Advanced_fluid_characterization_of_pauto (1).pdf

SPE 135085 Advanced Fluid Characterization of Pauto Complex, Colombia Sissel Øksnevad Martinsen, SPE, PERA; Iowany Castiblanco, SPE, BP Exploration Company Colombia (LTD); Raul Osorio, SPE, ECOPETROL S.A.; Curtis Hays Whitson, SPE, NTNU/PERA

Copyright 2010, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Florence, Italy, 19–22 September 2010. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract This paper presents a detailed fluid characterization study for a group of compartmentalized gas condensate & volatile oil reservoirs making up the Pauto Complex. Fluid oil-gas ratios range from 30 to 350 bbl/MMscf, producing from depths of 12,000-16,000 ft, with estimated IGIP>1 Tcf and significant oil/condensate resources. A single equation of state (EOS) was developed to describe the complex phase and volumetric behaviour of eight reservoir fluids undergoing depletion. The EOS also describes accurately the vaporization of retrograde condensate from one of the richer reservoir fluids when contacted by a leaner hydrocarbon gas. Years of experience in this thrust-belt basin have consistently demonstrated that the quality of fluids samples, PVT test design, lab-data QC, EOS tuning, and fluid initialization are essential to understand the hydrocarbon systems, reservoir compartmentalization, and recovery mechanisms. We show how integration of fluid characterization assists in the evaluation of depletion versus gas-injection strategies, development decisions, and ongoing reservoir management. The methodology used in this study has general application to PVT data acquisition and developing fluid characterizations exhibiting complex phase behaviour. We also provide standardized methods to QC the thermodynamic consistency of an EOS model, particularly when binary interaction parameters (BIPs) are used extensively. A single EOS model was developed successfully for describing the phase and volumetric behaviour of a wide range of fluids undergoing depletion and vaporization. This was achieved by modifying heavy-component properties, but more importantly, modifying BIPs extensively; these BIP modifications were possible only because considerable equilibrium compositional data was available. We believe this paper is significant as a proven case history. With the PVT data provided in our paper, the same problem can be used to test other fluid characterization methodologies and serve as a complex phase behaviour benchmark. Introduction A consistent equation of state (EOS) model is often needed in modelling field developments, e.g. reservoir simulation, pipe flow and process design. Developing a single, consistent EOS model is often a challenge when in-situ reservoir fluids show significant spatial variation, and when reservoirs undergo gas injection processes. This paper describes a successful EOS model development for the Pauto Field Complex, Colombia. The methods to develop a single EOS model for multiple samples are discussed in this paper, together with PVT data quality control procedures which are necessary to identify outlier data which should not be included in model development. Guidelines are given to check whether a developed EOS model is thermodynamically consistent using equilibrium K-value assessment. The Pauto Field Complex study is used to illustrate these methods and guidelines, providing the final EOS model, compositions, and measured PVT data. Justification and drivers for spending effort in EOS model development are highlighted from the operator’s point-of-view, together with a summary of the issues and impact that the developed Pauto EOS model has on field operations, planning, and design of gas injection production strategies. Pauto Field Complex The Pauto Field Complex is located in the Andean Foothills basin of Colombia north east of the giant Cusiana and Cupiagua fields. A series of stacked structural sheets, originated by the Andes uplift, have been drilled and tested to contain gas condensate & volatile oil reservoirs. The geological complexity of the overburden, the presence of different fluids with oil-gas ratios ranging from 30 to 350 bbl/MMscf, producing from depths of 12,000-16,000 ft TVDSS, and different pressure regimes

2

SPE 135085

encountered, presented a challenging scenario to develop the Pauto Field Complex resources estimated with IGIP>1 Tcf and significant oil/condensate resources. Reduction of subsurface uncertainties of the Pauto Field Complex has been possible through a staged development plan which allowed production of the first well in 2001 and the completion of five additional wells since then. During this period a rigorous information gathering process (3D seismic survey in the area, extended well production test for each horizon per well, fluids sampling, regular pressure transient testing, production logging and tracers injection) has been the key to get a better understanding of compartmentalization, possible connectivity, and ultimately a development plan flexible enough to account for uncertainties. Experience in the development and reservoir management strategies of other fields in the same basin has proved the value added by a robust fluids initialization model and EOS development supported by a large enough database containing PVT experiments that are relevant to the mechanisms that the field will undergo in its productive life. On top of that the possibility to validate/discard compartmentalization models based on fluid variations both laterally and vertically was important in the Pauto Field Complex understanding. To obtain a field-wide fluids understanding, separator sampling and subsequent PVT studies for all tested horizons in all wells have been conducted. A typical PVT study includes composition determination of separator streams, recombination at producing GOR, two constant composition expansion (CCE) experiments (one at reservoir temperature), one constant volume depletion (CVD) experiment, and one multistage separation test. A condensate revaporization study was conducted as well for the richer gas condensate systems. Sampling Program The fluids sampling approach used in the Pauto Field Complex has historically been surface separator sampling, given the operational constraints imposed at near-saturated initial conditions (i.e. pR ~ pS). Bottomhole sampling was not considered prudent and the acquisition of uncontaminated MDT samples was difficult because the wells were drilled with oil based mud. One MDT sample shows clear consistency with a surface recombined sample from the same formation and well. Surface separator sampling requires a minimum rate to guarantee stabilized producing gas-oil ratio. For Pauto Field the stabilization time could vary from 12 to 36 hr. Separator conditions are registered while testing and shrinkage factors are measured on site at regularly-spaced intervals. This data is important input for the proper recombination of surface fluids in the PVT laboratory. PVT Data Program PVT studies were performed on eleven samples from different structural sheets and different formations in the Pauto Field Complex. All samples are recombined separator samples (RSS) except for two duplicate samples which are a MDT and a bottomhole sample (BHS), respectively. The PVT measurements included standard depletion type experiments (CCE, CVD, viscosity and separator tests) on all of the recombined separator samples, and a multi contact vaporization (MCV) study was performed in 2008 on a medium-rich gas condensate sample. A true-boiling-point (TBP) analysis was performed on the stock tank oil from sample 1, a volatile oil sample; all other samples are lean to medium-rich gas condensate samples. Table 1 lists all the PVT measurements performed on the samples from Pauto Field Complex. TABLE 1—AVAILABLE PVT DATA FROM PAUTO FIELD COMPLEX. Sample

