Gas Condensate PVT

Gas Condensate PVT –
What’s Really Important and Why?

Curtis H. Whitsona,b,c
Øivind Fevang b
Tao Yang a,b
(a) Norwegian U. of Science and Technology (NTNU), (b) PERA a/s

(c) Corresponding author, [email protected] / [email protected]
Paper (© Curtis H. Whitson) presented at the IBC Conference

“Optimisation of Gas Condensate Fields”, London, Jan. 28-29, 1999, Cumberland Hotel.
IBC UK Conferences Ltd., Gilmoora House, 57-61 Mortimer Street, London W1N 81X, UK
http://www.ibc-uk.com
ABSTRACT
This paper gives a review of the key PVT data dictating recovery and well
performance of gas condensate reservoirs. The importance of specific PVT data are
put in the context of their importance to specific mechanisms of recovery and flow
behavior. Phase behavior important to gas cycling projects is also covered. Modeling
gas condensate reservoir fluid systems with an equation of state is discussed, as is
EOS modeling of complex fluid systems with strongly varying compositions and PVT
properties.
INTRODUCTION
It could be argued that the engineering of a gas condensate field is 80% traditional
“gas” engineering, and 20% “extra” engineering. The numbers could be 90/10 or
70/30 – but the majority of engineering of any gas condensate field is always the
same as the engineering of a gas reservoir without condensate.
The main difference between a gas condensate field and a “dry” gas field is the
additional income derived from surface condensate production. Condensate
production evolves from produced reservoir gas (= produced “wet gas” = produced
wellstream) as the wellstream is processed at the surface. The production of
reservoir gas can, for the most part, be handled with traditional gas engineering
tools.
Gas Condensate PVT – What’s Really Important and Why? 2
C.H. Whitson, Ø. Fevang, and T. Yang
From an engineering point of view, the two “extra” issues which must be addressed
in a gas condensate reservoir are:
• How the condensate “yield” will vary during the life of a reservoir, and
• How two-phase gas/oil flow “near” the wellbore affects gas productivity.
Both of these issues are strongly related to the PVT properties of the fluid system
(though productivity is more affected by relative permeability effects).
PVT properties important to the engineering of all gas condensate reservoirs

includes:
• Z-factor
• Gas viscosity
and a few “extra” properties needed to handle the “condensate” part of a gas-
condensate reservoir:
• Compositional (C7+) variation with pressure

• Oil viscosity and liquid dropout
Dewpoint pressure is implicitly defined by the pressure dependence of compositional

variation. As discussed below, the dewpoint is less importance than is commonly
thought.
The PVT properties listed above are particularly important to reservoirs produced by
pressure depletion. For gas condensate reservoirs undergoing gas cycling it may
also be important to quantify phase behavior (vaporization, condensation, and near-
critical miscibility) which develops in gas cycling below the dewpoint.
Compositional grading in gas condensate reservoirs may be important to the design

of well placement, estimation of in-place surface volumes, reserves, and prediction
of fluid communication vertically (between geologic layers) and areally (between fault
blocks). Prediction of a potential underlying oil is often required in discovery wells
which are drilled upstructure and encounter only gas which is near-saturated. Here,
accurate sampling and PVT modeling are paramount.
A PVT modela should describe accurately the key phase, volumetric, and viscosity
behavior dictating the key processes affecting rate-time performance and ultimate
recoveries of surface gas and oil. Unfortunately, a PVT model may not be capable of
accurately describing all PVT properties with equal accuracy. EOS models often
have difficulty matching retrograde phenomena (compositional variation of gas, and
liquid dropout), particularly when the system is near-critical, or only small amounts of
condensation occur just below the dewpoint (“tail-like” retrograde behavior). Oil
a
We define a PVT (pressure-volume-temperature) model to include the EOS (equation of state)
model describing phase and volumetric behavior, together with the viscosity model (e.g. the Lorentz-
1 2
Bray-Clark or Pedersen et al. ); the viscosity model is not formally linked with the EOS model but
makes use of EOS-calculated density.
viscosities are also difficult to predict for reservoir condensates, and measured oil
viscosities are not usually available for tuning the viscosity model.
Consequently, it is important to determine which PVT properties are most important

to the accurate engineering of reservoir and well performance for a given field
development. Different fields require different degrees of accuracy for different PVT
properties, dependent on field development strategy (depletion vs. gas cycling), low
or high permeability, saturated or highly undersaturated, geography (offshore vs.
onshore), and the number of wells available for delineation and development.
Consider the following examples.
Example 1. A small offshore “satellite” reservoir with high permeability (kh=4,000

md-m), initially undersaturated by 400 bar, and with a test yield of 300 STB/MMscf.
Example 2. A large offshore deep-water reservoir with moderate permeability

(kh=1000 md-m), initially saturated or near-saturated (?), large structural relief, and a
test yield of 80 STB/MMscf. A single (and very expensive) discovery well has been
drilled.
Example 3. An onshore “old” undeveloped gas cap with well-defined initial volume
(by production oil wells and pressure history), uncertain initial composition (estimated
initial yield of 120 STB/MMscf), partially depleted due to long-term production of
underlying oil, and low permeability (kh=300 md-m).
These three examples require significantly different emphasis in the treatment of

PVT data. Why? The reason lies in a coupling of the reservoir and well performance
with PVT properties. Every gas condensate reservoir provides a new example with a
different set of conditions requiring different emphasis on which PVT data are
important.
This paper will attempt to explain when and why various PVT properties are
important to the development of a particular gas condensate field.
PVT EXPERIMENTS
Constant Composition (Mass) Expansion Test
The constant composition expansion (CCE) test, sometimes referred to as a
constant-mass expansion test, is used to measure dewpoint pressure, single-phase
gas Z-factors, and oil relative volume below the dewpoint (“liquid dropout curve”). A
sample of reservoir fluid is charged in a visual PVT cell and brought to reservoir
temperature and a pressure sufficiently high to ensure single-phase conditions.
Pressure is lowered by increasing cell volume until a liquid phase is visually detected
(through a glass window). Total cell volume and liquid volume are monitored from
the initial reservoir pressure down to a low pressure (dictated by cell and sample
size).
Constant Volume Depletion Test

The constant volume depletion (CVD) test is an extremely important laboratory test
which monitors the phase and volumetric changes of a reservoir gas sample (at
reservoir temperature) as the pressure drops below the dewpoint and equilibrium
gas phase is removed. The CVD test simulates closely the actual behavior of a gas
condensate reservoir undergoing pressure depletion, and results from the lab
measurements can be used directly to quantify recoveries of surface gas and
condensate as a function of pressure below the dewpoint. Combined with single-
phase Z-factors from the CCE test, a complete prediction of depletion behavior
(recoveries and liquid-yield variation) can be accurately predicted from initial
pressure to abandonment.
The CVD test involves stepwise lowering the pressure below the dewpoint, with an
associated increase in cell volume. After equilibration at each pressure, enough
equilibrium gas is removed from the top of the cell to bring the cell back to the
original volume occupied at the dewpoint. The amount of gas removed, its
composition and Z-factor, and the remaining oil volume in the cell are measured and
reported. A “two-phase” Z-factor is also reported for use with the “gas” material
balance (see discussion below).
Accurate measurement of the removed gas composition is very important to the

prediction of condensate recovery and liquid-yield variation – much more important
than accurate measurement of retrograde oil volumes. Special laboratory procedures
should be followed to ensure accurate CVD compositional measurements (e.g.
appropriate heating of tubing used to remove equilibrium gas from the cell).
Measurement of the final low-pressure condensate composition allows an important
“backward” material balance check.b
INITIAL FLUIDS IN PLACE AND DEPLETION RECOVERIES

