You are on page 1of 136

Hydrocarbon Fluid Properties

and Phase Behaviour

Prepared By

Garry A. Gregory, P.Eng.

© May, 2002

Notes for a Professional Development Course


presented in

Calgary, Alberta, Canada

December 4 - 6, 2002

by

NEOTECHNOLOGY CONSULTANTS LTD.


510, 1701 Centre Street N.W.
Calgary, Alberta, Canada T2E 7Y2
Tel: (403) 277-6688 Fax: (403) 277-6687
Internet: www.neotec.com
Disclaimer

Although all of the information contained in these Notes is believed to be accurate at the time it is prepared, it
is presented without representation or warranty of any kind. Neither Neotechnology Consultants Ltd. nor the
author shall assume any liability of any kind whatsoever, either collectively or individually, arising from the use
or application of any of the technology, descriptions, or expressed opinions contained herein.
Table of Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Definitions

2.1 Base Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2


2.2 Gas Gravity, Relative Density (G) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.3 Specific Gravity (SG) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.4 API Gravity (API) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.5 Stock Tank Barrel (STB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.6 Gas Formation Volume Factor (BG) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.7 Oil Formation Volume Factor (Bo) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.8 Watson’s K Factor (K) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

3. Phase Behaviour (General) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

4. P-V-T Behaviour Of Dry Gas

4.1 The Engineering Gas Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8


4.2 Standing and Katz (1942) Z-factor Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.3 Fitted Equations for the Z-Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.4 Correction for Acid Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.5 Three Parameter Correlations for Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.6 Analytical Equations of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.7 Volumetric Gas Flow Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.8 Enthalpy, Entropy, and Heat Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

5. P-V-T Behaviour Of Compositional Systems

5.1 Flash Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26


5.2 Pseudo-Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.3 Liquid Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

6. Phase Behaviour Of Black Oil Systems

6.1 Characteristics of Black Oil Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34


6.2 Standing (1947) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.3 Lasater (1958) Correlation for Rs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.4 Vazquez and Beggs (1977) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . 42
6.5 Glaso/ (1980) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
`

-i-
6. Phase Behaviour Of Black Oil Systems (cont)

6.6 Al-Marhoun (1985, 1988, 1992) Correlations for Rs and Bo . . . . . . . . . . . . . . . 45


6.7 Abdul-Majeed and Salman (1988) Correlations for Bo . . . . . . . . . . . . . . . . . . . 47
6.8 Asgarpour et al (1989) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . . . . 48
6.9 Dokla and Osman (1992) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . . 49
6.10 Petrosky and Farshad (1993) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . 51
6.11 Kartoatmodjo and Schmidt (1994) Correlations for Rs and Bo . . . . . . . . . . . . . 51
6.12 Velarde et al (1999) Correlations for Rs and Bo . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.13 Brief Comparison of Various Bubble Point Correlations . . . . . . . . . . . . . . . . . 54
6.14 Recommended Calibration Procedure for Rs and Bo Correlations . . . . . . . . . . . 57
6.15 Correction to Bo for Undersaturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.16 Density of the Gas and Oil Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.17 In Situ Flow Rates of Gas, Oil and Total Liquid . . . . . . . . . . . . . . . . . . . . . . . . 62

7. Gas Viscosity

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
7.2 Carr et al (1954) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.3 Dempsey (1965) / Standing (1977) Approximation for the
Carr et al (1954) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.4 Dranchuk et al (1986) / Standing (1977) Approximation for the
Carr et al (1954) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.5 Lee et al (1966) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.6 Dean and Stiel (1965) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.7 Pedersen and Fredenslund (1986) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.8 A Brief Comparison of Gas Viscosity Correlations . . . . . . . . . . . . . . . . . . . . . . 70

