Predicting Reservoir Fluid Phase and
Volumetric Behavior From Samples
Contaminated With Oil-Based Mud
F. Gozalpour, SPE, A. Danesh, SPE, D.-H. Tehrani, SPE, A.C. Todd, SPE, and B. Tohidi, SPE, Heriot-Watt U.
Summary
The impact of sample contamination with oil-based mud filtrate on
phase behavior and properties of different types of reservoir fluids,
including gas condensate and volatile oil, has been investigated.
Two simple methods are used to determine the uncontaminated
fluid composition from contaminated samples. The capability of
the methods is demonstrated against highly contaminated samples.
An equation-of-state (EOS)-based method also has been devel-
oped to predict the phase and volumetric properties of the retrieved
composition. The method determines the required parameters of
the EOS for the uncontaminated fluid using the developed phase-
behavior models from contaminated-sample data. The method has
been examined against experimental data of different types of
reservoir fluids with successful results.
Introduction
Accurate reservoir fluid composition and properties are essential
for reservoir management and development. Reliable reservoir
fluid samples are therefore required; however, major challenges
can render the fluid analysis limited in value. The reservoir fluid
samples for pressure/volume/temperature (PVT) tests can be col-
lected by bottomhole and/or surface sampling techniques as ap-
propriate. During the drilling process, owing to overbalance pres-
sure in the mud column, mud filtrate invades the formation. If an
oil-based mud is used in the drilling, it can cause major difficulties
in collecting high-quality formation fluid samples. Because the
filtrate of oil-based drilling mud is miscible with the formation
fluid, it could significantly alter the composition and phase behav-
ior of the reservoir fluid. Even the presence of a small amount of
oil-based filtrate in the collected sample could significantly affect
the PVT properties of the formation fluid.
Oil-based mud is used widely in the petroleum industry. Con-
tamination with oil-based mud filtrate could affect reservoir fluid
properties such as saturation pressure, formation volume factor,
gas/liquid ratio, and stock-tank liquid density. Because collecting
a reservoir fluid sample is expensive, and accurate reservoir fluid
properties are needed in reservoir development, it is highly desir-
able to determine accurate composition and phase behavior for the
reservoir fluid from contaminated samples.
This study investigates the impact of sample contamination
with oil-based mud filtrates on composition and phase behavior
properties of different types of reservoir fluids, including volatile
oil and gas condensate samples. The samples were purposely con-
taminated with a known amount of oil-based mud filtrates in the
laboratory. The methods developed in this study were then applied
to determine the original composition of the reservoir fluid from
contaminated samples. The phase behavior of the contaminated
samples was also investigated by performing constant composition
expansion (CCE) tests at reservoir and surface conditions. The
measured experimental data were used to tune EOSs by adjusting
their parameters. The determined parameters of EOS tuned to the
contaminated samples were used to calculate the parameters of
EOS for the uncontaminated sample.
Copyright © 2002 Society of Petroleum Engineers
This paper (SPE 78130) was revised for publication from paper SPE 56747, first presented
at the 1999 SPE Annual Technical Conference and Exhibition, Houston, 3–6 October.
Original manuscript received for review 28 July 2000. Revised manuscript received 14
February 2002. Paper peer approved 21 February 2002.
June 2002 SPE Reservoir Evaluation & Engineering
EOS
EOSs are used extensively to simulate the volumetric behavior and
phase equilibrium of petroleum reservoir fluids. Among different
types of EOSs, cubic EOSs have enjoyed considerable success in
modeling because they are simple and give reliable results in phase
equilibrium calculations. Two EOSs, the Valderrama1 modifica-
tion of the Patel-Teja (VPT) EOS and a modified Peng-Robinson2
(mPR) EOS, were used in this study to perform phase equilibrium
calculations. All binary interaction parameters (BIP) in the mixing
rule were set to zero, and the temperature dependency of the
attractive term was used as the tuning parameter to fit the mea-
sured data.3
Extended compositional analyses (up to C20+) of fluids were
used in phase equilibrium calculations. The required critical prop-
erties of petroleum fractions to calculate parameters of EOS were
determined by perturbation expansion correlations.4 The required
boiling-point temperatures were calculated from the Riazi-
Daubert5 correlation using the molecular weight and specific grav-
ity of petroleum fractions. The Lee-Kesler6 correlation was used to
calculate the accentric factor of compounds.
Contaminated Reservoir Fluids
Hydrocarbon-based fluids (natural or synthetic oils) are generally
used in oil-based drilling muds. Because these fluids are soluble in
the reservoir fluid, they can render the fluid analysis limited in
value. Determination of the original fluid composition from the
analysis of a contaminated sample is feasible, but isolating the
properties of the reservoir fluid free from contamination is not
easily accomplished. Despite the recent improvements in sampling
reservoir fluids,7,8 obtaining a contamination-free formation fluid
is a major challenge, particularly in openhole wells. Therefore,
modeling techniques are required, along with the laboratory stud-
ies, to determine the composition and PVT properties of the un-
contaminated fluid.
We have demonstrated, as have other investigators,9,10 that an
exponential relationship exists between the concentration of com-
ponents in the C8+ portion of real reservoir fluids and the corre-
sponding molecular weights. For example, if the molar concentra-
tion of single carbon number groups is plotted against their mo-
lecular weights, it will give a straight line on a semilogarithmic
scale. Based on this feature of natural fluids, two methods have
been developed in this study to retrieve the original composition of
reservoir fluid from contaminated samples. The composition of the
C8+ portion of contaminated sample is plotted against molecular
weight on a semilogarithmic scale. The plotted data will show a
departure from the line over the range affected by the contaminants
(see Fig. 1). The concentrations of the contaminants are then
skimmed from the semilog straight line, presumed to be valid for
the uncontaminated reservoir fluid. The fitted line is used to de-
termine the composition of the uncontaminated fluid.
The above method, referred to as the Skimming method, gives
a reliable composition of the uncontaminated fluid if the contami-
nant comprises a limited hydrocarbon range. MacMillan et al.11
developed a similar method. They fitted a gamma distribution
function to the composition of the C7+ portion of contaminated oil
samples, excluding the composition of contaminants from the data-
fitting procedure.
In the second method, called the Subtraction method, a known
amount of drilling fluid with known composition is subtracted
197
 10 TABLE 1ÑCOMPOSITION AND PHYSICAL
PROPERTIES OF BASE OIL DMF-1