Sample Type

CCE

CVD

SEP

Viscosity

TBP

MCV

1

RSS

3

1

1

1

1

0

2

RSS

2

1

1

1

0

0

3

RSS

2

1

1

1

0

0

4

RSS

3

1

1

1

0

0

5

RSS

3

1

1

1

0

0

6

MDT

DP and ρ @ res. press.

0

0

0

0

0

7

RSS

1

1

0

0

0

0

8

RSS

2

1

1

1

0

0

9

BHS

1

0

0

0

0

0

10

RSS

2

1

1

1

0

0

11

RSS

1

0

0

0

0

1

Extended GC analyses of the Pauto samples are performed, and compositions up to C30+ are available for all samples except for sample 7 where compositions up to C20+ are available. Compositions up to C36+ were reported for sample 11. The compositions are reported both in mass and mole amounts. No measured molecular weights were measured. Katz-Firoozabadi molecular weights were used to convert mass amounts to mole amounts, as verified from the reported mass and mole fractions. EOS Fluid Characterization A pseudoized 12-component EOS fluid characterization was to be developed for the Pauto Complex reservoirs from a single carbon number (SCN) characterization. The SCN characterization consists of single carbon number fractions from C6 to C29

SPE 135085

3

and a remaining C30+ residue. The development consists of four parts (after the assessment and quality checking of PVT data was performed): 1. Characterization of the heptanes plus fraction. This involves (a) fitting the Gamma distribution model to measured weight fractions from extended GC and, if available, TBP distillation data, and (b) tuning the Søreide correlation. The Søreide correlation gives a relationship between molecular weights and specific gravities of the heptanes plus fraction. 2. Tuning a SCN characterization to all measured PVT data, except for viscosity data. 3. Tuning critical Z-factors of hexanes plus to match measured viscosity data and SCN liquid viscosities. 4. Pseudoizing the SCN characterization. Quality Check of Laboratory PVT Data Checking the quality of measured PVT data is an important first-step in EOS tuning. It is important because a single “wrong” input data might conceivably drive the EOS model in the wrong direction if the data “problem” is not identified at an early stage. Recommended quality-control methods include graphical plots, material balance calculations, use of empirical property correlations, and comparison with EOS predictions. Comparison of measured PVT data with EOS predictions will verify the qualitative PVT behaviour and identify “outlier” data. Graphical plots may include liquid API gravity versus C7+ gravity, saturation pressure and solution GOR versus C7+ mole fraction, saturated oil FVF versus solution GOR, and saturated liquid viscosity versus saturated liquid density. The more data, the better statistical ability to QC data. Material balance calculations can often be made in both directions. A forward material balance will start with a known composition and check the change in composition during the experiment, and a backward material balance will start with the final experimental composition(s) and work backwards toward the initial fluid charged to the PVT cell. Table 2 gives a comparison between the initially charged fluid system of sample 5 with the calculated recovered fluid system, and it shows a good agreement between measured and calculated data. TABLE 2—BACKWARD MATERIAL BALANCE OF CVD TEST FROM SAMPLE 5. Charged

Recovered

Measured

Calculated

mol-%

mol-% 0.49

N2

0.52

CO2

2.82

2.7

C1

68.01

67.96

C2

9.24

9.2

C3

4.71

4.79

i-C4

1.31

1.36

n-C4

3.01

2.95

i-C5

0.76

0.78

n-C5

0.51

0.54

C6

0.7

0.73

C7+

8.42

8.5

Quantitative assessment of quality in gas condensate depletion tests is simply to compare the error-% in C6+ material, as this reflects closely the initial CGR. A forward material balance of the vaporization study performed on sample 11 is plotted in Fig. 1. The procedure of the vaporization test was as follows: Separator oil and separator gas samples were recombined with field-producing GOR. The recombined fluid was depleted to 4000 psia and all equilibrium gas was removed. The collected equilibrium oil was used for the vaporization test. Volumes of equilibrium gas and oil were measured. Injection gas was injected into the oil in five stages. At each stage all equilibrium gas was removed and volumes and gas composition were measured. The final liquid composition was measured at stage 5. The figure shows the relative amounts of C7+ in the liberated gases to the amount in the initial equilibrium liquid at 4000 psia. The material balance check shows that the reported recovered C7+ amounts exceed the initially charged C7+ amount.