Gas Z-factor
Muz Standing would be happy to hear that the Z-factor is (still) the only PVT property
which always needs accurate determination in a gas condensate reservoir (as in a
“dry” gas reservoir). The reason is (1) to get an accurate and consistent estimate of
the initial gas (and condensate) in place, and (2) to accurately predict the gas (and
condensate) recovery as a function of pressure during depletion drive.c
Single-phase gas Z-factors are measured experimentally at reservoir temperature

and pressures from the initial reservoir pressure to the dewpoint. These data are
b
The backward material-balance check is made by starting with the final condensate composition and
amount (using the reported final oil relative volume and properties), adding incrementally the removed
gas from each CVD step, and ending up with a check of the original fluid composition. This check is
insensitive to oil relative volume (except the final value), a big advantage over the traditional “forward”
material balance which is extremely sensitive to all relative oil volumes and, consequently, can not be
used for leaner gas condensates.
c
It also is important to quantify recovery by depletion for a reservoir which is to be gas cycled,
because the evaluation of economics will always involve comparison of the gas cycling project with
depletion drive.
reported as part of the constant composition expansion test. Z-factors used in the
material balance equation below the dewpoint are “back-calculated” from data in the
constant volume depletion test. These so-called two-phase Z-factors are “pseudo”
(not-physical) properties which should only be used in the traditional gas material
balance equation.
It is not commonly recognized that condensate recovery is strongly related to the

recovery of gas. Gas Z-factors dictate gas and oil recoveries during depletion
(together with the amount of water expansion and influx), because recoveries are
proportional to (Zi/Z) – i.e. the term [1 - (Zi/Z)(p/pi)]. In fact, at average reservoir
pressures above the dewpoint, condensate recovery “exactly” equals the gas
recovery. Consequently, condensate recoveries are strongly dependent on accurate
description of gas-phase Z-factors (both above and below the dewpoint).
As an example for a high-pressure reservoir, a +5% error in Zi and a –5% error in Z

at the dewpoint will result in (a) a +5% error in initial gas and initial condensate in
place, and (b) a +5 to +10 recovery-% error in recovery of gas and condensate at the
dewpoint.
Compositional (C7+) Variation During Depletion

As mentioned earlier, the distinguishing characteristic of a gas condensate field is
the added value from condensate production, in addition to gas. Surface condensate
is, for practical purposes, the C7+ contentd of the produced wellstream. This
simplification makes the treatment of many engineering calculations easier to
understand without loosing engineering accuracy.
The condensate rate profile is easy to convert into an economic profile, and
engineers can readily relate the two. But how can we readily forecast the condensate
rate profile for a gas condensate reservoir?
For a surface gas-rate production profile qg(t), the profile of oil rate versus time is
given (approximately) by
( y 7 + ) CVD 1
qo ( t) ≅ q g (t ) ⋅ ⋅ ........................................................................ (1)
1 − ( y 7 + ) CVD (C og ) CVD
R Tsc ρo
Cog = ⋅ ....................................................................................................... (2)
Psc Mo
where (ρo /Mo) = (ρ7+/M7+)CVD. Time dependence of CVD properties must be
correlated with cumulative wet-gas volumes produced,
G pw = ∫ q w dt ............................................................................................................ (3)
where
qw = qg + qo Cog ........................................................................................................ (4)
d
The simplification of surface condensate being “essentially” C7+ is a subjective choice. One could
easily have chosen C6+ or C5+ without any change in our comments.
Given the qw(t) profile and Gpw(t), this can be translated into cumulative wellstream
produced from the CVD test, (np/nd)CVD,
(np/nd) = Gpw/Gw – [1 – (p/z)d/(p/z)i] ; p Sd ............................................................. (5)
where all CVD properties = initial gas properties for p > pd. Cog represents the
surface gas equivalent of one surface oil volume.
Equations for converting CVD results to approximate surface product recoveries,

including the depletion recovery from initial to dewpoint pressure are:
 (p / Z)d  (p / Z )d N  ∆n p  (1 + rsi ⋅ C og )
(p / Z )i  (p / Z)i ∑
RFgD = 1 − + ⋅   ⋅
 (1 + r ⋅ C ) ........................................... (6)
 k =1  n d k sk og
 (p / Z )d  (p / Z )d N  ∆n p  (1 / rsi + C og )
(p / Z )i  (p / Z)i ∑
RFoD = 1 − + ⋅   ⋅
 (1 / r + C ) ........................................... (7)
 k =1  n d k sk og
z 7+ 1
rs ≅ ⋅ ...................................................................................................... (8)
1 − z 7 + C og
A simple spreadsheet calculation using these equations allows quick translation of
laboratory CVD data to important engineering and commercial quantification of in-
place surface volumes, reserves, and production forecasts. Fig. 1 and Table 1
shows an example calculation using Eqs. 6-8 for a rich-gas condensate.
Compositional Variation with Depth

Numerous field case histories have shown that composition varies with depth in
petroleum reservoirs. Component “segregation” due to gravitational forces is usually
given as the physical explanation for the variation in composition. A theoretical
model for such variation was already defined by Gibbs in the late 1800’s for an
isothermal system under the influence of a constant force field such as gravity. The
result of gravitational segregation is that a gas condensate gets richer at greater
depths, with increasing C7+ mole fraction (and dewpoint pressure)3.
Not all fields show compositional gradients with depth as predicted by the isothermal
model. Some fields show practically no gradient over large depths, such as the
Cupiagua field in Columbia4 where a near-critical gas condensate with more-or-less
constant composition is found over an interval of some 2000 m. Some oil fields have
gradients larger than predicted by the isothermal model. Høier5,6 made
comprehensive calculations using a number of thermal diffusion models which
indicate that thermal gradients typically reduce compositional gradients in gas
condensate fluids, while for oils a thermal gradient may cause either a reduction or
an increase in compositional gradients.
Our concern with compositional gradients in gas condensates will be limited to three
topics: (1) assessing the effect of a gradient on in-place surface volumes, (2)
assessing the prediction of a gas-oil contact using a theoretical gradient model, and
(3) the impact of compositional gradients on depletion (and cycling) recoveries.
Variation in C7+ composition with depth will obviously affect the calculation of initial
surface condensate in place, compared with a calculation based on a constant
composition. Depending on the “location” of the sample used in the constant-
composition model, either smaller or larger initial condensate volumes in place can
result when compared with a gradient modele.
The gradient model will typically give an “optimistic” in-place sensitivity when the gas
condensate reference sample is upstructure. If the reference gas sample is
downstructure then the gradient model will provide a “pessimistic” in-place
sensitivity. For reservoirs with limited control of fluid variation with depth, we
recommend using the “in-situ representative” samples available, linear interpolation
between these samples, and for sensitivities use both (a) the gradient model for
extrapolation beyond the samples and (b) a constant-composition extrapolation.
Another interesting feature of near-saturated reservoirs with compositional gradients

is that the recoverable condensate volumes by depletion are insensitive to whether
the model is initialized with or without a compositional gradient. This lack of
sensitivityf will not be apparent if comparisons are made using recovery factors
(because initial volumes in place can be quite different for the two models of
compositional initialization).
A gradient model may also predict a transition from gas to oil which can dramatically
affect the initial oil in place volumes (see discussion below). However, a predicted
GOC from an upstructure gas sample is, at best, a “possibility”. Results of predicted
“oil zones” from gas samples should only be used for sensitivity analysis of a new
discovery, or in a reservoir where additional delineation wells are not planned.
Dewpoint Pressure
Strictly speaking the dewpoint pressure is the pressure where an incipient liquid
phase condenses from a gas phase. Practically, the dewpoint marks the pressure
where (1) reservoir gas phase composition changes and becomes leaner, and (2)
condensate accumulation starts in the reservoir. These two changes can have a
profound effect on reservoir and well performance – or, they may have little impact.
The importance of the actual dewpoint pressure will vary from reservoir to reservoir,
but in most situations accurate dewpoint determination is not important.
Why? First, in the context of compositional variation with pressure (and associated
variation of condensate yield with pressure) accurate determination of the
thermodynamic dewpoint pressure is not of particular importance. In fact, we don’t
need to know the “specific” dewpoint at all as long as the variation of composition
(C7+ content) with pressure is well defined “near” the thermodynamic dewpoint.
e
The gradient model requires that a composition be specified at a reference depth with a reference
pressure and temperature.
f
This behavior is readily understood in a black-oil model using a solution oil-gas ratio rs(p) function.
The initial variation of rs (and dewpoint) with depth is short-lived when reservoir pressure drops below
the dewpoint, as rs of the equilibrium gas through the reservoir becomes (more-or-less) constant at
the average reservoir pressure. Because only reservoir gas flows into wells, the producing OGR will
reflect the equilibrium rs at average reservoir pressure, which becomes independent of depth when
pR<pd.
Second, when the bottomhole flowing pressure (BHFP) drops below the dewpoint
and two phases start flowing near the wellbore, gas relative permeability drops and
well productivity drops. However, as long as BHFP is “anywhere near” the dewpoint
the well will have excess deliverability – i.e. we simply reduce the BHFP to produce
more gas (even though the well productivity is lower).
Only when the BHFP reaches a minimum value (dictated by some delivery-pressure
constraint) will the well no longer be able to deliver the desired rate. At this point,
well productivity is important. However, this occurs at a BH flowing pressure much
lower than the dewpoint. A typical minimum BHFP might be 50-150 bara, while
dewpoint pressures are typically 250-400 bara. Whether the BHFP drops below the
dewpoint at 400 or 350 bara has little impact on what the well will (or won’t) produce
when BHFP reaches a minimum constraint of 100 bara.
Another (less common) need for dewpoint pressure is when an underlying saturated
oil zone may exist and a PVT model is used to predict the existence and location of
the gas-oil contact (GOC). In this case, the PVT model dewpoint should be tuned
precisely to an accurately measured dewpoint pressure. It is not uncommon that a
predicted GOC may vary 10’s of meters per bar of uncertainty in the (PVT-model)
dewpoint pressure. Thus an accurate description of the dewpoint pressure will have
an impact on the prediction of initial oil and gas in place, placement of delineation
wells, and potential field development strategy. In this situation, accurate dewpoint
measurement and equally-accurate modeling of the measured dewpoint should be
given due attention.
On the other hand, if accurate treatment of the dewpoint pressure is not required (for
estimating a GOC), then we recommend using little if any weighting of the measured
dewpoint pressure when tuning the PVT model. Instead, priority should be given to
matching the variation of C7+ with pressure in the removed gas from a constant
volume depletion test.
CONDENSATE BLOCKAGE
When BHFP drops below the dewpoint and two-phase gas/oil flow stabilizes in the
near-wellbore region, relative permeability to gas (the primary flowing phase) may
drop dramatically and the well deliverability is lowered accordingly. Saturations in the
near-wellbore region can reach 40-60%, with gas permeability reductions of 0.05-
0.2.
Flow in the near-wellbore region reaches a steady-state condition in a relatively short