SCN Group Wt% Mol% Molecular Weight*

C11s 0.03 0.05 147
Composition, wt% C12s 0.12 0.17 161
C13s 1.37 1.75 175
1
C14s 11.38 13.38 190
C15s 25.46 27.60 206
C16s 20.01 20.13 222
Uncontaminated Region
Contaminated Region
C17s 17.87 16.84 237
C18s 10.41 9.26 251
C19s 6.44 5.47 263
C20s 3.26 2.65 275
0.1
100 200 300 400 500
Molecular Weight C21s 1.72 1.32 291
C22s 0.94 0.70 300
Fig. 1—Contaminated dead condensate LTB with base oil
DMF-1, C10–C30. C23s 0.54 0.39 312
C24s 0.25 0.17 324
from that of the contaminated sample. The C8+ portion of the
resultant composition is then used to fit an exponential distribution
C25s 0.10 0.07 337
function. The procedure is repeated for different levels of drilling C26s 0.04 0.03 349
fluid subtracted from the contaminated sample. The composition C27s 0.03 0.02 360
that gives the best fit to the exponential distribution function is
treated as the retrieved original reservoir fluid composition. This
C28s 0.02 0.01 372
approach can retrieve the original composition reliably even when C29s 0.01 0.01 382
the mud filtrate is composed of a wide range of natural hydrocar- C30+ < 0.01
bon components.
Measured molecular weight = 232 (± 1%); measured density (15.5¼C) = 812 ( ± 1) kg/m3.
In this study, two base oils, known as DMF-1 and DMF-3, have * Generalized single carbon number (SCN) physical properties.12
been used to investigate the effect of contamination on the com-
position and phase behavior of different types of reservoir fluids.
DMF-3 is a linear alpha olefin (LAO) drilling fluid composed of
69 mol% C14 (tetradecene) and 31 mol% C16 (hexadecene). The
the Skimming method in determining the level of contamination.
base oil DMF-1 consisted of a wide range of petroleum fractions,
The predicted level of contamination by the Skimming method
and its compositional analysis revealed a distribution of C11 to C30.
(18.5%) can be a good initial estimate for the iterative process of
Because DMF-1 is similar to a real reservoir hydrocarbon mixture,
the Subtraction method. With this initial guess, 18.5 wt% mud
determination of its contamination level is more demanding in
filtrate DMF-1 was subtracted from the contaminated composition
comparison to DMF-3. The original and contaminated samples
of LTB. The resultant composition was normalized, and the C10+
were analyzed by gas chromatography (GC) in this study. Tables
1 and 2 show the concentration of constituents of DMF-1 and portion was used to fit an exponential distribution function. The
DMF-3, respectively. R2, or the quality of fit, was 0.9619. By adjusting the amount of
subtracted mud filtrate slightly, the optimum level of contamina-
Retrieving Original Fluid Composition. The capability of the tion was found to be 16.4% by weight, with R2⳱0.9701 for the
developed methods in retrieving the uncontaminated composition best fit. The AAD% of the retrieved composition by the Sub-
was examined for a large number of samples. The results are traction method was 3.3; the composition is also presented in Ta-
summarized in Table 3. An example of applying the methods to a ble 4. It should be noted that the mud filtrate composition is re-
deliberately contaminated dead condensate is presented. The com- quired when the Subtraction method is used for the decontamina-
position of a dead condensate (LTB) was measured in the labora- tion process.
tory before experimentally contaminating it with 16.9 wt% oil-
based drilling fluid DMF-1. The gravimetric compositions of the Volatile-Oil LRA. Fluid LRA was a real, volatile reservoir oil
contaminated sample, along with GC analyses of the original and with bubblepoint pressure values of 4,965 and 4,550 psia at the
contaminated samples, are given in Table 4. The two developed reservoir (100°C) and separator (37.8°C) temperatures, respec-
methods, the Skimming and the Subtraction methods, were applied tively. The LRA was used to carry out CCE tests at reservoir and
to the measured composition of the contaminated dead condensate separator conditions.
to determine the composition of the uncontaminated fluid. The LRA was purposely contaminated with DMF-3 in the
The weight fraction of C10–C30 of the contaminated dead con- range of 5 to 20% on a volumetric basis at 6,000 psia and 100°C.
The composition of the original and contaminated samples is pre-
densate is plotted against the molecular weight on a semilogarith-
sented in Table 5. The contaminated samples were subjected to
mic scale in Fig. 1. It shows a departure from the linear trend over
CCE tests at the reservoir and separator temperatures in the labo-
the range affected by the contaminants. Excluding the contaminant
components from the contaminated composition, an exponential
distribution function (linear on a semilogarithmic scale) was then
fitted to the rest of the components. The calculated level of con- TABLE 2ÑCOMPOSITION AND PHYSICAL PROPERTIES
tamination for the contaminated dead condensate, using the Skim- OF BASE OIL DMF-3
ming method, is 18.5% (by weight). The retrieved (contamination-
free) composition of condensate LTB using the Skimming method
Component Wt% Mol% Molecular Weight
is also given in Table 4. The average absolute percentage deviation Tetradecene 66.10 69.02 196.4
(AAD%) of the retrieved composition is 7.3. Hexadecene 33.90 30.98 224.4
Because the contaminant DMF-1 comprises a wide range of Measured molecular weight = 218 (± 1%); measured density (15.5¼C) = 817 (± 1) kg/m3.
petroleum fractions, the Subtraction method is more suitable than

