P V T analysis has been routinely used by reservoir engineers to characterise the physical properties of a reservoired fluid as well as the changes in volume and phase state occurring during production (Pedersen et al., 1989). This characterisation is gen- erally performed using software packages which cal- culate P V T and phase behaviour based on fluid composition, as determined in specialised labora- tories on reservoir fluid samples at reservoir press- ure and temperature, using equations of state (EOS). These EOS provide a mathematical descrip- tion of the fluid behaviour for reservoir simulation and reserve estimation.
The phase behaviour of a fluid is commonly described using either specific physical properties of the fluids such as formation volume factor (Bo), sat- uration pressure (Psat) or gas to oil ratio (GOR), or by means of a pressure vs. temperature diagram, as shown in Fig. 1. The phase envelope shown in this figure consists of a region where fluids occur in a single phase state and a region where they exist as two separate phases. The latter one is enclosed by a bubble point curve and a dew point curve. The bubble point curve marks the PT-conditions where separation of a gas phase from a supercritical liquid phase takes place, while the dew point curve is defined as the PT-area where separation of a liquid phase from a supercritical gas phase occurs. The critical point, located where bubble point and dew point curves meet, characterises fluid conditions in-
*To whom correspondence should be addressed.
termediate between those of a liquid and a vapor phase. P V T simulation software, using EOS, predict such phase envelopes. It must, however, be taken into account that EOS predictions are usually only fitted to reservoir pressure and temperature. The rest of the phase envelope is a prediction whose ac- curacy is difficult to assess as measurements are not commonly made at these pressures and tempera- tures. The shapes of the phase envelopes predicted depend strongly on the composition of the fluid analysed, and hence offer a simple graphical rep- resentation of a complex dataset.
The strongest compositional influence on subsur- face fluid phase behaviour is the quantity of light hydrocarbons (gas) dissolved in the liquid phase (oil), i.e. the gas to oil ratio (GOR). The GOR is defined in this paper as the proportion of gas to oil of the reservoir fluid under surface conditions (1 bar, 15°C). Calculation of GOR using the P V T software was performed in all cases as a flash of the reservoir fluid to surface conditions (i.e. instan- taneous change in pressure and temperature per- formed mathematically). This type of GOR calculation is generally within 10% of the field GOR, and since GOR measurements in the field are dependant on a series of variables (production rate, choke size, etc) it is assumed to be accurate enough for the purposes of this paper.
The concentration, density and molecular weight of individual compounds or compound classes of petroleum also have a strong effect on the phase behaviour of petroleum systems. Therefore, in order to characterise the phase relationships of a pet- roleum fluid correctly, precise knowledge of the pet-
207
208 R. di Primio et al.
400,
300-
P/Bar 200 -
100-
Temperature / C
Fig. 1. Phase envelope nomenclature. The area enclosed by the dotted lines represents the extent of naturally occurring PIT conditions of reservoirs.
roleum composition is necessary. The data used for P V T analysis consists of detailed gas composition (molar contents of methane, ethane, propane, n- and i so-butane , n- and iso-pentane) and a bulk characterisation (molar abundance, molecular weight and density) of the C6 to (commonly) C9 boiling ranges as well as of the residual fraction of the petroleum (called plus fraction, i.e. C10+). An example of such a set of input data is shown in Table 1 for a normal North Sea black oil.
The P V T software package used for all phase behaviour calculations in this study (PVT-Sim) uses the Soave-Redlich-Kwong EOS (Soave, 1972). To be able to use this EOS on an oil or gas condensate the critical temperature (To), critical pressure (Pc)
Table 1. Example of the input data necessary for PVT-Sim calcu- lations
Compound Molar (%) Molecular weight Density
N2 1.1 CO 2 0.22 Ci 36.8l C 2 8.69 C3 8.39 i so 'C 4 1.19 n-C4 4.21 iso-C5 1.35 n-C5 2.03 C6 2.61 C7 4.02 C8 4.12 C9 2.06 Cio+ 23.2
92.9 0.732 106.2 0.755 118.4 0.777 269.0 0.869
and acentric factor (~o) of each component of the fluid must be known. For the well defined com- ponents (N2, CO2, C1, C2, C3, iso-C4, n-C4, iso-Cs, n-Cs, n-C6) accurate measured data are available. The input for the higher molecular weight com- pounds, however, represents averaged data for boil- ing ranges or retention time ranges. For example the molar proportion of C7 includes all samples eluting after n-C6 up to and including n-CT. Each of these higher molecular weight fractions thus include many different compounds, each having different critical temperatures, critical pressures and acentric factors. The software package characterises these mixtures' properties using correlations developed by Pedersen et al. (1989) based on the molecular weight and density of the fractions at surface P and T conditions. The predictions of the software (sat- uration pressure, GOR, formation volume factor, etc.) for black oils and condensates usually have an error margin of around 5%. A near perfect fit of predicted and measured data can be achieved by "tuning" the prediction using the molecular weight of the plus fraction as an adjustable parameter (see Pedersen et al., 1989 for an extended discussion of EOS tuning). This type of tuning was not per- formed on the data presented in this study since the predictions were consistently good. Care was taken to ensure the high quality of the data used, es- pecially with respect to the determination of the molecular weight of the plus fractions. Two pro- cedures are common for the determination of
PVT and phase behaviour 209
2,5
. • 3,5
o
% Surface gas expelled
20 40 60 80 100
•t9
4.5
Fig. 2. GOR evolution with maturity for an open system (modified after England and Mackenzie, 1989).
molecular weight, the most accurate is based on an actual measurement (e.g. freezing point depression of a pure solvent), whereas determination by means of mathematical extrapolations are less accurate. Data where the molecular weight of the plus frac- tion had been determined through extrapolation was not used in this study.
The amount and composition of gas in the fluid strongly influence its saturation pressure and GOR, whereby a methane-rich gas composition results commonly in a higher saturation pressure than a comparably "wetter" gas composition. Molecular weight and density of the plus fraction have a strong influence on the location of both cricon- dentherm and critical point. Additionally they in- directly influence the GOR and saturation pressure. Fluids with high molecular weights and densities are characterised by a high cricondentherm, a low GOR and low saturation pressure. The margin of error in the determination of molecular weight and density is usually below 1%. The influence of this error margin on the data predicted by P V T soft- ware is very small for the cricondentherm ( < 2 % ) whereas for GOR, Psat and Bo variation of molecu- lar weight over _+1% leads to shift of between 5 and 10% of the predicted values. Hence the cricon- dentherm temperature shows the lowest sensitivity to the data error-margin.
