-
PETROLEUM SOCIETY OF CIM AND CANMET PAPER NO. 95-09
TOWARDS OPTIMIZING GAS CONDENSATE RESERVOIRS
F .B. ThomasX.L. Zhou
D.B. BennionD.W. Bennion
Hycal Energy Research Laboratories Ltd
ABSTRACT 2.3.
Viscosity ratioThe healing of fractures with its concomitant
effect onabsoluJe permeabilityIn the last year the authors have
fielded many
questions from companies, both international and
domestic,concerning gas condensate reservoirs. It appears that
gascondensates are becoming more important throughout theworld Many
international petroleum societies are beginning tohave conferences
specifically oriented to gas condensatereservoirs and discussing
all parameters germane to suchsystems. In light of this increased
interest, the authors havemade a short list of questions which are
most often asked.Indeed; these questions point to two specific
areas which governthe production and future exploitation plans for
gas condensatesystems. These two areas are characterization and
retrogradecondensate influences on relative permeability.
In order to adequately forecast such systems, asimulator must
incorporate these effects.
SAMPLING CONDENSATE RESERVOIRS
Condensate reservoirs are inherently more difficult
tocharacterize correctly. The literature shows many
differencesbetween gas condensate reservoirs and dry gas
reservoirs(I-6).One question often asked is during and after
sampling. Figure1 provides a fairly typical GOR versus total flow
rate responsefrom a gas condensate reservoir. One sees that, at
very lowflow rates, one has a high producing GOR and, beyond
thecertain minimum value in GOR, the trend is again upwards. Itis
easy to identify why this occurs, but sometimes, when facedwith the
possibility of having extra sampling runs and spendingmore time in
the field, the generation of a plot such as Figure1 is not
easy.
It has been found that the characterization of the gascondensate
fluid\' can be strongly influenced by two mainfactors:
1.
2.
Any degree of contamination by a free liquid phase
in-situ;Hold-up of the retrograde condensate in the
formationresulting in excessive producing GOR's. In the same plot
one compares the response which
would nonnally be seen for an oil reservoir. With the
oilreservoir, the sampling technique is fairly easy to specify.
Allone must do is try to produce the well in the domain lowenough
so that a constant GOR is produced. Since thebehaviour is
asymptotic as a function of decreasing total flowrate from the
well, it is easy to identify what production levelone needs to
apply for taking the gas and liquid samples. Suchis not the case
with gas condensate reservoirs however. At lowflow rates, as shown
in Figure I, one may be producing enoughliquid in the wellbore
that, unless the flow rate is high enough,the liquid hold-up will
increase and slugging may result. Insuch a case, the GOR, depending
upon at which interval thesample is taken, may fluctuate and result
in an excessively highGOR. The increased GOR to the left of the
vertical line is dueto the low flow rate not providing enough lift
to transport theliquids in the wellbore.
Care must be taken when sampling gas condensatewells in order to
produce representative recombinedfluids. Inorder to gain an
appropriate evaluation of the gas condensatereservoir one must be
able to adequately characterize the fluidsin-situ. Experimental and
theoretical work performed onevaluating retrograde condensate
effects has pointed to the factthat the influence of retrograde
condensate is much moredeleterious in tighter formations and higher
interfacial ftuw.The ability to identify the influence of
retrograde liquid on gasphase production rates is a difficult task
and data are providedherein which compare the retrograde condensate
effects atlwolevels of interfacial tension and as a function of
rockpermeability.
It has been found that in a review of four gascondensate
reservoirs, one of which included a fractured system,there was a
coupling of a multiplicity of factors including: By contrast, one
may be inducing liquid dropout in the
reservoir at high flow rates to the right of the vertical line
inFigure I. In such a case, as the pressure drops below
theInterfacial tension effects1
-
dewpoint, the liquid will drop out and begin to collect in
thenear wellbore region. In so doing, produced hydrocarbons
willcontain less liquid than they should and therefore the GOR
willbe high. Thus, when one asks the question, "At what
producingrate should a well be sampled?", the only way that one
canadequately respond is if one already knows the character of
thefluid and the dynamics of the production well. Since
thisinformation is not available, sampling gas condensate
reservoirsin an optimal manner can sometimes be non-linear and
includesome trial and error.