4

SPE 135085

140 C7-10 C11-14 C7+

Recovery Factor (mass basis), %

120

100

80

60

40

20

0 1 contact

2 contact

3 contact

4 contact

5 contact

Fig. 1—Material balance of vaporization experiment on sample 11. The figure shows the relative amounts of C7+ in the liberated gases to the amount in the initial equilibrium liquid at 4000 psia.

Laboratory reported Z-factors may be compared against calculated Z-factors using the Hall and Yarborough (Hall and Yarborough 1973 and 1974) equation. The Hall and Yarborough equation is an accurate representation of the Standing-Katz (Standing and Katz 1942) Z-factor chart. The Standing-Katz Z-factor chart gives Z-factors within an accuracy of 2-3%. All reported Z-factors from CVD and CCE experiments performed on Pauto samples were compared against Hall and Yarborough calculated Z-factors using measured equilibrium gas compositions at each stage. Fig. 2 shows the laboratory reported Z-factors of sample 10 compared with calculated Z-factors. The reported Z-factors are 10% higher than the Z-factors calculated with the Hall and Yarborough equation, indicating errors in the reported Z-factors. All the Z-factors reported by the laboratory that performed PVT studies on samples 1, 2, 3, 6, 7, 8, and 10 were more than 5% higher than calculated Z-factors at pressures above 5000 psia. The reported Z-factors by the two other laboratories, however, are in agreement with the calculated Z-factors. 1.30 Lab Correlation

1.25 1.20 1.15

Z-factor

1.10 +/- 5%

1.05 1.00 0.95 0.90 0.85 0.80 0

1000

2000

3000

4000

5000

6000

7000

Pressure, psia Fig. 2—Laboratory reported Z-factors for sample 10 compared with calculated Z-factors using the Hall and Yarborough equation and measured equilibrium gas compositions at each stage.

SPE 135085

5

From the quality control of the measured data on the Pauto samples it seems the PVT data are of good quality except a material balance error in the vaporization study and erroneous Z-factors and viscosities reported by one of the three laboratories used. Characterizing the Heptanes Plus Fractions A proper characterization of heptanes and heavier fractions is essential and this is well documented (Whitson 1983 and 1984). The methods described below are used as implemented in PhazeComp (Zick Technologies 1995-2009) with TWUMW=0. A true-boiling-point (TBP) analysis was performed on the crude assay from sample 1. The TBP data included eleven distillation cuts defined by fixed boiling point ranges. For each cut, lower boiling point (TbL), upper boiling point (TbU), specific gravity (γ), molecular weight, and mass amount were reported. The TBP data was used to develop a relationship between specific gravity and molecular weight of C7+ fractions by regressing on numerical constants and characterization factor Cf in the Søreide correlation (Søreide 1989): γ i = a + C f (M i − b )c The molecular weights were calculated with the Twu correlations (Twu 1984) from the boiling point temperatures. A modified version (TWUMW=0) of the Twu correlations were used, in which the specific gravity is ignored and a normalparaffin-like relationship between molecular weight and boiling point were assumed, which will give a molecular weight – boiling point relationship very close to the Katz-Firoozabadi relationship. After fitting the TBP data, the following Søreide constants were found: Cf = 0.226 a = 0.410 b = 80 c = 0.13

(parameter indicating relative “paraffinicity”) (default 0.2855) (default 66) (default)

The result of matching the experimental TBP data is found in Fig. 3. 1000

1.00 MW EOS reg MW EOS no reg MW TBP SG EOS reg SG EOS no reg SG TBP

800

Molecular Weight

700

0.95 0.90 0.85

600

0.80

500

0.75

400

0.70

300

0.65

200

0.60

100

0.55

0 0

200

400

600

800

1000

1200

Specific Gravity

900

0.50 1400

o

Tb, F Fig. 3—TBP analysis of crude assay from sample 1, comparing experimental and calculated molecular weights, specific gravities and boiling points.

The Gamma distribution (Whitson 1983) model was used to fit the measured TBP data as well as the extended GC data of the separator oil sample from sample 1. Gamma fitting these compositions resulted in shape factors of 0.861 and 0.856 for the TBP data and the GC data respectively. The Gamma distribution parameters from the gamma fit of the extended GC analysis of sample 1 separator oil (Shape factor = 0.856, Bounding MW = 90.812, MWC7+ = 208.9) was used together with the Søreide and Twu correlations to calculate the molecular weights, specific gravities, boiling point temperatures and critical properties of the Pauto Complex fluid characterization. The acentric factors, ωi, were calculated to match the boiling point of each C7+