England et al. (1987) and England and Mackenzie (1989) discussed the evolution of GOR with maturation in detail. They stated that in the course of maturation the composition of a gener- ated petroleum changes systematically, depending also on the predominant kerogen type in the source rock. An example of these changes with maturity is shown in Fig. 2 for the GOR of a source rock con- taining marine type II kerogen (England and Mackenzie, 1989). As shown in Fig. 2, the GOR of petroleum increases systematically with (laboratory simulated) source rock maturity. If source rock and reservoir are in continuous communication during source rock maturation the GOR of the petroleum accumulation reflects the cumulative GOR of the products expelled from the source rock. Cumulative GOR relationships are shown in Fig. 3 (England
and Mackenzie, 1989). A sporadic source-reservoir connection could lead to the isolation of individual maturity intervals resulting in a GOR maturity dependence as shown in Fig. 2 (England and Mackenzie, 1989). Both trends are, of course, only valid for a single source rock-reservoir system, i.e. assume that no mixing occurs with oils or gases from other source rocks.
Since the GOR has a strong control on fluid P V T relationships the phase behaviour of fluids generated can be expected to change systematically with increasing maturity. Additionally, updip fluid migration or uplift of trapped fluids can be expected to affect the phase envelope of an original fluid, since it will reequilibrate by degassing and precipitation of oversaturated compounds from sol- ution to its new pressure/temperature conditions (Sokolov, 1964; Thompson, 1987, 1988). This paper attempts to characterise systematically the changes of fluid P V T and phase behaviour associated with maturation and migration or uplift and demonstrate the use of P V T analysis in a regional petroleum exploration context. All interpretations presented here are based on the analysis of an extensive data- base of P V T measurements from the entire Norwegian Continental Shelf as well as on exper- imental data.
The analysis of phase behaviour of oil and gas samples from the North Sea was performed using " P V T - S i m " version 7.0, a P V T and phase beha- viour simulation software system (CALSEP A.S.- DewPoint A.S.). The database used in this case con- sisted of over 100 samples, including both oils and gas/condensates. All data input into the program was derived from P V T analyses performed on natu- ral samples from exploration wells by the operating companies and covered most of the compositional variation observed in the Norwegian North Sea area.
THEORETICAL BACKGROUND
The GOR of petroleum generated from a source rock is known to increase systematically with increasing source maturity. This effect has been
0,1
3.00
3.50
A 4.00 g 4.50
o 5.00
5.50
6.00
Expelled GOR (kg/kg)
0,2 0.3 0.4 0,5 0,6
i iii/ii!:<+ ~: :/~: ~ i
Fig. 3. GOR evolution with maturity in a closed system (modified after England and Mackenzie, 1989).
210 R. di Primio et al.
//" 8 5 4 3 2 1
dr/gas Gas~/indo;wet~ga~ Ita~ Oil~do~j~y [ .... : , Temperature ~ z >
Fig. 4. Proposed evolution of phase envelopes of pet- roleums generated from a marine source rock based on
theoretical considerations discussed in the text.
recognised and described in a multitude of publi- cations based both on natural datasets as well as on laboratory experiments (e.g. Hughes et al., 1985; England and Mackenzie, 1989; Dueppenbecker and Horsfield, 1990; Karlsen et al., 1995). The molecular weight and density of petroleums of increasing maturity also change in a regular fashion, a fact which is again well known from natural systems and from artificial maturation experiments (Tissot and Welte, 1984; Hughes et al., 1985; Baskin and Peters, 1992; di Primio et al., 1993). GOR, molecu- lar weight and density of petroleums are the main factors controlling phase behaviour (Pedersen et al., 1989). High GOR oils are generally characterised by bubble point curves extending into high press- ures. Petroleums with high proportions of high mol- ecular weight material (and high densities) are characterised by dew point curves extending into high temperatures in the P I T diagram. For liquids, a systematic increase in GOR and a decrease in molecular weight and density (which as stated above occurs in the course of source rock matu- ration) can be expected to result in a shift of the phase envelope from a "high dew-point tempera- ture/low bubble-point pressure" biased shape to a "high bubble-point pressure/low dew-point tem- perature" biased shape with increasing maturity (phase envelopes 1-5, Fig. 4).
For gases and condensates generated at higher maturities from the source rock the phase envelope consists commonly only of the dew-point curve. As gas composition shifts towards a methane domi- nated mixture, i.e. becomes "drier", the dew point curves shrink again back towards lower pressures and temperatures (phase envelopes 6-8, Fig. 4).
This theoretical evolution of phase envelopes of fluids generated from a maturing source rock was tested in the first instance against a laboratory data- set from two closed system artificial maturation ex- periments. Micro-scale-sealed-vessel pyrolysis (MSSV, Horsfield et al., 1989) was performed on
one immature type II source rock sample from the upper Devonian Duvernay Formation, located in the western Canada sedimentary basin. The exper- imental procedure has been described by Horsfield et al. (1992). 30 aliquots of each sample were sealed in glass tubes and heated at 0.1 K/m up to a final temperature of 560°C using the furnace described by Schaefer et al. (1990). In the course of the heat- ing program sample tubes were removed from the furnace roughly every 10°C. The composition of the generated fluids was determined by a single step on- line gas chromatographic analysis (Horsfield et al., 1989). Fluid composition was quantified by means of an external standard (n-butane) and the compo- sition input into a P V T software package tailor- made for this type of experimental data (Geoflash, a modified version of P V T S i m , CALSEP A.S.). Input consisted of detailed compositional infor- mation, as exemplified in Table 2, as well as the molecular weight and density of the C7+ fraction. In contrast to produced oils and natural source rock bitumen, whose composition is affected by mi- gration and sampling phenomena, these artificial products are thought to be better indicators of pet- roleum generated originally from source rocks at increasing maturities.
The physical properties of these hydrocarbon mixtures are calculated in the Geoflash program from a combination of Cl-40 single resolved com- pounds as well as the C7+ totals (including resolved and unresolved compound mixtures). While the properties of the resolved compounds are known in detail from the literature, the properties of unre- solved compounds have to be calculated according to their density and average molecular weight.