for GOR which in and of itself is also logical in tenns of
thehigher gas mobility as mentioned earlier. A more difficult
casewould be that shown 1n Figure 3 where there is a
substantialdeviation between the saturation pressure at the
separator GORand the reservoir pressure. In the experience of the
authors,often when the saturation pressure of the system shows
apositive deviation from the reservoir pressure of more than
1500psi, it is usually due to the fact that something has occurred
inthe liquid phase. Work has been perfonned by the authors inthe
past, wherein a very slight change in the heavy endcharacter of a
gas condensate system can result in significantchanges in the
saturation pressure. This again is much differentthan for a
conventional oil system. For example, if one has aslight mixing of
two different oil zones in the reservoir, theinfluence of the two
may be fairly insignificant on thesaturation pressure properties.
This is due to the fact that theoil is already a bubblepoint system
and, therefore thecontaminants which also have a tendency to remain
in the liquidphase may modify the bubblepoint behaviour, but
thecomponents themselves are both consistent with a
bubblepointresponse. On the other hand, if one were to have
acontamination due to a free oil phase in-situ, which exhibits
abubblepoint response, then the dewpoint behaviour of thedominant
gas condensate phase can be radically different. Thisis consistent
since one must increase the pressure of the systemto a high enough
level in order to change the behaviour of thecontaminant components
from a natural bubblepoint response toa dewpoint response. This is
tantamount to having to vaporizeall of the heavier components and
in order to do so a significantdeviation in pressure from the
reservoir pressure is oftenencountered. This can therefore be used
as a criterion by whichone can judge if the deviation from the
reservoir pressure is dueto a liquid phase contaminant or simply a
perturbation in thelevel of producing GOR.
How does it influence the overall characterization ifone is
producing in other than the optimal zone of theproduction scenario?
Figure 2 shows a typical responseobtained when recombining
separator gas and liquid from a gascondensate well. In this case,
the asterisk corresponds to theseparator GOR which was observed in
the field during thesampling. As is often the case, the
relationship in Figure 1was not scoped during sampling, and
therefore Figure 2 maycorrespond to producing the well in the
region to the right ofthe vertical line in Figure 1. Figure 2
indicates that thesaturation pressure is higher than the reservoir
pressure andtherefore something is wrong. For systems like this
where thedeviation between the observed saturation pressure of
therecom bination and the reservoir pressure is less than or equal
to1000 psi, one can normally trust that a manipulation of theGOR
will result in satisfying the necessary condition for
therecombination. The necessary condition is that the
saturationpressure of the recombined sample must be equal to or
less thanthe reservoir pressure at reservoir temperature. By
decreasingthe GOR by adding additional separator liquid, one
willsuppress the dewpoint down to a level which meets thereservoir
pressure criteria. Should the real dewpoint pressurebe lower than
the reservoir pressure, one really has no way ofknowing that unless
the owner was prepared to return to thefield and measure the GOR
dependency on the flow rate of thesystem. One may also observe
situations where, by increasingthe GOR slightly, the saturation
pressure may converge quicklyto the reservoir pressure. This is an
option but, in light of thefact that the gas phase is more mobile
than the liquid phase, thisis usually not the procedure which is
followed.
The ways of remedying the deviant behaviour such asthat in
Figure 3 is also more challenging. Obviously, once thecontaminating
liquid components have entered into thecondensate components, it is
extremely difficult, if not totallyimpossible, to separate the two.
Therefore, if that is the onlyseparator liquid obtained, which is a
combination of gascondensate components and oil components, then
one musteither be prepared to re-sample or reduce the GOR to a
lowenough value to simply meet the necessary condition of
thesaturation pressure being equal to the reservoir pressure. In
sodoing however, one will only exacerbate the
characterizationproblems. Two reasons are responsible. The first is
that bydecreasing the GOR, one will only emphasize the liquid
phase,and therefore if the liquid phase is the one which
iscontaminated, one will be emphasizing the contamination. Thiswill
serve to make the subsequent testing with this fluid
lessrepresentative of the bulk of the reservoir, assuming that the
oilphase is only a very small portion of the overall reservoir
and
Nevertheless, with the response as in Figure 2, themodified GOR
results in a saturation pressure equal to thereservoir pressure and
the recombined fluid properties can thenbe measured including such
things as overall composition,constant composition expansion data
as well as constant volumedepletion properties.
