Professional Documents
Culture Documents
An MSc Thesis
by
Mihály Gajda
MASTER OF SCIENCE
in Petroleum Engineering
Miskolc, 07/05/2014
Table of Contents
List of Figures iv
Nomenclature vi
1. Introduction 1
2.3. Comparison 5
4.1.6. Grouping 17
5.3. Wells 29
6. Volatile Oil 30
6.3. Results 31
7. Gas Condensate 34
7.3. Results 35
8.3. Results 39
ii
8.4. Discussion and Conclusion 42
9.3. Results 44
References 47
Acknowledgement 51
Appendix A 52
Appendix B 53
Appendix C 58
Appendix D 59
Appendix E 60
iii
List of Figures
iv
Figure 9.1-Simulation Results for Sample VO - Case 1...................................................... 44
Figure 9.2-Simulation Results for Sample SO - Case 2 ...................................................... 45
v
Nomenclature
vi
1. Introduction
PVT calculations are used to describe the phase behaviour and to determine the
thermodynamic properties of hydrocarbon systems at the given pressure and temperature.
PVT properties of reservoir fluids are required by the most of petroleum engineering
calculations including : reservoir simulation, well testing, pipeline flow calculations and
separator design and the accurate prediction of phase behaviour is essential in case of
planning some tertiary recovery methods like gas injection or in situ combustion. As this is
an input data for the mentioned calculations its accuracy is crucial, wrong PVT properties
lead to erroneous calculation results, so the applied calculation method must be chosen
carefully.
Beside accuracy, simplicity and calculation speed are also important aspects in the
selection. Due to technological improvements in the latest decades the calculation speed is
not as important as it was, except reservoir simulation where it is still an issue. In the
solution of a problem, it is a common practice to select the simplest method that provides
the desired accuracy [1], so simplicity can also be a favourable property of a calculation
method.
The first objective of this thesis is to briefly describe the creation of PVT calculation
models, equation of state and black-oil tables used in reservoir simulation. The two
calculation models will be compared in respect of accuracy and calculation speed. The
main objective of this thesis is to give some advice about which PVT model should be used
in case of different fluid types or recovery methods.
The EOS fluid characterizations are developed with PETEX PVTp, but black-oil tables
are created by hand based on simulation of PVT experiment with PVTp. Eclipse100 and -
300 were used for the simulations. All of the figures are created by me with MS Visio,
Excell or Eclipse.
1
2. Basic Description of PVT Calculation Models
In the petroleum industry two approaches are used to calculate the PVT properties of
reservoir fluids. The first and the simplest is the black-oil formulation that determines the
necessary properties of the fluid based on separation data. The second is equation of state
(EOS) which provides both the compositional and the volumetric properties of the
reservoir fluid at given conditions based on its overall composition. Both methods have
their advantages and disadvantages that have to be taken into consideration in the
selection. The methods will be detailed in the next subchapters.
2
The main disadvantages of the application of EOS are:
Significantly increased calculation time compared to other methods.
Development of an EOS fluid characterization requires detailed PVT reports that
seldom available in case of old fields.
Creation and application of EOS fluid characterization need more experience.
(2.1)
(2.2)
(2.2)
3
well designed black-oil table, therefore they will not be discussed in the following
chapters. The proper creation of these tables is the topic of chapter 4.
The main advantages of using traditional black oil formulation:
Low calculation time and memory requirement.
Simple to use and create.
The main disadvantages of the method:
It cannot describe fluids with complex phase behaviour like gas condensates.
It cannot treat recovery methods where vapour-liquid equilibrium is crucial like in
gas injection.
Surface Oil
Reservoir Oil
Despite its drawbacks traditional black-oil formulation can be efficiently applied for
less volatile oil, dry gas and any oil reservoirs depleted above its bubble point.
(2.4)
The schematic of modified black oil formulation can be seen in Figure 2.2. The modified
gas FVF, called as "dry-gas" FVF is defined differently taking liquid dropout into
consideration [7].
These modifications greatly improve the accuracy of the method and make it capable to
treat volatile oil and gas condensate reservoirs. Several efforts were taken to take the
density change of surface gas and oil into consideration [8], but they are rarely available in
commercial software's, so they will not be detailed here. Therefore the surface oil and gas
4
densities are assumed to be constant during the depletion of the reservoir. In addition, the
density of the surface oil is the same whatever it originates from: reservoir gas or reservoir
oil and it is also true for the surface gas. Other modifications were also proposed to make
this method capable to treat gas injection [9, 10], but it is out of the coverage of this thesis.
Reservoir Gas
Surface Oil
Surface Oil
Reservoir Oil
2.3. Comparison
EOS is more sophisticated calculation model than black oil formulation. Although the
EOS requires more calculation time and memory, it can provide the necessary accuracy in
a wide range of circumstances. Black oil tables can be generated from simple separation
data, but the best way is that if it is generated with a well tuned EOS characterization.
Therefore the creation of an adequate black oil table also requires detailed PVT report, so
this does not make difference between them. The most crucial factors are the calculation
time and accuracy in the decision. To save computation time, the simpler model - black-oil
table - should be chosen if it can provide the desired accuracy. Otherwise EOS must be
used regardless of the increased calculation time and memory requirement.
5
3. Fluid Samples Used for Comparison
Four fluid samples were chosen for the comparison: highly volatile under saturated oil,
rich gas condensate, near critical gas condensate and saturated oil. These four samples
represent the four most commonly occurring type of fluid systems whose phase transition
is significant. This chapter will briefly introduce these samples. The first three samples are
real, but the saturated oil sample is synthetically made from the volatile oil sample.
The C1 content of this oil sample is 62.64 %, while its C7+ content is 18.44%, its
intermittent content is not significant. The reservoir temperature is 118 °C and the reservoir
pressure is 610 bars. The bubble point pressure is 379 bars at reservoir temperature. The
zero flash GOR of the oil is 416 sm3/m3 and its API gravity is 35.8. More detailed
information can be found in Appendix A (short name of the sample in the followings :
Sample VO).
The C1 content of this gas condensate sample is 66.9 %, it has a significant intermittent
content, 9.7 % CO2, and its C7+ content is 9.6%. The reservoir temperature is 191 °C and
the reservoir pressure is 505 bars. The dewpoint pressure is 338.3 bars at reservoir
temperature. The zero flash GOR of the sample is 1218sm3/m3 and the API gravity of the
stock-tank oil is 40.85. More information can be found in Appendix A (Sample GC).
The C1 content of the sample is 69.03 %, its C7+ content is 13.11 % and its intermittent
content is not significant. The reservoir temperature is 91 °C and the reservoir pressure is
534 bars. The Dewpoint pressure of the sample is 467 bars at reservoir temperature. The
zero flash GOR of the sample is 776 sm3/m3 and the API gravity of the stock tank oil is
33.8. More detailed information can be found in Appendix A (Sample NC).
There was not available sample from a saturated oil reservoir, so it is made synthetically
by flashing of the volatile oil sample at 300 bars. Therefore the saturation pressure of the
oil and gas phase is 300 bars. The received gas phase represents the gas cap and the oil
phase represents the oil body. The composition of the phases can be found in Appendix A
(Sample SO).
6
4. Development of PVT models
Both PVT models can provide the necessary accuracy, but only then if they are
sufficiently developed. The creation of an accurate PVT table needs a well tuned EOS
characterization thus it is better to start with its description. The development of an EOS
characterization requires more experience and time. Presently lots of guidelines and case
studies are available in the literatures that make our work easier.
Peng-Robinson EOS is chosen for the calculations, because this is the most widely used
EOS in the petroleum industry. The main steps of the development of an EOS
characterization are the following:
1. Checking the PVT measurement and sampling
2. Characterizing C7+ fraction
3. Tuning of EOS to measurements
4. Checking the result
5. Reducing components
These steps are essential to gain reliable EOS characterization. However many practicing
engineers omit some of them, but this can lead to inaccurate or inconsistent model.
7
Hoffmann-Crump-Hocott proposed a method for checking the consistency of the
samples [11]. Equation 4.1 is the basic equation of the method, it says that if log(Ki*p)
values are plotted vs. Fi characterization factor (defined by eq. 4.2), the points have to be
on a straight line in the case of light components.
logK i p A0 A1 Fi (4.1)
1 Tbi 1 T p
Fi log ci (4.2)
1 Tbi 1 Tci p sc
where:
A0, A1 - parameters of the line
Temperature is in °K or °R.