The transition from kerogen to primary formed liquid and gaseous hydrocarbons is marked by a progressive decrease of average molecular weight and an increase of the GOR at higher conversion temperatures.
Since the MSSV technique does not produce the amount of pyrolysate required to physically deter- mine the average molecular weight and density of the C7 + totals a series of assumptions and simplifi- cations had to be made for the estimation of these properties from the gas chromatographic analysis of the individual heating steps.
In order to calculate the average molecular weight of the C7+ totals the GC trace of this frac- tion was divided into 7 different areas representing boiling ranges. The average molecular weight of the identified hydrocarbons within each of the seven boiling ranges was assumed to be representative of the molecular weight of the unresolved compounds. Taking into consideration the yields (quantified by external standardisation) of these different boiling ranges the average molecular weight of the C7+ totals could be calculated with respect to increasing thermal maturity.
P V T and phase behaviour 211
Table 2. Example of the input necessary for "Geoflash" calcu- lations
Molar weight Molar amount (g/mol) Density (g/cm 3)
The determination of the average density of the total C7+ mixture was performed through elemental analysis of the C7+ compounds at different matur- ity steps going from low to high conversion tem- peratures. Whilst aliphatic compounds are hydrogen rich and lower in density, aromatics have a high density and a low H/C ratio. Taking into consideration the quantities of the resolved and unresolved compounds, the H/C-ratio of the unre- solved compounds can be calculated from the com- bination of H/C-ratio of total C7+ measured by elemental analysis and the known H/C-ratio of the resolved compounds. As shown in Fig. 5 the re- lationship between H/C ratio and density in the resolved fraction was used as a characterisation tool in extrapolating the given trend (calculated through
best fit nonlinear regression) to the low H/C-ratios which were calculated for the unresolved compound mixture. The high density values which correspond to these low H/C-ratios were fnally used to calcu- late the total average density of the liquid hydrocar- bons at different stages of kerogen conversion.
The calculated phase envelopes for selected steps of this MSSV experiment are shown in Fig. 6 together with the calculated equivalent vitrinite reflectance (%easy Ro, Sweeney and Burnham, 1990) in combination with compositional kinetics (Dieckmann et al., 1998). The evolution of the phase envelopes with increasing maturity from the Duvernay sample mimic the evolution postulated in Fig. 2, which was based purely on theoretical con- siderations. Only the high temperature/high matur- ity phase envelopes (equivalent Ro of > 1.4%) show anomalously high pressure and temperature shapes. This could possibly be due to the fact that the high temperature MSSV pyrolysates are characterised by an abundance of high density aromatic compounds, whereas in nature low density aliphatic moieties tend to be more predominant (Horsfield, 1997). Reduction of the aromaticity of the samples in the input dataset for Geoflash results in a significant shift of the calculated dew-point curves to lower pressures and temperatures, thus approximating the theoretical development much more closely. The observed anomalous phase envelopes from the MSSV experiment could, therefore, be an artefact of the experimental technique used and not directly correlatable to nature. However, the general phase envelope trend with increasing artificial maturation indicates that the theoretical considerations can be assumed to be correct.
The evolution of the GOR of the Duvernay Formation sample with increasing maturation is shown in Fig. 7. This type of diagram can be directly compared to the cumulative GOR evolution of England and Mackenzie (1989) shown in Fig. 3. Interestingly, despite the fact that the Duvernay Formation sample is a good example of a type II source rock it shows a distinctly different GOR trend with maturity when compared to the type II data of England and Mackenzie (1989). The
1
E 0.95
o9
.~ O.B5 &
0.8
~= 0.75 0.5 0.75 1 1.25 1.5 1.75 2
"hump'] H/C average HIC - ratio
Fig. 5. Determination of C7+ fraction density (line) based on measured H/C vs density relationships from MSSV
Fig. 6. Phase envelopes of petroleums of increasing maturity generated by MSSV of a Duvernay Formation sample.
6 0 0
Duvernay sample (Fig. 7) is characterised by a rela- tively constant, almost linear, increase in GOR with maturity up to 1.4% Ro. At values higher than 1.4% Ro the GOR increases at a higher rate. The sudden increase of GOR at 1.4% Ro is due to the predicted onset of secondary gas generation reac- tions at this maturity stage in the experimental, closed system, set-up used (Dieckmann et al., 1998). This is a feature which, we assume, is not relevant for the Duvernay Formation source rock studied here, since under natural conditions expulsion would have removed 60-90% of the fluids from the system long before they could have cracked to gas (Cooles et al., 1986). Due to this uncertainty in the comparability to natural systems the high maturity samples can be disregarded in the following discus- sion. In the comparison of the MSSV GOR data
0.2
0.4
0.6
n, 0.8
1 "~ 1.2
5; 1.4
1.6
1.8
Fig. 7.
GOR (kg /kg)
0 0.5 1 1.5 2 2.5 3
I I [ [ I
GOR evolution with maturity determined for the Duvernay source rock sample by MSSV.
with the data of England and Mackenzie (1989) of cumulative GOR evolution with depth (Fig. 3), it is interesting to note that the MSSV dataset presented here reaches a significantly higher final GOR: 1.5 kg/kg for the Duvernay Formation as compared to 0.5kg/kg in England and Mackenzie (1989). Whether this is due to the different experimental procedures used or whether it is really due to differ- ent organic matter types can not be answered at this stage.
The physical properties of petroleum fluids are not only controlled by maturity. Secondary mi- gration of a fluid phase, or uplift of a reservoired fluid leads to a reequilibration of the fluid to the new pressure-temperature conditions. In the course of the upward movement of a fluid (towards lower P and T conditions) phase separation will even- tually happen, and continue to occur as long as pressure and temperature are reduced (Silverman, 1965; Schowalter, 1979; England et al., 1987; England and Mackenzie, 1989). Phase separation of a liquid and gaseous phase, as discussed by these authors, are not the only processes occurring. The reequilibration of the fluid to its new pressure and temperature conditions can be expected to lead also to a precipitation of a solid phase out of the liquid (waxes or asphaltenes) due to changes in solubility. Accordingly, the average molecular weight of the liquid phase changes, i.e. in this case, decreases. A quantification of the effects of this process has, to our knowledge, not been attempted yet. The P V T
software used in this study does not take any type
PVT and phase behaviour 213
•" - -", Deep
I Temperature .
Fig. 8. Proposed evolution of phase envelopes with uplift or vertical migration based on theoretical considerations
discussed in the text.
