Beruflich Dokumente
Kultur Dokumente
where
[
Rs = a gb oc T d p ]
e
(6)
X = A1 + A2 Bob + A3 g + A4 Bob2 where
1 T = Temperature, o F
+ A5 + A6 (
2
) (2) a e = Coefficients of the above equation having these
g Bob
g
values:
a =0.1769 b =1.0674
A1 = 1.5897 A2 = 0.2735 c =-5.0956 d =-0.1294
e =1.0857
A3 = 0.4429 A4 = 0.04692 The average error of the correlation was reported at 1.31%
with a standard deviation of 4.21%.
A5 = 0.1440 A6 = 0.1596
Statistical Error Analysis
and T = Temperature, F
o There are four main statistical parameters that are being
The above correlation was originated from studying PVT considered in this study. These parameters help to evaluate the
data on 287 data points. The average error of the correlation accuracy of the predicted fluid properties obtained from the
was reported at 1.32% with a standard deviation of 4.22%. black oil correlations.
maximum values are scanned to derive the range of error for Graphic Error Analysis
each correlation. The equations of these values are given Graphic means help in visualizing the accuracy of a
below: correlation. Two graphic analysis techniques were used.
Cross plot
Emax = max in=d1 Ei (10) In this technique, all the estimated values are plotted against
the experimental values, and thus a cross plot is formed. A 45
and [0.79-rad] straight line is drawn on the cross plot on which the
E min = min in=d1 Ei (11) estimated value is equal to the experimental value. The closer
the plotted data points are to this line, better the correlation.
The accuracy of a correlation can be examined by
maximum absolute percent relative error. A higher accuracy is Error Distribution
achieved when the maximum value is small.
The deviations, E i for a good correlation are expected to be as
Standard Deviation8 close as possible to the normal distribution. The equation of a
normal-distribution curve to fit any data set can be derived by
Standard deviation, s x , of the estimated values with respect to
the use of the mean and standard deviation of the data set. This
the experimental values can be calculated using the following technique involves presenting relative frequency of deviation
equation: in histograms and then fitting a normal-distribution curve to it.
nd The accuracy of the correlation is then judged by matching the
1
] E i
2 2
sx = [ (nd 1) i =1
(12) error distribution with the normal-distribution curve.
r2 = 1 m
(13) The cross plot of the experimental against the calculated
bubble point pressure using the new correlation is presented in
[( x exp x) i ] 2 Figure 1. Most of the data points of the new correlation fall
i =1
very close to perfect correlation of the 45 line. A graphical
plot of residual (the difference between experimental and
where
calculated bubble point pressure) and experimental bubble
1 nd point pressure (shown in Figure 1) demonstrate a uniform
x = ( ) ( xexp ) i (14) distribution of errors with most of the data points falling
nd i =1 within 500 psia residual line. The other correlation results
The square value of correlation coefficient is known as are depicted in Figures 2 through Figure 8. As shown in Figure
coefficient of determination. The coefficient of determination 4 the negative amounts related to low bubble point pressure
is defined as the proportion of the validity in the dependent might be due to the limitation range of Petrosky and Farshad
variable (predicted) that is encountered for by the independent correlation which is valid for the range of 1574 to 6523 psia.
variable (experimental). In the correlation, a new independent variable PVT
parameter of bubble point oil FVF is introduced into the
4 SPE 104543
equation. The value of bubble point oil FVF can either be distribution of errors with most of the data points falling
obtained experimentally or estimated from correlations. The within 150 SCF/STB residual line.
above bubble point pressure study was based on bubble point
oil FVF obtained from experimental values. Table 5 shows the Conclusions
statistical error analysis results of calculated bubble point Based on the results of this study, the following conclusions
pressure where the bubble point oil FVF was used in the are obtained:
equation. The statistical analysis results show no significant 1- New empirical correlations for Iranian crude oils have been
differences between the two values of calculated bubble developed for bubble point pressure, solution gas-oil-ratio, and
pressure. The calculated bubble point pressure produced an bubble point oil formation volume factor.
