**Unformatted text preview: **C7+ CHARACTERIZATION FOR FLUID PROPERTIES
PREDICTIONS
R.S. WU R.M. FISH this article begins on the next page F JCPT89-04-09 RESERVOIR PERFORMANCE AND OFFIMIZATION C + characterization for fluid properties predictions 7 RS. WU and R.M. FISH Esso Resources Canada Limited ABSTRACT A simple iterative procedure has been developed for characteriz- ing the Cl fraction of a hydrocarbon fluid with only one Pseudo- component. This procedure applies the Peneloax, Rauzy and Freze shifting factors and Watson's heavy end correlations to the Peng- Robinson equation of state. The procedure requires the follow- ing @nniental data. (1) bubble point pressure; (2) compositions of gas and liquid resulting from a room
condition flash of the reservoir fluid; (3) flashed gas-oil ratio, and (4) reservoir fluid den- sity at pressures above the bubble point [email protected] "en the equa- tion of state is correctly tuned, the results match not only the reservoir fluid phase behaviour and fluid density, but also the phase behaviour of m
of reservoir fluid blended with hydrocarbon and carbon dioxide solvents. Introduction In order to predict the fluid properties and phase behaviour of hydrocarbon fluids, either in the presence or absence of solvents, people have been increasingly fuming to equations of state (EOS).
Unfortunately a great deal of misinformation about EOS usage has been created because of misunderstandings regarding their applications and @tations. As an example, it is sometimes be- Reved that an EOS can be used to generate reservoir fluid properties such as density and @sity without fu-st tuning the EOS to labora- tory data. This is not so. A reservoir fluid is a complex multicom- ponent mixture, the properties of which depend significantly upon the interaction of the various components. Because every reser- voir oil has its own unique composition, these interactions vary from one oil
to the next. Hence, experimental data is always re- quired for the proper tuning of an equation of state. This data should include low-pressure gas and liquid analyses as opposed to high-pressure liquid analysis. As wig be shown later, high- pressure liquid analyses are less accurate. In order
to
employ
an
EOS,
one
must
characterize
the
C 7 fraction of the reservoir fluid. In this context, characterization is defined as the detemiination of the critical temperature, critical pressure, acentric factor and the interaction parameters. There are two approaches to this
characterization: (1) the application of a continuous component model(i-3); and (2) the grouping of com- ponents into one or more pseudocomponents(4-13). This report describes a technique which, using only one pseudocomponent, fails into the latter group. Two
procedures
are
used
for
determining
C
I
pseudocom7 ponent parameters. In one
procedure
these
parameters
are
con- Keywords: C ' characterization, Phase behaviour, Equation of state, Shift- 7 ,ng faCtorS, Pseudocomponent,
Continuous
component
model,
Chromato- graph analysis, Recombination, Carbon dioxide. tenuously adjusted until the EOS can match measured equilibrium data, e.g. bubble point pressure, gas-off ratio, density and equilib- rium compositions. This is a very tedious task. A common alter- native procedure involves the determination of these parameters using correlations based on one or more measureable property, such as specific gravity, or molecular weight, or normal boiling point of the pseudocomponent. Although many correlations exist(4-13), there is little instruction available regarding the over-afl procedure from data
acquisition through to tuned EOS. Further- more, little exists in the literature regarding the need to ensure that all aspects of the tuning procedure are internally consistent. This report attempts to fill these voids, and at the same tune describes a [email protected] for C' fraction characterization through the apph- cation of Watson's heavy end correlations along with the Peng- Robinson equation of state(15), as modified by the Peneloux, Rauzy and Freze (PRF) shifting factors(13).
Only one pseudocomponent is used to represent the C' frac- 7 tion. From experience(18),
one
pseudocomponent
is
sufficient
to accurately predict fluid phase behaviour. Although this procedure can easily be extended to more than one pseudocomponent, in doing so, one may run into a correlation @tation. This occurs when a [email protected] weight of greater than 250 is required. Because most correlations do not extend beyond a molecular weight of 250, an extrapolation is required and it has been experienced that extra- polations can give erroneous results.
The over-atl procedure can be divided into two parts: acquisi- tion of a set of experimental data and the employment of this
data in the equation of state characterization. This procedure has been successfully applied to many of Esso Resources Canada Limited's reservoir flwds, not only for reservoir fluid phase behaviour predic- tion but also for miscible solvent design. Required Experimental Data A critical piece of data for aD EOS applications is the over-all reser- voir fluid composition. Often, high-pressure liquid chromatograph analyses are used to fulfill this requirement. This technique unfor- tunately has its drawbacks, and for the following two reasons can provide an inaccurate composition:
1.
During sample injection, a pressure differential must exist between the sample bomb and the injection valve. Due to this pres- sure drop some flashing may occur, resulting in a disproportion- ate amount of low molecular weight components in the sample.
2.
Chromatographic results depend upon equipment calibration.
This calibration drifts as the surrounding temperature changes.
High-pressure liquids exhibit more sensitivity than do low-pressure fluids to this phenomenon.