time after the BHFP drops below the dewpoint. Flow theory shows that the produced
wellstream mixture is constant throughout the “steady-state” region, meaning that if
we “captured” the flowing mixture at any point within this region, its composition
would be the same as the producing wellstream mixture.g
g
The flowing mixture composition at some point within the steady-state region will not equal to the
overall composition occupying the pore volume (“in-situ” composition) at that point, as the in-situ
composition will vary throughout the steady-state region.
Permeability reduction in the near-wellbore (steady-state) region is particularly

important because most of the pressure drop will be greatest in this region. A relative
permeability reduction of 0.1 in the first 10 meters from the wellbore will have
significantly greater impact than a krg reduction of 0.1 at a 10-m radial interval some
100 m away from the wellbore.
The relative permeability ratio krg/kro in the steady-state (SS) region is given by
krg/kro=(1/Vro – 1)(µg/µo)............................................................................................ (9)
where Vro = Vo/Vt is the CCE oil relative volume of the produced wellstream at any
pressure within the SS region. The pressure in the SS region ranges from the BHFP
to the dewpoint pressure of the produced wellstream, and most deliverability loss
occurs nearest the wellbore where pressures are closer to the BHFP.
Given the krg/kro ratio throughout the SS region from Eq. 9, the relative permeability
to gas krg can be found directly from the relative permeability relationship. That is,
krg(S) = krg(krg/kro(S)). Fevang and Whitson7 have shown that the krg/kro ratio in the
SS region does not vary more than about one order of magnitude throughout the life
of a reservoir. With regard to uncertainties in PVT properties, and the need to
measure (or predict) their values for accurate description of condensate blockage,
we can conclude that oil viscosity should be given highest priority because it has the
largest uncertainty both experimentally and in predictions. Fig. 2 shows the effect on
krg caused by a ±20% error in Vro for values of Vro ranging from 0.5 for a near-critical
condensate to 0.005 for a very lean condensate (using µg/µo=0.02/0.2=0.1 and a
typical Corey relative permeability relation).
Oil Viscosity
As discussed above, oil viscosity is important in the proper modeling of “condensate
blockage” – i.e. the two-phase gas/oil flow effect on gas relative permeability in the
region around the wellbore. Oil viscosity is usually low for reservoir condensates,
ranging from 0.1 to 1 cp in the near-wellbore region.h Measurement of condensate
viscosities is not made in routine laboratory tests, and it may be difficult to obtain
measurements for lean condensates (where volumes of condensate are small).
Viscosity correlations are typically unreliable for predicting low oil viscosities, and
some approach is needed to ensure accurate and consistent modeling of this
important property.
We recommend that the oil viscosity model be tuned to measured viscosities of a

separator condensate samplei at reservoir temperature and pressures in the range of
100-400 bara. More “appropriate” condensate viscosity measurements can be
designed (at greater expense), but having oil viscosity data from a separator oil
h
The viscosity of condensate flowing in the near-wellbore region throughout depletion, and
particularly when a well goes on decline, will remain fairly constant at a value close to the condensate
viscosity of the original reservoir fluid in the pressure range 100-200 bara. This viscosity is typically
lower than the viscosity of condensate from a CVD test in the same pressure range, where the
difference may be as much as a factor of 2-3.
i
This idea was suggested by Dr. Jeff Creek, Chevron Oil Company, ca. 1997.
sample to tune the viscosity correlation should ensure reasonably accurate oil
viscosity predictions of the condensate actually flowing in the near-wellbore region
when bottomhole flowing pressures drop below the dewpoint.
Gas Viscosity
Gas viscosity for most systems will vary from 0.02 to 0.03 cp for all pressure
conditions. For near-critical gas condensates and high-pressure gases the viscosity
may initially be 0.05 cp, but in most of the near-wellbore region experiencing
significant pressure losses the viscosity will be in the lower range of 0.02-0.03 cp.
Consequently, the absolute value of viscosity does not vary greatly for a given gas,
or from gas to gas system. Viscosity correlations are fairly reliable at predicting
accurate gas viscosities, within 5-10% in most cases.
What is important with respect to gas viscosities is that consistent viscosities be

used in all engineering applications – e.g. well test interpretation, well performance
design, reservoir simulation, tubing calculations, pipeline calculations, etc. It is not
uncommon that a 15-25% difference in gas viscosities may result using different
correlations (by different engineering disciplines). This may result in similar
inaccuracies in well performance calculations, even where all flow is single-phase
gas!
Oil Relative Volume (Liquid Dropout Curve)

The oil relative volume or “liquid dropout curve” is perhaps the most familiar property
to engineers working with gas condensate fields. The maximum liquid dropoutj is
often used as a subjective measure to characterize the richness or leanness of a gas
condensate fluid system (perhaps even more common than the liquid yield itself!).
Two definitions of oil relative volume Vro are used,

Vro = Vo / Vd ........................................................................................................... (10)
Vro = Vo / Vt = Vo / (Vo+Vg) ..................................................................................... (11)

It is important to differentiate between the two definitions. The first and most
common definition is oil volume relative to the dewpoint volume, where this gives a
direct measure of the actual volume of oil condensed. The second and more
important definition (for engineering purposes) is oil volume relative to total gas+oil
volume, where the change in this Vro depends on two effects – the change in oil
volume itself and the change in total volume, Vro(p) = Vo(p)/Vt(p). This latter definition
is more important because it enters directly in the condensate blockage problem,
and at lower pressures (<250 bara) where condensate blockage is particularly
important, the change in total volume Vt(p) due to gas expansion becomes even
more important than the change in oil volume.
Ironically, the liquid dropout curve has little direct impact on reservoir and well
performance. Only the CCE liquid dropout Vro=Vo/Vt of a reservoir gas at “lower”
pressures has a (second-order) effect on the modeling of condensate blockage. The
j
The maximum liquid dropout often occurs at a pressure near 150-250 bara (though higher for near-
critical systems), and is approximately correlated to the minimum in equilibrium K-values (Ki=yi/xi) of
heavier components (C5+).
“average” oil saturation in a gas condensate reservoir during depletion, given

approximately by the CVD experiment, is seldom important.
Interestingly, the magnitude of maximum liquid dropout does not determine whether
condensate blockage will or will not be a problem for a given reservoir. It only has a
second-order effect on the relative degree of severity. For example, one reservoir
with 35% maximum liquid dropout may have a condensate blockage effect which has
no importance to well deliverability, while another reservoir with 2% maximum liquid
dropout may have condensate blockage causing a dramatic well deliverability loss.
The importance of condensate blockage on well performance is dictated by the

relative importance of “reservoir” pressure losses compared to “pipe”
(tubing+flowline) pressure losses. For a high-kh (kh=10 000 md-m) rich condensate
well the blockage skin may be +30 with a resulting additional pressure loss of only 3
bar, where tubing pressure losses are 300 bar due to high deliverable rates. A lower-
kh (kh=500 md-m) lean condensate well may have a blockage skin of +15 with a
resulting additional pressure loss of 150 bar, where tubing+flowline pressure losses
are 150 bar. Clearly the lean condensate well has a more severe condensate
blockage problem than the rich high-kh well.
GAS CYCLING
Traditional gas cycling with full pressure maintenance is almost completely
unaffected by PVT properties. The gas-gas displacement process is fully miscible,
independent of the injection gas, so only the viscosity ratio of the two gases enters
into the displacement performance. For practical purposes, reservoir heterogeneities
(and mainly layering) dominate recovery performance of a gas cycling project –
almost totally for full-pressure maintenance cycling, but also if the reservoir is cycled
below the dewpoint.
As indicated by Coats8 and others, gas cycling projects below the dewpoint are also
affected by the vaporization characteristics of displacement gas on the retrograde
condensate. For most gas cycling projects the injection gas is fairly lean and
recovery efficiency of the reservoir retrograde condensate depends mostly on
vaporization.
If injection gas is rich in intermediate components C2-C5 and gas cycling occurs
below the (original) dewpoint, an efficient condensing/vaporizing mechanism can
develop and, in some cases, develop full miscibility with the gas+condensate
reservoir system. This mechanism is described by Høier and Whitson6.
Gas Cycling Recovery Efficiency in Swept Zone