8. Dead Oil Viscosity (Black Oil Models)

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.2 General Relationships for Effect of Temperature on Dead Oil Viscosity . . . . . 72
8.3 Beal (1946) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
8.4 Abbot et al (1970) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
8.5 Twu (1986) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
8.6 Beggs and Robinson (1975) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
8.7 Glaso/ (1980) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8.8 Ng and Egbogah (1983) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8.9 Amin and Maddox (1980) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
8.10 Beg et al (1988) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
8.11 Puttagunta et al (1992) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
8.12 Kartoatmodjo and Schmidt (1994) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . 84
8.13 A Brief Comparison of Some Dead Oil Viscosity Correlations . . . . . . . . . . . . 84

-ii-
9. Saturated Oil Viscosity

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
9.2 Chew and Connally (1959) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
9.3 Beggs and Robinson (1975) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
9.4 Khan et al (1987) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
9-5 Kartoatmodjo and Schmidt (1994) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . 90
9.6 A Brief Comparison of Saturated Oil Viscosity Correlations . . . . . . . . . . . . . . 90

10. Undersaturated Oil Viscosity

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
10.2 Beal (1946) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
10.3 Vazquez and Beggs (1977) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
10.4 Khan et al (1987) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.5 Abdul-Majeed et al (1990) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.6 Kartoatmodjo and Schmidt (1994) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . 93
10.7 A Brief Comparison of Undersaturated Oil Viscosity Correlations . . . . . . . . . 94

11. Viscosity Of Blended Hydrocarbon Liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

12. Hydrocarbon Liquid Viscosity (Compositional Models)

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
12.3 Dean and Stiel (1965) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
12.4 Van Velzen et al (1962) / Letsou and Stiel (1973) Correlations . . . . . . . . . . . . 97
12.5 Pedersen and Fredenslund (1986) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . 97
12.6 A Brief Comparison of Compositional Oil Viscosity Correlations . . . . . . . . . . 99

13. Viscosity Of Oil-water Emulsions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

14. Gas Enthalpy

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102


14.2 Compositional Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
14.3 Undefined Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

15. Liquid Enthalpy

15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102


15.2 Compositional Liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
15.3 Crude Oil and Other Undefined Hydrocarbon Liquids . . . . . . . . . . . . . . . . . . 104

-iii-
16. Properties Of Produced Water

16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106


16.2 Density of Produced Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
16.3 Volumetric Flow Rate of Produced Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
16.4 Viscosity of Produced Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
16.5 Enthalpy of Produced Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
16.6 Equilibrium Water Content of Natural Gas . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

17. Surface Tension

17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114


17.2 Baker and Swerdloff (1956) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
17.3 Gould (1972) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
17.4 Gray (1978) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
17.5 Gomez (1987) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
17.6 Compositional Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

18. Hydrates

18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121


18.2 Katz et al (1959) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
18.3 Hammerschmidt (1969) Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
18.4 Ng and Robinson Method for Hydrate Prediction and Suppresion . . . . . . . . . 124

19. Some Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

Literature References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

-iv-
HYDROCARBON FLUID PROPERTIES AND
PHASE BEHAVIOUR

1. Introduction

A primary task for a pipeline or production engineer is to predict accurately the pressure and
temperature changes that will occur in a pipeline or a well for particular sets of values of design parameters.
Among other things, calculating these changes requires the knowledge of various transport and
thermodynamic properties of the fluid(s) being transported or produced (e.g. density, viscosity, enthalpy) over
ranges of temperature and pressure which are sufficient to cover all of the conditions that might occur in the
pipeline or well.

Ideally, these data will all have been carefully measured using samples of the fluid or fluids to be
transported.

Unfortunately, experimental data are seldom ever available to the extent required. Most often, one
only has data which have been obtained for a single condition, or at best, a limited range of conditions.
Frequently, one has almost no data at all other than a few characterizing parameters such as specific gravity
or molecular weight, which might even have been estimated solely on the basis of experience. Thus, to
perform the flow calculations, one must first be able to calculate the required fluid properties. Such
calculations should of course make maximum possible use of whatever information is actually available for
the system.