In order to avoid potential errors caused by the use of high- pressure
liquid chromatography, the following procedure is Proposed: 1.
Measure the reservoir fluid bubble point pressure at
reser- Paper reviewed and accepted for publication by the Editorial Board
of
the
Journal
of
Canadian
Petroleum
Technology. 112 The
Journal
of
Canadian
Petroleum
Technology RESERVOIR PERFORMANCE AND OPTIMIZATION Ct characterization for fluid properties predictions
R.S. WU and R.M. FISH
Esso Resources Canada Limited
ABSTRACT
A simple iterative procedure has been developed jor characterizing the C; fraction of a hydrocarbon fluid with only one pseudo- component. This procedure applies the Peneloux. Rauz;y and Preze
shifting factors and Watson's heavy end correlations to the PengRobinson equation oj state. The procedure requires the following experimental data: (1) bubble point pressure; (2) compositions of gas and liquid resulting from a room condition flash of the
reservoir fluid; (3) flashed gas-oil ratio. and (4) reservoir fluid density 01 pressures above the bubble pOint pressure. When the equation 0/ state is correctly tuned, the results match not only the
reservoir fluid phase behaviour and fluid density, but also the phase
behaviour oj mixtures of reservoir fluid blended with hydrocarbon
and carbon dioxide solvents. Introduction
In order to predict the fluid properties and phase behaviour of
hydrocarbon fluids, either in the presence or absence of solvents,
people have been increasingly turning to equations of state (EOS).
Unfortunately a great deal of misinformation about EOS usage
has been created because of misunderstandings regarding their
applications and limitations. As an example, it is sometimes believed that an EOS can be used to generate reservoir fluid properties
such as density and viscosity without fIrst tuning the EOS to laboratory data. This is not so. A reservoir fluid is a complex multicomponent mixture, the properties of which depend significantly upon
the interaction of the various components. Because every reservoir oil has its own unique composition, these interactions vary
from one oil to the next. Hence, experimental data is always required for the proper tuning of an equation of state. This data
should include low-pressure gas and liquid analyses as opposed
to high-pressure liquid analysis. As will be shown later, highpressure liquid analyses are less accurate.
In order to employ an EOS, one must characterize the C{
fraction of the reservoir fluid. In this context, characterization is
defined as the determination of the critical temperature, critical
pressure, acentric factor and the interaction parameters. There are
two approaches to this characterization: (I) the application of a
continuous component model(l.); and (2) the grouping of components into one or more pseudocomponents(4-13). This report
describes a technique which, using only one pseudocomponent,
falls into the latter group.
Two procedures are used for determining
pseudocomponent parameters. In one procedure these parameters are con- q Keywords: ~+ characterization, Phase behaviour. Equation of state, Shift- ing factors, Pseudocomponent, Continuous component model, Chromato- graph analysis, Reoombination, Carbon dioxide. tinuously adjusted until the EOS can match measured equilibrium
data, e.g. bubble point pressure, gas-oil ratio, density and equilibrium compositions. This is a very tedious task. A common altcr~
native procedure involves the determination of these parameters
using correlations based on one or more measureable property,
such as specific gravity, or molecular weight, or normal boiling
point of the pseudocomponent. Although many correlations
exist(4-l3), there is little instruction available regarding the over-aIl
procedure from data acquisition through to tuned EOS. Furthermore, little exists in the literature regarding the need to ensure that
all aspects of the tuning procedure are internally consistent. This
report attempts to fill these voids, and at the same time describes
a technique for
fraction characterization through the appli~
cation of Watson's heavy end correlations along with the PengRobinson equation of state(lsl, as modified by the Peneloux,
Rauzy and Freze (PRF) shifting factors ( 3).
Only one pseudocomponent is used to represent the C 7+ f raction. From experience{l8), one pseudocomponent is sufficient to
accurately predict fluid phase behaviour. Although this procedure
can easily be extended to more than one pseudocomponent, in
doing so, one may run into a correlation limitation. This occurs
when a molecular weight of greater than 250 is required. Because
most correlations do not extend beyond a molecular weight of 250,
an extrapolation is required and it has been experienced that extrapolations can give erroneous results.
The over-all procedure can be divided into two parts: acquisi~
tion of a set of experimental data and the employment of this data
in the equation of state characterization. This procedure has been
successfully applied to many of Esso Resources Canada Limited's
reservoir fluids, not only for reservoir fluid phase behaviour prediction but also for miscible solvent design. C; Required Experimental Data
A critical piece of data for all EOS applications is the over-all reservoir fluid composition. Often, high-pressure liquid chromatograph
analyses are used to fulfill this requirement. This technique unfortunately has its drawbacks, and for the following two reasons can
provide an inaccurate composition:
1. During sample injection, a pressure differential must exist
between the sample bomb and the injection valve. Due to this pressure drop some flashing may occur, resulting in a disproportionate amount of low molecular weight components in the sample.
2. Chromatographic results depend upon equipment calibration.
This calibration drifts as the surrounding temperature changes.