Figure 2 provides the best case scenario in terms of thecredence
which can lend to the resulting recombination. Themodified GOR
fluid still possesses all of the characteristicswhich are
consiStent with expectation with a slight modification
-
that the bulk of the reservoir is a gas condensate system.
Bydecreasing the GOR to a level where the Pial is equal to
thereservoir pressure, the doubt will increase as to whether the
dataproduced from any studies will then be meaningful.
in-situ. Remember that, although the reservoir was almosttotally
dominated by the gas condensate phase, the productionof the oil
phase was enough to result in approximately 100 kgof solids
produced per 1000 barrels of liquid phase recovered.Although in a
laboratory perspective this mass percentage isextremely small, from
an operating perspective it was more thanan annoyance.
Secondly, as one sees some liquid phase contaminationwhich
requires a distinct drop in the GOR for therecombination, one will
often see a response where thebehaviour will change from a dewpoint
to a bubblepoint.Therefore, the implications of this decrease in
GOR are veryserious. Moreover, the approach to be taken for
theexploitation of such a reservoir may also become obscured
inthat, rather than being a gas condensate system, it may becomea
light oil system, thereby bringing other questions into the
fraysuch as, "Would this be a better candidate for a gas
injectionprocess or waterflooding instead of primary depletion
orpossibly a gas cycling project?".
Another system recently analyzed showed a very waxycondensate
phase containing components as high as Cso'Because of the very high
temperature and pressure of thisreservoir, these components could
actually exist in the gas phaseat reservoir conditions and, upon
depletion and liquid phasedropout, the system exhibited vapour
liquid solid behaviour. Inthis case, die cloud point temperature
was exactly the same asthe melting point of the solid components
which werecentrifuged out of the liquid phase at the cloud
pointtemperature. In other words, it was a totally
reversiblephenomenon normally associated with nothing but a pure
waxsystem. That was a fairly unique fluid but. in that case.
therewas vapour liquid solid equilibrium with a system
whichexhibited a dewpoint pressure 200 psi less than the
reservoirpressure and no signs of contamination. The
differentiatingfeature in that case was that the molecular weight
of the solidphase was on the order of 450 and the solid was
reversible.For these reasons it is more credible that it was
associated witha gas condensate system with no free oil
contamination effects.
One other behaviour often associated with a gascondensate system
which shows liquid phase contaminationproblems is that there may be
some solid phase instability.Typically, there are very few solids
which will precipitate froma gas condensate system. Rarely one sees
diamondoidprecipitation and, although more frequently the
precipitation ofsulphur-containing compounds might be a problem,
one canusually trace those to the composition and the conditions of
theproduction string. The situations, however, showing dark
highmolecular weight solids being produced in the production
wellsare usually representative of solid phase components
beingprecipitated from the phase which is contaminating the
gascondensate components and is not an inherent problem with
thelight gas condensate components themselves. It is usually a
ruleof thumb that if the molecular weight of the produced solids
isgreater than 500 - 700 and the solubility of those solids is
highin an aromatic solvent, such as toluene, then there is a
goodchance that the solids originate with a separate oil phase
in-situ,and not only is there a problem with the characterization
of thegas condensate phase but the oil is also causing
productionproblems due to solid components incompatibilities.
In summary, therefore, when characterizing the fluidobtained
from sampling a gas condensate reservoir, one needsto be aware of
the following practices:
There will be an optimal producing flow rate for thewell in
question. There is little possibility of beingable to predict a
priori what the appropriate flow rateis for sampling the well. This
is due to the fact thatthe separator response will be a function of
theinterfacial tension, the viscosity ratio, the reservoirrock
characteristics, the tubing size and type as wellthe depletion
parameters of the gas condensate itself.The best possible thing to
do if one needs to obtain arepresentative characterization without
having torepeatedly return to the well, if the fluids are
notrepresentative, is to perform a GOR versus flow ratesequence
while sampling so that the best possiblesamples can be taken.