0
50
Pressure 1 0 0 (bar)
150
200
250
0.00 0.50 1.00 1.50 2.00
[ * Density Gas (glcm a) • ~ ~ ~ D e n s i t y Oil {g/cm 3)
l" Viscosity Gas (cp) ~ Viscosity Oil (cp)
Fig. 9. Evolution of density and viscosity of equilibrium oil and gas phases of an upwards migrating black oil. Calculated using Snorre oil composition and PVT-Sim
simulation package.
of solid precipitation out of a liquid during phase separation into account. It is, however, possible to calculate the condensation of a liquid phase out of a vertically migrating vapour phase, which leads to a reduction in average molecular weight and density of the vapour phase. The results of this type of cal- culation, performed as a series of flashes to succes- sively shallower burial conditions, indicated that the decreasing CGR (condensate to gas ratio) of the vapour phases correlated more or less linearly to the molecular weights of both vapour and liquid phases in equilibrium for burial conditions deeper than roughly 1500 m. At shallower depths an expo- nential relationship was monitored. Based on these results the expected precipitation of solid matter out of a vertically migrating liquid, and consequent reduction of the average molecular weight and den- sity of the liquid phase was approximated assuming a linear decrease with decreasing GOR for burial depths deeper than 1500 m. The assumption that the reduction of molecular weight of the liquid frac- tion is a linear function of the decreasing GOR leads to a calculated phase envelope evolution of an upwards moving liquid as depicted in Fig. 8. Figure 8 shows schematically the evolution of the phase envelope of the liquid phase and its equilibrium gas phase for four pressure/temperature reduction steps. In each case the phase envelopes of the fluids in equilibrium intersect along the pressure/temperature gradient. The original liquid phase envelope, from which the liquids and gases shown in this fgure have separated, has been omitted. Figure 8 shows that, with pressures and temperatures decreasing along an assumed regional pressure/temperature trend, the phase envelopes of the associated gas and liquid phases at each individual step shrink in a sys- tematic fashion. The critical points of the liquids move towards lower pressures and temperatures. The molecular weight of the liquids and gases decreases constantly during the described pressure and temperature decrease, whereas bulk density and
viscosity are calculated to increase for the liquid phase and decrease for the vapor phase, as shown in Fig. 9. The implications of these calculations for migrating fluids are evident, the drastic difference in density and viscosity between physically segregated vapor and liquid phases should lead to differential migration as a consequence of the differences in capillary pressure required to restrain the fluid movement. At a pressure of 150 bar, for example, an oil column would need to be twice as large as a gas column to overcome the capillary pressure of a 10 #m large pore throat.
APPLICATION TO THE NATURAL SYSTEM
Matur i ty sequences and petroleum populations
The changes in the physical properties of pet- roleum fluids as well as the changes of the corre- sponding phase envelopes as calculated by P V T software during maturation and migration discussed above, were based solely on theoretical consider- ations and experimental data. The observed sys- tematics were then tested and calibrated to natural systems. This is discussed below.
Greater Ekof isk area. Hughes et al. (1985) dis- cussed in detail the petroleums found in the greater Ekofisk area. They recognised a distinct maturity sequence, as well as minor source effects in the dis- tribution of the fluids based on organic geochemical methods. The authors noticed that for this sequence principal component analysis indicated that the bulk properties of the oils (such as API gravity, sul- fur content, etc.) were primarily controlled by maturity. P V T data was available for five of the samples which Hughes et al. (1985) used in their study, these being 2/7A-2 (Hughes et al. #25) and 2/7B-14 (#20) from the Eldfisk Field, 2/7-4 (Hughes et al. #17) from the Edda Field, 2/4-5 (#11) from the West Ekofisk Field and 1/6-1 (#7) from Albusk- jell. The location of these samples in the sulfur vs. API gravity plot of Hughes et al. (1985) is shown in
214 R. di Primio et al.
0.40
0.35
0.30
0.25
=. 0.20
u) 0.15
0.10
0.05
0.00
[ ] •
[ ]
[ ]
Z~
A [2
• Tor
A NW Tor
Albuskjell
• W Ekofisk
[ ] Ekofisk
I-II 0 Edda
I~ [ ] i-t Eldfisk
0 • Hod
O @
25 30 35 40 45 50 55
Gravity (% API)
Fig. 10. AP1 versus sulfur content for sample suite analysed by Hughes et al. (1985). Modifed after Hughes et al. (1985).
Fig. 10. P V T data from Hughes et al. (1985) sample #27 (well 2/7-3), which represented the lowest maturity Eldfisk sample was not available. How- ever, we did have data from well 2/7-8 (East Eld- risk) which has the lowest maturity in the Eldfisk Field (Stoddart et al., 1995), and included it in the P V T analysis of the greater Ekofisk area as corre- sponding to sample #27 from Hughes et al. (1985).
The phase envelopes of the samples listed above are shown in Fig. 11. An evolution of the phase envelopes from high temperature-low pressure biased shapes to high pressure-low temperature
biased shapes with increasing maturity is evident in this figure. The systematic decrease in cricon- dentherms and increase in cricondenbars together with the shift of the critical point towards higher pressures and lower temperatures with increasing sample maturity (according to Hughes et al., 1985), fits quite well to the proposed evolution of phase envelopes with maturity discussed above (Figs 4 and 6).
Snorre .field. Horstad et al. (1995) demonstrated that Snorre Field oils (Tampen Spur) represent a maturity sequence, albeit on a smaller scale than
600
500
400
200
100 '
0 -100
2/4-9 DST-2 (Albuskjiell) #8
1 / 6 - ~ c~t P 2 (W.Ekofisk) #11
2/7B.1~ ( ] l ~ , d f i s k ) ~ ~ T ~ p (2E:da)#17
0 100 200 300 400 500 TemperaturelC
=;
,.~0 :.~.-.
,.36'
).25"
L20'
).16"
LI0"
~.05"
).00" - - 25
L~
[] A Q@On eu
r t
30 35 40 45 50
Gravity (% API)
600
Fig. l 1. Phase envelopes of samples from the greater Ekofisk area.