Let us look at the recovery mechanisms in a gas cycling project in the volume of the
reservoir swept by injection gas – the “microscopic” or “pore-level” recovery. At a
given time and position in the swept zone, the pressure is either above or below its
original dewpoint when the injection gas front arrives.
If above the dewpoint, a gas-gas miscible displacement will yield 100% recovery of
the current condensate in place. A miscible displacement is guaranteed,
independent of the injection gas used, even though the injected gas may be first-
contact “immiscible” with the original reservoir gas. Miscibility develops by a simple
vaporizing mechanism.
If reservoir pressure is below the dewpoint when the displacement front arrives,
ultimate recovery of condensate is dictated by two processes: (1) gas-gas miscible
displacement of the reservoir gas, and (2) partial vaporization of the retrograde
condensate. The condensate recovery by gas-gas miscible displacement is 100% of
the condensate in solution in the reservoir gas at the time the front arrives. The
recovery efficiency of retrograde condensate by vaporization (Ev) increases gradually
as increasing volumes of injection gas sweep this point in the reservoir.
We will try to discuss the two mechanisms of condensate recovery for a reservoir
undergoing cycling below the dewpoint. Before the gas front arrives, (1) the amount
of “condensate in place” in the gas-filled pores continuously decreases below the
dewpoint; and (2) the “cumulative retrograde condensate” in the oil-filled pores
continuously increases at decreasing pressures.
After the gas front arrives, (1) the gas-gas miscible displacement has a 100%
efficiency of the condensate remaining in solution in the reservoir gas; and (2) the
recovery efficiency of retrograde condensate by vaporization (Ev) rises quickly but
then flattens quickly after the front passes. This behavior of Ev is due to the
preferential vaporization of “light” C7+ first, leaving behind a heavier condensate
which is less efficiently (more slowly) vaporized. The more volumes of injection gas
passing over the condensate, the less efficient vaporization becomes. Even if
pressure continues to decline, new condensation will not occur because the gas
behind the front is lean and has little dissolved condensate.
Accurate prediction of the changing vaporization efficiency requires (1) an accurate

description of the C7+ molar distribution of the condensate, and (2) the K-values of
C7+ components as a function of pressure and overall composition. Because total
condensate recovery efficiency in the swept region may be dependent on an
accurate description of the component-by-component vaporization process, effort
should be made to obtain compositional data which describes the vaporization
process. Extra effort should also be given towards fitting these compositional data
with the EOS model. However, it is worthwhile to first evaluate the potential for
recovery by vaporization of retrograde condensate below the dewpoint prior to
obtaining extensive (and expensive) laboratory data of the type described above.
Evaluating Gas Cycling Potential

Defining the “target” of condensate recovery by gas cycling is important to economic
evaluation and field development strategy. The following definitions are useful for
defining the target of gas cycling:
1. RFoD is, as previously discussed (Eq. 7), the condensate recovery by pressure
depletion to some reservoir pressure – e.g. the end of production pend, or at the
pressure of gas cycling pcycle.
2. RFoM is the recovery of condensate which would be expected at the end of

cycling due to gas-gas miscible displacement with 100% sweep efficiency
(Es=100%); it is assumed the gas cycling occurs at a constant pressure pcycling

above or below the original dewpoint.
3. RFoV (=100 – RFoM) is the recovery of condensate which would be expected at

the end of cycling due to vaporization of retrograde condensate with 100% sweep
efficiency (Es=100%); it is assumed the gas cycling occurs at a constant pressure
pcycling below the initial dewpoint. Note, RFoV = 100% for gas cycling above the
initial dewpoint.
4. RFoDx is the extra condensate recovery from pressure depletion of the reservoir
volume not swept by injection gas during cycling, (depletion from pcycling to pend).
Note, RFoDx = RFoD(pcycling) – RFoD(pend).
5. RFoult is the ultimate condensate recovery due to (a) depletion prior to cycling, (b)
cycling, and (c) depletion after cycling. For pcycling > pd,
RFoult = RFoD + ES ⋅ RFoM + (1 − ES ) ⋅ RFoDx ............................................................ (12)
while for pcycling < pd,
RFoult = RFoD + ES ⋅ (RFoM + E V ⋅ RFoV ) + (1 − ES ) ⋅ RFoDx ......................................... (13)
where ES defines the final areal-times-vertical sweep efficiency at the end of cycling,
and EV defines the final efficiency of vaporized retrograde condensate (for pcycling
below the dewpoint).
What components of the ultimate recovery are strongly dependent on PVT

properties?
We already have defined the dependence of RFoD on PVT properties (Eq. 7), where
Z-factors and variation of C7+ in the CVD produced gas phase determine RFoD. It is
also easy to show that RFoM is given exclusively by data in the CCE and CVD tests,
as is RFoV (= 100 – RFoM).
The only other PVT-dependent parameter is EV, which (as mentioned earlier) is
determined by (1) the C7+ molar distribution of retrograde condensate, and (2) the K-
values of C7+ components as a function of pressure and overall composition.
However, EV is only important for cycling below the dewpoint, and often the
contribution of vaporization to overall condensate recovery is relatively small.
Fig. 3 shows recovery of condensate versus pressure for a high-pressure offshore

gas condensate field. Initial pressure is 900 bara, and dewpoint is 400 bara. The
calculations are based only on CCE and CVD data, as shown in Table 2. The lower
curve gives RFoD(p), the recovery due to pressure depletion. The upper curve gives
RFoM(p), the recovery due to gas-gas miscible displacement with 100% sweep
efficiency but with no vaporization of retrograde condensate.
In this high-pressure reservoir, the additional recovery due to gas cycling will not be
realized until late into the life of the field, as pressure approaches the dewpoint and
recovery is about 30%. In terms of net present value, and depending on when the
investment for compressors etc. are required, cycling does not appear attractive.
For the same reservoir but initially saturated at the dewpoint of 400 bara, additional
condensate recovery by gas cycling is more attractive, as shown in Fig. 4. Here the
rapid decline in producing liquid yields will have a pronounced affect on project
economy, while successful cycling (high sweep efficiency ES) can provide prolonged
initial liquid yields and higher ultimate recovery. For this case, net present value is
more positive to a cycling project.
In summary, most of the primary evaluation for potential of gas cycling can be
quantified by CVD and CCE data. Vaporization effects are often less significant than
commonly thought, an observation which follows from the calculation of recovery
factors RFoD and RFoM based on CVD and CCE data. When vaporization recovery
(RFoV) is important, special multi-contact vaporization tests should be conducted and
fit with the PVT model, where variation of C7+ in the equilibrium gas is the most
important data.
Combined Condensing/Vaporizing Mechanism