In the last two years, the authors have encountered twogas
condensate systems which exhibited solid precipitationproblems. One
had the response shown in Figure 4 wheresolids precipitated at high
temperature as long as the pressurewas high whereas at lower
pressures no solids were fonned,even at low temperatures. A
response such as this wouldusually be representative of a liquid
phase contamination of thegas condensate since the solids in this
case were pressure-dominated and therefore corresponded to an
asphaltene type ofdeposit. Since the vapour pressures of aspbaltene
are extremelylow, it is highly unlikely that these components are
producedfrom the gas phase in-situ and therefore this is a telltale
signthat the incompatibility was caused by a contaminating oil
phase
2 In recombining the separator gas and liquid to achievea
representative reservoir fluid one must be verycautious about how
the GOR is being adjusted. Theapproach taken by the authors is that
as long as thedownward adjustment in the GOR does not change
thebehaviour of the fluid from a dewpoint to a
't
-
bubblepoint then the affect of that adjustment on thefluid phase
behaviour will not be significant. Indeed,if there are significant
changes in the liquid dropout bydoing this GOR modification then
equation of statetechniques can be used to conduct sensitivity
tests tosee how the different characterization will impact
thelong-term reservoir forecasting including theeconomics of the
system.
of the capillary pressure equation, which can be
viewedintuitively as equation 1, wherein in order to sweep through
apore in a two phase scenario, the differential pressure must
beequal to or exceed the capillary pressure.
AP ~ p ~ JDtmaci8/ T19"oo~ /U",. (1)
3 The saturation pressure deviation from reservoirpressure
should not usually be in excess of 1000 psi.If it is, then one may
expect to see some liquid phasecontamination and therefore the
resultingrecombination may bear no resemblance to the actualfluid
which is in-situ in the reservoir. Many timeswith a system which
exhibits liquid phasecontamination, a change from dewpoint to
bubblepointis evoked by decreasing the GOR to meet thenecessary
condition of P 181 equalling reservoir pressure.One needs to be
very vigilant and careful in systemsthat are like this.
4. If there are solid instabilities then this is usually astrong
indicator that a liquid phase contamination isoccurring in-situ,
particularly if the molecular weightof the solid components is high
and the solid phaseresponse is more sensitive to pressure than
totemperature and is irreversible.
RETROGRADE CONDENSATE EFFECTS ON RELATIVEPERMEABILITY
One can observe from Equation I that as the interfacialtension
decreases, the capillary pressure decreases. Converselyas the
radius of the pore throat which contains the retrogradeliquid
decreases, the capillary pressure holding the liquid in thepore
increases. Therefore, to be able to produce retrogradeliquid from
small pore throats, one must have either a very highdifferential
pressure driving force or low interfacial tension.
Extrapolations of this thinking would indicate,therefore, that
for gas condensate systems which exhibit highinterfacial tensions
where the pore throats are very smaIl, whichmay correspond either
to low permeability rock or higherpermeability rocks but with very
large coordination number, thesuccess of flowing the liquid from
the rock, once it hascondensed, will be limited. In such cases
vaporization (lean gascycling) or injection of 1FT reducing agents
(CO2) may be theonly option to enhance the performance. On the
other hand ifthe equilibrium gas and liquid exhibit low interfacial
tension,then the liquid may flow freely from most of the pore
throatsin the rock and very little retrograde condensate
relativepermeability reduction will be observed.
Conversely, where the pore throat diameters are muchlarger, even
though the interfacial tension may be high, it maybe easy to
overcome the capillary forces which are keeping theliquid in those
pore throats. Therefore, for larger pore throatsyStems which may
correspond to higher permeability rocksand/or smaIl coordination
numbers, the interfacial tension effectmay not be very
important.