AAPRE of 3.71 percent with 4.06 percent standard deviation. 2- All of the proposed correlations have a wide range of
In the absence of experimental bubble point FVF this value validity, and are superior to other published correlations in the
can be estimated from equation 3 and used to calculate bubble literature.
point pressure without losing the accuracy. 3- These correlations were developed specifically for Iranian
crude oils but can be used for estimating the same PVT
Bubble Point oil FVF parameters for all types of oil and gas mixtures with properties
The statistical error analysis results of the bubble point oil falling within the range of the data used in this study.
FVF correlation as compared with other known correlations 4- The new bubble point oil formation volume factor
are shown in Table 6. The correlation provides higher correlation provided the highest accuracy of the correlations
accuracy in estimating bubble point oil FVF for Iranian crude evaluated; however, the other published correlations also
oils. Amongst the correlations, this correlation gives the produced excellent estimates of bubble point oil formation
lowest values of AAPRE, standard deviation 1.08 percent, volume factors.
1.71 percent, respectively. The correlation coefficient is 0.993, 5- The PVT correlations can be placed in the following order
which is close to an ideal value of 1.0. This shows that the with respect to their accuracy:
new bubble point oil FVF correlation corresponds better with a) Bubble point pressure: this study, Al-Marhoun,
experimental data than any other correlations. Standing, Hanafy, and Dindoruk.
In the graphical error analysis, the cross plot of b) Oil formation volume factor: this study, Petrosky,
experimental and estimated bubble point oil FVF shows that Standing, Glaso, Dokla, Al-Marhoun, and Hanafy.
most of the data points fall along the perfect correlation of c) Solution gas oil ratio: this study, standing,
45 line (Figure 9). This is reflected with good Al- Marhoun, Dindorak, Glaso, and Petrosky.
r 2 (coefficient of determination) value of 0.986 .The residual 6- The Proposed correlations can be tuned for other
plot of the bubble point oil FVF obtained from the new basins/areas, or certain class of oils.
correlation and others are shown in Figure 9 through Figure
16. All the data points in Figure 9 (except three data points) lie Acknowledgment
This study has been sponsored by the National Iranian Oil
between 0.1 residual lines. A small residual value indicates
Company, Research and Development Directorate (NIOC,
a better accuracy of the new correlation in estimating bubble
R&D) which is gratefully acknowledged.
point oil FVF for Iranian crude oils.
Nomenclature
Solution GOR
The statistical error analysis results of the Solution GOR AAD=Abosolute Average Deviation
correlation as compared with other known correlations are API=Stock-tank oil gravity, API
shown in Table 7. It shows that the proposed correlation has ARE= Abosolute Relative Error
an average absolute deviation of 4.07% compared with over Bo = Oil Formation Volume Factor, bbl/stb
16% for Dindoruk and 20.7% for Glaso. The correlation
coefficient is 0.991 which is close to an ideal value of 1.0. Bob = Oil FVF at bubble point pressure, bbl/stb
This shows that the Solution GOR correlation correlates better
with experimental data than any other correlations. Dev. =Deviation
The maximum error and standard deviation of these
E a = Average absolute relative error, %
correlations are in Table 7. Figure17 through Figure 22 give
cross plots of the values estimated by the same correlations
versus the measured experimental values. It is clear from both Ei = Percent relative error
Table 7 and Figure 17 through Figure 22 that the proposed
correlation is quite superior for Iranian crude oils than other E max = Maximum absolute percent relative error
correlations. The cross plot of the experimental against the
calculated solution GOR using the new correlation is
presented in Figure 17. Most of the data points of the new E min = Minimum absolute percent relative error
correlation fall very close to the perfect correlation of
45 line. A graphical plot of residual and experimental E r = Average relative error, %
solution GOR (shown in Figure 17) demonstrated a uniform
SPE 104543 5
f = Function Subscripts
b = bubble point
m = Number of data sets est =estimated from correlation
exp =experimental
nd = Number of data points max =maximum
min =minimum
OFVF=Oil Formation Volume Factor g = gas
o= oil
P = Pressure, psia s = solution
References
Pb = Bubble point pressure, psia 1. Standing, M.B., A Pressure-Volume-Temperature Correlation for
Mixtures of California Oils and Gases, Drilling and Production
PVT=Pressure Volume Temprature Practice, API, 275-287.
2. Glaso, O.,Generalized Pressure-Volume-Temperature
r = Coefficient of correlation Correlations, JPT (May 1980), 785-795.