198 June 2002 SPE Reservoir Evaluation & Engineering

 TABLE 3—RESULTS OF APPLYING THE SKIMMING AND SUBTRACTION METHODS
TO VARIOUS CONTAMINATED SAMPLES

Level of Contamination
AAD% of Retrieved
Calculated Composition
Fluid Contaminant Actual Skimming Subtraction Skimming Subtraction
a
LTA DMF-3 5.1 4.8 — 1.3 —
LTA DMF-3 10.0 9.8 — 1.3 —
LTA DMF-3 1.1 0.0 — 2.5 —
LTA DMF-3 7.3 6.9 — 1.2 —
LTA DMF-3 2.5 1.9 — 1.3 —
b
LTB DMF-1 16.9 18.5 16.4 7.3 3.3
c
LRA DMF-3 5.1 4.7 4.8 0.7 0.4
LRA DMF-3 10.0 9.6 9.6 0.7 0.6
LRA DMF-3 20.0 19.7 19.7 0.7 0.6
d
GCB-1 DMF-3 5.0 4.7 4.9 0.3 0.0
GCB-1 DMF-3 15.0 14.9 15.1 0.3 0.2
GCB-1 DMF-3 30.0 30.0 30.1 0.4 0.2
d
GCB-2 DMF-3 20.0 20.3 20.1 0.5 0.0
GCB-2 DMF-3 40.0 40.2 40.1 0.5 0.1
GCB-2 DMF-3 49.0 49.1 49.0 0.6 0.1
GCB-2 DMF-3 55.0 55.1 55.1 0.5 0.1
d
GCB-2 DMF-1 15.0 14.2 14.4 4.4 0.5
GCB-2 DMF-1 30.0 29.3 29.6 4.5 0.6
GCB-2 DMF-1 46.0 45.5 45.4 5.5 1.4
GCB-2 DMF-1 50.0 49.6 49.6 4.9 0.9
GCB-2 DMF-1 56.0 55.6 55.8 4.7 0.7
GCB-2 DMF-1 70.0 69.7 69.9 5.3 0.8

(a) Level of contamination based on volume of sample.

(b) Level of contamination based on weight of sample.
(c) Level of contamination based on volume of single-phase sample at 6,000 psia and 100ºC.
(d) Level of contamination based on volume of liquid dropped out at stock tank (STLV).

ratory. The measured data are illustrated in Tables 6 and 7 for the
CCE tests at 100 and 37.8°C, respectively. The measured liquid
fraction values of the original, 5, 10, and 20% contaminated
samples (by volume) are plotted against pressure at 100°C in
Fig. 2. The experimental results show that contamination with the
oil-based mud filtrate has decreased the bubblepoint pressure,
whereas it has increased the liquid fraction at all subsequent pres-
sures. For example, a 20% contamination by volume could reduce
the bubblepoint pressure of the system by as much as 22% at
reservoir conditions. An mPR EOS with no binary interaction
parameters was individually tuned to the measured experimental
data of the original, 5, 10, and 20% contaminated samples, using
the temperature dependency of the attractive term of the compo-
nents in the liquid phase as the regression variable.3 The tuning
results are also depicted in Fig. 2. It can be seen that the tuned EOS
predictions are in agreement with the experimental data.
The developed methods (Skimming and Subtraction) were ap-
plied to the composition of contaminated samples to retrieve the
composition of the original volatile-oil LRA. Both methods re-
trieved the original composition with a deviation within the error
bands of fluid analysis. The retrieved composition is also given in
Table 5.
The EOS models tuned to the experimental data of 5, 10, and
20% contaminated samples were used to predict the phase behav-
ior of the original reservoir fluid at reservoir conditions, using the
retrieved composition from fluids with different contamination
levels. The predicted results are compared with those of the ex-
perimentally measured values in Fig. 3. The results of the EOS
tuned to the measured liquid fraction values of the original fluid
are also shown in Fig. 3. It shows that the predicted liquid fractions
are in agreement with the experimental values, particularly at pres-
sures lower than 3,500 psia. Approaching the bubblepoint pres-
sure, the predicted liquid fractions deviate slightly from measured
experimental ones. However, the predicted liquid fractions, using
the retrieved compositions, match the tuned results of the original
fluid. Therefore, the deviation in predicting the liquid fraction of
the original fluid at pressures close to the bubblepoint is caused by
the inability of the EOS to give a perfect match to the experimental
data of the original fluid.
The deviation in predicting the bubblepoint pressure of the
original LRA, applying the tuned mPR EOS to contaminated
samples, increases with the level of contamination. For example,
when using the tuned mPR EOS to experimental data of the 20%
contaminated sample, the bubblepoint pressure of the original fluid
was determined within 3.5% error. This results from the tuning of
the EOS because the retrieved compositions at different contami-
nation levels were very similar to those of the original composi-
tion. As an example, the mPR EOS had been tuned to the experi-
mental data of the 20% contaminated sample, which behaved like
a low-volatile mixture. Removing the contaminants, the decon-
taminated sample behaved like a high-volatile oil. The parameters
tuned to the experimental data of the low-volatile oil are not nec-
essarily valid for a high-volatile oil system. Hence, when using the
adjusted EOS to predict the phase behavior and volumetric prop-
erties of the original reservoir fluid, the predictions can be affected
by the extent of EOS tuning.