Figure 4.1 depicts the Hoffmann plot for Sample NC, it can be seen that the light
components are on the line except C6. This is an anomalous behaviour, but the adequacy
of the sampling is still acceptable regardless of this anomaly. In case of the other samples
the accuracy is within the error range at all components.
3 N2
C1
CO2
2 C2
Log(K∙p)
C3
1 C4
C5
C6
0
C7
C8
-1
C9
-2
-2 -1 0 1 2 3 4
F Characterization Factor
Figure 4.1-Hoffmann Plot for Sample NC
8
The mathematical recombination of the sample pairs requires their mole fraction that
can be determined from the production data. First the measured gas rate must be corrected
with the measured gas properties, because assumed properties were used in the
measurement of the gas rate. The appropriate formula for this purpose is:
field Z field
Q gc Q g (4.3)
lab Z lab
Thereafter the next step is the determination of gas-oil rate related to separator
conditions. It can be done in the following way from the production gas-oil ratio:
Rs Qcg
Rsp (4.4)
Bo Qo Bo
The mol fraction of separator gas can be calculated in the knowledge of the separator
gas-oil ratio, the density and the average molar mass of separator liquid:
1
2130 osp
Fgsp 1 (4.5)
M o Rsp
where units are ρosp in lbm/ft3, Mosp in g/mol and Rsp in scf/bbl.
Then everything is given for the mathematical calculation of the overall composition of
wellstream by the following equation:
zi Fgsp yi 1 Fgsp xi (4.6)
9
check. The short summary of the most important measurements can be found in Appendix
B.
Whitson and Torp published an appropriate method for the check of CVD
measurements [12]. It is basically a material balance written up for every component.
During the CVD experiment the composition of the removed gas is measured. With the
help of this method the composition of the liquid, which is dropped out in the PVT cell,
can be calculated and in the knowledge of that it is easy to determine the equilibrium
ratios. Thereafter the Hoffmann plot [11] can be used to check the received results. Figure
4.2 depicts the results for Sample NC.
3.5
N2
3
CO2C1
←C6+ C2
2.5 C5 C4 C3
2
log(K∙p)
1.5
425 bar
1
375 bar
325 bar
0.5
275 bar
0
225 bar
175 bar
-0.5 125 bar
75 bar
-1
-3 -2 -1 0 1 2 3 4
F Characterization Factor
Figure 4.2-Hoffmann plot for the CVD measurement of Sample NC
The equilibrium ratios must decrease as the F characterization factor decreases and
values of a pressure step must be on a straight line. In this case these criterions are not
satisfied at the four highest pressure steps and the C6 differs from normal behaviour at
every pressure step. It seems in Figure 4.2, that C6+ components have higher equilibrium
ratio than C5's and C4's, and they seem to be on a different straight line than lighter
components. This is physically unrealistic behaviour resulting from measurements error.
The removed gas is separated into "produced" gas and liquid, their composition is
measured and recombined mathematically to gain the composition of the removed gas. The
fact is that C6+ components are on a different straight line indicating that perhaps the
10
mathematical recombination was wrong, which may comes from the incorrect
measurement of liquid properties. Therefore these properties should not be used for the
tuning of the EOS fluid characterization of Sample NC, except liquid dropout and the
liquid density, because these are not affected by this error.
Other measurements have found to be correct.
11
Before the creation of C7+ sample, something must be cleared up. All of the
compounds with similar boiling points are grouped into single carbon number fractions
(SCN), therefore C8 means not only n-C8, but the mixture of compounds with boiling
point similar to n-C8 [15]. Grouping based on molar mass leads to inadequate
characterization. The molar mass of SCN fractions cannot be measured by GC, but Katz
and Firoozobadi published a table [16], which contains the average properties of the SCN
fractions and can be used for estimation.
When the composition is only determined up to C15 or so, it has to be extrapolated to
~C35. The best solution for the problem is to fit a distribution function on the C7+ fraction,
and then with the help of this function the molar mass and molar fraction of the heavier
components can be determined. The most recommended is the gamma distribution because
of its flexibility [17]. Three parameters of the distribution function used for the fitting are:
the smallest molar mass (η), the average molar mass of the C7+ fraction (MC7+) and α,
shape factor.
Figure 4.3 depicts the measured and the extrapolated composition of the C7+ fraction of
Sample NC. Measured composition is represented by red squares, blue squares represent
the values calculated by gamma distribution, and green triangles represent the composition
measured on another sample from the same reservoir. The values calculated by gamma
distribution show a decent match with the composition measured on another sample, which
proves the potency of gamma distribution.
12
After the composition is determined up to C35+, the SCN fractions have to be grouped
into pseudo components. Usually from two to seven pseudo components are required to
describe appropriately the phase behaviour of the fluid. Whitson published an empirical
rule that can be used to determine the number of required components and the molar mass
boundaries used to organise pseudo components [17]. Six pseudo components were formed
for all the three samples, because more complex phase behaviour requires more pseudo
components to describe.
Thereafter the densities of fractions have to be determined; hence a function is needed
to relate density to molar mass. Equation 4.7 is a flexible formula published by Søreide
that can be fitted to distillation densities [18].
i a C f M i bc (4.7)
where a, b, c and Cf are the parameters of the function. (standard values: a=0.2855, b=66,
C=0.13 and Cf is between 0.27 and 0.31)
For Sample NC the measured densities, the fitted correlation and its parameters can be
seen in Figure 4.4. The fitted function accurately approximates the measured values,
therefore it is sufficient to determine the densities of the pseudo components. For Sample
GC, there was no distillation data, so standard values are used, and stock-tank density is
matched by adjusting Cf parameter.
13
There are many correlations in the literature that can be used to determine the boiling
point temperature from density and molar mass. Søreides correlation [18] is the most
recommended [19, 20] therefore it was used for all the three samples. The formula is the
following:
Tbi 1928.3 1.695105 M i0.03522 i3.266
(4.8)
exp 4.922103 M i 4.7685 i 3.462103 M i i
where Tbi is in °R.
The other required properties can easily be calculated in the knowledge of molar mass,
density and boiling point temperature. Twu's correlation was used for critical properties
[21], Edmister's correlation for acentric factor [22], and Chue-Prausnitz's formula for BIP's
[23]. The pseudo-component properties of Sample NC (for example) can be found in
Appendix C. Now the EOS fluid characterization is ready for the calculations.
120
100
Liquid Dropout [%]
80
Measured
60 Calculated
40
20
0
0 100 200 300 400 500 600
Pressure [bar]
Figure 4.5-Simulation of CCE without Tuning (Sample NC)
14
The two main sources of prediction error are: assumptions of EOS and the uncertainty
in C7+ properties. Therefore, to receive accurate and reliable predictions the EOS fluid
characterization has to be matched to the measured data. The three main phase of the
tuning process:
Comparison of calculated and measured data
Matching measured data in several ways
Evaluation and choosing the best resulting model
Someone may think that it is meaningless to put so much energy into C7+
characterization, because the C7+ properties will be adjusted. It can be stated after many
attempts to match EOS to measurements, that if the C7+ characterization is inadequate, the
desired accuracy cannot be reached or only with a thermodynamically inconsistent model.
The matching is done by adjusting different properties of the fluid components. It can
be done with "trial and error" or regression. Trial and error is time consuming and tiring
[14], therefore it is seldom used nowadays. The tuning methods that use non-linear
regression are fast and automatic, but the result can be unrealistic, so they have to be used
carefully [24]. The regression based methods can be categorized into three groups [25]:
1. Modification of Ωa, Ωb numerical constants and BIP's.
2. Modification of critical properties and BIP's.
3. Modification of the inspection properties of C7+ fraction(molar mass, density,
boiling point or even molar distribution) and BIP's.
The method published by Coats and Smart can be classified into the first group [26]. It
changes the Ωa and Ωb numerical constant of methane and the heaviest fraction and BIP
between them. The method is really simple and efficient, it is recommended to choose if
someone does not have too much experience in fluid characterization. The phase behaviour
of a system is really sensitive for these parameters, so the search for optimal values is an
easier mathematical problem than in the other two cases.
There are lots of publications dealing with the second part including guidelines and case
studies [14, 27]. Therefore it is well documented and with a little bit more experience it
yields more accurate models than group one. These methods basically assume that the
basic properties of the C7+(molar mass, density and boiling point) are accurate, and the
error comes from the uncertainty of the critical property correlations. The method may
yields unrealistic results, so the checking after regression is crucial for this method.
