Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 36 (2013) 738 745

TerraGreen 13 International Conference 2013 - Advancements in Renewable Energy

and Clean Environment

Empirical and Correlative Study of PVT Proprieties of

Algerian Oil
Hadj Mohammed SIDROUHOU* and Mourad KORICHI
Hydrocarbon Processing Department, Kasdi Merbah University, Ouargla DZ- Algeria
sidrouhou.el@univ-ouargla.dz

Abstract
The knowledge of the Pressure - Volume - Temperature (PVT) parameters is of an major importance in calculations
of the properties of the fluids and the development of the layers, namely, the determination of the bubble pressures
(Pb) and the factor of volume of oil formation (Bo), the design of the treatment units of the fluids and the estimate of
the volumetric characteristics in the oil tanks.
In absence of PVT data, the use of the correlations is capital for the prediction of PVT properties, thus the properties
associated. Sight the geological effect, which represents the modification of the chemical composition of crude oil
from one area to another, it is difficult to obtain such a precise result thanks to empiricisms, from this point of view,
correlations PVT must be studied and modified according to the regional characteristics of oils.
The development of correlations PVT was the scientific research task object many, which contributed to a great
number of publications in differing, re-examined. Several graphic correlations (in the form of abacus) and
mathematics are developed during two last decades to determine, in particular, Pb and Bo. These correlations are
primarily based on the assumption that Pb and Bo are in functions direct with the rate of gas-oil solution (Rs), the
temperature of tank (Tf), the specific gravity of gas (g), and the specific oil gravity (o). Moreover, the success of
such a correlation depends mainly on the beach of the experimental data at which it was developed, as well as the
geographical area.

2013
2013The
The
Authors.
Published
by Elsevier

Authors.
Published
by Elsevier
Ltd. Ltd.
Selection
peer-review
underunder
responsibility
of the TerraGreen
Academy Academy.
Selectionand/or
and/or
peer-review
responsibility
of the TerraGreen
Keywords: Data PVT, PVT Correlations, Petroleum Engineering, Algeria Crude Oil.

* Corresponding author. Tel.: +213-771-337-224; Fax: +213-2971 - 26 - 27

E-mail address: sidrouhou.el@univ-ouargla.dz.

1876-6102 2013 The Authors. Published by Elsevier Ltd.

Selection and/or peer-review under responsibility of the TerraGreen Academy

doi:10.1016/j.egypro.2013.07.086

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745

1. Introduction :
The calculation of the reservations in oil reservoir or determination of its execution and economics scenes
requires a good knowledge of the physical properties of fluids. Bubble Pressure (Pb), Factor Volume of
Oil Formation (Bo) and Gas Oil Ration (GOR) are of importance primary in the calculation of matter
balance; the correlations on PVT which is generally used in oil industry are the important tools in
calculations of performance of tank.
For the development of a correlation, the geological state is considered important because the chemical
composition of crude oil differs from the area to the area. It is difficult to obtain the same precise results
by empirical correlations for different oil samples having different physical and chemical characteristics.
Consequently, to explain regional characteristics, correlations of PVT must be modified before their
application.
This study examines existing correlations of PVT with their original coefficients and the modified
coefficients based on data of PVT collected of different places in the south of Algeria. The statistical
exactitude and validity are determined for these correlations and finally more adapted correlations are
recommended for their application to crude oils of Algeria. In addition, this study can be employed like
effective directive for the application of the correlations with various crudes.

Nomenclature
Physical Properties (Parameters of Oil)
API

API degree oil

Bo

Factor Volume of Oil Formation (scf/STB)

GOR

Gas Oil Ration

Pb

Bubble Pressure (atm.)

Statistical Parameters
Ea
2

Average absolute error %

Correlation coefficient

Standard Deviation

2. PVT Correlation:
Last six decades, the engineers realized of the importance to develop and employ empirical correlations
for PVT properties. The most popular empirical correlations [1 - 6] used in this work are:

739

740

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745

x Standing (1947): a correlation for the forecast of the pressure bubble presented by correlating the
temperature of tank, report/ratio of gas-oil of solution, the relative density of gas and the degree of API
oil gravity. By using the same data used for its forecast of pressure of bubble the first correlation for
the factor of volume of oil formation developed [1].
x Lasater (1958): for its development of correlation for the pressure of point of bubble gathered data of
the fractions of gas of solution [2].
x Vasquez and Beggs (1980): brought back their work for the forecast of pressure of bubble and the
factor of volume of formation of a crude saturated with gas and they brought back their research
recommending a junction in the data with two ranges of API oil gravity [3].
x Glaso (1980): also presented a correlation for the forecast of the pressure of the point of bubble of a
comprising whole of data of the temperature of tank, report/ratio of gas of solution/oil, the relative
density of gas and the degree of API oil gravity. For the factor of volume of oil formation a correlation
published which was based on the correlation of Standing with minor modifications [4].
x Al-Marhoun (1988): its correlation published to determine the pressure of point of bubble and the
factor of volume of formation based on 160 benchmarks of the oil samples of the Middle East [5].
x Al-Marhoun (1992): updated its beginning correlation 1988 for the factor of volume of formation by
acquiring a great whole of data of 4012 benchmarks gathered of everywhere in the world [6].
3. PVT Data Acquisition :
The data used in this study are obtained from the analysis of 884 samples coming from 51 different wells
in the three popular fields in the south of Algeria, namely Bir Berkine, Hassi Messaoud (HMD) and
Ohanet). After preliminary treatment of data, only 384 point of are introduced in the modeling approach
to correlate Pb and Bo to reservoir parameters. Table 1 present all numbers of data points used study.
Table 1.Data Base
Areas

Bir Berkine

HMD

Ohanet

Total

Number of the samples

268

512

104

884

Number of the data

156

157

71

384

4. Process of evaluation:
Existing correlations of PVT are applied to the whole of data and an analysis of complete error is
carried out based on a comparison between the value envisaged and the original experimental value. For a
detailed analysis of the exactitude of the examined correlations, an analysis of error based on various
ranges of degree of API oil gravity is also carried out graphically. An analysis of error based on ranges of
degree of API oil gravity is considered an effective tool to determine the suitability of the correlation for
heavy, average or light oils
5. Results and Comparisons :
The statistical analysis shows an important improvement in the errors for all the correlations with the
new coefficients. Such a comparison will indicate that the correlation with the least error offers a better

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745

explanation of the parameters of PVT. After application of the existing correlations to the whole of
gathered data, average results in the form of relative error, the standard deviation and the correlation
coefficient are recapitulated in the tables 2a, 2b, 3a and 3b.
Another effective comparison of the correlations can be carried out by the chart of the errors according
to the ranges of degree of API oil gravity. The average absolute error is an important indicator of the
exactitude of an empirical correlation. It is employed here to examine the exactitude of the validation of
standard correlations and modified correlations (correlations with new constants adapted to the data of the
3 separate and grouped fields).
5.1. Bubble point pressure:
Table 2a and 2b proves that the correlation of Al-Marhoun [5] has the least errors for the data used of
the field of Bir Berkine and for the grouped data of the three fields, the correlation of Standing [1] for the
data used of the field of HMD and the correlation of Laseter [2] for the data used of the Ohanet field. The
statistical analysis shows an important improvement in the error with modified coefficients and the model
of network of neuron.
The correlation of Al-Marhoun [5] exceeds the remainder of the correlations studied with an average
absolute error of 31.73% and 20.31% after modification of the correlation for the field Bir Berkine, and
with an average absolute error of 11.89% for neural network model.
For the grouped data of the three fields, it gives an average absolute error of 27.28% and 25.45% after
modification and 10.97% by the neural network model. For HMD field, the correlation of Standing[1]
gives: 25.91%, 19.57% and 03.30% by the neural network model and for the Ohanet field the correlation
of Lasater [2] gives: 18.48%, 17.68% and 02.28% by the neural network model.
Figure 1 represents the comparison between experimental Pb and estimated Pb by modified correlations
for each fields and with coefficient of correlation, R=0.977, 0.950, 0.767 and 0.762 for Bir Berkine field,
HMD field, Ohanet field and grouped fields respectively.
5.2. Factor of volume of formation:
After the analysis of the errors, the correlations of Vasquez and Beggs [3], Standing [1] and AlMarhoun [6] have less error; 11.19%, 06.97%, 09.79% and 10.23% for the fields Bir Berkine; HMD,
Ohanet and for the three fields grouped respectively.
After the statistical analysis one sees an important improvement in the error with modified coefficients
and by the model of network of neuron.
Figure 2 represents the graphic comparison between Bo experimental and Bo estimated by the modified
correlations for each fields and with maximum coefficient of correlation, R =0.938; 0.741; 0.992 and
0.841 for the fields Bir Berkine, HMD, Ohanet and three fields grouped respectively.