6

SPE 135085

fraction exactly. The volume translation parameters (Peneloux et al. 1982; Jhaveri and Youngren 1988) of the plus fractions were calculated to match the specific gravities exactly by solving the Peng-Robinson EOS (Robinson et al. 1979). Regression To match the EOS with measured PVT and compositional data, some modifications of the EOS parameters and/or in the compositional data is usually necessary. All reliable PVT data from a given reservoir or a field complex, including as many samples and PVT data as available, should be included in the regression. When few measured PVT data are available, only a few modifications should be made to the untuned EOS. In this EOS study measured PVT data was available for 11 samples, including one vaporization study. The vaporization study includes measured equilibrium vapour/liquid data, which allows for more extensive tuning of regression variables. It is important that the regressed EOS can predict vapour/liquid behaviour accurately, especially in reservoirs where gas is injected. The EOS model development on Pauto Complex reservoirs was conducted twice; once in 2006 and once in 2008. Only depletion type PVT data was available in 2006 to tune the EOS fluid models. An untuned 34 component fluid characterization (EOS34-default) was first developed for quality checking of measured PVT data and as a starting point for regressions. A single 34-component single carbon number (SCN) EOS, and a single pseudoized 12-component EOS (EOS12-2006) were developed matching all depletion PVT data. The EOS12-2006 was used to design a vaporization test performed in 2008. A new 12-component EOS (EOS12-2008) was developed based on results of both the standard depletion type PVT experiments and the vaporization experiment. All the EOS fluid characterizations developed in this study are based on the Peng-Robinson (1979) EOS model. The development of EOS12-2008 will be described in this paper. All measured PVT data, with the exception of poor quality data, was included in the regressions. The saturation pressures and CVD liquid saturations at reservoir temperature were weighted with factors of 5 and 1 respectively. The liquid volumes, initial liquid compositions (equilibrium liquid compositions at 4000 psia), and removed gas compositions from the vaporization test were all weighted with factors of 1. All other PVT data were weighted to 0, but included for monitoring. This set of weight factors was set after much “trial-and-error”, and was found to give the overall best fit of all measured PVT data. Normally the C6+ content in CVD equilibrium gases will be weighted instead of the liquid saturations, as the C6+ would represent the produced surface liquids. However; in this study it proved more efficient to match liquid saturations. During viscosity regression, only single carbon number viscosities were weighted. This was because of uncertainties connected with oil viscosity measurements. Main Regression of PVT Data The following regression variables were used in tuning the EOS12-2008: 1. Average molecular weights of the C7+ fraction of each sample. 2. Boiling points of the C11+ fraction. Separate regression parameters for a. C11-14 b. C15-20 c. C21-29 d. C30+ 3. Critical temperatures of the C7+ fraction. Each pseudo-component was given its critical temperature by following o formula: Tctuned = fTc (Tcinitial − 950 ) + 950 , where fTc is the regression variable and critical temperatures are given in R. ,i ,i 4.

Binary Interaction Parameters. Default non-hydrocarbon hydrocarbon BIPs were used (Whitson and Brule 2000). Hydrocarbon hydrocarbon BIPs were calculated using the Chueh-Prausnitz equation (Chueh and Prausnitz 1968) with A as a variable and B=6. a. BIPs between methane and C5+. Separate regression variables for BIPs between i. C1-C5-6 ii. C1-C7-10 iii. C1-C11-20 iv. C1-C21+ b. BIPs between the “heavies”. Separate regression variables for BIPs between i. C7-10-C21+ ii. C11-14-C21+ iii. C15-20-C21+

The molecular weights of the Pauto Complex stock tank liquids were not measured. The C7+ mass distribution has a significant impact on saturation pressure calculations and the relative proximity to critical points. Gamma fitting the C7+ fraction of the separator liquid compositions gives a good match for all samples. The fit is excellent when fitting only the C13+ fraction of the samples. All samples show a similar, near exponential distribution of the C13+ fraction. An “irregular” behaviour of the C7-C12 fractions is observed. This behaviour is common for the Pauto Complex samples. The compositional behaviour of C7+ is shown in Fig. 4.

SPE 135085

7

0.1 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7 Sample 8 Sample 10

Mole Fraction

0.01

0.001

29 C

27 C

25 C

23 C

21 C

19 C

17 C

15 C

13 C

11 C

9 C

C

7

0.0001

Component Fig. 4—Normalized measured C7+ compositions of samples from Pauto Complex area.