55
P VT and phase behaviour 215
300
200
100
L u n d e F o r m a t i o n
S t a t f j o r d F o r m a t i o n
M a t u r i t y i n c r e a s e
410
• Lunde Fm. b E 39o
ss0
~ 370
'~ 350
3,0
0 330 I I ~ I 1 I I I 400 420 440 460 480 500 5:20 540 560 580 600
Crlcondenthean temperature (deg. C)
0 I .... I I I I "' "1 I -100 0 100 200 300 400 500 600
Temperature (deg. C)
Fig. 12. Phase envelopes of samples from Snorre field (a) and corresponding cricondentherm vs. critical point temperatures (b).
the greater Ekofisk area. The clearest maturity trend was reported for the oils hosted in the Lunde Formation, encountered predominantly in block 34/ 4. Oil family type differentiation performed by Hor- stad et al. (1995), using organic geochemical methods, also differentiated between the oils hosted in the Lunde and Statfjord formations, defining these as belonging to the same population but representing individual families (definitions of popu- lation and family after Horstad and Larter (1997)).
P V T data from 11 samples from the Snorre Field was available for this study. The phase envelopes of the oils hosted in the Statfjord Formation are shown in grey in Fig. 12, the oils from the Lunde Formation in black. The two oil families described by Horstad et al. (1995) can be clearly discrimi- nated in the P / T diagram. Lunde Formation oils show a systematic decrease in their cricondentherms with increasing cricondenbars, and the locations of the critical points move systematically towards lower temperatures and higher pressures. These cri- teria correspond to what was predicted for a matur- ity sequence. The maturities of these oils according to their C2v13/~, 17c~(H)-diasterane (20S)/[C2713~, 17~(H)-diasterane (20S) + C2714~(H)-17cffH)-ster- ane (20R)] ratios correlate to their GOR (Horstad et al., 1995) and hence saturation pressure, confirm- ing the maturity trend recognition through the phase envelope trend. The oils from the Statfjord Formation, however, do not show the same sys- tematics. Figure 12 shows that the increase in GOR of the individual samples is not accompanied by a
decrease in the cricondentherm. The critical points of the individual samples seem to remain at almost constant temperatures independent of the shift of the bubble point curves to higher pressures. These features are commonly observed when recombining oils of different source or maturity to new mixtures. Oils recombined in the lab show phase envelopes which lie between those of the end members of the mixture, whereby the critical point plots on a linear trend between the extremes. Accordingly, based on this phase envelope interpretation, we could pro- pose that Lunde Formation oils represent a matur- ity sequence generated from one specific basin and from a well defined and stratigraphically distinct source rock, whereas Statfjord Formation oils are generally of lower maturity, but represent mixtures of at least two original end members.
Eldfisk fieM. According to the interpretation pro- vided by the API vs. sulfur content diagram of Hughes et al. (1985) (Fig. 10) the Eldfisk Field should contain a maturity sequence of oils ranging from marginally mature (#27) to relatively high maturity (#22). Therefore a set of oils from Eldfisk should also show the typical maturity evolution of phase envelopes discussed above and recognised in the Snorre and greater Ekofisk area datasets. P V T data from the Eldfisk Field was available for 11 samples from both exploration and production wells. The phase envelopes of these samples are plotted in the PIT diagram in Fig. 13. The data falls clearly into two groups. One set of samples shows the expected maturity trend very clearly
216 R. di Primio et al.
,oo] 400 -
300-
350
30O
250
,Ip 150
' i 100
i 50'
0 400
I I I I I I I 410 420 430 440 450 460 470
C r l c o n d e n t h e l T n temperature (deg. C)
I I 480 490 500
-100 0 100 200 300 400 500 600
Temperature (deg. C)
Fig. 13. Phase envelopes of samples from Eldfisk field (a). Samples which show a clear crossover and are hence assumed to represent a maturity sequence are drawn in black. (b) shows the cricondentherm
vs. critical point temperature trend of the maturity sequence.
(shown in black in Fig. 13), and is in fact character- ised by a clear crossover point of the four phase envelopes. Such a crossover point or area in a set of phase envelopes from samples of increasing maturity seems to be characteristic for maturity sequences generated from one and the same source rock, as exemplified by its occurrence in the Snorre (Lunde Formation) dataset, which represents a dis- tinct geochemical family of oils (Horstad et al., 1995). In the greater Ekofisk area dataset, however, a crossover is not evident. This is most likely due to the fact that the petroleums shown were generated from a set of different basins where minor vari- ations in the depositional environment of the source rock may have occurred, or where secondary mi- gration and/or fractionation may have influenced the fluids to varying degrees. Interestingly this type of crossover also occurs (tentatively) in the labora- tory dataset (Fig. 6). The second group of fluids shown in Fig. 13 (in grey) is characterised by the fact that, besides the systematic shift of the critical points with increasing GOR, no obvious maturity relationship is evident. Accordingly, the wide vari- ation in phase envelopes of this dataset should be attributed to mixing different source rocks and/or different pathways of secondary migration of the in- dividual fluids. Nevertheless, when these two groups of samples are plotted on a map of the Eldfisk
Field, they show an interesting distribution, which matches closely the petroleum populations recog- nised by Stoddart et al. (1995) based on isotopic variations (Fig. 14). In their study of the Eldfisk Field Stoddart et al. (1995) discussed the complexity of Eldfisk Field and stated that up to four different potential petroleum-generating kitchens exist within the Eldfisk catchment area. Additionally petroleum emplacement was stated to have been fault con- trolled and largely sporadic. All of these obser- vations (Stoddart et al., 1995) are confirmed by the interpretation of P V T data of fluids from the field.
Tampen ,~pur. On a more regional scale an attempt was made to recognise petroleum popu- lations through their P V T properties and compare the results to organic geochemical interpretations. Horstad et al. (1995) published a first petroleum population map of the Tampen Spur area in the northern part of the Norwegian North Sea, which has in the meantime been expanded upon at Saga Petroleum ASA. The newest interpretation of oil populations and families, as well as the relevant mi- gration routes is shown in Fig. 15. The information shown in this figure is the result of geochemical analysis of, to date, over 200 samples (core extracts as well as RFT, FMT and produced oil and gas samples).
P V T and phase behaviour 217
6d'26
s ~ 4
S ~ 2
3°115
7 -
D-IS
' , , : ; 'AN::;:; ' •
3=2O
r ~ 2
y heavy
Isotopically light
0 2 4 K m [ l I I I
Fig. 14. Map describing petroleum populations in Eldfisk field based on P V T analysis (location of samples described by red phase envelopes of Fig. 13 shown here in black, location of samples described by green phase envelopes shown in white, arrows show maturity trends recognised) (a) and according
to the Stoddart et al. (1995) interpretation (b). Modified after Stoddart et al. (1995).