Unfortunately, other factors cloud the issue. One ofthe factors
which complicates this development is the mobilityeffect. If one
were to analyze the viscosity ratio between mostequilibrium gas and
condensate systems, the viscosity of the gaswould normally be at
least 10 - 20 times lower than that of thecondensate phase. Due to
the inherent nature of the less viscousphase to flow more readily,
it will tend to take the path of leastresistance and will
preferentially flow through the larger porethroats. Gardnef1>
has shown correlations wherein forgas/Iiquid flows, exponential
viscous finger growth is oftenseen. Therefore in light of the 1FT
criterion (equation I), onemay automatically conclude that the
larger pore throat sizes willcontribute to much easier production
from a reservoir whichexhibits liquid condensate effects. This is
usually true.However, even though the fluids may have the capacity
to flow
Retrograde condensation results in a number ofproblems. The most
obvious and serious of these is lostproductive capacity due to
accumulation of liquid in thereservoir. This has two facets: the
first is associated with notbeing able to produce the higher value
liquid components andsecondly, the increased liquid saturation
results in reduced gasflow rates. These factors work in concert and
the more seriousthe liquid dropout, the greater the reduction in
gas relativepermeability .
For a system which is single phase initially,
relativepermeability effects are absent. Relative permeability
should beviewed as a dependent variable determined by three
otherparameters or influences. These general influences
areassociated with:
I. interfacial tension effects2. viscosity ratio3. pore size
distribution
By definition. 1FT effects are only involved when twophases are
present. The interfacial tension is important because
4
-
In this case one sees the expected effect of a reduction in
gasrelative permeability very severe for the lowest
penneabilitysystem and least severe for the highest penneability
core. Thetechniques for measuring retrograde condensate themselves
arevery difficult from a laboratory perspective since all of
theinfluences which were mentioned earlier are at play.
Figure 6 shows an example of the influence of liquidsaturation
on the relative permeability. In this test which wasperfonned by
the authors for a very extreme condition reservoirfluid one sees
the drastic impact of small liquid saturations onthe gas
productivity as well as the influence of the interfacialtension. In
this case, the interfacial tensions of the fluids weremeasured
using the machine vision pendant drop technique asshown in Figure
6. This technique has been described byRotenberg('). Using this
technique, interfacial tensions down to10-2 dyne/cm can be measured
quickly and consistently.
With the interfacial tension at a level of 0.2 dyne/cm,one sees
a slightly lower critical condensate saturation than atthe 2
dyne/cm system. Moreover, the influence of the wr onthe endpoint
saturations and relative penneabilities is also asintuitively
expected. The influence of the critical condensatesaturation may be
very important depending upon the degree ofretrograde condensation
effect. For example, if the maximumretrograde condensation effect
is lower than the criticalcondensate saturation then the only way
in which the gasproductivity can be remedied is by extracting the
components.Thomas et a1(IO) defmed extraction effects for systems
such asthis. For retrograde condensate levels higher than the
criticalcondensate saturation one may be able to consistently
mobilizeand produce the condensate saturation above the
criticalsaturation. This therefore points to the importance of
beingable to accurately define the liquid dropout as well as
thecritical condensate saturation.
through most of the rock where the pore throats are larger,
thegas may preferentially "choose" only the largest pore throatsand
may bypass the rest. Therefore even though the gas mayhave a low
enough 1FT to potentially sweep all pore throats, itmay, due to
mobility effects, only contact those of largerdimension. Therefore
a compromise will be reached betweeninterfacial tension and
mobility. If the system ismobility-dominated. then the only way to
effectively reduce theliquid saturation may be to vaporize
components from the liquidinto the flowing gas phase. Whether the
free liquid will flowor improved recovery will have to rely on
extraction effectsdepends on the compromise reached between IFf and
mobilityin the presence of the actual porous media.
In order to assess the optimal way to produce a gascondensate,
one must be prepared to perform the testing in thepresence of the
porous media. If the actual reservoir rock is notused, one may have
the appropriate viscosity ratio andinterfacial tension, but the
conclusions drawn may beinappropriate in light of the fact that the
compromise betweenmobility and 1FT is not the one which would be
consistent withthe reservoir rock.