Because molecular weights of the stock tank liquids were (unfortunately) not measured, the average C7+ molecular weights used in the Gamma distribution model were chosen as regression variables. An exponential distribution was assumed for all samples. The non-exponential behaviour of C7, C8, C9, C11, C12 fractions was honoured with multipliers to the values from the exponentional distribution of 0.7, 1.29, 1.13, 0.74, and 0.71 respectively. The C7+ portion of each separator liquid and recombined fluid mixture was replaced by compositions given by separate Gamma distribution models (different average molecular weights for each fluid, but all using exponential distribution) and the common multipliers. The C6- and C7+ fractions were recombined with the same mass ratio as measured by the laboratory. After recombination, the components were lumped to the component slate of EOS12-2008. Viscosity Regressions The Lohrenz-Bray-Clark (LBC) (Lohrenz et al. 1964) viscosity model is used in this study. Tuning of the critical Z-factors for the C6+-fraction was done to match viscosities. This tuning of critical Z-factors was performed after all other EOS regression variables were “locked in”. The liquid viscosity of a single carbon number component increases monotonically with increasing carbon number, if the same “paraffinicity” is assumed for all components. The critical Z-factors were tuned to match the liquid viscosity of individual SCNs of the C6+ fraction at reservoir temperature and standard pressure. Liquid viscosities were calculated using the Orrick and Erbar method (Reid et al. 1987; Yang et al. 2007), which is based on a relationship between liquid viscosity, molecular weight, and specific gravity. C6 through C9 are gases at reservoir temperature and standard pressure, so these components’ saturated liquid viscosities were calculated with reduced temperature (close to boiling point temperature). The critical properties and binary interaction parameters of model EOS12-2008 are given in Tables 3 and 4. Thermodynamic Consistency of EOS Model Several regression variables were chosen in developing the EOS12-2008 model. It can be difficult to develop a fluid characterization that predicts consistent behaving K-values with this many regression variables. Consistent K-values implies: (1) monotonically decreasing Ki with increasing molecular weights, and (2) Ki(p) that don’t cross. From Table 4 it is observed that the binary interaction parameters of EOS12-2008 are both negative and positive, e.g. the BIP between C1 and C5-6 is 0.06773 and the BIP between C1 and C7-10 is -0.17864. Non-monotonic BIPs, and particularly contiguous BIPs with opposite signs, may lead to non-physical K-value behaviour. Fig. 5a and b show calculated K-values with model EOS12-2008 for a CCE experiment on sample 4. From these figures it is seen that the K-values are consistent as a function of both molecular weight and pressure. The same consistency check was performed for all Pauto Complex samples; all with consistent behaving K-values.

8

SPE 135085

TABLE 3—EOS PARAMETERS FOR MODEL EOS12-2008. Component

Mi

Tc,i ( R)

Pc,i (psia)

ωi

si

Zc,i

CO2

44.010

547.42

1069.51

0.225

0.00191

0.27433

N2

28.014

227.16

492.84

0.037

-0.16758

0.29178

C1

16.043

343.01

667.03

0.011

-0.14996

0.28620

C2

30.070

549.58

706.62

0.099

-0.06280

0.27924

C3

44.097

665.69

616.12

0.152

-0.06381

0.27630

C4

58.123

752.27

541.14

0.19416

-0.05728

0.27724

o

C5-6

77.638

873.00

479.33

0.24831

-0.02975

0.26939

C7-10

117.787

1057.95

410.00

0.36861

-0.00635

0.26950

C11-14

174.530

1199.21

320.18

0.57595

-0.00597

0.27320

C15-20

239.550

1327.72

238.61

0.86854

0.08320

0.25763

C21-29

338.129

1433.66

192.66

1.17796

0.04564

0.27705

C30+

545.598

1541.35

166.98

1.53167

-0.21371

0.36811

TABLE 4—BINARY INTERACTION PARAMETERS FOR MODEL EOS12-2008. CO2

N2

C1

C2

C3

C4

C5-6

C7-10

C11-14

C15-20

C21-29

C30+

CO2

0

0.00000

0.10500

0.13000

0.12500

0.11704

0.11500

0.11500

0.11500

0.11500

0.11500

0.11500

N2

0.00000

0

0.02500

0.01000

0.09000

0.09500

0.10710

0.11000

0.11000

0.11000

0.11000

0.11000

C1

0.10500

0.02500

0

0.00000

0.00000

0.00000

0.06773

-0.17864

-0.03843

-0.05061

0.00741

0.00811

C2

0.13000

0.01000

0.00000

0

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000

C3

0.12500

0.09000

0.00000

0.00000

0

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000 0.00000

C4

0.11704

0.09500

0.00000

0.00000

0.00000

0

0.00000

0.00000

0.00000

0.00000

0.00000

C5-6

0.11500

0.10710

0.06773

0.00000

0.00000

0.00000

0

0.00000

0.00000

0.00000

0.00000

0.00000

C7-10

0.11500

0.11000

-0.17864

0.00000

0.00000

0.00000

0.00000

0

0.00000

0.00000

-0.08291

-0.11186

C11-14

0.11500

0.11000

-0.03843

0.00000

0.00000

0.00000

0.00000

0.00000

0

0.00000

-0.05090

-0.08123

C15-20

0.11500

0.11000

-0.05061

0.00000

0.00000

0.00000

0.00000

0.00000

0.00000

0

-0.02136

-0.05873

C21-29

0.11500

0.11000

0.00741

0.00000

0.00000

0.00000

0.00000

-0.08291

-0.05090

-0.02136

0

0.00000

C30+

0.11500

0.11000

0.00811

0.00000

0.00000

0.00000

0.00000

-0.11186

-0.08123

-0.05873

0.00000

0

1.E+02 1.E+01 1.E+00 1.E-01

K-value

1.E-02 1.E-03 1.E-04 C1 C4 C5-6 C7-10 C11-14 C15-20 C21-29 C30+

1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 10

100

1000 Pressure, psia

Fig. 5a—Calculated K-values for a CCE-experiment on sample 4.