June 2002 SPE Reservoir Evaluation & Engineering 199

 An investigation of the developed phase behavior models for of the contaminated samples to the zero level of contamination, the
contaminated samples revealed that the correction factors of the predicted phase behavior of the original fluid is given in Fig. 7.
attractive term in EOS, determined by tuning the model to the The improvement is quite clear when compared to the use of the
contaminated data, changed linearly with the contamination level tuned VPT EOS to 20% contaminated samples in Fig. 6. It should
(Fig. 4). Hence, the correction factor of the attractive term for the be noted that although the proposed method adjusts the parameter
original fluid could be determined by extrapolating the values for of the fluid model at any contamination level, it becomes less
contaminated samples to the zero contamination level. It was also reliable when the amount of base oil (mud filtrate) in the collected
observed that when using the specific gravity of the plus fraction sample exceeds the reservoir oil.
(C7+) as a regression variable to match the measured phase behav- Gas condensate samples are more sensitive to contamination
ior of contaminated samples, the tuned value varied linearly with with oil-based mud filtrates in comparison to oil samples. A high
the level of contamination.13 The results of the model with the level of contamination can change gas condensate samples to
derived correction term using the above extrapolating technique oil systems.
and the retrieved composition are shown in Fig. 5. Note the im-
provement, in comparison with the results shown in Fig. 3. Gas Condensate GCB-1. Gas condensate GCB-1 was prepared in
The contamination of the volatile-oil LRA with the DMF-3 was this laboratory by blending separator liquid and vapor samples in
also investigated at separator conditions. The original, 5, and 20% the ratio of field gas/liquid ratio (GLR). The composition of origi-
(by volume) contaminated samples were used to perform CCE nal gas condensate is given in Table 8. To investigate the effect of
tests at 37.8°C in the laboratory. Contamination of the LRA with contamination on phase behavior and volumetric properties of res-
5% (volume) oil-based mud filtrate at this temperature reduced the ervoir fluids, three batches of gas condensate GCB-1 were experi-
bubblepoint pressure by 6.8%. The VPT1 EOS was individually
mentally contaminated with 5, 15, and 30% base oil DMF-3. The
tuned to the experimental data of the original and contaminated
gravimetric composition of the contaminated samples is also pre-
samples, using the temperature dependency of the attractive term
sented in Table 8. The level of contamination was based on the
of components in the liquid phase as the regression variable. The
volume of liquid dropped out in the stock tank at laboratory con-
tuned models were then applied to predict the phase behavior of
the original fluid at 37.8°C. The results are shown in Fig. 6. The ditions (STLV). The contaminated samples, as well as the original
tuning results of EOS, using the original composition, are also one, were used to carry out CCE tests at the reservoir (100°C) and
given in Fig. 6. The same deviation in predicting the bubblepoint separator (37.8°C) temperatures. The original (uncontaminated)
pressure of the original composition, which was observed at gas condensate GCB-1 behaved like a medium-rich gas condensate
100°C, can also be seen in Fig. 6. Using the model with its pa- with a dewpoint pressure of 5,525 psia and a maximum liquid
rameters, determined by extrapolating the correction to parameters fraction of 13.0% at the reservoir temperature (100°C).
Contamination with DMF-3 had a large effect on the dewpoint
pressure and the maximum liquid fraction of the system. A 30%
TABLE 4ÑRETRIEVED, ORIGINAL, AND CONTAMINATED STLV contamination reduced the dewpoint pressure by 9% at
COMPOSITION (WT%) OF DEAD CONDENSATE LTB reservoir conditions. However, it almost doubled the maximum
WITH DRILLING FLUID DMF-1 liquid fraction of the system at 100°C. The modified mPR EOS
was individually tuned to the experimental data of the original and
Contaminated Retrieved the contaminated samples, with reasonable results. The measured
and predicted phase behavior of the original and contaminated
Original Gravimetric GC Analysis Skimming Subtraction samples of the GCB-1 at 100°C are plotted in Fig. 8.
C8sÐ 22.99 19.10 19.11 23.48 22.87 The EOSs tuned to the experimental data of contaminated
samples were applied individually to the retrieved composition to
C9s 9.30 7.72 7.43 9.11 8.89
determine the phase behavior and volumetric properties of the
C10s 7.45 6.19 6.15 7.54 7.35 uncontaminated fluid. The results are depicted in Fig. 9 for the
C11s 5.85 4.86 4.93 6.05 5.90 mPR EOS at 100°C. Fig. 9 shows that the deviation in predicting
C12s 4.80 4.00 4.03 4.94 4.80 the dewpoint pressure and volumetric behavior of the uncontami-
nated fluid, using the tuned mPR EOS with contaminated samples,
C13s 5.83 5.18 5.48 4.99 6.29 increases with an increase in the contamination level. The devia-
C14s 4.94 6.31 6.70 4.54 5.78 tion is caused by the unreliability of the model, particularly that of
C15s 4.37 7.35 7.44 4.01 3.90 the 30% contaminated sample; the retrieved composition matches
the original one perfectly. Using the mPR EOS with its parameter,
C16s 3.29 6.25 5.98 3.67 3.23 determined by extrapolating the tuning parameters of the contami-
C17s 2.72 5.24 4.43 3.31 1.79 nated samples to the zero level of contamination, the predicted
C18s 3.35 4.83 4.76 2.92 3.65 phase behavior of the original fluid at 100°C is illustrated in Fig.
10. It shows that using the developed method, the predicted phase
C19s 3.35 3.78 4.13 2.67 3.67 behaviors of the retrieved compositions are in close agreement
C20s 2.11 2.28 2.28 2.27 2.09 with those of the measured values, and they match the tuned results
C21s 1.87 1.82 1.85 1.99 1.88 of the original composition. This illustrates the capability and re-
liability of the method developed in this study to determine the
C22s 1.77 1.62 1.65 1.78 1.79
phase behavior of the original fluid from contaminated samples.
C23s 1.60 1.41 1.43 1.67 1.60 The presented extrapolating technique will not be applicable when
C24s 1.34 1.15 1.18 1.53 1.36 contamination changes an originally uncontaminated gas conden-
C25s 1.24 1.05 1.07 1.31 1.26 sate sample to an oil system.*
The EOS that requires the least tuning should be used to model
C26s 1.15 0.97 0.98 1.20 1.16 the phase behavior of contaminated samples. The VPT EOS was
C27s 1.05 0.88 0.89 1.09 1.06 also tuned to the contaminated samples at 100°C. In this case, the
C28s 0.92 0.77 0.79 0.97 0.94 models needed very little tuning to match the measured experi-
mental data. Therefore, the VPT EOS tuned to the contaminated
C29s 0.84 0.70 0.71 0.87 0.85 samples could reliably predict the phase behavior of the original
C30s 0.77 0.64 0.68 0.83 0.81
C31+ 7.10 5.90 5.92 7.26 7.08
Total 100.00 100.00 100.00 100.00 100.00 * Gozalpour et al.: “Contamination of Gas Condensate Samples With Oil-Based Drilling
Fluid.”