The selection of appropriate methods for calculating fluid properties depends on both the type of fluid
and the nature of available data. The fluid may be a gas (or vapour), dense phase fluid, or liquid. The
composition of the fluid is also important. It may consist of a single pure component, or a mixture of two or
more components. When dealing with a multicomponent mixture, it can also be important to know whether
it is a mixture of similar or dissimilar components.

Regardless of the composition of the fluid system, it may enter the pipeline as either a single phase
or as a two phase mixture. In fact, if water is one of the components, it could enter as a three phase system.
Even when the fluid enters as a single phase however, changing conditions in the pipeline may result in the
separation of the fluid into a gas phase and one or more liquid phases. Hence methods are also required for
the prediction of conditions under which

(i) gas will begin to evolve from a liquid (i.e. the bubble point)
(ii) liquid will begin to drop out from a gas stream (i.e. the dew point)

or in other words, the phase behaviour.

When a multicomponent mixture exists as two or more phases, one must be able to predict, for a
given temperature and pressure,

(i) the fraction of the total system which constitutes each phase
(ii) the composition (or some other suitable characterizing parameters) for each phase
(iii) the transport and thermodynamic properties (e.g. density, viscosity, and enthalpy) of
each phase

This is obviously a very broad subject, a thorough treatment of which is well beyond the scope of
these Notes and only a brief description of some methods that are particularly useful for design calculations
related to pipeline systems and producing wells is presented. These selected methods can be used to perform
all of the required fluid property evaluations for one, two, or three phase systems involving various mixtures
of any of the following:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 2

- pure hydrocarbons and related compounds


- natural gas
- crude oil
- condensate liquid
- formation water

For additional information on these systems, and all other fluids, reference should be made to more
detailed sources, such as the API 44 Tables (1975), Reid et al (1977), the GPSA Engineering Data Book
(1977), SPE Handbook of Petroleum Engineering (1987), and other recent technical literature.

2. Definitions

Before we can proceed, we need some basic definitions. The following terms will be used throughout
these Notes:

2.1 Base Conditions

The pressure, P o, and temperature, T o, are the base conditions (also referred to as stock tank
conditions) to which standard gas volumes and stock tank liquid volumes are referred. Throughout
these Notes, the base conditions given in Table 2.1 are assumed to apply.

In different geographical regions, gas flow rates may be expressed relative to base conditions
different from those in Table 2.1. These flow rates can be adjusted to correspond to the base
conditions in Table 2.1 using the appropriate multiplying factor from either Table 2.2 or Table 2.3.

Table 2.1

Base Pressure ( P o ) and Base Temperature ( T o ) Used in these Notes

System of Units Po To

Field units 14.65 psia 60 °F


SI units 101.325 kPa 15 °C

2.2. Gas Gravity /Relative Density (G)

(2-1)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 3

Table 2.2

Multiplying Factors to Convert ft3 of Gas at Specified Reference Conditions


to ft3 of Gas at 14.65 psia and 60 °F

Pressure (psia) Temperature (°F) Factor

14.400 60 0.982935
14.650 60 1.000000
14.696 60 1.003140
14.700 60 1.003413
14.730 60 1.005461
14.900 60 1.017065
15.025 60 1.035597

Table 2.3

Multiplying Factors to convert ft3 of Gas at Specified Reference Conditions


to m3 of Gas at 101.325 kPa and 15 °C

Pressure (psia) Temperature (°F ) Factor

14.400 60 0.028693
14.650 60 0.028174
14.696 60 0.028262
14.700 60 0.028270
14.730 60 0.028328
14.900 60 0.028655
15.025 60 0.028895

Note: To convert gas flow rates from the base conditions used in these Notes to
any of the various reference conditions in Tables 2.2 and 2.3, one must
divide the flow rate by the corresponding factor.