Historically it has been assumed that any gas cycling project in a gas condensate
reservoir was miscible only by the vaporizing gas drive mechanism. Consequently,
the MMP has always been assumed equal to the dewpoint pressure. Cycling
projects where reservoir pressure drops below the dewpoint were considered
"inferior" because only partial vaporization of the retrograde condensate could be
expected. For most separator injection gases these traditional assumptions are valid.
However, Høier and Whitson6 show that miscible displacement of gas condensates
(by the condensing/vaporizing mechanism) can be obtained at pressures far below
the dewpoint for continuous or slug injection of gas enriched with components C2-C5.
Whether a below-dewpoint miscible displacement can develop in a gas condensate

depends on (1) pressure, (2) composition of the injection gas, (3) composition of the
retrograde condensate ahead of the front, and (4) physical dispersion or fingering
(for slug injection). Although the same conditions also apply for enriched-gas
miscible displacement of an oil reservoir, conditions (3) and (4) are particularly
important for gas condensates.
The most likely candidate for enriched-gas miscible gas cycling below the dewpoint
would be rich or near-critical condensates where injection gas is not available in
sufficient quantities to maintain reservoir pressure.
“ REPRESENTATIVE” SAMPLES
Before a field development starts, the primary goal of sampling is to obtain
"representative" samples of the fluids found in the reservoir at initial conditions. It
may be difficult to obtain a representative sample because of two-phase flow effects
near the wellbore. This occurs when a well is produced with a flowing bottomhole
pressures below the saturation pressure of the reservoir fluids. It is also commonly
thought that “bad” fluid samples result if gas coning or oil coning occurs during
sampling.
The most representative insitu samples are usually obtained when the reservoir fluid
is single phase at the point of sampling, be it bottomhole or at the surface. Even this
condition, however, may not ensure representative sampling. And, as shown by
Fevang and Whitson9, samples obtained during gas coning in an oil well can provide
accurate insitu representative samples if a proper laboratory procedure is followed.
Because reservoir fluid composition can vary areally, between fault blocks, and as a
function of depth, we are actually interested in obtaining a sample of reservoir fluid
that is representative of the volume being drained by the well during the test.
Unfortunately, the concept of a "representative" sample is usually a sample that
correctly reflects the composition of reservoir fluid at the depth or depths being
tested.
If we suspect or know that a sample is not "representative" (according to this

definition), then we tend to do nothing with the sample. Or we question the validity of
the PVT analysis done on the "unrepresentative" sample, and consequently don’t
include the measured data when developing the PVT model.
We strongly recommend against using this definition of "representivity." First of all, it

is a definition that costs our industry in terms of wasted money and time, and lost
opportunity. An important point to keep in mind is that:
Any fluid sample that produces from a reservoir is automatically

representative of that reservoir. After all, the sample is produced from the
reservoir!
The final EOS fluid characterization of a field should match all (accurate) PVT
measurements of all (uncontaminated) samples produced from the reservoir,
independent of whether the samples are representative of insitu compositions.
Accuracy of PVT Data ,QVLWX5HSUHVHQWLYLW\RI6DPSOH
Accurate PVT measurements can be made on both representative and

unrepresentative samples. Inaccurate PVT measurements can also be made on
both types of samples; bad PVT data should be ignored.
Furthermore, an EOS fluid characterization is used to predict compositional changes

during depletion which represent a much greater variation than the compositional
differences shown by "representative" and "unrepresentative" samples.
Another misconception in "representative" fluid sampling of gas condensates is that it

is difficult to obtain insitu-representative samples in saturated gas condensate
reservoirs (with underlying oil). The exact opposite is true! Fevang and Whitson9
have shown that if a gas condensate is initially saturated and in contact with an
underlying oil zone, then a near-perfect insitu-representative sample can be obtained
(at the gas-oil contact) – independent of whether the reservoir gas and reservoir oil
samples collected are insitu-representative.
In summary, all “uncontaminated” samples collected from a reservoir are reservoir

representative and, accordingly, should be described accurately by the PVT model.
Insitu-representative samples may be difficult to obtain. But even when collected,
they may not represent more than a “local” volume of the reservoir, where significant
variations in fluid composition exist vertically and areally away from the point of
sampling.
EOS MODELING
To make EOS calculations, the minimum required input are (1) molar composition, and
(2) molecular weight and specific gravity of the heaviest fraction (usually C7+ or C10+).
With this minimum information, an EOS can calculate practically any phase and
volumetric property of the mixture - e.g.,
· Bubblepoint or dewpoint pressure at a specified temperature

· Pressure-temperature phase envelope
· Densities and compressibilities of oil and gas phases
· Separator gas-oil ratio and surface gravities
· Depletion PVT experiments
· Multicontact gas injection experiments
Splitting the Plus Fraction

Usually, three to five C7+ fractions (or 2 to 3 C10+ fractions) should be used. The
Whitson et al.10 splitting/characterization procedure is recommended for the Peng-
Robinson EOS. The Pedersen et al.11 method is recommended for the Soave-Redlich-
Kwong EOS, where each plus fraction has equal mass fraction.
When true boiling point distillation data are available, these data should be used
directly, or to define parameters in the splitting model. TBP data can be used, for
example, to define the molar distribution parameters DQG LQWKHJDPPDGLVWULEXWLRQ
model, and constants in the specific gravity correlation.
Tuning the EOS Model

If measured PVT data are available, and they have been checked for accuracyk, the
EOS characterization can be modified to improve the predictions of measured data.
Manual adjustments of EOS parameters such as binary interaction parameters (BIPs)
and heavy component critical properties can be used, though this approach is time
consuming (and often frustrating). Nonlinear regression can be used to mathematically
minimize the difference between EOS predictions and measured PVT data. A critical
aspect of the “tuning” procedure is to properly weigh individual data (and data types)
based on the importance of individual data to reservoir and well performance.
k
Material balance methods are often useful for checking consistency of depletion experiments (CVD
and DLE). The material balance starts at the final stage of the experiment, with known amounts and
compositions. Removed gas is then added back from each depletion stage, arriving at the initial fluid.
Comparison with the initial composition gives a direct measure of consistency. Another approach is to
start the material balance with the initial fluid, back-calculating the oil phase properties and
compositions as depletion progresses; this approach is only useful for medium-rich to rich
condensates because small errors in Vro have a dramatic effect on the back-calculated oil properties.
Developing a “ Common” EOS Model for Multiple Reservoir Fluids

An important requirement in the development of an EOS model is the need to have one
set of components to describe all reservoir samples in a given field. This is particularly
important if the reservoir fluids from different parts of the reservoir (layers or fault
blocks) mix in the reservoir. It also may be important if mixing only occurs at the
surface.
Our experience has shown that a single set of components and a single set of EOS
component properties can be used to describe a wide range of reservoir fluids, ranging
from leaner gas condensates to low-GOR oils – fluids which may or may not be in fluid
10
communication initially. Whitson et al. propose one method for developing a
“common” EOS model for multiple reservoir fluids. Another approach is to develop the
EOS model based on a single sample, and then “generate” the other reservoir fluids by
flash calculations (saturation pressure, two-phase split, or isothermal gradient).
Generating Black-Oil PVT Tables

Once the EOS characterization has been developed, a primary application of the EOS
is to generate black-oil PVT tables for reservoir simulation, material balance and flow
calculations (also pipeflow calculations). The most common application of black-oil
PVT is black-oil simulation.
The procedure proposed by Whitson and Torp12 is recommended for generating black-
oil PVT tables. They suggest using a reference fluid to conduct a depletion test (e.g.
CVD), sending the equilibrium reservoir phases separately through a surface
separation to obtain (Rs, Bo o) for the oil phase and (rs, BgdDQG g) for the gas phase.
Surface densities are taken from the surface separation of the reference fluid.
The definition of black-oil PVT properties are:

V o = oil formation volume factor
Bo =
V OO
V go
Rs = = solution gas- oil ratio
V oo
Vg
Bgd = = dry gas formation volume factor
V gg
V og
rs = = solution oil- gas ratio
V gg
where subscripts are defined as:
o : reservoir oil phase at p and T

g : reservoir gas phase at p and T
oo : surface oil from reservior oil (“solution” oil)

go : surface gas from reservoir oil ("solution" gas)
og : stock-tank oil (condensate) from reservoir gas
gg : surface gas from reservoir gas
The important black-oil PVT properties for IFIP calculations in gas condensate
reservoirs are rs/Bgd = the surface oil in place per reservoir gas volumel, and 1/Bgd = the
surface gas in place per reservoir gas volume. The term rs/Bgd is the quantity required
by “geologists” to convert reservoir gas pore volumes to surface oil – a kind of “oil FVF
(Bo)” for the reservoir gas phase. In fact, for compositionally-grading reservoirs with a
transition from gas to oil through an undersaturated (critical) state, the term rs/Bgd
should equal Bo – exactly – at the undersaturated gas-oil contact, thereby ensuring
continuity and consistency.
Some special problems related to generating black-oil PVT tables with an EOS include:
1. How to extrapolate saturated PVT properties to pressures higher than the original
saturation pressure of the reference fluid.
2. Non-monotonic saturated oil properties Bo and Rs for gas condensate systems.
3. Consistency requirements for comparison of black-oil and EOS simulations.
4. Handling saturated reservoirs with a gas overlying an oil, where black-oil PVT
properties from the two fluid systems can be significantly different.
5. Modeling reservoirs with compositional gradients, and how to initialize these
reservoirs in black-oil simulators.
Extrapolating Saturated Tables