200 June 2002 SPE Reservoir Evaluation & Engineering

 TABLE 5—MOLAR COMPOSITION OF ORIGINAL, CONTAMINATED, AND RETRIEVED
VOLATILE-OIL LRA, FROM SAMPLES CONTAMINATED WITH BASE OIL DMF-3

5% Contamination* 20% Contamination*

Component Original Gravimetric GC Analysis Gravimetric GC Analysis Retrieved

C1 55.90 54.74 54.04 50.96 51.35 55.85

C2 9.88 9.68 9.55 9.01 8.76 9.87
C3 5.67 5.56 5.49 5.17 4.98 5.66
iC4 1.18 1.16 1.14 1.08 1.07 1.18
nC4 2.03 1.99 1.94 1.85 1.78 2.03
iC5 1.10 1.08 1.08 1.00 1.01 1.10
nC5 1.92 1.88 1.87 1.75 1.65 1.92
C6s 1.54 1.51 1.49 1.40 1.36 1.54
C7s 2.16 2.12 2.30 1.97 1.83 2.16
C8s 2.69 2.64 2.70 2.45 2.47 2.69
C9s 1.99 1.95 1.96 1.81 1.75 1.99
C10s 1.70 1.67 1.66 1.55 1.53 1.70
C11s 1.30 1.27 1.24 1.19 1.18 1.30
C12s 0.95 0.93 1.07 0.87 0.93 0.95
C13s 1.22 1.20 1.00 1.11 1.05 1.22
C14s 0.86 2.24 2.16 6.87 7.56 0.91
C15s 0.87 0.85 0.99 0.79 0.81 0.87
C16s 0.68 1.29 1.31 3.36 3.48 0.70
C17s 0.50 0.49 0.71 0.46 0.42 0.50
C18s 0.57 0.56 0.66 0.52 0.51 0.57
C19s 0.60 0.59 0.59 0.55 0.50 0.60
C20+ 4.69 4.60 5.05 4.28 4.02 4.69

Properties of the Original Volatile-Oil LRA

Molecular weight = 65.2
Bubblepoint pressure at 100ºC = 4,965 psia
Bubblepoint pressure at 37.8ºC = 4,550 psia
*Level of contamination based on volume of sample at 6,000 psia and 100ºC.

TABLE 6—CCE TEST OF VOLATILE-OIL LRA, ORIGINAL, AND CONTAMINATED SAMPLES

WITH DMF-3 AT 100ºC (CONTAMINATION BASED ON VOLUME AT 6,000 PSIA AND 100ºC)

Original 5% Contamination 10% Contamination 20% Contamination

Liquid Liquid Liquid Liquid
Pressure Fraction Pressure Fraction Pressure Fraction Pressure Fraction
(psia) (Vliq/Vtotal) (psia) (Vliq/Vtotal) (psia) (Vliq/Vtotal) (psia) (Vliq/Vtotal)
Pb=4,965 100.00 Pb=4,630 100.00 Pb =4,355 100.00 Pb =3,850 100.00

4,500 85.81 4,000 83.56 3,600 81.96 3,000 80.32

4,300 81.45 2,000 46.28 3,400 78.08 2,000 57.85
4,100 78.24 3,200 74.33 1,200 36.07

3,800 73.83 3,000 70.55

3,500 69.13 2,500 60.63
3,000 61.37 2,000 49.89
2,500 52.71 1,200 30.37
2,000 42.75
1,200 25.40