2.3 Specific Gravity (SG)

(2-2)

Note: At 60 oF and 14.65 psia, density of water = 62.37 lb/ft3

At 15 oC and 101.325 kPa, density of water = 999.1 kg/m3

In SI units, the parameter analogous to specific gravity is referred to as Relative Density, and is the
density of the substance relative to the density of water at 15 °C and 101.325 kPa.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 4

2.4 API Gravity (API)

The API gravity, expressed in oAPI, is related to the specific gravity (or relative density) of a liquid
as follows:

(2-3)

2.5 Stock Tank Barrel (STB)

A stock tank barrel is a reference quantity, and is equal to a volume of 1.0 barrel (42 US gal),
measured at base conditions (stock tank conditions). In SI units the corresponding reference is 1.0
m3 at standard conditions, which can also be written as 1.0 sm3 .

2.6 Gas Formation Volume Factor (BG )

The gas formation volume factor, BG , is the ratio of the volume occupied at a specified pressure and
temperature by a given mass of gas to the volume of the same mass of the gas at base conditions.
Common units for BG are ft3/scf or m3/m3 at standard conditions (i.e. m3/sm3 ).

2.7 Oil Formation Volume Factor (BO )

The oil formation volume factor, Bo , is the ratio of the volume of one stock tank volume of oil plus
any dissolved (i.e. solution) gas at a specified pressure and temperature to that of one stock tank
volume of oil (i.e dead oil at base conditions).

2.8 Watson K Factor (K)

The Watson K factor, also known as the UOP characterization factor, or UOPK, is defined as
(Watson et al, 1935),

(2-4)

where Tb = the mean boiling temperature for the particular oil or fraction, (°R)
SG = specific gravity of the oil or fraction

This oil characterization parameter is often used in correlations for predicting thermodynamic
properties.

3. Phase Behaviour (General)

For a single pure component, a phase diagram on pressure-temperature co-ordinates appears as shown
in Figure 3.1. At any temperatures to the left of the vaporization curve, only single phase liquid exists until
the fusion curve is reached. At any temperature to the right of the vaporization curve only single phase gas
or vapour exists. At any point on the curve above the triple point, a two phase mixture of gas and liquid
exists in equilibrium. The relative amounts of each phase must be determined from an enthalpy balance. The
two phase region ends at the critical point. At pressures higher than the critical pressure, only a single phase
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 5

exists; this is known as the dense phase region and it has some unique characteristics. When the temperature
is below the critical temperature, the properties of the dense phase are similar to those of a liquid. However,
when the temperature is above the critical temperature, the properties of the dense phase resemble those of
a gas.

CRITICAL
POINT

Pc

E
Liquid Region

FUSION CURV
PRESSURE

E
RV
U
C
N
IO
Solid Region

AT
IZ
R
PO
VA
Gas Region
C URVE
ATION
SUBLIM
TRIPLE
POINT

Tc
TEMPERATURE

Figure 3. 1

Typical Phase Diagram for a Single Component

When two pure components are mixed together, the phase behaviour changes quite markedly from
that of a single pure component. Instead of a single line locus of equilibrium two phase mixtures, there is
now a two phase region or envelope on a P-T diagram. Within the two phase envelope, gas and liquid
phases of varying composition and varying amount are in equilibrium with each other. The location of the
critical point for the binary mixture is a complex function of the composition of the mixture, as shown in
Figure 3.2. Typically, the critical pressure of the mixture is always higher than a simple weighted mean based
on the mole fractions of the components.

With multicomponent mixtures, the phase behaviour becomes progressively more complex.
However, for a given composition, a P-T diagram will typically exhibit a two phase envelope similar to that
shown in Figure 3.3. Within the two phase locus, there are loci along which the mole percent of the system
appearing as a liquid phase remains constant. The boundary between single phase liquid and the two phase
envelope is termed the bubble point locus and corresponds to 100% liquid (mole basis). The boundary
between single phase gas and the two phase envelope is called the dew point locus and this corresponds to
100% gas. The dew point locus and the bubble point locus are joined together at the critical point for the
particular system. Note that all of the constant liquid fraction loci also converge at the critical point. The
shaded area in Figure 3.3 is called the retrograde region, and it is characterized by showing an increase in
the amount of liquid with a decrease in pressure

The maximum pressure at which two phases exist is called the cricondenbar. The cricondenbar is
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 6

also the lower pressure bound for the dense phase region. For single component fluids, the cricondenbar is
thus equal to the critical pressure. It is apparent from Figure 3.3 that this is not necessarily the case for
multicomponent mixtures (in fact, it is seldom the case).