10000

SPE 135085

9

100

10

K-value

1

C1 C4 C5-6 C7-10 C11-14 C15-20 C21-29 C30+

0.1

0.01

0.001 1000

10000 Pressure, psia Fig. 5b—Calculated K-values for a CCE-experiment on sample 4.

Fig. 6 shows the EOS calculated K-values at reservoir temperature and saturation pressure for all available Pauto Complex samples. The K-values are decreasing monotonically with increasing molecular weights. The same monotonically decreasing K-values are observed for saturation pressures at lower temperatures than reservoir temperature. 10

K-value

1

0.1 Sample 5 Sample 7 Sample 3 Sample 11 Sample 4 Sample 8 Sample 2 Sample 10

0.01

0.001 0

100

200

300

400

500

600

Molecular Weight Fig. 6—Calculated K-values at reservoir temperature and saturation pressure for the samples of Pauto Complex reservoirs.

10

SPE 135085

Fig. 7 shows separator gas and liquid compositions of sample 4. The laboratory measured compositions are compared with compositions calculated with EOS12-2008. Model EOS12-2008 predicts separator composition accurately for all samples. Some error is observed in the amount of C7+ in the separator gases. The EOS over-predicts this amount for three samples, and under-predicts the amount for six samples. It should be noted that the amount of C7+ in the separator gases is only about 0.2 mol-%. 1

Mole Fraction

0.1

0.01

0.001

Lab SepLiq

0.0001

Lab SepGas EOS SepLiq EOS SepGas

30 + C

29 21 C

20 C

15 -

14 11 C

10 7C

C

56

4 C

2

3 C

C

C

1

2 N

C

O

2

0.00001

Fig.7— Separator gas and liquid compositions of sample 4.

A consistency check on EOS predicted K-values are very important when extensive regressions on BIPs are performed, but it is good practice to do this consistency check for all developed fluid characterizations. Results of EOS tuning We present in this paper the EOS match of PVT data from samples 1, 5, 10, and 11. Sample 1 is a volatile oil sample, sample 5 is a medium-rich gas condensate sample, sample 10 is a lean gas condensate sample, and sample 11 is the medium-rich gas condensate sample used for the vaporization study. The sample compositions are given in Table 5. TABLE 5—RECOMBINED SAMPLE COMPOSITIONS FOR MODEL EOS12-2008. Sample 11 Component

Sample 1

Sample 5

Sample 10

EQL@4000 psia

Inj Gas

CO2

0.033564

0.028259

0.077737

0.027516

0.031502

N2

0.004529

0.005164

0.004392

0.002386

0.005000

C1

0.668671

0.680944

0.792996

0.526118

0.806760

C2

0.062180

0.092521

0.056769

0.079770

0.089807

C3

0.034468

0.047206

0.018721

0.043481

0.039403

C4

0.026307

0.043209

0.008944

0.031561

0.019501

C5-6

0.031758

0.019660

0.007541

0.033753

0.007026

C7-10

0.057327

0.041681

0.018501

0.096542

0.000999

C11-14

0.028959

0.018103

0.007124

0.059661

0.000000

C15-20

0.027758

0.014600

0.005014

0.059041

0.000000

C21-29

0.016746

0.006783

0.001897

0.031983

0.000000

C30+

0.007732

0.001871

0.000363

0.008188

0.000000

SPE 135085

11

Figs. 8-16 show results from CCE, viscosity, and CVD experiments performed on samples 1, 5, and 10. Calculated Zfactors using the Hall and Yarborough equation are plotted with the laboratory reported and EOS calculated Z-factors. The Hall and Yarborough equation is an accurate representation of the Standing-Katz Z-factor chart. Model EOS12-2008 predicts the Z-factors within the accuracy of the Standing-Katz Z-factors, which is 2-3%. 1.2 Lab EOS34-default EOS12-2006 EOS12-2008

Liquid Saturation

1

0.8

0.6

0.4

0.2

0 0

1000

2000

3000

4000

5000

6000

7000

8000

9000

Pressure, psia

o

Fig. 8—CCE liquid saturation behaviour of sample 1 at a temperature of 261 F, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

100 Lab

90

EOS34-default EOS12-2006

80

EOS12-2008

Removed, mol

70 60 50 40 30 20 10 0 0

1000

2000

3000

4000

5000

6000

7000

Pressure, psia Fig. 9—CVD removed moles behaviour of sample 1, comparing experimental data with the results of models EOS12-2006 and EOS122008.

12

SPE 135085

15

85 Lab EOS34-default EOS12-2006

80

EOS12-2008

9

75

6

70

3

65

0 0

1000

2000

3000

4000

C1 y, mol-%

C7+ y, mol-%

12

60 6000

5000

Pressure, psia Fig. 10—C1- and C7+-content behaviour of the liberated CVD gases of sample 1, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

1 Lab

0.9

EOS34-default EOS12-2006

Liquid Viscosity, cp

0.8

EOS12-2008

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

1000

2000

3000

4000

5000

6000

7000

8000

9000

Pressure, psia Fig. 11—Liquid viscosity behaviour of sample 1, comparing experimental data with the results of models EOS12-2006 and EOS122008.