The third group is the most novel, only some recently proposed method can be
classified into this one [19,20]. It can be interpreted as the method suppose that the
15
correlations calculating critical properties are accurate and the error comes from the
erroneous inspection properties of the C7+ fraction. It means that the regression re-
characterize the whole C7+ fraction in every iteration step. Although these methods require
the most experience, there is much lower chance for receiving unrealistic results. In spite
of these, methods are "superior" compared to the other two groups; they are still not
available in commercial software's.
The measurements were matched by adjusting critical pressure and temperature of C7+
components and BIP's between them and the methane. The results are evaluated in the next
subchapter. The modified critical properties of Sample NC can be found in Appendix C.
120
100 Measured
Liquid Dropout [%]
Original
80
After Matching
60
40
20
0
0 100 200 300 400 500 600
Pressure [bar]
Figure 4.6-CCE Liquid Dropout After Matching for Sample NC
To accept a fluid model, that is not enough to inspect only its accuracy, but also its
thermodynamic consistency should be checked. As a result of regression, the model is
physically unrealistic many times, even though it satisfies the problem mathematically.
Therefore the check of thermodynamic consistency is necessary.
16
Thermodynamic consistency desires the increase of critical temperature and the
decrease of critical pressure as the molar mass increases. When other parameters are also
changed like BIP's and numerical constants, the check is not so simple. Whitson published
a method, which is sufficient for this purpose[19]. A CCE experiment has to be simulated
with the matched EOS fluid characterization and the equilibrium ratios have to be plotted
in the function of pressure. The plot has to fulfil two criterions:
The equilibrium ratios have to monotonically decrease as the molar mass increases
The equilibrium ratio line must not cross each other (mostly caused by alternating
BIP's)
Figure 2.1 shows the mentioned plot for Sample NC, the fluid characterization satisfy
every expectation, therefore it is thermodynamically consistent. The other two tuned EOS
fluid characterization also fulfil these criterions.
4.1.6. Grouping
However, the technology has improved a lot in the previous decades, the computation
time and memory are still crucial issues in case of reservoir simulation. Flash calculations
have to be performed in every block during every time step, this means many million
times. The number of components is the main factor that affects the calculation time (and
can be changed). On the other hand, the number of components also has a significant effect
on accuracy; therefore the reduction of components has to be done carefully.
17
The accurate simulation of some recovery methods requires more components, while
others require less; there is no general rule for it. To reduce the number of components,
two or more of them are grouped into pseudo components. The grouping has to be based
on volatility [15]. Generally the N2 can be lumped with C1, and CO2 with C2, if the
intermittent content is not too high. The n-C4 can always be grouped with i-C4 and also n-
C5 with i-C5. Thereafter the grouping should be made by "trial and error" and step by step
followed by regression. That means one component grouped with another component or
group, and if the accuracy is deteriorated considerably, the component should be grouped
with other component or should not be grouped at all. After every step of grouping
matching with regression is suggested [14, 27].
The second task is to determine the properties of newly grouped pseudo components.
Mixing rules are used for this purpose. The most simple is Kay's mixing rule [28], which is
given by equation 4.9, it is basically a mole-fraction average. The molar mass of the
components should always be calculated by this rule.
z i i
I iI
(4.9)
z
iI
i
z M i i
I iI
(4.10)
z M
iI
i i i
k IJ zi z j kij (4.11)
iI jJ
The critical properties can also be calculated with Kay's mixing rule. Some authors
suggest weight-fraction average [29], whereas others suggest the mix of weight- and
molar-fraction average [30]. Lee-Kesler proposed different mixing rules for every
property, that needed by EOS or viscosity calculation [31] and it has many advantages over
the simple mixing rules. Based on personal experience[25] and the recommendations of
some authors[14] the method published Coats [32] gives the best solution for the problem.
The approach is simple and accurate; it tries to eliminate the error caused by pseudoization.
The method forces the pseudoized fluid characterization to reproduce the volumetric
18
behaviour of the original fluid characterization in undersaturated condition, this is
succeeded by the determination of ΩaI and ΩbI with equation 4.12 and 4.13. For the
calculation of other critical properties and accentric factor any method can be used.
2
z i z j ai a j 1 k ij z i
aI
iI jI iI (4.12)
R TcI p cI I
2 2
z i bi z i
bI iI iI (4.13)
R TcI pcI I
Coats' method and Kay's mixing rule were used to determine the properties of the
pseudo component. The number of components was reduced from 15-16 to 7 at each
samples, the further decrease of components seriously deteriorated the accuracy. The
accuracy of the pseudoized fluid characterization of Sample NC can be seen in Figure 4.8.
The properties of the pseudo-components can be found in Appendix D (for Sample NC).
Figure 4.8-CCE - Liquid Dropout - Comparison of the Original and Pseudoized Fluid
Characterization for Sample NC
19
each reservoir fluid type will be detailed. The check of tables are also recommended [33],
because a black-oil table may contain inconsistencies that cause non-physical behaviour in
reservoir simulation.
20
Most of the mentioned methods are not implemented in commercial PVT software's, for
this reason the method published by Whitson and Torp will be detailed and applied to
create the black-oil tables.
Saturation Pressure
Stage 1 Stage 2 Stage 3
It can happen that the pressure increases in the reservoir because of water injection for
example. In this case the saturated fluid becomes undersaturated due to pressure increase,
so variable saturation pressure model is needed. Therefore Rs becomes the primary
independent variable. For oils Rs vs pressure relationship determines that fluid is saturated
or not at specified Rs and pressure. Figure 4.10 depicts schematically the Bo vs pressure
relationship calculated for oil phase. At point A the oil is saturated, as the pressure
decreases to point B the oil remains saturated and its Rs will decrease. As the pressure
starts to increase due to water injection, the oil becomes undersaturated and its FVF will
decrease on B-C path to point B and while its Rs stays unchanged. If the pressure decreases
again, first the oil FVF will increase to Point B, the oil will be saturated again and while its
Rs remains unchanged. If the pressure decreases further, the oil FVF will decrease to point
D and its Rs will also decrease.
21
Rs1
A
Rs2
B
D C
Bo
Rs3
Figure 4.11 plots schematically the other two oil properties in the function of pressure.
Viscosity
RS
Rs3
Rs2
Rs1
Pressure Pressure
The gas properties depicted in Figure 4.12. The problem for gas phase is similar, but the
function is not single-valued between dewpoint pressure and vaporised oil-gas ratio. This
means that vaporised-gas oil ratio does not identify the condition of the gas phase
certainly. Therefore pressure is the primary independent variable and the properties plotted
in the function of vaporised gas oil ratio.
22
Viscosity
Bgd
p3
p4
p3 p2
p2
p1 p1
rs rs
PDEW
rs
Different depletion experiment(s) is/are required for every fluid type to create accurate
an black-oil table, therefore in the following subchapter every fluid type will be detailed
with respect to the suggested depletion experiment .
The black-oil table created for Sample GC (CVD) can be found in Appendix E.
23
on gas samples. Whitson et al. recommend the use of CCE for gas condensates [35], in this
thesis the two possible way of creation of black-oil tables will be compared.
Saturated Oil Reservoir
In Saturated oil reservoirs there are two phases initially at reservoir: gas cap and oil
body. The general recovery strategy for these reservoirs that first the oil body is depleted
with or without pressure maintenance, then the gas cap is depleted. Whitson et al. suggest
two different experiments for the two phases as it can be seen in Figure 4.13 [35]. The
properties of reservoir gas are determined with CVD experiment, while the properties of
the reservoir oil are determined with DLE. This recommended method was used to create
black-oil tables for Sample SO (Saturated Oil Reservoir Sample).
Reservoir Gas
Dewpoint
Gas
Oil
Water
Reservoir Oil
Bubblepoint
P1=Saturation Pressure P2 P3
24
are adequate, the calculated properties can be inaccurate. The source of the error can be
different and sometimes cannot be perfectly eliminated.
The error in density originates from the assumption that the specific gravity of stock-
tank oil and gas are constant during the depletion. Figure 4.14 illustrates the problem in
case of Sample VO. The error varies with the pressure, for reservoir gas the maximum
error is 4.5% at bubblepoint, for reservoir oil the maximum error is 2% at low pressure.
Whitson suggests the adjustment of surface densities to gain accurate reservoir
densities[35]. With the application of this method the maximum error in reservoir densities
is reduced below 1%. If the error is still unacceptable in a particular case, then other PVT
method should be chosen.