While all multicomponent mixtures of hydrocarbons will exhibit a two phase envelope over some
range of pressure and temperature, it is possible to characterize such systems, at least approximately,
according to the relative amount of liquid that would be expected at typical production and/or transportation
conditions. Thus, a system for which the dew point locus occurs only at temperatures that are less than those
expected in normal production and transport operations is termed a dry gas. Systems in which there is some
liquid, but which are still predominately in the gas phase are known as gas-condensate systems. Systems
for which the mole fractions of gas and liquid are both large are often referred to as volatile oil. Two phase
systems in which the liquid phase dominates under most conditions are referred to as black oil systems.
Figure 3.4 illustrates typical slopes of the two phase envelopes for these various cases. As noted above, there
is no rigorous way to distinguish between them; however, they can be approximately differentiated according
to composition as shown in Table 3.1.

2000

1800

1600

1400
PRESSURE (psia)

1200

1000

800

600

400

200

0
-300 -200 -100 0 100 200 300 400
TEMPERATURE (°F)

Figure 3.2

True Critical Pressure Loci for Some Binary Mixtures


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 7

Retrograde Region
Critical Point

Constant Liquid
Volume % Lines
PRESSURE

s
cu
Lo
int
Po
e
l
bb
Bu

us
oc
L
int
Po
w
De
TEMPERATURE

Figure 3.3

Typical P - T Diagram for a Retrograde Gas-Condensate System

Table 3.1

Approximate Criteria for Various Types of Produced Hydrocarbon Systems *

Mole Percent Typical Gas/Oil Ratio


Type of System
C1 C2 - C7 C7+ (scf/stb) sm3/sm3

Dry Gas 9e+07 9e+06 < 1 - -


Gas-Condensate < 12.5 > 3,200 > 570
Volatile Oil 12.5 - 20 1,750 - 3,200 310 - 570
Black Oil > 20 < 1,750 < 310

* Reference: McCain (1994a)

McCain (1993) notes that there is an important difference between the characteristics of black oil and
volatile oil systems. In the case of the former, the gas that is produced with the oil tends to behave as a dry
gas. That is, it remains as a gas as the pressure and temperature are reduced throughout the production system
to some ultimate separator conditions. In the latter case, however, the produced gas tends to be a retrograde
gas, and may result in a large volume of condensate liquid being released in the surface facilities. McCain
(1994b) further notes that, over the life of a volatile oil reservoir, often more than one-half of the total stock
tank liquid produced actually left the reservoir as free gas. Moses (1986) notes that volatile oils should more
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 8

properly be referred to as near-critical oils, since the reservoir fluid typically exhibits the behaviour of a fluid
near its critical point. In extreme cases, a shrinkage of up to 45% can occur just below the bubble point for
a pressure change as small as 10 psi (69 kPa)!

Typical operating range

C Critical point

Dry Gas
Gas-Condensate

C C
Volatile Oil
C
PRESSURE

Black Oil

TEMPERATURE

Figure 3.4

Relative Phase Envelopes for Various Types of Reservoir Fluids

4. P-V-T Behaviour of Dry Gas

4.1 The Engineering Gas Law

The most commonly used P-V-T relation for a dry gas is the familiar engineering gas law,

(4-1)

where P = absolute pressure


V = volume
n = number of moles
Z = compressibility factor
R = gas law constant
T = absolute temperature
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 9

For an ideal gas, Z = 1.0 . At 492 °R (32 °F ) and 14.696 psia (1.0 atm), the volume of 1.0 lb mole
of a gas is 359 ft3. In SI units, the corresponding temperature and pressure of 273 K (0 oC) and 101.325 kPa
result in a volume of 22.414 m3 for 1.0 kmol of gas. These values, and others derived from them, can be used
to calculate the value of the gas law constant in any particular set of units that might be required. Some of
the more common values are shown in Table 4.1 .