SPE 135085

13

30

Liquid Saturation, %

25

20

15

10 Lab EOS34-default EOS12-2006 EOS12-2008

5

0 0

1000

2000

3000

4000

5000

6000

7000

Pressure, psia Fig. 12—CVD liquid saturation behaviour of sample 5, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

1.3 Lab EOS34-default Z Corr

1.2

EOS12-2006 EOS12-2008

Gas Z-Factor

1.1

1 +/-3%

0.9

0.8

0.7 0

1000

2000

3000

4000

5000

6000

7000

Pressure, psia Fig. 13—CVD gas Z-factor behaviour of sample 5, comparing experimental data with the results of models EOS12-2006 and EOS122008.

16

80

14

75

12

70

10

65

8

60

6

55

4

C1 y, mol-%

SPE 135085

C7+ y, mol-%

14

50

Lab EOS34-default

2

EOS12-2006

45

EOS12-2008

0 0

1000

2000

3000

4000

5000

40 7000

6000

Pressure, psia Fig. 14—C1- and C7+-content behaviour of the liberated CVD gases of sample 5, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

7 Lab EOS34-default EOS12-2006 EOS12-2008

6

Liquid Saturation, %

5

4

3

2

1

0 0

1000

2000

3000

4000

5000

6000

7000

8000

Pressure, psia Fig. 15—CVD liquid saturation behaviour of sample 10, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

8

86

7

84

6

82

5

80

4

78

3

76

2

C1 y, mol-%

15

C7+ y, mol-%

SPE 135085

74 Lab EOS34-default

1

72

EOS12-2006 EOS12-2008

0 0

1000

2000

3000

4000

5000

70 7000

6000

Pressure, psia Fig. 16—C1- and C7+-content behaviour of the liberated CVD gases of sample PD-VL-VLC2-MIR, comparing experimental data with the results of models EOS12-2006 and EOS12-2008.

Fig. 17 shows the initial liquid compositions at 4000 psia and 252 oF from the vaporization study on sample 11. Model EOS12-2006 over-predicts the amount of C7-14 by 6.9 wt-% and under-predicts the amount of C21+ by 8.9 wt-%. This error is reduced with model EOS12-2008 to 0.5 and 2.8 wt-% respectively. 100

1

0.1 Lab EOS34-default EOS12-2006 EOS12-2008

30 + C

21 -2 9 C

15 -2 0 C

11 -1 4 C

C

710

56

C

3 C

2 C

1 C

4

C

C

O

2

2

0.01

N

wt-%

10

Component o

Fig. 17—Initial liquid compositions at 4000 psia and 252 F in vaporization study of sample 11.

16

SPE 135085

Fig. 18 shows the liquid compositions of the final, fifth stage in the vaporization study of sample 11. Model EOS12-2006 under-predicts the amount of C7-14 by 2.0 wt-% and over-predicts the amount of C21+ by 7.9 wt-%. This error is reduced with model EOS12-2008 to 0.7 and 0.5 wt-% respectively. 100

wt-%

10

1

0.1 Lab EOS34-default EOS12-2006 EOS12-2008

30 + C

21 -2 9 C

15 -2 0 C

11 -1 4 C

C

C

710

56

4 C

3 C

2 C

1 C

2 O C

N

2

0.01

Component o

Fig. 18—Final liquid compositions at 4000 psia and 252 F in vaporization study of sample 11.

Model EOS12-2006 over-predicts liquid vaporization by gas injection. Figs. 19 and 20 show the C7+ content in the liberated gases and the liquid volumes from the vaporization experiment. EOS12-2008 accurately matches the C7+ content in the liberated gases from contacts 2-5. There are errors in the reported measured compositions from contact 1. The calculated liquid volume of contact 5 was under-predicted by 20% with EOS12-2008 compared with 62% with EOS12-2006. 60 Lab EOS34-default EOS12-2006 EOS12-2008

Liquid Volume, cm3

50

40

30

20

10

0 0 contact

1 contact

2 contact

3 contact

4 contact

5 contact

Fig. 19—Liquid volume behaviour in the vaporization study of sample 11, comparing experimental data with results of models EOS122006 and EOS12-2008.

SPE 135085

17

25 Lab EOS34-default EOS12-2006 EOS12-2008

y-C7+, wt-%

20

15

10

5

0 0 contact

1 contact

2 contact

3 contact

4 contact

5 contact

Fig. 20—C7+-content behaviour of the liberated gases in the vaporization study of sample 11, comparing experimental data with results of models EOS12-2006 and EOS12-2008.