Figure 4.14-Density of Reservoir Oil and Gas Calculated with EOS and BO Table
The compressibility's (apparent) can be determined with the following formulas [33]:
1 dB Bgd rs Bo dRs
co o (4.16)
Bo dp 1 rs Rs dp
1 dBgd Bo Rs Bgd drs
cg (4.17)
Bgd dp 1 rs Rs dp
Figure 4.15 depicts the calculated compressibility values for Sample VO. The basic black-
oil properties were calculated at every 25 bar from 450 to 50 bars, plus at bubblepoint.
Negative oil compressibility can occur near the bubblepoint, this can be avoided by
including additional data close to saturation pressure. In this case the saturation pressure is
379 bars and the closest point at 375 bars, which is quite close; therefore there was no need
for additional pressure point. The discontinuities are caused by the change in derivates at
25
pressure nodes. As the difference between derivates increases the magnitude of
discontinuity also increases. The discontinuity influences the performance and stability of
simulation. The discontinuity can be reduced by including more pressure nodes in the
saturated region.
After the check is performed on the created black-oil tables they are ready to be used in
simulation. Some other consistency checks are also recommended, but those are trivial or
important only for tables generated to simulate gas injection [33].
The two most important PVT calculation models were reviewed in this chapter. Both
models rely on PVT measurements preformed on reservoir fluid samples, this implies the
importance of the quality check of sampling and measurements. The methods used for the
creation of PVT models have to be chosen carefully in order to receive accurate and
consistent model. The consistency of each PVT model has to be checked as well, otherwise
it can lead to unexpected errors in simulation, material balance or pipe flow calculations.
Different tasks in reservoir engineering desire different PVT calculation methods.
Simulation of gas injection requires sophisticated and accurate PVT model. History
matching can be tiring and time consuming, where fast calculation models are favourable.
Memory requirements can also be a limiting factor in the simulation of huge fields.
Therefore there is no superior calculation method. For every particular case the appropriate
PVT model has to be chosen based on comparison.
26
5. Description of the Numerical Reservoir Model
A static reservoir model is artificially created for the comparison. The reservoir can be
found in an anticline structure, the grid generated for the reservoir can be seen in Figure
5.1. The reservoir rock is siliciclastic with single porosity and permeability. The top of the
reservoir is at -2800m, while the water-oil/gas contact is at -2900m. The reservoir is 2000
m long and 1500 m wide. Five wells were created for the depletion of the reservoir. Further
details will be described in the following subchapters.
Block centred grid is used for the problem; it consists of 14260 cells with 50*50*5
dimension (the height of the cells is slightly varying). Grid refinement was applied around
the wells. The reservoir model consists of two different layers. Two sealing faults are also
defined in the model; their position can be seen in Figure 5.2.
The reservoir rock is consolidated sandstone, rock properties defined as the most
common values specific for this rock type (at the boundary of medium and good reservoir
rock.). The porosity is 21% in the top layer and 19% in the bottom layer. The net-gross
ratio varies from 0.75 to 0.85. Horizontal permeability changes between 120 and 80 mD,
while vertical permeability changes between 50 and 30 mD. Hall correlation was used to
27
take compaction into consideration [36]. The permeability curve will be described in the
next subchapter. The capillary pressure was neglected in the absence of measurements.
The cells found under -2900m are saturated with water. In some cases, a Carter-Tracy
aquifer model [37] is assigned to the reservoir with the following parameters : 3000
external radius, 40 md permeability, 0.2 porosity and 50 m thickness.
The Corey formula was used to describe the permeability curves [38]. The characteristic
points of the curves were taken from the literature [39]. The connate water saturation is
30%, the residual gas saturation is 20% and the residual oil saturation is 25%, critical gas
saturation is 5%. The end-point of oil and gas is 0.7, while the end-point of water is 0.25.
28
The Corey exponent of the oil phase is 3, it is 6 for gas and 4 for water. Figure 5.3 shows
the oil-water relative permeability curves.
5.3. Wells
As it mentioned earlier five wells have been assigned to the reservoir. The location of
the wells can be seen in Figure 5.2. The well completion consist of a 5" casing and a 2 7/8
tubing, the perforation intervals are different in case of different reservoir fluid types. The
modified Hagerdon-Brown method [40] was used to calculate the vertical-lift performance
curves, because it is considered to be accurate over a wide range of liquid rate and gas-oil
ratio [41].
The development strategy is different in all cases. The object of this thesis is to compare
the PVT models and not to find the development strategy providing the highest recovery,
therefore in many cases the aim is to reach the highest drop in the average reservoir
pressure.
29
6. Volatile Oil
This chapter deals with the comparison of different fluid models developed to simulate
the phase behaviour of a volatile oil sample. The volatile oil sample was introduced in
chapter 3.1. The fluid sample has a high degree of phase transition due to pressure drop,
which makes it a good choice for comparison. The following two subchapters describe the
different fluid models and development strategies used for the comparison. Thereafter the
results will be detailed and discussed.
The process of PVT model development has been already described in chapter 4. This
subchapter will briefly introduce the four PVT models developed for Sample VO with
respect to the most important differences.
First an EOS fluid characterization was made, which consists of 10 normal components
and 6 pseudo components. Thereafter this model was used to create the other three models.
The EOS model with 16 components resulted unacceptable computation time in simulation
runs, therefore it was not used in the comparison.
With the pseudoization of the previous fluid characterization an EOS model with 8
components was created (Comp). This provided almost the same accuracy, but required
much less calculation time.
Two black-oil tables were created by using the 16 component EOS. The first uses the
traditional gas FVF (BO), while the second uses the modified black-oil formulation,
therefore it contains dry-gas FVF instead of the traditional (MBO). The formulation of the
second table is suggested in the literature, but many programs like PVTp create PVT tables
using traditional gas FVF, which makes it necessary to include them in the comparison.
Vaporised oil-gas ratio is included in both tables.
Two cases were defined, one without aquifer (Depl) and another with aquifer (Aquifer).
The aquifer makes a great difference between the two cases, so two different development
strategies were worked out for them. The manifold pressure is 20 bars in both cases and the
applied separator train is also (first stage: 20 bars 20 °C, second stage : standard condition).
In the first case, when no aquifer was added , the target liquid rate was 100 m3/day/well
and the gas production limit was 1 106 m3/day. The production period was 10 years. Full
implicit solution scheme was used for both cases with maximum time step of 15 days.
30
In the second case Carter-Tracey aquifer model was added to the reservoir. The target
liquid rate was 150 m3/day/well; the limitation for water production was 450 m3/day and
0.97 WOR. The production period was 25 years. Adaptive implicit solution scheme was
applied in both cases with maximum time step of 15 days.
6.3. Results
The OOIP is 6.37 106 sm3 and the OGIP is 2.354 109 sm3. The results of two cases will
be exhibited separately.
Another useful diagram that can help to understand the difference can be seen in Figure
6.2, where oil and gas in place are plotted against production time. The line styles are the
31
same, but orange represents gas in place, green - oil in place, pink - liquid oil in place and
pale blue - free gas in place. The difference can be perceived only in free gas in place and
liquid oil in place.
Figure 6.2- Oil and Gas in Place During Depletion (Sample VO - Case 1)
32
Figure 6.3-Simulation Results of Sample VO - Case 2
33
7. Gas Condensate
After the volatile oil fluid models were compared the comparison of gas condensate
models is coming. It is a common belief among engineers that the gas condensate reservoir
cannot be treated with black-oil PVT models, which makes this comparison interesting.
The following subchapters will summarize the fluid models and development strategies
used for the comparison and finally the simulation results will be detailed and discussed.
Five different fluid models were developed for the problem, two EOS characterization
and three differently created black-oil tables.
Firstly a detailed EOS fluid characterization is developed from the measurements, that
consists of 15 components. This model was used to create the other four models. 15
components lead to high computation cost, accordingly this model was not included in the
comparison.
Subsequently a 7 component EOS model (Comp) is created by the lumping of
components. This model provided both decent accuracy and calculation time that made it
adequate for the simulation runs.
As it was mentioned, three different black-oil tables were created. The first was
generated by simulating a CCE (MBO-CCE), while the second was by simulating CVD
experiment (MBO-CVD). Both tables use dry-gas FVF instead of its traditional definition.
The third table was created by simulating CCE experiment (BO-CCE), but it contains
traditionally defined gas FVF (this table was automatically created by PVTp). All of the
tables include vaporised oil-gas ratio.