Extrapolating saturated black-oil PVT tables can be done in a number of ways,
depending on the reservoir process and why the extrapolation is needed. Extrapolation
is usually required for (a) gas injection studies, (b) reservoirs with compositional
gradients where the reference sample is undersaturated, or (c) ensuring numerical
stability for near-critical fluid systems where pressures may exceed the original
saturation pressure during iteration.
Methods for extrapolating black-oil PVT tables include: (a) mixing the incipient phase
from a saturation pressure of the reference fluid to increase the saturation pressure,
usually in a number of steps (“reverse DLE”), (b) using a compositional gradient
algorithm, or (c) adding an injection gas in increments and determining the PVT
properties of each incremental mixture, or (d) adding injection gas to a “maximum”
saturation pressure and then conducting a depletion test – either stopping at the
original saturation pressure or continuing all the way to a low (minimum) pressure.
The most appropriate method for extrapolating saturated properties may not be
obvious. It will often depend on the reservoir process. Several extrapolation methods
may be tested in a realistic reservoir simulation model, where results are compared
with a fully-compositional EOS model. The extrapolation method which consistently
gives results most similar to the EOS model can be said to “best represent” the
reservoir process. Comparisons should include initial fluids in place, recoveries, and
GOR profiles of individual wells.
Non-monotonic Saturated Oil Properties

For medium to lean gas condensates we have often found that the saturated oil black-
oil PVT properties Bo and Rs are not monotonic – first increasing just below the
l
rs/Bgd is equivalent to the compositional “equivalent” C7+ content in the reservoir gas, y7+.
dewpoint pressure, reaching a maximum, and then decreasing with pressure in a

“normal” fashion. The physical explanation for this behavior is that the first condensate
which appears is (for lean and medium-lean condensates) quite heavy, with a high
surface-oil density (e.g. low API gravity < 40). The “low-gravity” or heavy condensate
has, as expectedm, a relatively low Bo and Rs.
As more condensate evolves from the reservoir gas, the total reservoir condensate
becomes lighter with higher (API) gravity. The change in the gravity has a stronger
influence than the decreasing pressure, so the net change in Bo and Rs is to increase
with decreasing pressure.
As pressure continues to decrease, the condensate gravity stabilizes and “normal”

pressure dependence of saturated oil properties results, with Bo and Rs both
decreasing at decreasing pressures.
One solution we have found useful in this situation is to generate a separate set of
black-oil PVT tables, starting with the incipient oil phase of the original reservoir gas at
dewpoint, and using a depletion (DLE or CVD) test with this oil. The “oil phase” PVT
table (Bo, Rs, and o) from depletion of the incipient oil can then be used together with
the “gas phase” PVT table (Bgd, rs g) from depletion of the original reservoir gas.
Consistency Between Black-Oil and EOS Models

Coats8 addresses the need for compositional EOS models for gas condensate
reservoirs. He shows that gas cycling below the dewpoint is the only situation when
black-oil modeling may be inadequate. He suggests that an EOS model with at least 3
C7+ fractions should be used to properly model vaporization recovery of retrograde
condensate. His results indicate that depletion, single-well modeling, and gas cycling
above the dewpoint are properly modeled with a black-oil model.
Fevang and Whitson7 show that some modification of black-oil saturated oil viscosities
may be needed to ensure accurate single-well modeling of condensate blockage. They
show that the oil viscosity in the near-wellbore blockage region is the liquid which
condenses from the flowing wellstream, and not a cumulative CVD-type liquid. The
flowing blockage-area liquid can be significantly lighter than the CVD-type condensate,
and with correspondingly lower viscosity (1.5 to 5 times lower).
Handling Saturated Gas/Oil Systems

Black-oil PVT properties for a saturated reservoir gas/oil system (gas cap overlying oil)
may be difficult to generate using a consistent approach. Traditionally we generate a
complete set of PVT tables separately for the reservoir oil and reservoir gas, using a
depletion test for the reservoir oil (e.g. DLE) and a depletion test for the reservoir gas
(e.g. CVD). From the depletion test of each reservoir phase, the complete set of black-
oil PVT tables are consistent only at the initial saturation pressure. That is, the incipient
oil from the dewpoint of the reservoir gas is identical to the reservoir oil; and the
incipient gas from the bubblepoint of the reservoir oil is identical to the reservoir gas.
The saturated oil and gas phases which form from the two depletion tests are different
below the original saturation pressure. This leads to differences in PVT properties
m
Expected, for example, based on a Standing-type saturated correlations for Bo and Rs.
which are not handled consistently in a black-oil simulator. One solution is to use two
PVT regions, one for the cells originally in the gas cap, and another region for cells
originally in the oil. This solution is incorrect for cells which originally are defined as one
“phase” but become the other “phase” due to movement of the gas-oil contact. Still, this
may be the best solution in some reservoirs.
Initializing Reservoirs with Compositional Gradients

Two problems arise when trying to use black-oil simulators for reservoirs with
compositional gradients. The one problem is obtaining correct initial surface fluids in
place (compared with initialization using an EOS model). The second problem is
analogous to that discussed above for saturated gas/oil systems, where PVT
properties of the different reservoir fluids are not the same.
The best way to ensure accurate initialization of surface gas and surface oil in place
is to initialize using Rs and rs versus depth, instead of using saturation pressure
versus depth. The more-common practice of initializing with saturation pressure
versus depth leads to problems of initial fluid in place because of the second
problem mentioned above – i.e. though only one PVT table is used, the black-oil
PVT properties of the different reservoir fluids are not the same.
Our recommended method of using Rs an rs versus depth for initialization may lead
to a small error in recoveries near the initial saturation pressures. However, this error
is usually insignificant and always less than errors introduced by wrong initial fluids in
place caused by initialization with saturation pressure versus depth.
Pseudoization (Grouping Components)

Some reservoir processes can not be adequately modeled with a black-oil PVT
formulation. Gas injection, near critical oil and gas condensate systems, and laboratory
simulations may require fully compositional EOS simulation. The mathematical
complexity of integrating an EOS in a reservoir simulator is many times that of using a
simple black-oil PVT formulation. The result is a simulator that runs much slower than
a black-oil simulator. It may be necessary to economize the number of components
used in compositional simulation by "pseudoization" (i.e. reducing the number of
components in an EOS characterization).
The number of components used in an EOS characterization depends both on

computational restraints, and on the desired level of accuracy from the EOS. Some
balance between these two requirements is needed to determine the final number of
components for solving a given problem.
An initial fluid characterization will typically contain from 13 to 20 components, and

sometimes more. For best results, a stepwise pseudoization procedure is
recommended, whereby several pseudoized characterizations are developed
sequentially (e.g. 15, 12, 10, 7, and 5 pseudocomponents). The goal with each
pseudoization is to maintain PVT predictions as close to the original full
characterization as possible. With this stepwise approach, it is readily determined how
few pseudocomponents are necessary to maintain a required similarity to the original
full characterization.a Reducing the number of components in a stepwise fashion has

three main advantages:
1. It is possible to establish when a further reduction in number of components results

in predicted properties that deviate unacceptably from the original N-component
characterization.
2. The procedure usually results in several alternative characterizations with a

common basis. One simulation might require more components than another (e.g.
radial single-well study versus full-field simulation). Because several
characterizations are available, and they are "related" through the original N-
component characterization, more consistency can be expected.
3. Experience has shown that better results are obtained in going from the N-
component characterization to (for example) a 7-component characterization in
several steps, than going from an N-component to a 7-component characterization
in a single pseudoization.
The recommended stepwise pseudoization procedure is given below:
1. Use regression to develop an EOS characterization so that all relevant and

accurate PVT data are adequately matched. (This is probably the most difficult part
of any fluid characterization).
2. Using this tuned EOS, simulate several PVT experiments. Save the results of these
calculations as “data”. The experiments should cover as large as possible the
pressure-, temperature-, and composition-space expected in the reservoir during its
development. If gas injection is being considered, multicontact gas injection
experiments should be included, perhaps several with different injection gas
compositions.
3. Reduce the number of components by 2 or 3 by grouping components, e.g., group

iso- and normal-alkanes of butanes and pentanes.
4. Fine tune (by regression) the newly-created pseudo-component EOS parameters.

