23-26 Sepk., 1979).
*
11. Bergman, D.F., Tek, M.R. and Katz, D.L., ?975, “Retrograde
Co.ndensatlonIn Natural Gas Pipelines”,Mer. Gas Proc. Ass. Arlington.

12. Oobson, C.R., 1953, J. Pet. “iech.Trans., AIME, 19$3: 287.

13. Cornish, R.E., July 1976, J. Pet. Tech. : 825. ‘

14. Passut, C.A. and Oanner, R.P., 1972, Ind. Eng. Chem. P.D.D., 11, 2, : 543.

15. Cavett, R., 1962, API’Div. of Refintng~42, 3, : 351.

16. Erbar, J.H., 1964, NGPA k and H Value Program, Gas PYYW Ass. Tulsa.

17. Hopke, S.W. and Lin, CA., June 1974, Hydrwarbon I%%m#ssing : 136.

18. Erbar, J.H., Prediction of Absorber Oi’1k-Values and Enthatpies, 1~~~~

Gas Proc. Ass. Tulsa.

19. Peng, D.Y. and Robinson, O.B., 1978, GPA.RR 28, Gas Proc. Ass. Tulsa.

20. Technical Data Book - Petroleum Refining. 1977. 3rd cd., API,
NashtngtonD.C.
,

tieptanes+ ..
Methane Intermediates
~(ml %) (mo7 %) (mol %)
.—

l!laclc
Oil 30 35 35
..__—

Volatile Oil 55 30 15

Gas Condensate 70 22 8

Dry Gas 90 9 1

Table 1 Typical compositionsof hyd~carbon fluids.

.—--

.
 ●
● .@’-

.
. . ,.
.
- ‘d

FE. 1. Flomhart for the Algoritiun of an EQUi 1ibrlun flash Calcuhtion

dlsti

&

‘+
I EVALUATE IHITIM
$ frax Equ4tian
k V&USS,
( m“)
I

I . FRACTION OF FEEII
VAPO RWO. a = 0.S I
CAUULATE PNAsE’cawsmORs
*i
‘i 8
1.0+0.5(fit
- 1)

I Yi =Ei xi

-QL-
ruPOATE
QliPOSITIONS
‘i
8 ~o
Yf 9 Y’f
i
EVAIUATE Contrition
kVALUES, ki f = 1, ~
LIQUXD EIJWY,
ANO WQUR
OEPSNCENT

ENTMLPY,
hL
hv .

UPDATE a MO T
SV WASI -NEMTON
HE’fN~

THE THE

1
@JTER lflNER
LOOP LOOP

.. ... .
2.
,.

.
 .

FIG. 1. (continwd)

).

REACNSO SOLUTION

L-
PM a,T

‘k
SYAIUATE NSM PHASE CWOS1TIONS
et
xl’ .),” IT
I
+ ‘w -
—!

.
q ‘Skf%’
i
i ml, NC

NO

==j .. .
.
—.

,
4

SOLUTIONS REACttEO FOR

FRACTION OF FEW VAm RISEU a
TEMPERATURE .T
LIWIO CWOSITIONS * xi’
VAPWR txMPosmoNs ●Yf’
1=1*NC

I
.
o m

.
.

—
 8 ●

.
m!f!!u Dewpoint curve fora dry gas as predicted by FLASHMAN
(PR aquation) ustng different characterisationsof the C6+.
. . ,..
C6+ as n-paraffin
C()+as pseudo-component(TCMOD = l.(f)
Cg+ as ten pseudo-components(TCMOD =+).0
C6+0as ten pseudo-components(TCMOCI= 6.97)
!teasured(ref. 5)

..

27*579

20.684

n
y \

?’ Feed composition &.

13.789

—
.

,.
..