Two cases were created; the first model doesn't have aquifer model assigned (Depl.),
whereas the second include a Carter-Tracy aquifer model (Aquifer). The second model
requires the treatment of water production, that makes working out a different development
plan necessary for it. The initial reservoir pressure is 370 bars (the dewpoint pressure is
338.3 bars) and the reservoir temperature is 191°C in both cases. The manifold pressure is
the same, 20 bars. The separator train used is also the same, the condition of the first stage
is 20 bars and 20 °C, the second stage is atmospheric.
When no aquifer is assigned to the reservoir model, the target rate was 100 103sm3/day/.
The lack of water production makes further constraints unnecessary.
34
When aquifer was linked to the reservoir the target gas rate was the same as it was in
the first case. 200 m3/day was set as the water production limit of the field. If the water cut
of an individual well exceeded 0.1, then the worst perforation section was shut off.
Full implicit solution scheme was used for all black-oil simulation runs, whereas
adaptive implicit method was sufficient for compositional runs, since saturation changes
were not so high, the oil phase was not mobile, and big difference in densities makes water
conning less likely to happen [42].
7.3. Results
The resources are the following : OGIP is 1.867 109 sm3 and OOIP is 1.89 106 sm3
(vaporised). The next subchapters will introduce the results of simulation runs.
35
The end of depletion period is zoomed in Figure 7.2, which makes the detection of the
difference between the well corresponding fluid models easier. The black-oil table created
by simulating CVD experiment provides slightly more accurate results.
The changing in oil and gas in place is depicted in Figure 7.3, gas in place is marked
with orange, oil in place is represented by pale blue. A great difference can be noticed in
the values between black-oil table with traditional gas FVF(BO) and the others (MBO).
36
7.3.2. Case 2 (Aquifer)
Due to some initialization errors; Eclipse300 had to be also used for black-oil
simulations, therefore the difference in calculation time was not as big as it was in the first
case, 268 seconds were required for compositional and 197 seconds for black-oil
simulation runs. The black-oil table using traditional gas FVF has been left out, because it
gave disastrous results in the first case including a huge error in OGIP and OOIP. Figure
2.1 shows the result for case 2 with the same notation as before. The results are nearly the
same for all the three fluid models.
The assumption in the definition of traditional gas FVF makes it incapable to simulate
gas condensate reservoirs. On the other hand black-oil tables using dry-gas FVF performed
well both in accuracy and calculation time. However, the black-oil table created by
simulating CVD experiment provided slightly better accuracy, black-oil table created by
simulating CCE experiment also give satisfactory results. The difference occurs at low
pressure and originates from the re-vaporisation of oil, that is higher in case of CVD. This
could be expected, because CVD experiment approximate more better depletion processes
where oil phase is immobile. As a consequence modified black-oil tables created by
simulating CVD experiment can be used to substitute EOS fluid model in the simulation of
gas condensate reservoirs, where gas injection is not applied.
37
8. Near Critical Gas Condensate
Modelling the phase behaviour of a near critical gas condensate is a challenging task.
The current PVT calculation methods including EOS are not considered to be capable of
describing near critical fluids theoretically, although no one have ever published a better
method that can be applied in practice. Therefore the engineers are forced to treat them
with the currently available methods. Despite these theoretical problems, some recently
published papers show off promising results in the topic. Many recently found reservoirs in
great depths can be classified into this category, so the problem is getting more important.
The following subchapters will introduce the fluid models and development strategies used
in the simulation. After that the results will be detailed and discussed.
Four fluid models were developed for the problem, this time the black-oil table using
traditional gas FVF has been left out, because it is proved in the case of previous problems
that it is both theoretically and practically unable to treat the gas phase correctly. Therefore
the four models include two EOS models and two black-oil tables.
In the case of previous problems, a detailed EOS fluid characterization is created first
that having 15 components (EOS15). This fluid model was used to create other models.
Although the detailed EOS model was not included in the comparison in other cases, this
time it is considered to be necessary because of the complex phase behaviour.
The second step was the creation of the pseudoized EOS fluid characterization, the
number of components was reduced from 15 to 7 to save computation time and keep
accuracy accordingly (EOS7).
Two different black-oil tables were created by simulating CCE (MBO-CCE) and CVD
(MBO-CVD) experiments. Both include vaporised oil-gas ration and use dry gas FVF
instead of the traditional definition.
For this problem two different development strategies were created as well. The first
model represents a reservoir without aquifer model (Depl.), the second model has a Carter-
Tracy aquifer model (Aquifer). The initial reservoir pressure was 500 bars (where the
dewpoint pressure was 467 bars) and the initial reservoir temperature was 91 °C. The two
different development strategies were developed for the two cases which differ in the
treatment of water production. The minimal manifold pressure was 20 bars in both cases,
38
the applied separator train is also the same, the condition of the first stage was 20 °C and
20 bars, the second stage is atmospheric. The production time was 10 years.
In the first case where no aquifer model was assigned, the target gas rate was 500 103
sm3/day/well. The maximum allowed water production was 300 m3/day.
In the second case (Aquifer), he target gas rate was the same as it was in the first case.
The maximum water production was set to 200 m3/day and if it was exceeded the worst
interval of perforation was shut off.
Significant phase transitions that cause sharp saturation changes were expected.
Accordingly, full implicit solution scheme was applied for all cases and fluid models [42].
8.3. Results
The resources are the following : OGIP is 2.355 109 sm3 and the OOIP is 3.72 106 sm3.
The next subchapters will show the most important plots depicting the results.
39
Figure 8.1-Simulation Results for Sample NC - Case 1
40
component-EOS fluid model. The 15-component-EOS fluid model resulted in hours of
calculation time and around at the beginning of the 8th year convergence errors were
occurred, so only few days were simulated in an hour, therefore the simulation was stopped
after 8 hours of total.
Figure 8.3 compares the results of simulations using detailed EOS fluid model and
black-oil table, created by simulating CVD experiment. The two black-oil models provided
almost identical results, therefore the other one has been left out, the comparison of the
detailed and pseudoized EOS fluid model can be seen in Figure 8.4 (the notation is the
same that it was in the case 1). The black oil-models show decent accuracy, the difference
in pressure does not exceed 2.5 bars and even the water production is correct.
The accuracy of the pseudoized EOS fluid model is unsatisfactory. As in the first case
when the pressure decreased below the dewpoint pressure, the inaccurate phase behaviour
prediction yielded wrong pressure and production predictions.
41
Figure 8.4-Simulation Results of Sample NC - Case 2 (EOS15 vs EOS7)
The results have disproved the common belief that rich/near critical gas condensates
cannot be simulated with black-oil fluid model. Near critical condition increases the
number of iteration required for the flash calculation, which also increases the computation
cost of simulation. On the other hand, the complex phase behaviour has a slight effect on
the computation time of black-oil fluid model. As a result the difference between the
computation cost of the two different fluid models increases with the complexity of the
phase behaviour.
The black oil tables provided appropriate accuracy in simulations. The two differently
created black-oil tables performed almost identical results. The pseudoized EOS fluid
model performed poorly, which is surprising. There were also some problems with the
detailed EOS model, when the pressure started to increase and exceeded the dewpoint it
caused serious convergence error, which affected both calculation time and accuracy.
As a consequence properly created black-oil tables can provide the desired accuracy
even if the reservoir fluid is a near-critical gas condensate. Meanwhile its computation time
is much lower, which is an important factor when someone would like to choose the
adequate fluid model for history matching. Therefore the idea of the application of black-
oil tables for near-critical fluids should not be rejected, but few simulation runs are
recommended to compare the performance of different fluid models before making the
final decision.
42
9. Saturated Oil Reservoir
Saturated oil reservoirs are well known as oil reservoirs with gas cap. They are
commonly found in small and average depth, Due to this most of these reservoirs were
discovered decades ago, therefore the industry has a lot of experience about them. These
reservoirs are mostly depleted by water flooding that means slight or no pressure change
during the life of the reservoir. As a consequence it is a commonly approved fact that these
reservoirs can be easily and accurately treated with black-oil tables. But, as it was pointed
out in chapter 4.2 the density prediction of these tables is not always satisfactory, which
can be a problem, hence gravity plays a crucial role in the recovery of these reservoirs.
Therefore the way of the black-oil table creation is important.
Three different fluid models were created for this problem. The two EOS models are
identical, those which were created for the simulation of the volatile oil reservoir (Comp).
The third fluid model is a modified black-oil table (MBO).