3. Al-Marhoun, M.A.," PVT Correlations for Middle East
SCF Crude Oils", JPT (1988) 650-665.
Rs = Solution gas-oil-ratio, 4. Hanafy, H.H., Macary, S.A., Elnady, Y. M., Bayomi, A.A. and El-
STB Batanoney, M.H., Empirical PVT Correlation Applied to
Egyptian Crude Oils Exemplify Significance of Using Regional
S x = Standard deviation Correlations, SPE 37295, SPE Oilfield CHEM.INT.SYMP
(Houston, 2/18-21/97) PROC (1997) 733-737.
Std. =Standard 5. Dindoruk B., Peter G. Christman,"PVT Properties and Viscosity
Correlations for Gulf of Mexico Oils", SPE Annual Technical
T = Temperature, o F Conference and Exhibition, New Orleans, Louisiana, 30 September-3
Octobor.
x = Any physical quantity 6. Dokla, M, and Osman, M.E., Correlation of PVT Properties for
the UAE Crudes. SPE Formation Evaluation (1992) 41-46.
7. Petrosky, G.E.Jr., and Farshad, F.F., Pressure-Volume-
x = Average value of xexp Temperature Correlations for Gulf of Mexico Crude Oils, SPE
26644 presented at 68th Annual Meeting of the Society of
Petroleum Engineers, Houston, Texas (1993) 3-6.
g = Gas specific gravity (air=1) 8. Almehaideb, R.A., "IMPROVED PVT CORRELATIONS FOR
UAE CRUDE OILS", SPE 37691-MS, Middle East Oil Show and
o = Oil specific gravity (water=1) Conference, 15-18 March, Bahrain (1997)
Table 2: Data Description for Bubble Point Pressure, GOR, and Bubble Point Oil FVF.
Number of points PVT Property Minimum Maximum Mean
287 Bubble point Pressure, psia 348 5156 2780.4
287 Solution GOR, SCF/STB 125 2189.25 783.3
287 FVF, bbl/STB 1.091 2.54 1.44
287 Temperature, F 77.5 290 165
o
287 Stock tank oil gravity, API 18.8 48.34 30.1
287 Gas Gravity,(air=1) 0.523 1.415 0.924
6 SPE 104543
Table 3: Range for Bubble Point Pressure, Solution GOR, and Oil FVF Correlations.
Al-
Standing1 Glaso2 Petrosky7
Marhoun3
o
Tank oil gravity, API 16.5 to 63.8 22.3 to 48.1 19.4 to 44.6 16.3 to 45
Bubble point pressure, psia 130 to 7000 165 to 7142 130 to 3573 1574 to 6523
o
Reservoir temperature, F 100 to 258 80 to 280 74 to 240 114 to 288
Oil FVF at bubble point 1.024 to 1.025 to 1.032 to
1.1178 to 1.6229
,bbl/STB 2.15 2.588 1.997
Solution GOR, SCF/STB 20 to 1425 90 to 2637 26 to 1602 217 to 1406
Total surface gas 0.65 to 0.752 to
0.59 to 0.95 0.5781 to 0.8519
gravity(air=1) 1.276 1.367
Separator pressure, psia 256 to 465 415(mean) - -
o
Separator temperature, F 100(mean) 125(mean) - -
Reservoir pressure, psia - - 20 to 3573 1700 to 10692
Table 4: Summary of Statistical Measures for Pb for New and Common Correlations
7 2 1 3 5 6 4
Proposed Petrosky Glaso Standing Al-Malhoun Dindoruk Dokla Hanafy
%AAD 3.67 15.13 22.43 8.28 8.77 42.46 18.50 16.88
%ARE 1.32 -6.99 -22.43 -4.36 8.00 -42.45 7.51 -12.06
%Max.
10.95 161.96 73.13 45.68 26.04 91.01 197.81 192.34
Dev.