Table 4.1

Selected Values of the Gas Law Constant, R

10.732 psia-ft3/lbmole-°R
1545.4 lb-ft/lbmole-°R
8.314 kPa-m3/kmol-K
1.986 cal/mol-K
0.082 l-atm/mol-K

At the base (i.e. reference) conditions defined in Section 2, most gases can be assumed to behave
ideally, and the molar volumes can be adjusted accordingly; thus

at 14.65 psia, 60 °F Vo = 380.6 ft3/lb-mole


at 101.325 kPa, 15 °C Vo = 23.64 m3/kmol

4.2 Standing and Katz (1942) Z-factor Correlation

The number of moles of a gas is simply,

(4-2)

where m = mass of the gas


M = molecular weight of the gas

Since the density is the mass per unit volume, Equation (4-1) can be written as,

(4-3)

or, in terms of the gas gravity, G,

(4-4)

where ρG = gas density at pressure P and temperature T


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 10

For a real gas, the compressibility factor Z must thus be known if the gas density is to be calculated.
Z is generally determined using the familiar Standing and Katz (1942) chart which is shown in Figure 4.1.
This chart is based on a two parameter corresponding states model,

(4-5)

(4-5a)

(4-5b)

where Pr , Tr = reduced pressure and temperature respectively


P c , Tc = critical pressure and temperature respectively

For most pure fluids of interest, the critical properties are readily available (e.g. GPSA Data Book).
Pr , Tr and consequently, Z , can therefore be easily determined.

When the gas is a mixture of components, a mixing rule is required to take into account the effect of
all of those components on the critical properties of the mixture. For the Standing and Katz correlation, Kay's
(1936) rule is used, which is expressed by,
(4-6a)

(4-6b)

where yi = mole fraction of component i in the mixture


Pci = critical pressure of component i
Tci = critical temperature of component i
Ppc = pseudo-critical pressure of the mixture
Tpc = pseudo-critical temperature of the mixture

As previously noted in Section 3, Ppc and Tpc are not the true critical properties for the mixture,
which is why they are designated as pseudo-parameters. They are however the appropriate critical properties
to use in Equations (4-5a) and (4-5b). Throughout the rest of these Notes, for sake of simplicity, we will drop
the distinction between the two types of critical properties, at least with respect to nomenclature, and will
simply refer to Pc and Tc .

When the actual composition of the gas is unknown but the gas gravity is known, one can use the
simple correlation presented by Standing (1977), and shown in Figure 4.2, to estimate the pseudo-critical
properties.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 11

REDUCED PRESSURE (PR )

COMPRESSIBILITY FACTOR (Z)

COMPRESSIBILITY FACTOR (Z)

REDUCED PRESSURE (PR )

Figure 4.1

Standing and Katz (1942) Z-Factor Correlation

Gregory (2000) has examined a number of analytical expressions that have also been proposed for
estimating Pc and Tc, based on the gas gravity, including equations proposed by Standing (1977), Sutton
(1985), Thomas et al (1970), and Elsharkawy et al (2000). He demonstrated that the simple, linear
relationships of Thomas et al (1970), shown below, are comparable to the other more complex expressions
that all involve a G2 term,

(4-7)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 12

(4-8)

In Equations (4-7) and (4-8), Pc and Tc are expressed in psia and °R respectively.