June 2002 SPE Reservoir Evaluation & Engineering 201

 100
TABLE 7ÑCCE TEST OF VOLATILE-OIL LRA, ORIGINAL,
AND CONTAMINATED SAMPLES WITH DMF-3 AT 37.8¼C

Liquid Fraction ( liq/ total), percentage

(CONTAMINATION BASED ON VOLUME
AT 6,000 PSIA AND 100¼C)
80
Original 5% Contamination 20% Contamination
Liquid Liquid Liquid
Pressure Fraction Pressure Fraction Pressure Fraction

V V
(psia) (Vliq/Vtotal) (psia) (Vliq/Vtotal) (psia) (Vliq/Vtotal)
60

Pb =4,550 100.00 Pb =4,241 100.00 Pb =3,408 100.00

Experimental, Original
Experimental, 5% Contaminated

4,250 93.94 4,000 96.24 3,000 94.06 40 Experimental, 10% Contaminated

Experimental, 20% Contaminated

4,000 90.36 3,800 92.94 2,500 84.29 EOS, Tuned

3,000 77.70 3,500 89.03 2,000 72.33 20

2,000 60.91 3,000 80.87 1,500 58.61 1,000 2,000 3,000

Pressure, psia
4,000 5,000

2,000 61.81 1,000 40.45

1,500 49.23 Fig. 2—Measured and predicted phase behavior of volatile-oil
LRA, contaminated with 5, 10, and 20% (volumetric) DMF-3,
1,250 40.08 mPR EOS, 100°C.

100 0.000
100
Liquid Fraction ( liq/ total), percentage

Correction to Alpha Parameter (F)

90
80 –0.005

80
4,000 4,250 4,500 4,750 5,000
V V

60 –0.010

Experimental, Original
Tuned EOS, Original
40 Retrieved, 5%
–0.015
Retrieved, 10%
Retrieved, 20%

20
–0.020
1,000 2,000 3,000 4,000 5,000
0 5 10 15 20 25
Pressure, psia
Level of Contamination, vol%

Fig. 3—Predicted phase behavior of the original volatile-oil

Fig. 4—Correction to the alpha parameter of supercritical com-
LRA, using the retrieved composition and EOS tuned to con-
ponents in the liquid phase, volatile-oil LRA at 100°C, mPR EOS,
taminated samples, mPR EOS, 100°C.
[F=alpha(tuned)/alpha(std)–1].

100 100

Experimental, Original
Liquid Fraction ( liq/ total), percentage
Liquid Fraction ( liq/ total), percentage

Experimental, Original
Tuned, Original
90 Tuned EOS, Original
Retrieved, Adjusted EOS
80 Retrieved, 5%
Retrieved, 20%
80
V V
V V

60

70

40
60

20 50
1,000 2,000 3,000 4,000 5,000 1,000 2,000 3,000 4,000 5,000
Pressure, psia Pressure, psia

Fig. 5—Predicted liquid fraction of the uncontaminated volatile- Fig. 6—Predicted phase behavior of the original volatile-oil
oil LRA, using EOS with adjusted tuned parameter and retrieved LRA, using EOS tuned to contaminated samples and retrieved
composition, mPR EOS, 100°C. composition, VPT EOS, 37.8°C.

202 June 2002 SPE Reservoir Evaluation & Engineering

 100
TABLE 8ÑMOLAR COMPOSITION OF ORIGINAL,
Experimental, Original

Liquid Fraction ( liq/ total), percentage

CONTAMINATED, AND RETRIEVED GAS CONDENSATE
Tuned EOS, Original GCB-1, FROM SAMPLES CONTAMINATED
90
Retrieved, Adjusted EOS WITH BASE OIL DMF-1

Contaminated* Samples
V V 80 (Gravimetric)
Component Original 5% 15% 30% Retrieved
70
C1 71.99 71.81 71.43 70.66 71.98
C2 11.81 11.77 11.71 11.58 11.81
60 C3 4.76 4.75 4.72 4.67 4.76
iC4 0.81 0.81 0.80 0.79 0.81
50 nC4 1.72 1.72 1.71 1.69 1.72
1,000 2,000 3,000 4,000 5,000
iC5 0.62 0.62 0.62 0.61 0.62
Pressure, psia
nC5 0.78 0.78 0.77 0.77 0.78
Fig. 7—Predicted liquid fraction of the uncontaminated volatile- C6s 0.90 0.90 0.89 0.88 0.90
oil LRA, using EOS with adjusted tuned parameter and retrieved C7s 1.36 1.36 1.35 1.33 1.36
composition, VPT EOS, 37.8°C.
C8s 1.32 1.32 1.31 1.30 1.32
C9s 0.73 0.73 0.72 0.72 0.73
C10s 0.54 0.54 0.54 0.53 0.54
C11s 0.40 0.40 0.40 0.39 0.40
30 C12s 0.31 0.31 0.31 0.30 0.31
C13s 0.27 0.27 0.27 0.26 0.27
0% Contam., Exp.
Liquid Fraction ( liq/ total), percentage

0% Contam., EOS
5% Contam., Exp.
5% Contam., EOS C14s 0.27 0.43 0.80 1.55 0.27
15% Contam., Exp. C15s 0.23 0.23 0.23 0.23 0.23
20
C16s 0.18 0.25 0.42 0.75 0.19
15% Contam., EOS
30% Contam., Exp.