145
189
220
!!
105
250 116 \
295 131 &
320 138
369 155 I
ail420 175
470 193
!E 365 231
●For lumped G calcd -1=
.. 1299 moi% BFr=27s*F. PI=
~ nCO
59.7S, moi wc = 124.31, L299 ,
;. mohMb aGpropaKd*
—

I 1
/“”
%

f
311 366 422
 W5ERVOIR FLUID Cf)MPOWIOl!

8:
;OMPONENT MOLE PERCENT

N2 1.32
CO* 15,99

% 42.33 ‘
C* 8.37
c~ 8.60
ic~ 1.29
.
nC4 3.46
. —. ..-. ,.
ic~ ‘ 0.98
nC5 1.55
c~ 1.76
Cy 2.80
c~ 2.62
Cg 1.74
7.14
Clo+
L-
.

MOLECULAR MEIGHTOF CIO+ = 381 ., ..

SPECIFIC GRAVITY OF CIO+ = 0.89 (ass~ed)

t
SATURATION SATURATION % DIFFERENCE % DIFFERENCE
PRES URE DENS TY IN SATURATION IN SATURATION
MNm-3 at 39% kg dJ at 39% PRESSURE DENSITY

EXPERIMENTAL 32.56 574 .

PR 32.55 573.5 - 0.03 -0.09

Fig. 3 Comparison of experimentalsaturation pressure and density for a

volatile oil with those calculated by FLASHMAN using the PR equation.
 .
,.
“s
●
.
✎
✎✎ ✎
✎

●
✎✌

“f;

MOLE FRACTION VAPOUR MOLEFRACTIONLIQU ID

.
EXPERIMENTAL

*
CALCULATED EXPERItlENTAL CALCUM7ED I
0.0157 0.0157 0.0000
‘2
0.1901 0.1899 0.0031
0.5018 0.5037 0 0.0027
.
C2 0.0989 0.0990 ().0026 0.0036
0.0997 “ 0:0999. 0.0125 0.0132
iC4 0.0144 0.0143 0.0047 0.0053
. nC4 0.0370 . 0.0373 0.0216 0.0203
fc5 0.0087 0.0091 0.0149 0.0133,
nC5 0,0?29 0.0134 0.0296 0.0267
0.0088 0.0091 0.0651 0.0623
C6
0.0087 0.0062 0.1313 0.1420
C7
0.0031
C8 0.0019 0.1502 0.1530
0.0003 0,0004 , 0.1119 0.1093
C9
0.0000
. ... . . .
c10+ 0.4550 0.4448 ,
.

I I
DENSITY (kg ;3 )
.
VAPOUR LIQUID “

EXPERIMENTAL 1.6696
I I
845.6
CALCULATED 1.2232 804.1

Table 2. Comparison of experimental phase composition and densities

far a volatile otl at 290?K and 101-35 kNm-z with those
calculated by FLASl$lANusing the fftted PR equation.

.
 . . ..
$
.,

t
“1;

MIXTURE DENSITY (kgjii3)at395 K

;NDj~m’WFECE
PRESSURE EXPERIM&WAL CALCULATED
M~-2
mNSm ..

29.52 555 550.9 -0.7

23.48 503 491.8 -2.2
15.89 387 377.8 -2.4
a*f14 204 200.1 -1.9
4.07 98 97.7 -0.3
.

Hg. 4 Comparisonofexperhnental mixture density for a volatlle

oil with values calculatedby.FLASHMANusing the fitted
PR equation. Fitted parametersare Clo+ botling point =
664 Kand TCMOD = 1.0672. —
— ... ----

--

●
 .
b .

●
. ..

RESERVOIR FLUIt)COMPOSITION”

cQ\lpoNENT MOLE PERCENT

v
..
0.08.
‘2
CO* 2.44

c1 , 82.10

C2” 5.7a -
2*&! “
C3
ic4 0.56

flc4 1.23

ic5 “ 0.52

ncs ●
0.60
I ..
i ..
% 0.72
C6
t 0.75
~
0.75
C8
Cg 0.49 .
... . ..
. 0.31 ~ -.. . . .
Cyo.
0.22
c11
0.15
c12 I
0.12
C13
0.09
C14
0.24 “
C15

-.
Fig. 5 Reservoir fluid composition of %zanan 10 gas
—
condensate.