P V T data for samples from exploration wells in the Tampen Spur area was available for a total of 36 samples. P V T interpretation was performed using crossplots of GOR, formation volume factor
and saturation pressure as well as by means of the phase envelope interpretation discussed above. Five petroleum populations were recognised by these means (Fig. 16), whereby in the cases of the Snorre
2 ° O0 2020
61 30'
61 15'
Fig. 15. Petroleum populations in the Tampen Spur area, northern Norwegian North Sea, based on geochemical analysis. Arrows indicate directions of petroleum migration.
218 R. di Primio et al.
20 O0 2020
6t 30'
61 t5'
Fig. 16. Petroleum populations in the Tampen Spur area, northern Norwegian North Sea, based on P VT analysis. Arrows indicate increasing petroleum maturity.
Field (see above) and the Statfjord Fields different families were also evident. In most cases maturity gradients in the individual oil populations were recognised and are plotted as arrows in Fig. 16. The concordance to the results of the geochemical characterisation of oil populations and filling direc- tions is evident in the comparison of both maps (Figs 15 and 16), and underlines the potential of PVT data interpretation in regional correlation stu- dies.
Uplift~vertical migration
The Hammerfest basin, Barents Sea. The changes which a petroleum fluid undergoes upon uplift or vertical migration were discussed above from a theoretical point of view. The Hammerfest Basin, in the Norwegian part of the Barents Sea, is a perfect example of a petroliferous basin which has been subjected to significant uplift during its recent geo- logic evolution (Ronnevik et al., 1982; Nardin and Rossland, 1993). The magnitude of uplift in the Hammerfest basin during the Tertiary ranges from 300 m to over 1 km (Nardin and Rossland, 1993), increasing from west to east. Oils and condensates encountered in the Hammerfest basin are character- ised by being generally saturated under reservoir conditions, i.e. their saturation pressure equals the reservoir pressure at the gas-oil contact. In a few cases undersaturated oils were encountered, the occurrence of which can be explained by Plio-Pleis-
tocene reburial after equilibration of the fluids to shallower conditions. The phase envelopes of 20 oil and condensate samples from the Hammerfest basin, as shown in Fig. 17, match the expected uplift evolution of Fig. 8 perfectly. Especially for the liquid samples, the decrease in cricondentherm with decreasing reservoir pressure, accompanied by the decrease in pressure and temperature of the critical point and saturation pressure, underline the fact that high molecular weight material must have been removed from the fluid during uplift as pro- posed from theoretical considerations above.
The cooccurrence of oils and condensates in the Hammerfest Basin has been assumed to indicate that phase separation led to either gas exsolving from oil during uplift, or a condensate dropping out of a vapor phase. Gas expansion during uplift is furthermore implied to have caused a significant loss of liquids due to spilling, thus explaining the predominance of gas over oil in the Hammerfest basin. These arguments can be discussed in more detail if the volumetrics of the fluids under the pro- posed geological evolution are taken into account, i.e. using PVT data and simulating various uplift scenarios (see also Skagen, 1993).
An example of the results of volume calculations based on PVT data is given below for the Snohvit Field and shown in Fig. 18. The main assumptions are: total uplift of 750m, saturated oil or gas at preuplift conditions, hydrocarbon pore-volume
PVT and phase behaviour 219
200
300
100
0
.100 0 100 200 300 400 500
Temperature (deg. C)
Fig. 17. Phase envelopes of samples from the Hammerfest Basin, Barents Sea. Phase envelopes of con- densates are drawn in grey, oils in black.
I 600
(HCPV) of 1.55 x 109 m 3, gas volume of 6.42 x 108 m 3, live oil volume of 9.5 x 107 m 3, re- sidual oil volume of 8.07 x 108 m 3 (Fig. 18(a); re- sidual oil, i.e. "irreducible" o i l saturation, was observed both in the gas zone and below the oil- water contact), oil and gas compositions from Snohvit wells flashed to surface and recombined to saturation point at preuplift conditions.
If, in the first case, a saturated oil preuplift is assumed, an excessive oil volume (in addition to the presently observed oil volume) of almost 3 x 109 m 3 would have been necessary to explain the present day proportions of oil and gas in the reservoir (Fig. 18(b)). If a saturated gas phase at preuplift conditions is assumed a gas volume of 1.46 x 10 x2 m 3 would have been necessary to explain the present day composition (Fig. 18(b)). The large volumes required for both of these cases are due to the fact that the oil contains relatively small amounts of gas and the gas is only able to dissolve small amounts of oil under reservoir P and T. Accordingly the geologically most reasonable in- terpretation of the filling history of Snohvit is that oil and gas migrated into the reservoir at different times, possibly first an oil phase which was later displaced by inflowing gas which also released a condensate during uplift. The suggestion that gas derived from aquifer degassing might have contrib- uted to the abundance of gas in the Hammerfest basin (Oygard and Eliassen, 1989), is not sustained by the gas composition of the Snohvit Field. If pre-
dominantly methane derived from aquifer degassing had led to the formation of a gas cap in Snohvit, the gas composition in the reservoir would have to be at present much drier, even taking reequilibra- tion with the fluid phase in the reservoir into account. P V T modelling of a mixing event of methane and the Snohvit oil composition led to a free gas composition which contained over 7% higher methane contents than measured in the field.
Application in exploration
P V T analysis in the petroleum industry has up to now concentrated mainly on the evaluation of pro- duction related problems. The potential of including P V T analysis in the exploration process should, however, not be neglected. The experimental data of England et al. (1987) and England and Mackenzie (1989) as well as the application orien- tated considerations of Heum et al. (1986) or the data presented in this study can, for example, be used to attempt a direct prediction of petroleum properties expected in a reservoir prior to drilling based on source rock maturity and various geologic scenarios and their influence on the charging of the reservoir.