C17s 0.14 0.14 0.14 0.14 0.14

V V

30% Contam., EOS

C18s 0.14 0.14 0.14 0.14 0.14

10 C19s 0.12 0.12 0.12 0.12 0.12
C20+ 0.60 0.60 0.60 0.59 0.60
*Level of contamination based on the volume of liquid dropped out in the
stock tank (STLV).
0
0 1,000 2,000 3,000 4,000 5,000 6,000
Pressure, psia
fluid at reservoir conditions without having to adjust its param-
eters, as shown in Fig. 11.
The phase behavior and volumetric properties of the original
Fig. 8—Effect of contamination with DMF-3 on phase behavior and contaminated samples of GCB-1 with the DMF-3 also were
of gas condensate GCB-1, contamination based on volume of
liquid in stock tank, mPR EOS, 100°C.
investigated at separator conditions. Fig. 12 shows the results of
using the mPR EOS tuned to contaminated samples to predict the
phase behavior of the original composition at 37.8°C. As expected,
the mPR EOS tuned to the 30% contaminated sample is not reli-

15
15
Liquid Fraction ( liq/ total), percentage
Liquid Fraction ( liq/ total), percentage

10
10
V V
V V

5
5 Original, Experimental
Original, Experimental
Retrieved, 5%
Retrieved, 15% Original, Tuned EOS

Retrieved, 30% Retrieved, Adjusted EOS

0 0
0 1,000 2,000 3,000 4,000 5,000 6,000 0 1,000 2,000 3,000 4,000 5,000 6,000
Pressure, psia Pressure, psia

Fig. 9—Predicted liquid fraction of the uncontaminated gas con- Fig. 10—Predicted liquid fraction of the uncontaminated gas
densate GCB-1, using EOS tuned to contaminated samples and condensate GCB-1, using EOS with adjusted tuned parameter
retrieved composition, mPR EOS, 100°C. and retrieved composition, mPR EOS, 100°C.

June 2002 SPE Reservoir Evaluation & Engineering 203

 15
25

Liquid Fraction ( liq/ total), percentage

Liquid Fraction ( liq/ total), percentage
20

10

15

V V
V V

10
5
Original, Experimental
Original, Experimental
Retrieved, 5% 5 Retrieved, 5%
Retrieved, 15% Retrieved, 30%

0 0
0 1,000 2,000 3,000 4,000 5,000 6,000 1,000 2,000 3,000 4,000 5,000 6,000
Pressure, psia Pressure, psia

Fig. 11—Predicted liquid fraction of the uncontaminated gas Fig. 12—Predicted liquid fraction of the uncontaminated gas
condensate GCB-1, using EOS tuned to contaminated samples condensate GCB-1, using EOS tuned to contaminated samples
and retrieved composition, VPT EOS, 100°C. and retrieved composition, mPR EOS, 37.8°C.

able in predicting the phase behavior of the original fluid. Using 3. The EOS tuned to the experimental data of contaminated
the same approach, the predicted liquid fractions and the dewpoint samples, along with the retrieved composition, can be used to
pressure of the retrieved composition are given in Fig. 13. Note a reasonably predict the phase behavior of the uncontaminated oil
perfect match to the predicted results by the model tuned to ex- samples. However, the predicted results for gas condensate
perimental data of the uncontaminated sample. samples could be erroneous.
4. The temperature-dependency parameter of an EOS, used as the
regression variable, tuned to match the data of contaminated
Conclusions samples varied linearly with the level of contamination. There-
The impact of contamination with oil-based mud filtrates on com- fore, the required EOS parameters for the uncontaminated fluid
position and phase behavior of different types of reservoir fluids, can be determined by extrapolating the line to the zero level
including volatile oil and gas condensate samples, has been inves- of contamination.
tigated. Two methods have been developed to retrieve the original 5. Applying the determined parameters to the EOS, the developed
fluid composition from contaminated samples. EOSs have been model can accurately predict the phase behavior and volumetric
tuned to the experimental data of contaminated samples. The pa- properties of the uncontaminated fluid.
rameters of the tuned EOS have been adjusted to develop a phase 6. The method developed to determine the parameters of EOS for
behavior model for the uncontaminated sample. The following the uncontaminated fluid is general, and it can be applied to any
conclusions can be drawn based on the results of this work: EOS. This was successfully applied to the mPR and VPT EOSs
1. Using the linearity of concentration of single-carbon groups in this study.
(molecular weight, semi-logarithmic plot), two methods (Skim-
ming and Subtraction) have been developed to determine the Acknowledgments
original composition from contaminated samples. The experiments were performed by K. Malcolm and K. Bell, with
2. The Skimming and Subtraction methods can reliably predict the support from A. Reid. We gratefully acknowledge the sponsors of
level of contamination and the composition of uncontaminated this study: the U.K. Dept. of Trade and Industry, BP Exploration
fluid from contaminated samples. However, the Subtraction Co. Ltd., Edinburgh Petroleum Services (EPS) Ltd., Elf (UK) Ltd.,
method is preferred if the contaminant comprises of a wide Marathon International (GB) Ltd., Mobil (North Sea) Ltd., Oil
range of petroleum fractions. Phase Sampling Ltd., Phillips Petroleum Co. (UK) Ltd., Shell UK
Exploration and Production Ltd., and Statoil AS. Some of the fluid
samples used in this study were donated by Statoil AS and BP
25 Exploration Co. Ltd.

References
Liquid Fraction ( liq/ total), percentage

20
1. Valderrama, J.O.: “A Generalised Patel-Teja Equation of State for
Polar and Non-polar Fluids and Their Mixtures,” J. Chem. Eng. Japan
(1990) 23, No.1, 87.
15
2. Gozalpour, F. et al.: “Rapid and Robust Phase Equilibrium Calculation
V V

to Model Fluids in Reservoir and Surface Processing,” Chem. Eng. Res.