.
 . >

,.
.

..

‘P
CALCULATED o - \’
PRESSURE ~ EXPERIMEW4L .

(}1M”;2abs ) VALUE
FIROOZABADI ( 9“ ) F~S~A~

ilEli
POINT
PRS5~dAE - 28.1 26.6 ““27.3
.—
WELL STREA$lPROOUCED, % OF INITIALPRESSURE
..
.. .
23.5 11.91 9.56 “:” 11.30
18.0 30.77 29.07 28.19
12.5 52.05 50.79 53.78.
8.4 68.41. 67”.34 “69.49 “ ..

4*9 81,21 80.59 82.42

GAS CJXIPRESSXBILITY
FAC+OR; ZV

23.5 0.852 0.828 0.826

18.0 0.836 0.816 “ 0,814
12.5 0.858 0.$36 0.835
8.4 0.893 0.872 0.871
4.9 0.932 0.914 0.914.
HEPTANES PLUS IN GAS PHASE .

23.5 2.59 2.51 2.52

18.0 2.01 1.67 1.72
12.5 1.47 1.09 1.12
8.4 . 1.21 0.84 . 0.86
● .- .
4.9 .. 1.20 0.81 0.82
VOLUME % LIQUID IN.CELL - ‘
23,5 1.3 2.74 ~ 2.10
1~.(1 3.4 5.14 ‘ 3.63 “
12.5 4.6 5.74 3.68
8.4 4.7 5,50 ““ 3.78
4.9 4.4 5.02 3.78

Table ~ L Cunparism of @xp@rim@ntal and calculated depletlqn

behaviour of pazanan gas ,condensatesystem at 355.sK. “ ~ ~ :;
 ● ✌☛

. ..
.
.
PRESSURE MNm-2

A MODEL I
x MODEL II
* MEASURED

Figure 6 Comparisonof pressure profiles

 .
.

a b A B c o

768.017121 2.8240406 ~ ‘ 3.1130335

x 10-2 -8.186204X 10-’ 3.1073165
X 10-4 -5.6334748X lC
1.7133693 0.44120109x 10-3 3.5442502 -2.8568658 X 10-5 - S.8323815 X 10-8 9.1745019
x 10-1
-0.10834003 x 10-2 -0.30474749 x 10-5 -5.2716944X 10-2 1,7024654 X 10”6

-0.89212579 X 10-2 -0.2087611 X 10-4 3.5353777X 10-4 -1.4886832 X 104 “

0.38890584x 10-6 0.15184103X 10-8 ~0340~6~8 x 101 8.8395563 X 10-2

0.53094920x 10-5 0.11047899X 10-7 -8.9530784X 10 ~ 4.2934703

x 10-3
0.32711600x 10-7 -.0,4827?599 X 10-7 3.5118581
X 10-1 -3.0836492 X 104
<. :
0.13949619X 10-9 -5.9563889 x 10-3

TABLE 4- Constants
for Tc, Pc and H* usingCavett’sMethod.

*.“
.8 &
.’

*
● * ● .

● *
.
*

RESERVOIR FLUID COMPOSjfION

COMPONENT MOLEPERCENT

N2 1.25
co* ‘ 0.16
cl
23,81
C* 8.05
c~ 9.11
ic~ , 1.41
nC4 4.47
ic5 1.56
nC5 2.27
4.29
C6
43.62
C7+

Molecular weight ofC7+ , = 226

Specific gravity ofC7+ = 0.8684
Fitted boiling point ofC7+ = 555 K
TCMOII = 1.0148

SATURATIONPRESSURE, SATURATION DENSITY,

MNm-2at 352 K k~-3 at 352 K

EXPERIMENTAL 11.139 726.09

PR 11.138 725.13

Figure 7 Reservoir fluid compositionand fitted parameters

equation.