For the case in which a continuous connection is assumed between kitchen area and reservoir England and Mackenzies' cumulative GOR evol- ution diagram (Fig. 3) can be used to predict the ul- timate GOR accumulated in the trap, depending of course on the degree of maturity the source rock
220 R. di Primio et al.
1.61E+09
1.41E+09
1.21 E÷09
1.01E+09 E
8.10E+08
6.10E+08
4.10E+08
2.10E+08
1.00E+07 HCPV Gas Live Oil
~t
'1"
Residual Oil
E
J~
1.00E+13
1.00E+12
1.00E+I 1
1.00E+10
1.00E+09 HCPV Sat. Oil Sat Gas
b
Fig. 18. Volumetrics of reservoir fluids of Snohvit field (a) and calculated volumes of single phase satu- rated oil or single phase saturated gas required to explain observed fluid distribution (b), notice logar- ithmic scale. The assumption of a single phase fluid leading, after uplift and phase separation, to the present day fluid distribution in the reservoir requires fluid volumes much higher than the HCPV of the
field.
has reached. If a sporadic source-reservoir connec- tion existed, such as can be postulated for some reservoirs charged through faults, the GOR evol- ution of Fig. 2 is probably more appropriate. In both cases however, depending on how secondary migration took place, an estimation of GOR can be made. The analysis of P V T data from neighbouring finds helps define a basic petroleum composition (from e.g. discoveries in similar plays, tectonic situ- ations, stratigraphic intervals) which is then recom- bined to the predicted GOR value.
This type of approach depends to a large extent on the GOR prediction, which, as demonstrated in the comparison of the data of England and Mackenzie (1989) and the results of the MSSV ex- periment performed on the Duvernay Formation sample discussed above, seems to be quite variable depending on the source rock type. Hence this form of P V T application in exploration requires a
detailed GOR-evolution database for the different source rocks of interest.
The more common approach in exploration is to compile a regional P V T database and to try to understand compositional relationships of fields sur- rounding a prospect before attempting a prediction of the contents of the prospect itself.
One situation encountered consisted of a struc- ture (termed "prospect" in Fig. 19) which lay on a trend surrounded by a series of different discoveries (Fig. 19). Wells A, B and C contained undersatu- rated fluids of increasing GOR. Well D contained a saturated oil with a gas cap, whereas well E con- tained a gas-condensate. Conventional Bo vs. Psat crossplots indicated that only wells A, B and C showed a certain degree of correlation (i.e. were genetically related to each other). Therefore, three different populations of fluids were present in the study area. The prospect could accordingly contain
P V T and phase behaviour 221
W(
g i l l
Led oil
ID
Fig. 19. Depth to top reservoir map of exploration area discussed in the text.
fluids related to any of the three discoveries sur- rounding it.
Analysis of the phase envelopes helped in the in- terpretation of the fluid relationships (Fig. 20). In a former geological interpretation of the discoveries in this area the condensate find of well E had been explained by phase separation and differential mi- gration of oil and gas from well D, whereby the condensate was expected to represent the exsolved gas phase from well D. P V T considerations, how- ever, clearly pointed out that such a relationship is impossible, since an undersaturated gas can not be released by phase separation from a saturated oil. In fact the condensate of well E was impossible to relate in any manner to either wells A, B and C nor to well D. P V T modelling revealed that the wells from A to D represent a mixing sequence, in which oil from well D migrated along the ridge reaching wells C and B and mixing with the oil present in these wells (originally of a composition as observed in well A) to different degrees. This mixing of oils is evident in the location of the critical points of the
450, =
4 0 0
3 5 0
2 5 0
w 2 0 0
~" 150
100
5o
Well D
1:1 tool/tool Rec. A g O 1:0.65 mol/rnol
0 100 200 300 400 500 600 Temperature (d~l. C)
Fig. 20. Phase envelopes of fluids from wells A to E shown in Fig. 19 (bold lines) as well as those of samples recombined using well A and D fluid compositions to march well B and C compositions (dotted and dashed lines
respectively).
individual fluids as discussed for the Snorre/ Statfjord Formation liquids.
Recombination of the oils from wells A and D to a ratio of 1:1 (mol/mol) lead to the near perfect reconstruction of well C fluid composition; recombi- nation at a ratio of 1:0.65 (mol/mol) leads to a very close match of the well B fluid (Fig. 19). In the case of the first recombination the error between recom- bined and measured physical properties of the fluids was below 2%. For the second recombination (well B) the error was slightly larger (up to 5%), but still the match was remarkably good. By thus proving that mixing of two different oil populations had taken place it was concluded that since the mi- grating fluids would have had to cross through the prospect in order to reach wells C and B the pro- spect itself should accordingly be filled to spillpoint (given structure, seal and reservoir) by a liquid at saturation pressure.
C O N C L U S I O N S
The work discussed in this paper clearly points out the potential of using P V T analysis in conjunc- tion with organic geochemical techniques for a bet- ter understanding of petroleum movement and ultimate distribution. P V T analysis must, in this context, be understood as a tool complementary to organic geochemistry, which allows the integration and interpretation of data reflecting fluid response to geological conditions on a regional scale. In this field of work P V T analysis can be used as a screen s ing tool for oil population characterisation, relative maturity estimation, volumetric evaluation of com- plex reservoir filling histories and recognition of mixing effects.
The use of P V T analysis in petroleum exploration in a predictive fashion depends to a large extent on a very limited experimental dataset. Further research in the area of phase envelope evolution during maturation is required in order for this tech- nique to become fully applicable.
222 R. di Primio et al.
Acknowledgements--We would like to thank Saga Petroleum ASA for making the data available and for per- mission to publish. V. Dieckmann thanks his supervisor B. Horsfield for advice and support as well as Forschungszentrum Juelich GmbH for his Ph.D. stipend. The samples from the Duvernay Formation were obtained from the Geological Survey of Canada with the kind as- sistance of M. Fowler and L. D. Stasiuk. The authors wish to express their gratitude to W. A. England and P. Muelbrock for their reviews which improved the paper.
REFERENCES
Baskin, D. K. and Peters, K. E. (1992) Early generation characteristics of a sulphur-rich Moneterey kerogen. American Association of Petroleum Geologists Bulletin 76, 1-13.
Cooles, G. P., Mackenzie, A. S. and Quigley, T. M. (1986) Calculation of petroleum masses generated and expelled from source rocks. Organic Geochemistry 10, 235-245.
di Primio, R., Horsfield, B., Krooss, B. and Leythaeuser, D. (1993) Experimental simulation of hydrocarbon gen- eration and expulsion by open-system hydrous-pyrolysis of a sulphur-rich marine source rock. In Organic Geochemistry, Poster Sessions from the 16th International Meeting on Organic Geochemistry, Stavanger 1993, ed. K. Oygard, pp. 220-228. Falch Hurtigtrykk, Oslo.