700
PSEUDO-CRITICAL PRESSURE (psia)

Miscellaneo
us gases
Conde
650 nsate w
ell fluids

600

550

500
PSEUDO-CRITICAL TEMPERATURE ( °R)

s
se
s ga
ou
450 ne
ell a
sc
Mi
s
fluid
well
ate
400 ens
C ond

350

300
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

GAS GRAVITY

Figure 4.2

Generalized Correlation of Standing (1977) for Estimating Pseudo-Critical


Properties of Hydrocarbon Gases
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 13

4.3 Fitted Equations for the Z-Factor

The graphical Z-factor correlation has been fitted to equation form suitable for computers by a
number of authors, including Hall and Yarborough (1973), Dranchuk et al (1974), and Dranchuk and Abou-
Kassem (1975). These equations are complex and require an iterative solution, but are generally accurate,
with typical errors being in the order of 0.3 to 0.5% or less. For example, the correlation of Dranchuk and
Abou-Kassem is given by Equations (4-9) through (4-13), with the constants having the values shown in
Table 4.2,

(4-9)

(4-10)

(4-11)

(4-12)

(4-13)

For most cases of interest (e.g. naturally produced hydrocarbon fluids), Zc = 0.27, and this is the value
generally used in Equation (4-13). This is an implicit correlation in Z , which is also typical of the other
analytical methods, and an iterative solution procedure is required to determine Z. The solution is relatively
straightforward however, and convergence is seldom, if ever, a problem.

Table 4.2

Constants for Use in Equations (4-9) to (4-12)

A1 = 0.3265 A7 = -0.7361
A2 = -1.0700 A8 = 0.1844
A3 = 0.5339 A9 = 0.1056
A4 = 0.01569 A10 = 0.6134
A5 = -0.05165 A11 = 0.7210
A6 = 0.5475

An A.G.A. publication (Transmission Measurement Committee Report No. 8) describes a very


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 14

complex procedure, based on the Virial equation of state, for evaluating Z factors for gas mixtures. The
motivation for this work was to improve the accuracy of gas density calculations for use in custody transfer
applications. Most such situations involve sales quality gas, or similar, and this method is applicable only
to relatively light, sweet gases (e.g. total mole fraction of H2O, H2S, CO2, O2, He and Ar less than 0.01). In
any case, since the stated error expectations for the A.G.A. procedure are only marginally lower than those
observed for the other procedures described above, the additional complexity does not seem to be warranted,
and it is not recommended for general use. The procedures described above are sufficiently accurate for most
pipeline design purposes.

4.4 Correction for Acid Gases

The use of Figure 4.1, or Equations (4-9) through (4-13) can result in significant errors if the gas con-
tains more than trace amounts of the acid gas components H2S and CO2, or other non-hydrocarbon com-
ponents. A simple procedure that can be used to account for the presence of acid gas components has been
presented by Wichert and Aziz (1971, 1972).

The first step is to determine the value of a correction factor, ε , from Figure 4.3. This factor is then
used to adjust the pseudo-critical properties that were either calculated using Equation (4-5a) and (4-5b), or
estimated using either Figure 4.2 or Equations (4-7) and (4-8), as follows:

(4-14)

(4-15)

where * = denotes adjusted value


Xs = mole fraction of H2S in the gas mixture

4.5 Three Parameter Correlations for Z

Three parameter equations of state, which can lead to a significant improvement in accuracy over
that of the simpler two parameter models, have been proposed in various forms.

For example, Equation (4-5) can be extended to,

(4-16)

where

(4-17)

and Vc = critical molar volume

This correlation is typically presented in the form of several graphical relations, each similar to Figure 4.1,
with each separate graph corresponding to a particular value of Zc .
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 15

For a wide range of chemical compounds, Zc ranges from 0.20 to 0.30. However, for most of the
organic compounds produced as natural hydrocarbons, Zc ranges only from about 0.27 to 0.29. Also, since
Figure 4.1 corresponds to Zc = 0.27 , the use of this figure for oil and gas systems is already optimal, in any
case. It should be noted, however, that the procedure of Dranchuk and Abou-Kassem (1975) can be used with
any appropriate value of Zc .