It should be noted at this point that in light of thecoupled
nature of interfacial tension, mobility and the pore
sizedistribution of the porous media that laboratory testing
mustadequately represent each of these three factors. If any
portionof laboratory analysis does not adequately represent
theseparameters, then the credence which one can lend to
theconclusions drawn from a more routine laboratory study may
beminimal. That is, if viscosity ratio is used as a reference
foratmospheric relative permeability testing, but the
interfacialtension of the fluids is not matched. then one has a
priori biasedthe conclusions in favour of the system being
mobilitydominated.
One of the standard tests which is often applied todetennine the
response of a gas condensate system is theconstant volume depletion
test wherein the liquid phase as afunction of pressure is measured.
On this basis one candetennine if the liquid is of high enough
volumetric proportionto cause a problem. Usually, those fluids
which exhibit lessthan I % retrograde condensation show very little
tendency toreduce gas ~. However, in some cases, even at low
liquiddropouts, the liquid tends to migrate into the
productionwellbore and result in reductions in gas penneability.
Theevaluation via CVD, however, introduces one to how serious
theproblem may be and initiates the overall evaluation on
alaboratory scale.
The liquid dropout is fairly easy to do since it is aroutine
phase behaviour experiment. However, to determine thecritical
condensate saturation is much more difficult. Thecritical
condensate saturation is defmed as the condensatesaturation below
which the condensate will not be able to moveor to be moved and
above which the condensate will be mobile.Therefore, to measure the
critical condensate saturation strictlyone must deplete the
dewpoint fluid in the representativereservoir rock and then
subsequently flow equilibrium vapourthrough the specimen. Depending
upon the pore volume of thesystem, the dead volumes of the
experimental apparatus andinherent nature of the liquid phase being
dropped out, this canbe an extremely difficult procedure. If the
condensate is mobilebut is trapped in the accessories of the
experimental apparatus,one may have too high of critical condensate
saturation. Onemay then possibly get involved in a gas cycling
scheme when,in fact, the critical condensate saturation is lower
and would not
Following the perfonnance of a constant volumedepletion test one
often will quantify the influence of theretrograde condensate on
the gas relative penneability. Figure5 provides an example of some
data obtained by the authors.
s
-
necessitate a cycling operation. Because of the
inherentdifficulties in measuring this, some laboratories have
judged theresidual condensate saturation to be the same as the
criticalcondensate saturation. This, however, is not the case.
This brings us to another point wherein themeasurement of
retrograde condensate effects can be verysensitive. This is mainly
the case where the differentialpressure of the system is fairly
high. For a large differentialpressure, the thermodynamic behaviour
may be coupled in withthe fluid flow behaviour. Figure 9 shows a
comparison betweenan equilibrium assumption versus anon-equilibrium
assumption.For the equilibrium assumption following each saturation
levelusing the analog methane binary, the core was
depressurized.The gas was collected and the gas composition was
used tointerpolate the gas phase saturation. The pressure
effect,however, resulted in a compression of that gas phase over
whatwould normally be present if there had not been a
substantialdifferential pressure. It should be noted that in this
case thedifferential pressure was as high as 300/0 of the
backpressuresetting. Therefore, by including a compression effect
of thevapour phase only, the non-equilibrium values were
generatedin Figure 9. Compared to that, are the values calculated
fromthe equation of state based on the produced gas compositionupon
depressurizing the core. One can see that although theyare
different there is a reasonable comparison between the two.Knowing
this may be particularly important for some of the gascondensate
relative permeability testing that may have to bedone very close to
the saturation pressure corresponding to verylow interfacial
tensions. For example, if an 1FT required wasin the order of 0.1
dyne/cm which meant that the operatingpressure had to be within 50
psi of the saturation pressure, thenthe differential pressure would
result in a pressure higher thanthe saturation pressure. Therefore,
knowing the comparison andthe reasonable response between the
equation of state calculatedvalues and the non-equilibrium
assumption is an importantparameter to note for those performing
the experimentation.