Dieckmann, V., Schenk, H. J., Horsfield, B. and Welte, D. H. (1998) Kinetics of petroleum generation and cracking by programmed-temperature closed-system pyrolysis of Toarcian shales. Fuel 77, 23-31.
Dueppenbecker, S. J. and Horsfield, B. (1990) Compositional information for kinetic modelling and petroleum type prediction. In Advances in Organic Geochemistry 1989, ed. B. Durand and F. Behar, pp. 259-266. Pergamon, Oxford.
England, W. A. and Mackenzie, A. S. (1989) Some aspects of the organic geochemistry of petroleum fluids. Geologische Rundschau 78, 291-303.
England, W. A., Mackenzie, A. S., Mann, D. M. and Quigley, T. M. (1987) The movement and entrapment of petroleum fluids in the subsurface. Journal of the Geological Society London 144, 327-347.
Heum, O. R., Dalland, A. and Meisinger, K. K. (1986) Habitat of hydrocarbons at Haltenbanken (PVT model- ling as a predictive tool in hydrocarbon exploration). In Habitat of Hydrocarbons an the Norwegian Continental Shelf, ed. A. M. Spencer et al., pp. 259-274. Graham and Trotman, London.
Horsfield, B. (1997) The bulk composition of first-formed petroleum in source rocks. In Petroleum and Basin Evolution, ed. D. H. Welte, B. Horsfield and D. R. Baker, pp. 335-402. Springer, Berlin.
Horsfield, B., Disko, U. and Leistner, F. (1989) The microscale simulation of maturation: outline of a new technique and its potential applications. Geologische Rundschau 78, 361-374.
Horsfield, B., Schenk, H. J., Mills, N. and Welte, D. H. (1992) An investigation of the in-reservoir conversion of oil to gas: compositional and kinetic findings from closed-system programmed-temperature pyrolysis. In Advances in Organic Geochemistry 1991, ed. C. B. Eckardt, J. R. Maxwell, S. R. Larter and D. A. C. Manning, pp. 191-204. Pergamon, Oxford.
Horstad, I. and Larter, S. R. (1997) Petroleum migration, alteration and remigration within Troll Field, Norwegian North Sea. American Association of Petroleum Geologists Bulletin 81, 222-248.
Horstad, I., Larter, S. R , and Mills, N. (1995) Migration of hydrocarbons in the Tampen Spur area, Norwegian North Sea: a reservoir geochemical evaluation. In The
Geochemistry of Reservoirs, ed. J. M. Cubitt and W. A. England, pp. 159-183. Geological Society Special Publication No. 86.
Hughes, W. B., Holba, A. G., Miller, D. E. and Richardson, J. S. (1985) Geochemistry of greater Ekofisk crude oils. In Petroleum Geochemistry in Exploration of the Norwegian Shelf, ed. M. Bruce et al., pp. 75 92. Graham and Trotman, London.
Karlsen, D. A., Nyland, B., Flood, B., Ohm, S. E., Brekke, T., Olsen, S. and Backer-Owe, K. (1995) Petroleum geochemistry of the Haltenbanken, Norwegian continental shelf. In The Geochemistry of Reservoirs, ed. J. M. Cubitt and W. A. England, pp. 203-256. Geological Society Special Publication No. 86.
Nardin, T. R. and Rossland, K. G. (1993) Restoration of the eroded section in the western Barents Sea. In Arctic Geology and Petroleum Potential, ed. T. O. Vorren et al., pp. 607-618. NPF Special Publication No. 2.
Oygard, K. and Eliassen, P. E. (1989) Residual hydrocar- bons in the Barents Sea. Paper presented at the Norwegian Petroleum Society Conference: Arctic Geology and Petroleum Potential, 15-17 August 1989, Troms~.
Pedersen, K. S., Fredenslund, A. and Thomassen, P. (1989) Properties of Oils and Natural Gases, 252 pp. Gulf Publishing, Houston.
Ronnevik, H. C., Beskow, B. and Jacobsen, H. P. (1982) Structural and stratigraphic evolution of the Barents Sea. In Arctic Geology and Geophysics, ed. A. F. Embry, pp. 431-440. Canadian Society of Petroleum Geologists.
Schaefer, R. G., Schenk, H. J., Hardelauf, H. and Harms, R. (1990) Determination of gross kinetic parameters for petroleum formation from Jurassic source rocks of different maturity levels by means of laboratory exper- iments. Organic' Geochemistry 16, 115-120.
Schowalter, T. T. (!979) Mechanics of secondary hydro- carbon migration and entrapment. American Association of" Petroleum Geologists Bulletin 63, 723-760.
Silverman, S. R. (1965) Migration and segregation of oil and gas. American Association of Petroleum Geologists Memoir 4, 54-65Tulsa.
Skagen, J. I. (1993) Effects on hydrocarbon potential caused by Tertiary uplift and erosion in the Barents Sea. In Arctic Geology and Petroleum Potential, ed. T. O. Vorren et al., pp. 711 719. NPF Special Publication No. 2.
Soave, G. (1972) Equilibrium constants from a modified Redlich-Kwong equation of state. Chemical Engineering Sciences 27, 1197-1203.
Sokolov, V. A. (1964) Migration processes of oil and gas, their intensity and directionality. Proceedings of the Sixth World Petroleum Congress, pp. 493-505. John Wiley and Sons, Chichester.
Stoddart, D. P., Hall, P. B., Larter, S. R., Brasher, J., Li, M., and Bjoroy, M. (1995) The reservoir geochemistry of the Eldfisk Field, Norwegian North Sea. In The Geochemistry ~?[ Reservoirs, ed. J. M. Cubitt and W. A. England, pp. 257-279. Geological Society Special Publication No. 86,
Sweeney, J. and Burnham, A. K. (1990) Evaluation of a simple model of vitrinite reflectance based on chemical kinetics. American Association of" Petroleum Geologists Bulletin 74, 1559-1570.
Thompson, K. F. M. (1987) Fractionated aromatic pet- roleums and the generation of gas-condensates. Organic Geochemistry 11, 573-590.
Thompson, K. F. M. (1988) Gas-condensate migration and oil fractionation in deltaic systems. Marine and Petroleum Geology 5, 237 246.
Tissot, B. and Welte, D. H. (1984) Petroleum Formation and Occurrence, 2nd edn. Springer-Verlag, Berlin.