The detailed EOS fluid characterization was used to create the black-oil table. As for the
gas cap, a CVD experiment was performed to determine the black-oil properties of the gas
phase, while a DLE experiment was conducted on the oil column to determine the black-
oil properties of the oil-phase.
Two different cases were assumed, the first with no aquifer model assigned (Depl.),
while in the second case a Carter-Tracy aquifer model was attached to the reservoir model
(Aquifer). The reservoir pressure was 325 bars in both cases and the reservoir temperature
was 118 °C. The manifold pressure was 20 bars in both cases. The applied separator train
was the following : first stage 20 bars and 20°C, second stage is atmospheric.
When no aquifer is assumed, the target liquid rate was 60 m3/day/well, the production
time was 5 years. The minimum bottomhole pressure was set to 100 bars.
In the second case, when aquifer is added to the reservoir model, the target liquid rate
was the same; meanwhile the maximum water production was set to 200 m3/day. The
production time was assumed to be 15 years.
43
9.3. Results
The resources are the following : OOIP is 4.124 106 sm3 and OGIP is 1.702 109 sm3.
The next subchapters will introduce the results of simulation runs.
44
model the pressure started to increase after the third year, the highest difference exceeded 2
bars. The production rates shows decent match. No significant difference can be detected
in the solution gas, vaporised oil, free gas and liquid oil in place that implies the error may
comes from another source and not from PVT properties.
The performance of the modified black-oil table was satisfactory, although the
discrepancy between black-oil and EOS model was the highest in the case of this fluid
type. Perhaps, the way how the black-oil table was created is not the best, because no PVT
experiment adequately approximates the process takes place during the depletion of a
saturated oil reservoir. The compositional simulation runs required higher computation
cost in both cases as expected. As a consequence, black-oil tables can be used for the
simulation of saturated oil reservoirs in order to save computation time and without losing
accuracy.
45
10. Summary and Conclusion
The objective of this thesis was to compare the performance of black-oil and
compositional fluid models in reservoir simulation with respect to accuracy and
computation time. First, the two different fluid description methods were briefly outlined,
then four different fluid samples were introduced and used for the comparison. Thereafter
the developments of PVT models were described in details.
The same static reservoir model was used in every case, only the fluid model and the
development strategy were changed. Relative permeability curves were originated from
literature average values. In sandstone reservoirs the capillary pressure hardly affect the
result, but greatly reduce the calculation speed, therefore it is neglected.
In the subsequent chapters, the two different fluid description methods were compared
on the four different fluid samples. For each comparison, two different cases were created,
one without aquifer to represent simple depletion and one with aquifer to represent
pressure maintenance. The simulation results were introduced and discussed at the end of
each chapter.
The black-oil tables using traditional gas FVF lead to erroneous simulation results, but
the modified black-oil tables provided promising results which proves the crucial role of
the proper creation of black-oil tables. The computation cost of black-oil simulation runs
was much lower than compositional ones' in every case. On the other hand, the accuracy of
black-oil model was nearly the same; it is proved to be adequate even in the case of near-
critical gas condensate. Therefore the automatic rejection of black-oil tables in case of gas
condensates and volatile oils is proved to be a wrong practice.
As a consequence, modified black-oil tables are capable for the simulation of reservoirs
with complex phase behaviour and exploited by simple recovery method (no gas injection
or temperature change). The proper creation of the fluid models is also proved to be
important. Some assumptions can be fatal, black-oil tables neglecting vaporised oil in the
calculation of gas FVF have yielded catastrophic simulation results, therefore their use
should be avoided. At the beginning of every reservoir study require simulation, several
sensitivity runs should be done to select the fluid model that is the most adequate for the
objective of the study. Among the fluid models, that model should be chosen which is the
most simple and requires the least calculation time and provides the necessary accuracy.
46
References
1. Mattax C. C., Dalton R. L.: "Reservoir Simulation", SPE Monograph Volume 13,
Richardson, Texas (1990).
2. van der Waals, J.D.: Continuity of the Gaseous and Liquid State of Mater (1873).
3. Peng, D.Y. and Robinson, D.B.: "A New-Constant Equation of State," Ind. & Eng.
Chem. (1976) 15, No. 1,59.
4. Soave, G.: "Equilibrium Constants from a Modified Redlich-Kwong Equation of
State," Chem. Eng. Sci. (1972) 27, No. 6, 1197.
5. Martin, J.J.: " Cubic Equations of State-Which?," Ind. & Eng. Chem. (1979) 18,
No. 2, 81.
6. Dake, L.P.: Fundamentals of Reservoir Engineering, Elsevier Scientific Publishing
Co., Amsterdam (1978).
7. Spivak, A. and Dixon, T.N.: "Simulation of Gas-Condensate Reservoirs," paper
4271 presented at the 1973 SPE Annual Meeting, Houston, 10-12 January.
8. Kniazeff, V.J. and Naville, S.A.: "Two-Phase Flow of Volatile Hydrocarbons,"
SPEJ (March 1965) 37; Trans., AIME, 234.
9. Cook, R.E., Jacoby, R.H., and Ramesh, A.B.: "A Beta-Type Reservoir Simulator
for Approximating Compositional Effects During Gas Injection," SPEJ (October
1974) 471.
10. Whitson, C.H., da Silva, F.V., and Søreide, I.: "Simplified Compositional
Formulation for Modified Black-Oil Simulators," paper SPE 18315 presented at the
1988 SPE Annual Technical Conference and Exhibition, Houston, 2-5 October.
11. Whitson, C.H. and Torp, S.B.:"Evaluating Constant-Volume-Deplation Data," JPT
(March 1983) 610; Trans., AIME, 275.
12. Hoffmann, A.E., Crump, J.S., and Hocott, C.R.: 'Equilibrium Constants for a Gas-
Condensate System," Trans., AIME (1953) 198, 1.
13. Whitson, C.H.: "Effect of C7+ Properties on Equation of State Predictions," paper
SPE 11200 presented at the 1982 SPE Annual Technical Conference and
Exhibition, New Orleans, 26-29 September.
14. Whitson, C.H. and Brulé, M.R.: "Phase Behavior", SPE Monograph Volume 20,
Richardson, Texas (2000).
15. Behrens, R.A. and Sandler, S.I.: "The Use of Semicontinuous Description To
Model the C7+ Fraction in Equation of State Calculations," paper SPE 14925
47
presented at the 1986 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa,
Oklahoma, 23 April.
16. Katz, D.L. and Firoozabadi, A.: "Predicting Phase Behavior of Condensate/Crude-
Oil Systems Using Methane Interaction Coefficients," JPT (November 1978) 1649;
Trans., AIME, 265.
17. Whitson, C.H.:"Characterizing Hydrocarbon Plus Fractions," SPEJ (August 1983)
683; Trans., AIME, 275.
18. Søreide, I.: "Improved Phase Behavior Predictions Petroleum Reservoir Fluids
From a Cubic Equation of State," Dr.Ing. dissertation, Norwegian Inst. of
Technology, Trondheim, Norway (1989).
19. Martinsen, S.Ř., Castiblanco, I., Osorio, R. and Whitson, C.H.:"Advenced Fluid
Characterization of Pauto Complex, Columbia," paper SPE 135085 presented at the
2010 SPE Annual Technical Conference and Exhibition, Florence, Italy, 19-22
September.
20. Singh, K., Mantatzis, K., Whitson, C.H., and Benjamia, R.: "Reservoir Fluid
Characterization and Apllication for Simulation Study," paper SPE 143612
presented at the 2011 SPE EUROPEC/EAGE Annual Technical Conference and
Exhibition, Vienna, Austria, 23-26 May.
21. Twu, C.H.: "An Internally Consistent Correlation of Predicting Critical Properties
and Molecular Weights of Petroleum and Coal-Tar Liquids," Fluid Phase Equilibria
(1984) No. 16, 137.
22. Edmister, W.C.: "Applied Hydrocarbon Thermodynamics, Part 4: Compressibility
Factor and Equation of State," Pet. Ref. (April 1958) 37, 173.
23. Chueh, P.L. and Prausnitz, J.M.:"Calculation of High-Pressure Vapor-Liquid
Equilibria," Ind. Eng. Chem. (1968) 60, No 13.
24. Pedersen, K.S,, Thomassen, P., and Fredenslund, A.:"On the Dangers of Tuning
Equation of State Parameters," paper SPE 14487 available from SPE, Richardson,
Texas (1985).