Recommended parameters include multipliers to EOS constants A and B and
volume shift parameter s for each newly-created pseudocomponent separately; and
BIPs between methane and the C7+ fractions (collectively).
5. In a subsequent step, regress viscosities for the EOS model with the newly-created
pseudocomponents.
6. Return to step 3, selecting 2 or 3 new components to group.
Fig. 5 summarizes an example pseudoization procedure.
a
The number of pseudocomponents will vary according to the application. Simulation of depletion
processes and water flooding will generally require only 4 or 5 pseudocomponents; immiscible gas injection
may require additional pseudocomponents, and developed miscible gas injection will probably require at
least 6 to 8 pseudocomponents.
CONCLUDING REMARKS
We have tried to summarize and detail the importance of various PVT properties on
the reservoir and well performance of gas condensate fields.
1. For calculation of initial gas and condensate in place the key PVT data are (a)
initial Z-factor and (b) initial C7+ molar content. In terms of black-oil PVT
properties, the two “equivalent” PVT quantities are (a) Bgd and (b) rs/Bgd.
2. The constant composition and constant volume depletion tests provide the key
data for quantifying recovery of produced gas and condensate during depletion.
Above the dewpoint depletion recoveries of gas and condensate are equal and
are given by the variation of Z-factor with pressure.
3. For calculation of condensate recovery and varying yield (producing oil-gas ratio)
during depletion it is critical to obtain accurate measurement of C7+ (rs) variation
in the produced gas from a constant volume depletion test.
4. For near-saturated gas condensate reservoirs producing by pressure depletion,

cumulative condensate produced is insensitive to whether the reservoir is
initialized with or without a compositional gradient (even though initial condensate
in place can be significantly different for the two initializations).
5. Oil viscosity should be measured and modeled accurately to properly model

condensate blockage and the resulting reduction in gas deliverability.
6. For richer gas condensates, the oil relative volume (from a constant composition
expansion test) has only a “secondary” effect on the modeling of condensate
blockage; for lean condensates, Vro has a small effect on blockage.
7. For gas cycling projects above the dewpoint, PVT properties have essentially no
effect on condensate recovery because the displacement will always be miscible.
Only the definition of initial condensate in place is important. Gas viscosity has
only a minor effect on gas cycling.
8. For gas cycling below the dewpoint, the key PVT properties are Z-factor variation
during depletion, C7+ content in the reservoir gas during depletion, and C7+
vaporized from the reservoir condensate into the injection (displacement) gas.
Nomenclature
Bo = formation volume factor (FVF) of reservoir oil phase
Bgd = dry gas FVF of reservoir gas phase
Cog = conversion factor for gas-equivalent of surface oil
C5+ = Pentanes-plus
C6+ = Hexanes-plus
C7+ = Heptanes-plus
C10+ = Decanes-plus
ES = sweep (vertical and areal) efficiency
EV = vaporization efficiency of condensate recovery below the dewpoint
krg = gas relative permeability
kro = oil relative permeability
Ki = equilibrium ratio (K-value) of component i
Mo = surface oil molecular weight
M7+ = C7+ molecular weight
N = total number
nd = moles of gas at initial (dewpoint) pressure
∆npk = incremental CVD moles of gas produced in stage k
p = pressure
pb = bubblepoint pressure
pd = dewpoint pressure
pi = initial pressure
ps = saturation pressure
psc = standard pressure
qo = surface oil production rate
q7+ = Heptanes-plus production rate
R = universal gas constant
rs = solution oil-gas ratio of reservoir gas phase
rsi = solution oil-gas ratio at initial pressure
Rs = solution gas-oil ratio of reservoir oil phase
RFgD = depletion recovery factor of gas
RFoD = depletion recovery factor of condensate
RFoDx = extra depletion recovery factor of condensate (after gas cycling)
RFoM = gas-gas miscible recovery factor of condensate
RFoult = ultimate recovery factor of condensate
Tsc = standard temperature
Vd = oil volume at dewpoint pressure
Vg = gas volume
Vo = oil volume
Vro = oil volume divided by oil volume at saturation pressure
Vt = total (gas+oil) volume
xi = oil composition
yi = gas composition
y7+ = C7+ composition in the produced gas
z7+ = mole fraction of C7+ of produced wellstream
zi = produced wellstream or total mole fraction
Z = Z-factor
Zi = Initial Z factor
o = surface oil density at standard conditions

ρ7+ = surface density of C7+ at standard conditions
µg = gas viscosity
µo = oil viscosity
REFERENCES
1. Lohrenz, J., Bray, B. G., and Clark, C. R. : “Calculating Viscosities of Reservoir Fluids
From Their Compositions,” J. Pet. Tech. (Oct. 1964) 1171-1176; Trans., AIME, 231.
2. Pedersen, K. S., and Fredenslund, Aa., : "An Improved Corresponding States Model for
the Prediction of Oil and Gas Viscosities and Thermal Conductivities," Chem. Eng. Sci.,
42, (1987), 182.
3. Whitson, C. H. and Belery, P. : "Compositional Gradients in Petroleum Reservoirs,"
paper SPE 28000 presented at the University of Tulsa/SPE Centennial Petroleum
Engineering Symposium held in Tulsa, OK August 29-31, 1994.
4. Lee, S. T., and Chaverra, M. : " Modeling and Interpretation of Condensate Banking for
the Near Critical Cupiagua Field," paper SPE 49265 presented for SPE Annual Technical
Conference and Exhibition held in New Orleans, Louisiana, 27-30 September, 1998.
5. Høier, Lars: " Miscibility Variation in Compositional Grading Petroleum Reservoirs,"
Thesis for dr.ing., Norwegian University of Science and Technology, NTNU, Nov.,1997.
6. Høier, Lars and Whitson, C. H. : “Miscibility Variation in Compositional Grading
Reservoirs,” paper SPE 49269 presented at the SPE Annual Technical Conference and
Exhibition held in New Orleans, Sep 27-30, 1998.
7. Fevang, Øivind and Whitson, C. H. : “Modeling Gas Condensate Well Deliverability,”
SPE Reservoir Engineering, (Nov. 1996) 221.
8. Coats, K. H. : "Simulation of Gas Condensate Reservoir Performance," JPT (Oct. 1985)
1870.
9. Fevang, Øivind and Whitson, C. H. : "Accurate Insitu Compositions in Petroleum
Reservoirs," paper SPE 28829 presented at the EUROPEC Petroleum Conference held
in London, Oct. 25-27, 1994.
10. Whitson, C. H., Anderson, T. F., and Soreide, I. : "C7+ Chracterization of Related
Equilibrium Fluids Using the Gamma Distribution," C7+ Fraction Characterization, L. G.
Chorn and G. A. Mansoori (ed.), Advances in Thermodynamics, Taylor & Francis, New
York (1989) 1.
11. Pedersen, K. S., Thomassen, P. , and Fredenslund, A. : "Characterization of Gas
Condensate Mixtures," C7+ Fraction Characterization, L. G. Chorn and G. A. Mansoori
(ed.), Advances in Thermodynamics, Taylor & Francis, New York (1989) 1.
12. Whitson, C. H. and Torp, S. B. : "Evaluating Constant Volume Depletion Data," JPT
(March, 1983) 610; Trans AIME, 275.
C.H. Whitson, Ø . Fevang, and T. Yang
Table 1 – Approximate Depletion Material Balance Calculations

Based on CVD and CCE Test Results.
CVD Data Conversion to Surface Oil and Gas Recoveries

Based on Simplified Surface Flash (Surface Gas = C6- and Surface Oil = C7+)
© PERA a/s, programmed by Curtis H. Whitson (19981126)
C7+ Mole Weight 161 kg/kmol

C7+ Density 830 kg/m3
Cog 122 Sm3/Sm3 (assumed constant)
(p/z)i/(p/z)d 0.9120 Approx.
Solution
Input (red) OGR
P Z np/n dnp/nd y7+ rs RFg RFo
bara % % mol-% Sm3/Sm3 % %
Pi 532 1.2172 0.000 3.996 3.407E-04 0.00 0.00
Pd 430 1.0788 0.000 0.000 3.996 3.407E-04 8.80 8.80
408 2.710 2.710 3.339 2.827E-04 11.29 10.87
372 7.070 4.360 3.366 2.851E-04 15.29 14.22
320 14.720 7.650 2.875 2.423E-04 22.35 19.24
272 24.420 9.700 2.245 1.880E-04 31.36 24.21
221 36.060 11.640 1.742 1.451E-04 42.22 28.83
170 49.130 13.070 1.302 1.080E-04 54.48 32.72
121 62.630 13.500 1.055 8.727E-05 67.17 35.97
62 79.160 16.530 0.675 5.562E-05 82.76 38.52
Table 2 – Approximate Depletion and Gas Cycling Calculations