741

742

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745
Table 2a. Pb statistical study for each field

Bir Birkine Field

Ea
s
R

HMD Field
Ea
s
R

Standing (47)

46,35

44,35

25,91

38,05

Lasater (58)

45,46

42,93

32,49

24,71

18,48

07,13

Vasquez et Beggs (80)

47,37

49,45

31,02

42,06

30,57

36,07

Glaso (80)

41,25

44,66

32,04

42,58

36,34

40,72

Al-Marhoun (88)

31,73

26,51

26,50

34,68

19,24

24,51

20,31

26,96

0,977

19,57

107,5

0,950

17,68

09,47

0,767

11,89

0,756

03,30

0,778

02,28

0,996

Correlations

Al-Marhoun (88) modified

(This study)
Standing (47)
modified (This study)
Lasater (58)
modified (This study)
Neural Network Model
(This study)

Ohanet Field
Ea
s
R
28,39

34,08

Table 2b. Pb statistical study for all field data

Ea

Standing (47)

Correlations

34,48

42,81

Lasater (58)

34,30

28,06

Vazquez et Beggs (80)

37,53

47,17

Glaso (80)

36,76

46,70

Al-Marhoun (88)

27,28

31,28

Al-Marhoun (88) modified (This study)

25,45

46,99

0,762

Neural Network Model (This study)

10,97

Figure.1. Comparison between estimated Pb by the modified correlations and experimental Pb

743

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745

Table 3a. Bo statistical study for each field

Bir Birkine Field

Ea
s
R

Ea

Standing (47)

16,13

43,76

06,97

14,18

11,37

15,70

Vasquez et Beggs (80)

11,19

21,85

08,07

14,49

15,72

16,69

Correlations

HMD Field
s
R

Ohanet Field
Ea
s
R

Glaso (80)

13,79

30,12

07,64

14,02

10,88

14,24

Al-Marhoun (88)

14,55

34,82

07,67

14,32

10,76

15,24

Al-Marhoun (92)

13,54

32,52

07,44

14,35

09,79

12,76

08,86

17,20

0,938

06,45

13,14

0,741

08,91

12,36

0.992

Vasquez et Beggs (80)

modified (This study)
Standing (47) modified
(This study)
Al-Marhoun (92) modified
(This study)

Table 3b. Bo statistical study for all field data

Correlations

Ea

Standing (47)

11,13

29,79

Vazquez et Beggs (80)

11,02

18,75

Al-Marhoun (92)

10,23

23,05

Al-Marhoun (92) modified (This study)

10,09

22,33

0,841

Neural Network Model (This study)

09,26

Glaso (80)

10,80

22,13

Al-Marhoun (88)

10,42

23,21

Figure.2. Comparison between estimated Bo by the modified correlations and experimental Bo

744

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745

6. Conclusion
The following conclusions were drawn from this study on evaluation for data of the Algerian crude :
1) The correlations recommended give better predictions with original and new coefficients for the
prediction of the pressure of bubble (Pb):
x Al-Marhoun [5] Correlation is the best for the Bir Berkine field and the grouped data;
x Standing [1] Correlation for HMD field;
x Correlation of Lasater [2] for the Ohanet field.
2) Correlations recommended and because of its least errors and less standard deviation and with
original and new coefficients, they predicted better than of other correlations for the factor of
volume of formation (Bo) of oil:
x All the studied correlations showed a good prediction compared to the experimental data.
x Vasquez and Beggs [3] Correlation for the Bir Berkine field.
x Standing [1] Correlation for HMD field.
x Al-Marhoun [6] Correlation for the field Ohanet.
These correlations were compared for the crude oil of Algeria but they can be employed to estimate the
same parameters of PVT for all the types of mixture of oil and gas with properties belonging to the
margin of the data used in this study, and they are employed to produce differential tables of release for
the simulation of tank, and they can be granted for other basins/sectors, or for certain oil classes.
All the results of this study prove that a higher exactitude was obtained when the correlations are granted
to the regional data and the forecasts of models of neurological network are better for all properties PVT
studied compared to the correlations. In addition, the neurological model of network exceeds these
empirical correlations most common.
References
[1] ADEL M. ELSHARKAWY and al. Assessment of the PVT Correlations for Predicting the Properties of Kuwaiti Crude
Oils. Petroleum Science and Technology-1995. p: 219-232.
[2] TAREK H AHMED. "Equations of state and PVT analysis: applications for improved reservoir modeling". 2007 by Gulf
Publishing Company, Houston, Texas.
[3] DANESH ALI. "PVT and phase behaviour of petroleum reservoir fluids". 1998 Elsevier Science .B.V.
[4] MUHAMMAD AAMIR MEHMOOD et MUHAMMAD ALI AL-MARHOUN, Evaluation of empirically derived PVT
properties for Pakistani crude oils, Journal of Petroleum Science and Engineering-1996. vol.16 p :275-290.
[5] MUHAMMAD ALI AL-MARHOUN. Evaluation of Empirically Derived PVT Properties for Middle East Crude Oils.
Journal of Petroleum Science and Engineering-2004. vol.42. p:209-221.
[6] MUHAMMAD ALI AL-MARHOUN. New Correlations for Formation Volume Factors of Oil and Gas Mixtures.Journal
of Canadian Petroleum Technology-1992. vol 31, n3. p:22-30.
[7] SOUAHI F. Developing Correlations for Prediction of Petroleum Fraction Properties Using Genetic Algorithms. Oil and
Gas Science and Technology-2008. vol.63.n2.p:229-237.