10
& Des., Trans. IChemE (July 1998) 76(a), 594.
3. Gozalpour, F. et al.: “Integrated Phase Behavior Modeling of Fluids in
Original, Experimental
Reservoir-Surface Processes Using Equation of State,” paper SPE/DOE
5 39630 presented at the 1998 SPE/DOE Improved Oil Recovery Sym-
Original, Tuned EOS
posium, Tulsa, 19–22 April.
Retrieved, Adjusted EOS
4. Twu, C.H.: “An Internally Consistent Correlation for Predicting the
0 Critical Properties and Molecular Weights of Petroleum and Coal-Tar
1,000 2,000 3,000 4,000 5,000 6,000 Liquids,” Fluid Phase Equilibria (1984) 16, No. 2, 137.
Pressure, psia 5. Riazi, M.R. and Daubert, T.E.: “Characterization Parameters for Pe-
troleum Fractions,” Ind. Eng. Chem. Res. (1987) 26, No. 4, 755.
Fig. 13—Predicted liquid fraction of the uncontaminated gas 6. Lee, B.I. and Kesler, M.G.: “A Generalized Thermodynamic Correla-
condensate GCB-1, using EOS with adjusted tuned parameter tion Based on Three-parameter Corresponding States,” AIChE J.
and retrieved composition, mPR EOS, 37.8°C. (1975) 21, No. 3, 510.

204 June 2002 SPE Reservoir Evaluation & Engineering

 7. Morris, C.W. et al.: “Using Optical Fluid Analysis To Evaluate Down-
hole Fluid Sample Contamination,” paper SPE 50603 presented at the
1998 SPE European Petroleum Conference, The Hague, 20–22 Octo-
ber.
8. Akram, A.H., Fitzpatrick, A.J., and Halford, F.R.: “A Model to Predict
Wireline Formation Tester Sample Contamination,” paper SPE 48959
prepared for presentation at the 1998 SPE Annual Technical Confer-
ence and Exhibition, New Orleans, 27–30 September.
9. Katz, D.L.: “Overview of Phase Behavior in Oil and Gas Production,”
JPT (June 1983) 1205.
10. Pedersen, K., Thomassen, P., and Fredenslund, A.: “Phase Equilibria
and Separation Processes,” Report SEP 8207, Inst. of Kemiteknik,
Denmark Tekniske Højskole (July 1982).
11. MacMillan, D.J., Ginley, G.M., and Dembicki, H. Jr.: “How to Obtain
Reservoir Fluid Properties from an Oil Sample Contaminated with
Synthetic Drilling Mud,” paper SPE 38852 presented at the 1997 SPE
Annual Technical Conference and Exhibition, San Antonio, Texas, 5–8
October.
12. Whitson, C.H.: “Characterizing Hydrocarbon Plus Fractions,” SPEJ
(August 1983) 683
13. “Reservoir Fluid Studies,” 1996–1999 Programme, Report No. PVT/
99/2, Progress Report, Dept. of Petroleum Eng., Heriot-Watt U., Ed-
inburgh, U.K. (July 1999).
SI Metric Conversion Factors
°F (°F - 32)/1.8 ⳱ °C
psi × 6.894757 E+00 ⳱ kPa
Fathollah Gozalpour is a senior researcher in the Dept. of Pe-
troleum Engineering at Heriot-Watt U. in Edinburgh, U.K. His re-
June 2002 SPE Reservoir Evaluation & Engineering
search interests include phase behavior modeling of volatile
oils and gas condensates, HP/HT fluids, contamination of res-
ervoir fluids with oil-based drilling fluid, and gas-injection pro-
cesses. He holds a BS degree in chemical engineering, an MS
degree in reservoir engineering from Tehran U., and a PhD
degree in petroleum engineering from Heriot-Watt U. Ali
Danesh is a professor in the Dept. of Petroleum Engineering at
Heriot-Watt U. His research interests include reservoir fluids and
hydrocarbon recovery mechanisms, and he teaches courses
on fluids phase behavior and gas condensate reservoirs. He is
the author of the book PVT and Phase Behaviour of Petroleum
Reservoir Fluids (1998). Danesh holds a BS degree in petroleum
engineering from the Abadan Inst. of Technology (AIT), Iran,
and a PhD degree in chemical engineering from Manchester
U., U.K. Dabir H. Tehrani is an honorary professor of petroleum
engineering at Heriot-Watt U. He joined the oil industry (Iran) in
1957, working in different positions up to Petroleum Engineering
Manager (1978) and Reservoir Engineering Manager (1983–91)
with Britoil and BP Exploration. Previously, he taught part-time
at AIT (Iran) and Heriot-Watt U. (1983–92). He joined Heriot-Watt
U. full-time as a research consultant after retiring from BP in
1991. Tehrani holds a BS degree in mathematics from Tehran U.
and a postgraduate diploma in petroleum engineering from
Birmingham U., U.K. Adrian C. Todd is a professor and former
chairman of the Dept. of Petroleum Engineering at Heriot-Watt
U. He joined the department in 1972 and has been involved in
research of oilfield scale, phase behavior of reservoir fluids,
water injection, and hydrates. Todd holds BS and PhD degrees
in chemical engineering from Loughborough U. of Technology,
U.K. Bahman Tohidi is a reader in the Dept. of Petroleum Engi-
neering at Heriot-Watt U. He joined the National Iranian Oil Co.
(NIOC) in 1984, where he worked as an instructor and produc-
tion engineer for 7 years. He joined Heriot-Watt U. in 1991. His
research interests include gas hydrates, PVT and phase behav-
ior, and properties of reservoir fluids and production technol-
ogy. Tohidi holds a BS degree in chemical engineering from AIT
and a PhD degree in petroleum engineering from Heriot-Watt U.
205