By a residual condensate saturation, one defines thisvalue by
filling the specimen completely with a bubblepointequilibriwn phase
produced from the depletion experiment. Byfilling the core
completely with that phase, one has filled all ofthe pore throats
and pores with the liquid. Upon injecting thegas, even though it is
an equilibrium gas, the pore throats,which are so small that the
Iff between the gas and the liquidpreclude entry, will still be
filled with the condensate. In otherwords, it will only be a judge
of which pores the equilibriumgas has been able to sweep. For the
pore size distribution givenin Figure 7 at a level of 1FT which
corresponds to 20 micronsand the obviously artificial pore size
distribution contained inFigure 7 (for explanation purposes), the
residual condensatesaturation will be associated with all of the
pores lower than 20microns. In such a case, the critical condensate
saturation, ifmeasured as a residual condensate saturation, could
be veryhigh; the residual condensate saturation will be an
indication ofthe saturation of the condensate remaining in the core
that couldnot be swept.
If one compares that to the critical condensatesaturation, when
the condensate saturation builds up to a certainvalue, then the
fIrst sign of production of that condensate wouldbe called the
critical condensate saturation. This does not meanthat all of the
condensate will move from the core but onlyrepresents the
saturation at which the condensate has becomemobile. Necessarily,
therefore, the critical condensate saturationis going to be
governed by the gas condensate response in thelargest pore throats
whereas the residual condensate saturationis going to be governed
by the gas liquid interaction in thesmallest pore throats. The
critical condensate saturation will bethe most optimistic value and
the residual condensate saturationwill be the most pessimistic
value. Otherwise stated, the criticalcondensate saturation will
indicate the lowest saturation thatneeds to be built up before the
condensate in the largest porethroats will be mobilized. This,
however. does not mean thatthe condensate in the lower pore throats
will be mobilized and,in fact, will not be. Indeed, as the
condensate phase begins tomigrate throughout the rock one may
result in the net effectassociated with the residual condensate
saturation being morerepresentative than the so-called critical
condensate saturation.At a level of 1FT the smaller pore throats
will have thepossibility of imbibing the liquid phase and therefore
once thesmaller pore throats are filled with the condensate there
may beno possibility other than through gas cycling or through
1FTmanipulation of accessing those pore throats and mitigating
theinfluence of retrograde condensation on the
relativepermeabilities.
SIMULATION APPLICATIONS
The ability to produce representative experimental datais
inherently important. Nevertheless, this importance would
bediminished significantly if there was no ability to couple
thesedata into making forecast estimates for gas condensate
reservoirperformance. The constant volume depletion data are
normallyreadily imported into a simulation model. The ability,
however,to couple the fluid phase behaviour with the fluid
flowcharacteristics is more challenging from a
simulationperspective and one must be very aware of some of the
factorswhich need to be present in a simulator. A quick review
ofthese parameters, as mentioned earlier, are:
Influence of interfacial tension on relativepenneabilities.The
inclusion of viscosity/IFT compromise in therelative
penneabilities.An adequate means of representing the liquid
dropout.
2.
3.
6
-
4. 3 Retrograde condensate effects on relative permeabilityhave
been shown. These effects become more seriouswith increasing
interfacial tension and decreasing rockpenneability. This effect is
determined by the interplay ofinterfacial tension, pore size
distribution, viscosity ratioand, to some degree, connate water
saturation.
The influence of pore volume compressibility as afunction of
pressure.For fractured systems, the mechanism wherebyfractures heal
as the reservoir is depleted and the netoverburden pressure
increases.
s
The authors have worked in the last year with a systemwhich
required all of these parameters. For instance, upondiscovery and
initial depletion. the behaviour was above thedewpoint and
corresponded mainly to a simple gasdecompression problem. Once the
dewpoint pressure wasreached and liquid phase dropout commenced
then this had aninfluence, based upon the retrograde condensate
effect. Notonly did that occur but simultaneously with the decrease
inpressure the 1FT began to increase. This increasing 1FTresulted
in more drastic relative permeability decreases. In areverse
synergistic effect, the decreasing pressure and increasing1FT also
resulted in decreasing absolute permeability due to thefractured
nature of some of the systems which the authors havemodelled. This
calls into play, therefore, an influence in theabsolute
permeability of the model along with an ability tointerpolate
between re lati ve permeabiIities corresponding to highand low 1FT
regimes. The authors found it necessary to havethe simulator used
(Computer Modelling Group compositionalsimulator GEM) modified to
include all of these effects andevery one of these effects was
required in order to get anadequate response of the field data.