High-pressure liquids exhibit more seruitivity than do low-pressure
fluids to this phenomenon.
In order to avoid potential errors caused by the use of highpressure liquid chromatography, the following procedure is
proposed:
1. Measure the reservoir fluid bubble point pressure at reser- Paper reviewed and accepted for publication by the Editorial Board of the Journal of Canadian Petroleum Technology. 112 The Journal of Canadian Petroleum Technology .~ ,.
'.n ...... 1 u "• .n X u ~ 0 ~ MaasurDd _W1thPRF
--- Without PRF I '.. ". ."
'.n -----------------------------" PRESSURE - MPa " " " FIGURE 2. Comparison of density for Oil 1.
L-____________~N~o~~~~&
(/ of the liquid fraction are included in the calculations, this overall composition is affected by the specific gravity and molecular (/) Pis marchedfirsr.
(2) ":, is then marched. weight of the C:j fraction.
FIGURE 1. Flow diagram. Equation of State Characterization
voir temperature. Generally this measurement is very accurate
(within 20 kPa) because gas flashes ·to form a visible bubble in
the test apparatus very quickly.
2. Obtain several density measurements above the bubble point pressure.
3. Flash the reservoir fluid to atmospheric conditions.
4. Measure the gas-oil ratio (GOR).
5. Obtain low-pressure gas and liquid compositions separately.
(In this way stabilized gas and liquid samples analyses are obtained,
thereby minimizing the previously discussed analysis problems.)
6. Mathematically recombine the gas and oil to obtain an overall composition. An elaboration of this step is contained in the
hext section. Recombilllation Procedure
tively simple matter to convert the gas and liquid compositions
and GOR to an over-all composition. The basic equation is: + L '" .............................................................. (I) where x and y are known and z is the desired output. In order
to solve this equation, however, we need to convert the gas and
liquid volume ratio (Le. GOR) to mole fractions (Le. V and L).
This is accomplished by first assuming the liquid volume to be
1.0, and then solving:
ml=lxpl ..................................................................... (2) and
m,~GORxp, ............................................................... (3) Knowing the mass of the liquid and the gas, we can now solve
for moles/unit volume using:
Dr = m,/MI .................................................................... (4) n, ~ m,JM, .................................................................... (5) V and L can now be obtained by:
V ~ y(n, + n,) ............................................................... (6) and
L~I-V ....................................................................... (7) You will note that during this procedure, the GOR is explicitly
taken into aooount. Also, becanse the density and molecular weight
July-August 1989, Volume 28, No.4 r· These properties include: critical temperature, critical pressure,
acentric factor, aromaticity and interaction pariuneters. Although
three correlations are available in this program (Lin(l~, Watson(l3)
and Riazai(J7), it has been our experience that the Watson correlations provide, on average, the best results.
The phase behaviour program, with the over-all fluid compo- sition as input, uses the Peng-Robinson equation of state (see
appendix) and PRF shifting factors to calculate: bubble point pres- sure, gas and liquid.compositions and-the liquid densities above
the bubble point pressure. These results are then compared to the
experimentally derived data. If the bubble point pressure or den- sity matches are unacceptable, then the GOR and/or C; specif- Using the data obtained from the above procedure, it is a rela- Z; ~ V Yi Using the same C; specific gravity and molecular weight that was
used in the recombination procedure, a phase behaviour computer
program is employed to detennine the C; critical properties. ic gravity and molecular weight are adjusted (generally the bubble
point is most affected by changes to the GOR, while the density
is most affected by the other two parameters).
The rationale for adjnsting experhnental data is two-fold: frrst,
experimental results (particularly GOR) have intrinsic errors, and
second, the C; parameters, particularly molecular weight, are
derived from an empirical relationship with composition. In any
event, the adjnstments are not pennitted to 5OJo of the experimental
values for specific gravity or molecular weight or 15% of the experimental GOR. If data matches are unattainable inside these limits, then the data itself is reviewed for accuracy.
A critical aspect of this procedure is that the over-all composition be recalculated any time that a parameter is adjustoo. In this
way, an over-all consistency is maintained. As shown in the Figure
1 schematic, a summary of the over-all proce<Iure is as follows:
1. Obtain the gas and liquid compositions (Yi and .. and GOR
at room conditions.
2. Calculate V and L by using equations (2) to (6) baSed on .. ,
Yi, and GOR [incorporate changes in C; specific gravity (SO)
and molecular weight (M) in subsequent iterations].
3. Calculate", using equation (1).
4. Calculate bubble point and density with Peng-Robinson equation of state (use PRF shifting factors and
SO and M for
Watson correlations).
5. Check bubble point and density match with expetimental data.
6. If the match is acceptable, check the degree to which the GOR,
SO, and M have been adjusted. If greater than 15% for GOR,
or 5% for Sg or M, check data.
7. If match is unacoeptable, adjust GOR and/or C; SO and M;
repeat steps 2 through 7. .,
;. , f ,;'
,1.- , Cr 113 !» ..~.'
.. t· ...

View
Full Document

- Spring '10
- Dr.JackieChang