The accuracy of the EOS calculated results of the vaporization study was greatly improved with model EOS12-2008 compared with model EOS12-2006. The match of the depletion PVT experimental results is about the same (or better) than the EOS12-2006 model. Fluid Characterization and Reservoir Management The two main products obtained from the fluid characterization process described above are: (1) fluid initialization and (2) a single EOS capable of accurately predicting fluid properties for depletion and gas injection reservoir processes. The fluid initialization work integrated observed spatial compositional variations and EOS calculated compositional gradients for all available sample compositions from Pauto Complex. The result of the initialization study provided validation and/or highlighted inconsistency of isolated, non-communicating fluid regions. This allowed more-reliable assessments of initial condensate/oil and gas in place, and development strategies to account for possible compartmentalization. The EOS model is important for proper modelling of near-well condensate blockage and value added by EOR gas injection processes. There may be several reasons why gas injection is desirable in Pauto Field Complex, e.g. to avoid reservoir pressure dropping below saturation pressure, mitigating water encroachment and geomechanical risk factors. A thorough reservoir uncertainty study, including all of above-mentioned issues (and others not pertinent to this paper), has been conducted to provide management with means to take informed decisions about the production strategy of Pauto Field Complex. These recommendations include drilling additional wells, facilities expansion, gas injection implementation, hydraulic fracturing, implementation of permanent downhole surveillance devices, continued surveillance strategy, and the acquisition of new fluid samples in potentially non-penetrated compartments, to test the robustness of the developed EOS and fluids models. Conclusions A single, consistent EOS has been successfully developed to describe the complex phase and volumetric behaviour of eight reservoir fluids in the Pauto Field Complex, with oil-gas ratios ranging from 30 to 350 bbl/MMscf. The EOS accurately matches PVT data describing depletion and vaporization processes. EOS model development included modifying heavycomponent properties, binary interaction parameters, and slight changes in sample-specific compositions. The BIPs were modified extensively, this being possible because equilibrium composition data were abundant. The final EOS shows thermodynamic consistency in K-value behaviour. Acknowledgements We would like to thank BP Exploration Company (Colombia) Ltd and ECOPETROL S.A. for support and permission to publish this paper.

18

SPE 135085

Nomenclature M = p = s = T = Zc = γ = ω =

molecular weight, g/gmol pressure, psia volume translation factor temperature, oR critical Z-factor specific gravity acentric factor

Subscripts C = i = R = S = wf =

critical component i reservoir saturation well flowing

References Chueh, P.L. and Prausnitz, J.M.: “Calculation of High-Pressure Vapor-Liquid Equilibria,” Ind. Eng. Chem. (1968) 60, No. 13. Hall, K.R. and Yarborough, L.: “A New EOS for Z-factor Calculations,” Oil & Gas J., June 1973, 82. Jhaveri, B.S. and Youngren, G.K.: “Three-Parameter Modification of the Peng-Robinson Equation of State To Improve Volumetric Predictions,” SPERE (August 1988) 1033. Lohrenz, J., Bray, B.G., and Clark, C.R.: “Calculation Viscosities of Reservoir Fluids From Their Compositions,” JPT (October 1964) 1171; Trans., AIME, 231. Peneloux, A., Rauzy, E., and Freze, R.: “A Consistent Correlation for Redlich-Kwong-Soave Volumes,” Fluid Phase Equilibria (1982) 8, 7. PhazeComp, Zick Technologies, 1995-2009. Reid, R.C., Prausnitz, J.M., and Polling, B.E.: “The Properties of Gases and Liquids,” Fourth Edition, McGraw-Hill Book Co. Inc., New York City (1987) 388-485, Section 9-11. Robinson, D.B., Peng, D.Y., and Ng.H.-Y.: “Capability of the Peng-Robinson Programs, Part 2: Three-Phase and Hydrate Calculations,” Hydrocarbon Proc. (1979) 58, 269. Søreide, I.: “Improved Phase Behavior Predictions of Petroleum Reservoir Fluids from a Cubic Equation of State,” Dr.Ing. dissertation, Norwegian Inst. of Technology, Trondheim, Norway (1989). Standing, M.B. and Katz, D.L.: “Density of Natural Gases,” Trans., AIME, 1942, 146, 140. Twu, C.H.: “An Internally Consistent Correlation for Predicting the Critical Properties and Molecular Weights of Petroleum and Coal-Tar Liquids,” Fluid Phase Equilibria (1984) No.16, 137. Whitson, C.H.: “Characterizating Hydrocarbon Plus Fraction Fractions,” SPEJ (August 1983) 683; Trans., AIME, 275. Whitson, C.H.: “Effect of C7+ Properties on Equation-of-State Predictions,” SPEJ (December 1984) 685; Trans., AIME, 277. Whitson, C.H. and Brule, M.R.: “Phase Behavior”, SPE monograph (2000). Yarborough, L. and Hall, K.R.: “How to Solve EOS for Z-factors”, Oil & Gas J., February 1974, 86. Yang, T., Fevang, Ø., and Christoffersen, K.: “LBC Viscosity Modeling of Gas Condensate to Heavy Oil”, paper SPE 109892 presented at the 2007 SPE Annual Technical Conference and Exhibition, California, 11-14 November.

SI Metric Conversion Factors o API 141.5/(131.5+oAPI) bbl x 1.589 873 E-01 cp x 1.0 E-03 ft x 3.048 E-01 o F (oF - 32)/1.8 MMscf x 2.831 685 E+04 psi x 6.894 757 E+00 o R x 5/9 Tcf x 2.831 685 E+10

= = = = = = = = =

g/cm3 m3 Pa.s m o C m3 kPa K m3