25. Gajda, M.:"Egy kritikus közeli kondenzátum karakterizációs lehetőségei Peng-
Robinson állapotegyenlettel" Scientific Student's Association, University of
Miskolc, Miskolc, Hungary (2013).
26. Coats, K.H. and Smart, G.T.:"Application of a Regression-Based EOS PVT
Program to Laboratory Data," SPERE (MAY 1986) 277.
48
27. Smith R.W., Bard, W.A., Lugo, C., Yemez, I., Guerini, A., Whitson, C.H. and
Fevang, Ø.:"Equation of State of a Complex Fluid Column and Prediction of
Contacts in Orocual Field, Venezuela," paper SPE 63088 presented at the 2000
SPE Annual Technical Conference and Exhibition, Houston, 1-4 October.
28. Kay, W.B.:"Density of Hydrocarbon Gases and Vapors at High Temperature and
Pressure," Ind. Eng. Chem. (1936) No. 28, 1014.
29. Pedersen, K.S., Thomassen, P., and Fredenslund, A.:"Characterization of Gas
Condensate Mixtures," C7+ Fraction Characterization, L.G. Chorn and G.A.
Mansoori (eds.), Advences in Thermodynamics, Taylor & Francis, New York City
(1989) 1.
30. Wu, R.S. and Batycky, J.P.:"Pseudocomponent Characterization for Hydrocarbon
Miscible Displacement," paper SPE 15404 presented at the 1986 SPE Annual
Technical Conference and Exhibition, New Oreans, 5-6 October.
31. Lee, B.I. and Kesler, M.G.: "A Generalized Thermodynamic Correlation Based on
Three-Parameter Corresponding States," AIChE J. (1975) 21, 510.
32. Coats, K.H.:"Simulation of Gas-Condensate-Reservoir Performance," JPT (October
1985) 1870.
33. Singh, K., Fevang, R. and Whitson, C.H.:"Consistent Black-Oil PVT Table
Modification," paper SPE 109596 presented at 2007 SPE Annual Technical
Conference and Exhibition, Annaheim, California 11-14 November.
34. Drohm, J.K. and Goldthorpe, W.H.:"Black-Oil PVT Revisited - Use of
Pseudocomponent Mass for an Exact Material Balance," paper SPE 17081
available from SPE, Richardson, Texas (1987).
35. Fevang Ø., Singh K., Whitson C.H.:"Guidelines for Choosing Compositional and
Black-Oil Models for Volatile Oil and Gas Condensate Reservoirs." paper SPE
143612 presented at 2000 SPE Annual Technical Conference and Exhibition,
Dallas, Texas 1-4 October.
36. Hall, H.N.: "Compressibility of Reservoir Rocks." Petroleum Transactions, AIME
(1953), pp 309-311.
37. Carter, R.D. and Tracy, G.W.: "An Improved Method for Calculating Water
Influx." Trans., AIME(1960), p. 415.
38. Corey, A.T., Rathjens, C.H., Henderson J.H., Wyllie M.R.J.:"Three-Phase Relative
Permeability." JPT (1956), November, pp 63-65.
49
39. Felsenthal M.:"Correlation kg/ko Data with Sandstone Core Caracteristics." Trans.
AIME (1959), Vol. 198, p 258.
40. Hagedorn, A.R., Brown, K.E.:"Experimental Study of Pressure Gradients Occuring
During Continuous Two-Phase Flow in Small Diameter Vertical Conduits." JPT,
April 1965, pp. 475-84.
41. Takács, G.:"Gas-Lift Manual" PennWell Books, Tulsa, Oklahoma (2005).
42. Turgay, E., Abou-Kassem, J.H., King, G.R.:"Basic Applied Reservoir Simulation"
SPE Textbook Series Vol. 7, Richardson, Texas (2001).
50
Acknowledgement
51
Appendix A
Composition [mol %]
Component Sample SO
Sample Sample Sample
VO GC NC Gas Oil
Cap Column
N2 0.83 0.89 0.57 1.20 0.72
CO2 0.66 9.68 1.44 0.72 0.64
C1 62.18 66.92 68.78 79.47 56.78
C2 6.43 7.37 6.71 6.24 6.49
C3 4.65 2.78 4.03 3.89 4.89
i-C4 1.05 0.82 0.94 0.79 1.12
n-C4 2.17 0.98 1.79 1.56 2.37
i-C5 0.99 0.52 0.84 0.64 1.10
n-C5 1.13 0.46 0.80 0.70 1.26
C6 1.47 1.20 0.99 0.80 1.68
C7 2.41 1.35 1.53
C8 2.59 1.19 1.86
3.99 22.95
C9 1.72 1.08 1.62
C10+ 11.71 4.77 8.10
Sum 100.00 100.00 100.00 100.00 100.00
Molar Mass [g/mol] 242.5 150.7 227.5
C10+ Density [g/cm3] 0.8744 0.7766 0.8668
52
Appendix B
Solution Liquid
Pressure Oil FVF
GOR Density
psia bbl/STB scf/STB g/cc
10015 2.469 0.582
9015 2.504 0.574
Pres=8846 2.51 0.573
8015 2.548 0.564
7015 2.602 0.552
6015 2.678 0.537
5733 2.695 0.533
5557 2.714 0.53
Pb=5500 2.719 2933 0.529
5300 2.553 2591 0.538
4500 2.209 1929 0.569
3700 1.953 1415 0.6
2900 1.762 1062 0.631
2100 1.605 774 0.664
1300 1.463 521 0.7
500 1.325 290 0.74
14.7 1.095 0 0.783
g/cc 0.857
Residual Oil Desnity on 60°F
°API 33.4
Specific Molar
Solution Liquid
Pressure Temperature Gravity Mass Oil FVF
GOR Density
(gas) (gas)
psia °F Scf/STB g/cc - g/mol bbl/STBB
Pb=5500 244 2.247
1500 110 1632 0.726 0.662 19.16 1.313
14.7 60 607 0.828 0.943 27.3 1
Molar Mass g/mol 171.88
Stock-Tank Oil g/cc 0.827
Density
°API 39.3
53
Table 4-CVD Experiment (Sample GC)
Well-
Pressure Gas (atm sep.) Condensate (atm sep.)
stream
Molar Molar Density Molar Density
bar Rel.Density MC10+ MC7+
Mass M. [kg/m3] M. [kg/m3]
Pr=338.3
321.1 33.79 23.327 0.9867 0.8052 154.3 200.4 163.8 794.2
301.1 32.78 23.099 0.9771 0.7973 151.1 198.6 160 791.1
281.1 31.5 23.158 0.9796 0.7993 150.5 196.1 159.1 790.3
251.3 28.88 22.961 0.9712 0.7925 143.5 188.7 152 782.8
220.7 28.64 23.175 0.9803 0.7999 140 182.9 148.2 778.8
190.5 27.62 23.144 0.979 0.7988 136.5 177.4 143.3 774.5
54
Table 7-Composition of the Produced Fluid (CVD)(Sample GC)
Pressure [bar]
Comp.
321.1 301.1 281.1 251.3 220.7 190.5
N2 1.82 1.21 1.18 1.15 1.12 1.14
CO2 10.54 10.27 10.55 10.12 10.99 10.88
C1 66.12 67.38 68.07 69.26 69.5 70.1
C2 7.58 7.75 7.76 7.85 7.79 7.92
C3 2.54 2.56 2.56 2.57 2.55 2.57
C4 1.98 1.86 1.79 1.77 1.76 1.73
C5 1.07 1.04 1.03 1.01 0.97 0.94
C6 1.27 1.23 1.24 1.22 1.21 1.2
C7 1.05 1.07 0.92 0.88 0.71 0.66
C8 0.94 1.02 0.83 0.79 0.67 0.61
C9 1.01 0.95 0.87 0.81 0.69 0.62
C10+ 4.08 3.66 3.2 2.57 2.04 1.63
Sum 100 100 100 100 100 100
55
Table 9-Properties of the Condensate (CVD) (Sample NC)
Molar
Pressure GCR Density
Mass
bar m3/m3 kg/m3 g/mol
Pres=534 586 820.7 157
Pdew=467 586 820.7 157
425 895 806.5 144.5
375 1260 792.5 133
325 1680 779.5 124
275 2270 767.5 116.5
225 3520 755 111
175 6130 745 107
125 11900 735 103.5
75 16600 726.5 101
56
Table 11-Composition of the Produced Fluid (Sample NC)
Pressure [bar]
Comp.