Hadj Mohammed Sidrouhou and Mourad Korichi / Energy Procedia 36 (2013) 738 745
[8] AVAULLEE. L.Thermodynamic Modeling for Petroleum Fluid. III. Reservoir Fluid Saturation Pressures. A Complete PVT
Property Estimation. Application to Swelling Test. Fluid Phase Equilibria-1997. vol.141.p:87-104.
[9] TAGHIZADEH MAZANDARANI M. et S. MOHAMMAD ASGHARI. Correlations for Predicting Solution Gas-Oil
Ratio, Bubble point Pressure and Oil Formation Volume Factor at Bubble point of Iran Crude Oils. European Congress of
Chemical Engineering (ECCE-6)-2007.
[10] OMOLE O, O.A. FALODE et DENG A. DENGET. Prediction of Nigerian Crude Oil Viscosity Using Artificial Neural
Network. Petroleum and Coal-2009. vol.51.n3. p:181-188.
[11] R .ABEDINI, A. ABEDINI et N. ESLAMI YAKHFROUZAN. A New Correlation for Prediction of Undersaturated
Crude Oil Viscosity. Petroleum and Coal-2010. p:50-55.
[12] EL-SEBAKHY EMAD A. Estimating PVT Properties of Crude Oil System Based on Functional Networks.Petroleum
science and engineering-2008.
[13] EL-SEBAKHY EMAD A. Forecasting PVT Properties of Crude Oil Systems Based on Support Vector Machines
Modeling Scheme. Petroleum Science and Engineering-2009. vol.64, p:2534.
[14] GHARBI RIDHA B. Universal Neural-Network-Based Model for Estimating the PVT Properties of Crude Oil Systems.
Energy & Fuels-1999. vol.13,p:454-458.
[15] HEMMATI M.N. et R.KHARRAT. Evaluation of Empirically Derived PVT Properties for Middle East Crude Oils.
Scientia Iranica-2007. vol. 14, n4, p:358-368.
[16] MALALLAH ADEL. Accurate Estimation of the World Crude Oil PVT Properties Using Graphical Alternating
Conditional Expectation. Energy & Fuels-2006, vol.20, n2, p:688-698.
[17] MUHAMMAD AAMIR MEHMOOD. "Evaluation of Empirically Derived PVT Properties for Oil Field Brines". These
Master of Science in petroleum engineering, JUNE 1996.
[18] EL-SEBAKHY EMAD A. Data Mining in Forecasting PVT Correlations of Crude Oil Systems Based on Type1 Fuzzy
Logic Inference Systems. Computers and Geosciences-2009.vol.35.p:1817-1826.
[19] DUTTA SARIT et J.P. GUPTA. PVT Correlations for Indian Crude Using Artificial Neural Networks. Petroleum
Science and Engineering-2010. vol.72, p:93109.
[20] MUHAMMAD ALI AL-MARHOUN. Reservoir Fluid Properties State of the Art and Outlook for Future Development.
Society of Petroleum engineers SPE-2002.
[21] MACCAIN WILLIAM D.. "The Properties of Petroleum Fluids. Second edition. 1990 by PennWell Company Publishing
Company, Tulsa, Oklahoma.

745