Based on CVD and CCE Tests.
Data from CVD and CCE Calculated
Pressure rs Bgd Visg npw/nd Z-factor Vro RFoD RFgD RFoM RFoV
bara Sm3/Sm3 m3/Sm3 cp % % % % % %
900 8.07E-04 3.69E-03 0.062265 0.0 1.928 0.000 0.0 0.00 100.0 0.0
800 8.07E-04 3.83E-03 0.057694 0.0 1.778 0.000 3.6 3.64 100.0 0.0
700 8.07E-04 4.01E-03 0.053105 0.0 1.628 0.000 7.9 7.89 100.0 0.0
600 8.07E-04 4.24E-03 0.04843 0.0 1.477 0.000 13.0 12.97 100.0 0.0
500 8.07E-04 4.56E-03 0.043548 0.0 1.325 0.000 19.2 19.17 100.0 0.0
450 8.07E-04 4.78E-03 0.040966 0.0 1.249 0.000 22.8 22.84 100.0 0.0
398 8.07E-04 5.07E-03 0.038125 0.0 1.171 0.000 27.2 27.21 100.0 0.0
375 7.62E-04 5.21E-03 0.035959 3.3 1.138 2.581 29.5 29.43 96.0 4.0
350 7.09E-04 5.39E-03 0.033592 7.1 1.105 5.222 32.1 32.14 91.5 8.5
325 6.45E-04 5.61E-03 0.031119 11.4 1.075 7.973 34.7 35.22 86.3 13.7
300 5.75E-04 5.89E-03 0.028633 16.1 1.049 10.401 37.3 38.73 80.8 19.2
275 5.06E-04 6.24E-03 0.026314 21.2 1.027 12.042 39.8 42.64 75.8 24.2
250 4.46E-04 6.70E-03 0.024242 26.6 1.008 12.933 42.1 46.89 71.7 28.3
225 3.95E-04 7.28E-03 0.022415 32.5 0.993 13.297 44.3 51.45 68.4 31.6
200 3.52E-04 8.04E-03 0.020816 38.6 0.980 13.317 46.4 56.27 65.9 34.1
175 3.16E-04 9.06E-03 0.019427 45.1 0.970 13.108 48.4 61.35 64.0 36.0
150 2.86E-04 1.04E-02 0.018235 51.9 0.963 12.738 50.2 66.62 62.6 37.4
125 2.64E-04 1.24E-02 0.017226 59.0 0.958 12.246 52.0 72.06 61.7 38.3
100 2.52E-04 1.55E-02 0.016382 66.2 0.957 11.674 53.6 77.62 61.3 38.7
600
Surface Gas
500
Surface Oil
CVD (Reservoir) Pressure, bara
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100
Surface Gas and Surface Oil Recovery Factors
Fig. 1— Approximate Material Balance Calculations

Based on CVD Test Results.
0.6
Lean GC Rich GC
0.5 Normal Range for Near-Wellbore Blockage Region
0.4
krg
0.3
EOS Vro is 20% too low
0.2
EOS Vro is 20% too high
0.1
0.0
0.1 1 10 100
Vro = Vo / Vt of Flowing Mixture (=Produced Wellstream), %
(Vro not equal to Normalized Oil Saturation !)
Fig. 2 – Effect of error in Vro on gas relative permeability

in near-wellbore condensate blockage zone.
100
90
80 RFoV
Recovery of Condensate, % IOIP
70
60
RFoM
50 RFoDx
40
30
RFoD Initial
20 Pressure
10
0
0 100 200 300 400 500 600 700 800 900 1000
Pressure, bara
Fig. 3 – Condensate recoveries for pressure depletion and

gas cycling below the dewpoint in a high-pressure undersaturated reservoir.
100
Recovery of Condensate, % IOIP
80
RFoV
60
40 RFoM
RFoDx Initial
Pressure
=
20 Dewpoint
RFoD
0
0 50 100 150 200 250 300 350 400 450
Pressure, bara
Fig. 4 – Condensate recoveries for pressure depletion and

gas cycling below the dewpoint in a saturated reservoir.
Component EOS22 EOS19 EOS12 EOS10 EOS9 EOS6 EOS4 EOS3
N2 N2 C1N2 C1N2 C1N2 C1N2 C1N2
CO2 CO2 CO2 CO2 CO2
C1 C1 CO2C2 C02C2
C2 C2 C2 C2 C2 C1N2CO2C2-C6 C1N2CO2C2-C6
C3 C3 C3 C3 C3 C3
IC4 IC4
IC4NC4 IC4NC4 IC4NC4 IC4NC4
NC4 NC4
C3-C6
IC5 IC5
IC5NC5 IC5NC5 IC5NC5 IC5NC5
NC5 NC5
C6 C6 C6 C6 C6 C6
C7 C7 C7
C7C8
C8 C8 C8
C9 C9 C9 C7C8C9F1F2 C7C8C9F1F2 C7C8C9F1F2 C7C8C9F1F2
C10+ F1 F1 C9F1F2
F2 F2
C7C8C9F1-F8
F3 F3
F4 F4 F3-F5
F5 F5
F3-F8 F3-F8 F3-F8 F3-F8
F6 F6
F7 F7 F6-F8
F8 F8
F9 F9 F9 F9 F9 F9 F9 F9
Fig. 5 – Example pseudoization procedure reducing an original EOS

characterization with 22 components to multiple pseudoized characterizations.

Gas Condensate PVT

Uploaded by

Copyright:

Available Formats

Gas Condensate PVT

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gas Condensate PVT

Uploaded by

Copyright:

Available Formats

Gas Condensate PVT –

What’s Really Important and Why?

(a) Norwegian U. of Science and Technology (NTNU), (b) PERA a/s

Paper (© Curtis H. Whitson) presented at the IBC Conference

PVT properties important to the engineering of all gas condensate reservoirs

• Compositional (C7+) variation with pressure

Dewpoint pressure is implicitly defined by the pressure dependence of compositional

Compositional grading in gas condensate reservoirs may be important to the design

Consequently, it is important to determine which PVT properties are most important

Consider the following examples.

Example 1. A small offshore “satellite” reservoir with high permeability (kh=4,000

Example 2. A large offshore deep-water reservoir with moderate permeability

These three examples require significantly different emphasis in the treatment of

Constant Volume Depletion Test

Accurate measurement of the removed gas composition is very important to the

INITIAL FLUIDS IN PLACE AND DEPLETION RECOVERIES

Single-phase gas Z-factors are measured experimentally at reservoir temperature

It is not commonly recognized that condensate recovery is strongly related to the

As an example for a high-pressure reservoir, a +5% error in Zi and a –5% error in Z

Compositional (C7+) Variation During Depletion

Equations for converting CVD results to approximate surface product recoveries,

Compositional Variation with Depth

Another interesting feature of near-saturated reservoirs with compositional gradients

Flow in the near-wellbore region reaches a steady-state condition in a relatively short

Permeability reduction in the near-wellbore (steady-state) region is particularly

We recommend that the oil viscosity model be tuned to measured viscosities of a

What is important with respect to gas viscosities is that consistent viscosities be

Oil Relative Volume (Liquid Dropout Curve)

Two definitions of oil relative volume Vro are used,

Vro = Vo / Vt = Vo / (Vo+Vg) ..................................................................................... (11)

“average” oil saturation in a gas condensate reservoir during depletion, given

The importance of condensate blockage on well performance is dictated by the

Gas Cycling Recovery Efficiency in Swept Zone

Accurate prediction of the changing vaporization efficiency requires (1) an accurate

Evaluating Gas Cycling Potential

2. RFoM is the recovery of condensate which would be expected at the end of

(Es=100%); it is assumed the gas cycling occurs at a constant pressure pcycling

3. RFoV (=100 – RFoM) is the recovery of condensate which would be expected at

RFoult = RFoD + ES ⋅ RFoM + (1 − ES ) ⋅ RFoDx ............................................................ (12)

while for pcycling < pd,

RFoult = RFoD + ES ⋅ (RFoM + E V ⋅ RFoV ) + (1 − ES ) ⋅ RFoDx ......................................... (13)

What components of the ultimate recovery are strongly dependent on PVT

Fig. 3 shows recovery of condensate versus pressure for a high-pressure offshore

Combined Condensing/Vaporizing Mechanism

Whether a below-dewpoint miscible displacement can develop in a gas condensate

If we suspect or know that a sample is not "representative" (according to this

We strongly recommend against using this definition of "representivity." First of all, it

Any fluid sample that produces from a reservoir is automatically

Accuracy of PVT Data ,QVLWX5HSUHVHQWLYLW\RI6DPSOH

Accurate PVT measurements can be made on both representative and

Furthermore, an EOS fluid characterization is used to predict compositional changes

Another misconception in "representative" fluid sampling of gas condensates is that it

In summary, all “uncontaminated” samples collected from a reservoir are reservoir

· Bubblepoint or dewpoint pressure at a specified temperature

Splitting the Plus Fraction

Tuning the EOS Model

Developing a “ Common” EOS Model for Multiple Reservoir Fluids

Generating Black-Oil PVT Tables

The definition of black-oil PVT properties are:

Accuracy of PVT Data ,QVLWX5HSUHVHQWLYLW\RI6DPSOH