4, Detennination of retrograde condensation effects onrelative
penneability is a challenging measurement tomake experimentally and
some procedural concerns areidentified with possible solutions.
ACKNOWLEDGEMENTS
The authors would like to express appreciation to thepersonnel
at Hycal Energy Research Laboratories Ltd. for theirtime and effort
in producing the experimental results for thispaper.
REFERENCES
Dake, L.P.: "Fundamentals of Reservoir Engineering",Elsevier
Publishing, Oxford (1978).
2. craft, B.C. and Hawkins, M.F.: "Applied PetroleumReservoir
Engineering", Prentice-Hall Inc., EnglewoodCliffs, NJ, (1959)
Chapter 2.Only in using a simulator with this much detail and
based upon detailed experimental data can one hope to close
thegap between the initial characterization and the initial
hopesassociated with discovery and the ultimate dollar value
whichcan be realized from such reservoirs.
Gilchrist, R.E. and Adams, J.E.: "How to Best UtilizePVT
Reports", Petroleum Engineer International (July1993).
3,
Moses, P.L.: "Engineering Applications of PhaseBehaviour of
Crude Oil and Condensate Systems", JPT(July 1986) pp. 715-723.
4CONCLUSIONS
Characterization of the gas condensate is often a
verychallenging endeavour. Many times the separator GOR' swill be
too high depending upon whether the well is beingproduced at too
high or too Iowa flow rate.
s, Strong, J., Thomas, F.B., Bennion, D.B.: "Reservoir
FluidSampling and Recombination Techniques For
LaboratoryExperiments", paper presented at the CIM 1993
AnnualTechnical Conference in Calgary, May 9-12, 1993.
2. Techniques for modifying GOR difficulties are routinewhereas
if there is any contamination of the producedliquid phase these
problems require resampling. A rule ofthumb for saturation pressure
deviation from reservoirpressure is defined. Usually, if a GOR
manipulation isrequired the saturation pressure is normally less
than 1000psi higher than the reservoir pressure. If there is
greaterthan 1000 psi deviation from reservoir pressure thennormally
this is associated with a liquid phasecontamination problem.
6. McCain, W.D., and Alexander, R.A.: "Sampling GasCondensate
Wells", Society of Petroleum EngineersReservoir Engineering (Aug.
1992) pp. 358-362.Gardner, J. W." Ypma, J.G.J.: "An Investigation
of PhaseBehaviour - Macroscopic Bypassing Intersection in
CO2Flooding", SPE 10686, 1992 SPE DOE Symposium onEnhanced Oil
Recovery, Tulsa, OK., April 4-7, 1992.
7.
'f
-
8 Rotenberg, Y., Boruvka, L., Neumann, A.W.:"Determination of
Surface Tension and Contact Anglefrom the Shapes of Axisymmetric
Fluid Interfaces",Journal of Colloid and Interface Science, Vol.
93, No.1(May 1983).
9. Bennion, D.B., Thomas, F .B.: "Recent Improvements
inExperimental and Analytical Techniques for theDetermination of
Relative Permeability Data fromUnsteady State Flow Experiments",
presented at the SPE10th Technical Conference and Exposition held
in Port ofSpain, Trinidad, June 26-28, 1991.
10. Thomas, F.B., Holowach, N., Zhou, X.L., Bennion. D.B.:"Optim
izing Production from Gas Condensate Reservoirs"CIM Paper 94-04,
1994.
-
IINiIi,~I~II~.~
'.tit!~ ,
~-
/ \..I
~~
--
tl .
.IIII'..7
I0
],~fa
Iila;I~il~II
\."'"
~.G
J.. I
. I
-
1-
.1s~~Iti;J-!IIII~
-.:_
~~
-~
1111
irlil
=IQ~i0000-II.1.fc!
I.~I~IiI~il~I~i4~
tlt~l~
~,,~ III,1IIII
.J "l
-I
iIIJ.-
.
IIIIIJII;.
I!
-
Ii
;1~ t
~I~
I