%Min Dev. 0.00012 0.185 1.591 0.009 0.105 0.070 0.108 0.192
Std. Dev. 4.22 24.58 25.69 11.28 10.57 44.65 29.08 30.15
r 0.993 0.91 0.84 0.95 0.97 Negative 0.8549 0.94
Table 6: Summary of Statistical Measures for Bob for New and Common Correlations
3 1 2 7 6 7 4
Proposed Al-Marhoun Standing Glaso Petrosky Dokla Dindoruk Hanafy
%AAD 1.08 2.25 1.88 1.96 1.35 2.55 3.61 8.89
%Max Dev. 8.05 15.25 11.21 12.28 13.76 13.08 26.94 22.65
%Min. Dev. 0.006 0.003 0.0008 0.0164 0.015 0.0045 0.0091 0.135
Std Dev. 1.706 3.30 2.65 2.59 2.35 3.35 4.92 9.69
r 0.9929 0.965 0.985 0.9849 0.9870 0.9805 0.9495 0.7676
SPE 104543 7
Table 7: Summary of Statistical Measures for Solution GOR for New and Common Correlations.
3 1 2 7 5
Proposed Al-Marhoun Standing Glaso Petrosky Dindorak
%AAD 4.07 14.95 8.73 20.70 15.70 16.80
%ARE -0.33 13.43 -4.23 -20.70 -6.70 +7.53
%Max Dev. 10.86 101.7 32.99 38.29 116.41 126.48
%Min Dev. 0.0003 0.1471 0.0009 2.0813 0.1522 0.0138
Std. Dev. 4.95 19.98 11.01 22.06 23.17 24.21
r 0.9911 0.9014 0.9536 0.8920 0.8839 0.8950
Residual ,psia
pressure,psia
4100
3100 0
2100
1100
-500
100
100 1100 2100 3100 4100 5100 0 2000 4000 6000
Figure 1: Cross plot and residual plot of Pb (this study) based on Iran PVT data.
Standing Standing
5100 1500
Calculated Bubble point pressure,psia
1000
4100
Residual ,psia
500
3100
0
2100 -500
1100 -1000
-1500
100 0 2000 4000 6000
100 1100 2100 3100 4100 5100
Measured Bubble point pressure,psia
Experimental bubble point pressure ,psia
Figure 2: Cross plot and residual plot of Pb (Standing) based on Iran PVT data.
8 SPE 104543
Marhoun Marhoun
Calculated Bubble point pressure,psia
1000
5100
4100 500
2100
-500
1100
100 -1000
100 2100 4100 0 1000 2000 3000 4000 5000 6000
Figure 3: Cross plot and residual plot of Pb (Marhoun) based on Iran PVT data.
5500
Calculated Bubble poin
1000
500
pressure,psia
Residual ,psia
3500 0
-500
-1000
1500
-1500
-2000
-500 0 2000 4000 6000
0 2000 4000 6000
Experimental bubble point pressure ,psia
Measured Bubble point pressure
Figure 4: Cross plot and residual plot of Pb (Petrosky) based on Iran PVT data.
Glaso Glaso
1500
Calculated Bubble point
5100 1000
Residual ,psia
pressure,psia
4100 500
3100 0
2100 -500
1100 -1000
100 -1500
100 2100 4100 0 2000 4000 6000
Experimental bubble point pressure ,psia
Measured Bubble point pressure,psia
Figure 5: Cross plot and residual plot of Pb (Glaso) based on Iran PVT data.
SPE 104543 9
Hanafy Hanafy
1000
5100
500
Calculated Bubble point
4100
Residual ,psia
pressure,psia
0
3100
-500
2100
-1000
1100
100 -1500
100 2100 4100 0 2000 4000 6000
Figure 6: Cross plot and residual plot of Pb (Hanafy) based on Iran PVT data.
Dokla Dokla
2000
1500
5100
Calculated Bubble point
1000
Residual ,psia
pressure,psia
4100
500
3100
0
2100
-500
1100
-1000
100 0 2000 4000 6000
100 2100 4100
Experimental bubble point pressure ,psia
Measured Bubble point pressure,psia
Figure 7: Cross plot and residual plot of Pb (Dokla) based on Iran PVT data.
Dindoruk Dindoruk
1000
5100 500
Calculated Bubble point
0
4100
Residual ,psia
pressure,psia
-500
3100 -1000
-1500
2100
-2000
1100
-2500
100 -3000
100 2100 4100 0 2000 4000 6000
Measured Bubble point pressure,psia
Experimental bubble point pressure ,psia
Figure 8: Cross plot and residual plot of Pb (Dindorak) based on Iran PVT data.