467 425 375 325 275 225 175 125 75
CO2 1.44 1.49 1.52 1.54 1.55 1.57 1.61 1.66 1.7
N2 0.57 0.72 0.78 0.84 0.82 0.8 0.78 0.75 0.69
C1 68.78 73.48 75.97 77.85 79.51 81.13 82.63 83.5 83.4
C2 6.71 6.63 6.59 6.59 6.61 6.68 6.76 6.86 6.98
C3 4.03 3.45 3.36 3.34 3.37 3.39 3.4 3.42 3.48
IC4 0.94 0.73 0.68 0.66 0.65 0.64 0.62 0.62 0.66
NC4 1.79 1.41 1.25 1.18 1.13 1.08 1.05 1.04 1.14
IC5 0.84 0.63 0.54 0.5 0.48 0.44 0.4 0.37 0.39
NC5 0.8 0.62 0.54 0.5 0.48 0.43 0.37 0.34 0.34
C6 0.99 0.89 0.8 0.73 0.66 0.55 0.42 0.33 0.32
C7 1.53 1.34 1.15 1.02 0.86 0.69 0.48 0.33 0.32
C8 1.86 1.64 1.4 1.19 0.99 0.75 0.47 0.29 0.26
C9 1.62 1.41 1.18 0.98 0.79 0.55 0.34 0.19 0.16
C10 1.24 1.08 0.9 0.71 0.53 0.36 0.2 0.11 0.07
C11 0.8 0.71 0.56 0.42 0.31 0.2 0.11 0.05 0.03
C12 0.65 0.54 0.43 0.31 0.22 0.14 0.07 0.03 0.02
C13+ 5.41 3.23 2.35 1.64 1.04 0.6 0.29 0.11 0.04
MC13+ 268 247 229 220 212 208 205 203 202
57
Appendix C
z M γ Tb Kw
Component
% g/mol - °R -
C6-7 2.52 86.4 0.7104 173.3 11.98
C8-10 4.57 112.2 0.7677 261.3 11.67
C11-14 2.95 161.1 0.8229 407.4 11.62
C15-21 2.41 233.4 0.8698 571.3 11.73
C22-29 1.04 336.7 0.9126 739.5 11.91
C30+ 0.61 516.1 0.9610 932.2 12.17
58
Appendix D
z Tc pc ω ΩA ΩB vc M
Comp.
% °C bar - - - m3/kg*mol g/mol
PS1 69.35 -83.0 46.30 0.0112 0.43555 0.07344 0.0991 16.14
PS2 8.15 31.9 53.28 0.1237 0.45453 0.07933 0.1387 32.56
PS3 6.76 116.6 40.50 0.1694 0.46233 0.07843 0.2270 49.75
PS4 1.64 191.9 33.52 0.2387 0.45340 0.07779 0.3050 72.20
PS5 7.09 284.7 31.04 0.3047 0.46371 0.07760 0.4800 103.03
PS6 6.40 461.0 20.00 0.5560 0.48553 0.08044 0.8824 216.84
PS7 0.61 672.1 11.40 1.0053 0.52224 0.07309 1.5785 516.14
59
Appendix E
50 0.00045 0.03334 0.0163 200 0.00050 0.00828 0.0227 275 0.00068 0.00638 0.0296
0.00000 0.03300 0.0162 0.00045 0.00828 0.0224 0.00061 0.00636 0.0286
75 0.00045 0.02193 0.0169 0.00045 0.00830 0.0223 0.00055 0.00634 0.0278
0.00040 0.02187 0.0170 0.00042 0.00828 0.0222 0.00050 0.00633 0.0272
0.00000 0.02181 0.0170 0.00040 0.00828 0.0221 0.00045 0.00631 0.0268
100 0.00045 0.01632 0.0177 0.00040 0.00827 0.0221 0.00045 0.00633 0.0265
0.00040 0.01629 0.0177 0.00039 0.00827 0.0220 0.00042 0.00631 0.0264
0.00039 0.01627 0.0177 0.00000 0.00827 0.0220 0.00040 0.00630 0.0262
0.00000 0.01626 0.0177 225 0.00055 0.00747 0.0246 0.00040 0.00630 0.0262
125 0.00045 0.01302 0.0186 0.00050 0.00746 0.0242 0.00039 0.00630 0.0261
0.00040 0.01300 0.0186 0.00045 0.00746 0.0238 0.00000 0.00630 0.0260
0.00040 0.01299 0.0186 0.00045 0.00748 0.0237 300 0.00078 0.00602 0.0329
0.00039 0.01299 0.0186 0.00042 0.00745 0.0236 0.00068 0.00598 0.0314
0.00000 0.01299 0.0186 0.00040 0.00746 0.0234 0.00061 0.00596 0.0303
150 0.00045 0.01088 0.0197 0.00040 0.00745 0.0234 0.00055 0.00594 0.0294
0.00042 0.01085 0.0197 0.00039 0.00745 0.0234 0.00050 0.00592 0.0287
0.00040 0.01086 0.0196 0.00000 0.00745 0.0233 0.00045 0.00590 0.0282
0.00040 0.01085 0.0197 250 0.00061 0.00685 0.0269 0.00045 0.00591 0.0280
0.00039 0.01085 0.0196 0.00055 0.00684 0.0262 0.00042 0.00589 0.0279
0.00000 0.01085 0.0196 0.00050 0.00683 0.0257 0.00040 0.00589 0.0276
175 0.00045 0.00936 0.0211 0.00045 0.00682 0.0253 0.00040 0.00588 0.0276
0.00045 0.00938 0.0210 0.00045 0.00684 0.0251 0.00039 0.00588 0.0275
0.00042 0.00936 0.0209 0.00042 0.00682 0.0250 0.00000 0.00588 0.0274
0.00040 0.00937 0.0208 0.00040 0.00682 0.0248 325 0.00092 0.00576 0.0371
0.00040 0.00936 0.0208 0.00040 0.00681 0.0248 0.00078 0.00570 0.0355
0.00039 0.00936 0.0208 0.00039 0.00681 0.0247
0.00000 0.00936 0.0207 0.00000 0.00681 0.0246
60
Table 18-MBO - Gas Properties (continued) (Sample GC)
Pres rs Bg ηg Pres rs Bg ηg
335 0.00099 0.00569 0.0392 350 0.00101 0.00556 0.0410
0.00092 0.00565 0.0379 0.00099 0.00554 0.0405
0.00078 0.00559 0.0355 0.00092 0.00550 0.0391
0.00068 0.00554 0.0338 0.00078 0.00543 0.0365
0.00061 0.00551 0.0325 0.00068 0.00539 0.0348
0.00055 0.00548 0.0316 0.00061 0.00535 0.0335
0.00050 0.00546 0.0308 0.00055 0.00532 0.0325
0.00045 0.00544 0.0303 0.00050 0.00529 0.0317
0.00045 0.00545 0.0300 0.00045 0.00527 0.0311
0.00042 0.00543 0.0298 0.00045 0.00528 0.0308
0.00040 0.00542 0.0295 0.00042 0.00526 0.0307
0.00040 0.00542 0.0295 0.00040 0.00525 0.0303
0.00039 0.00541 0.0294 0.00040 0.00525 0.0303
0.00000 0.00541 0.0293 0.00039 0.00525 0.0302
338.3 0.00101 0.00567 0.0400 0.00000 0.00524 0.0301
0.00099 0.00565 0.0395 375 0.00101 0.00534 0.0430
0.00092 0.00562 0.0381 0.00099 0.00533 0.0424
0.00078 0.00555 0.0357 0.00092 0.00528 0.0409
0.00068 0.00551 0.0340 0.00078 0.00521 0.0383
0.00061 0.00547 0.0328 0.00068 0.00515 0.0364
0.00055 0.00544 0.0318 0.00061 0.00511 0.0351
0.00050 0.00542 0.0310 0.00055 0.00508 0.0340
0.00045 0.00540 0.0304 0.00050 0.00505 0.0331
0.00045 0.00541 0.0302 0.00045 0.00503 0.0325
0.00042 0.00539 0.0300 0.00045 0.00504 0.0322
0.00040 0.00538 0.0297 0.00042 0.00501 0.0320
0.00040 0.00538 0.0297 0.00040 0.00501 0.0316
0.00039 0.00538 0.0296 0.00040 0.00500 0.0317
0.00000 0.00537 0.0294 0.00039 0.00500 0.0315
0.00000 0.00500 0.0314
61
Table 19-MBO - Oil Properties (Sample GC)
62