10 SPE 104543
This study
This study correlation
3
0.2
2.8
2.6
Calculated Bo,bbl/STB
2.4 0.1
Residuals,bbl/STB
2.2
2
0
1.8
1.6
1.4 -0.1
1.2
1 -0.2
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 9: Cross plot and residual plot of Bo (this study) based on Iran PVT data
2.6
Residuals,bbl/STB
2.4
0.1
2.2
2 0
1.8
1.6 -0.1
1.4
-0.2
1.2
1 -0.3
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 10: Cross plot and residual plot of Bo (Standing) based on Iran PVT data.
SPE 104543 11
Residuals,bbl/STB
2.4
0.2
2.2
2
1.8
1.6 0
1.4
1.2
1 -0.2
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 11: Cross plot and residual plot of Bo (Al-Marhoun) based on Iran PVT data.
2.6
Residuals,bbl/STB
2.4
0.2
2.2
2
1.8
1.6 0
1.4
1.2
1 -0.2
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 12: Cross plot and residual plot of Bo (Petrosky) based on Iran PVT data.
2.6
Residuals,bbl/STB
2.4
0.2
2.2
2
1.8
0
1.6
1.4
1.2
1 -0.2
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 13: Cross plot and residual plot of Bo (Glaso) based on Iran PVT data.
12 SPE 104543
Dokla
Dokla correlation
3 0.4
2.8
Calculated Bo,bbl/STB
2.6
Residuals,bbl/STB
2.4
0.2
2.2
2
1.8
1.6 0
1.4
1.2
1 -0.2
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 14: Cross plot and residual plot of Bo (Dokla) based on Iran PVT data.
2.6
Residuals,bbl/STB
2.4
2.2 0.1
2
1.8 -0.1
1.6
-0.3
1.4
1.2
-0.5
1
0 1 2 3
1 1.5 2 2.5 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 15: Cross plot and residual plot of Bo (Dindoruk) based on Iran PVT data.
Hanafy
Hanafy correlation
3 0.2
2.8
0.1
2.6
Calculated Bo,bbl/STB
Residuals,bbl/STB
2.4 0
2.2
-0.1
2
1.8 -0.2
1.6 -0.3
1.4
-0.4
1.2
1 -0.5
1 1.5 2 2.5 3 0 1 2 3
Measured Bo,bbl/STB Experimental bubble point oil FVF,bbl/STB
Figure 16: Cross plot and residual plot of Bo (Hanafy) based on Iran PVT data.
SPE 104543 13
1700
Residuals,SCF/STB
1300
0
900
500
100 -200
100 600 1100 1600 0 500 1000 1500
Measured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 17: Cross plot and residual plot of Rs (this study) based on Iran PVT data.
2100
200
Residuals,SCF/STB
1600
0
1100
-200
600
100 -400
100 600 1100 1600 2100 0 500 1000 1500 2000
Measured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 18: Cross plot and residual plot of Rs (Glaso) based on Iran PVT data.
14 SPE 104543
1650
Residuals,SCF/STB
1250 400
850
0
450
50 -400
50 550 1050 1550 0 500 1000 1500 2000
Measured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 19: Cross plot and residual plot of Rs (Standing) based on Iran PVT data.
1700
Residuals,SCF/STB
0
1300
900
-400
500
100 -800
100 600 1100 1600 2100 0 500 1000 1500 2000
Measured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 20: Cross plot and residual plot of R s (Al-Marhoun) based on Iran PVT data.
2100
Calculated Solution GOR,SCF/STB
400
Residuals,SCF/STB
1600
0
1100
-400
600
100 -800
100 600 1100 1600 2100 0 500 1000 1500 2000
Measured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 21: Cross plot and residual plot of Rs (Dindoruk) based on Iran PVT data.
SPE 104543 15
1750 400
Residuals,SCF/STB
1350 0
950 -400
550 -800
150 -1200
150 650 1150 1650 2150 0 500 1000 1500 2000
Meas ured Solution GOR,SCF/STB Experimental solution GOR,SCF/STB
Figure 22: Cross plot and residual plot of Rs (Petrosky) based on Iran PVT data.