Figure 4.3

Wichert and Aziz (1971) Critical Temperature Correction Factor ε for Sour Gases

Another three parameter model is given by the relations,

(4-18)

(4-19)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 16

where yi = mole fraction of component i in the gas mixture


ωi = Pitzer acentric factor for component i

Values of Z (o) and Z (1) , as functions of Pr and Tr , are available in both tabular and graphical form
in a number of sources (e.g. API Data Book, GPSA Data Book, Reid et al (1977)). The Pitzer acentric factor
is also tabulated for a wide range of components in these sources. However, it can also be computed using
the relation,

(4-20)

where (Pvpr)i* = the reduced vapour pressure of component i at Tr = 0.7

4.6 Analytical Equations of State

Many analytical expressions have been proposed for predicting P-V-T behaviour. One of the more
widely used of these, at least historically, is the Benedict-Webb-Rubin (1940) equation,

(4-21)

where Ao, Bo, Co, a, b, c, α, and γ are constants which are different for each gas and which must be
determined experimentally. The BWR equation is thus an 8 parameter equation of state. Various authors
have reported sets of values for many pure components, based on their analyses of extensive P-V-T data.
Because of the complexity of the equation, however, the set of values of the constants determined for a given
pure component is not necessarily unique, and significantly different sets of values have been reported by the
various authors, depending on the data base that they used.

For mixtures of gases special mixing rules must be used to combine the values of the constants for
pure components. Variations in these mixing rules have given rise to a number of modifications of the BWR
equation over the years.

In more recent times, the tendency has been to use the simpler two-parameter equations of state. The
most common of these are the Soave (1972) modification of the Redlich-Kwong (1949) equation, and the
Peng-Robinson (1976a, 1976b) equation. The Soave-Redlich-Kwong, or SRK equation is given by,

(4-22)

and the Peng-Robinson, or PR equation is,

(4-23)

In this case, a and b are the two parameters that are characteristic of the particular gas. For multicomponent
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 17

mixtures, the appropriate values of a and b are determined using the mixing rules,

(4-24)

and

(4-25)

where the subscripts i and j refer to the individual pure components in the mixture. For pure components,
the parameters ai and bi are defined by,

(4-26)

(4-27)

(4-28)

The numerical values of the constants α, β, γ, δ, and ε for the PR equation are different from those for the
SRK equation.

The kij parameters in Equation (4-25) are called binary interaction parameters, and account for the
effect of each particular component on the P-V-T behaviour when it is in the presence of another particular
component. For example, the effect of pentane on the equilibrium behaviour of the system depends on
whether or not methane is also present, and the value of the methane-pentane interaction parameter is greater
than zero. The kij are tuning parameters, and they must be evaluated from P-V-T data for all possible binary
mixtures of the components. The numerical values of the kij used in any particular computer version of the
PR equation may thus differ from those used in another because of the data base used to determine them. Dif-
ferences in predicted P-V-T behaviour between two computer versions of the same equation of state are not
uncommon, but such differences should be relatively small in most cases. Where the effect of one component
is independent of whether or not another component is present, the corresponding kij = 0. In general, the
interaction parameters become less significant as the component molecules become more similar. Thus, the
parameter for C1 - C10 interaction is important ( e.g. k 1,10 = k 10,1 = 0.035), whereas that for C6 - C10
interaction is effectively 0.

If Equation (4-1) is used to replace V in the PR equation, the result can be re-arranged to give a
cubic relationship for the Z-factor, i.e.

(4-29)

where
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 18

(4-30)

(4-31)

Note that the acid gas correction described in Section 4.4 is not required for an analytical equation
of state, since the appropriate compensation for acid components is implicit in the corresponding binary
interaction parameters.

A somewhat more complex (but generally accurate) procedure for dry gas has been proposed by Lee
and Kesler (1975). It has been found to be particularly applicable in the dense phase region. The method is
based on a generalized version of the BWR equation and the principle of corresponding states. To use the
procedure, the following steps must be carried out:

Step 1. Calculate:

(4-32)

(4-33)

(4-19)

where:

(4-34)

(4-35)

Step 2. Calculate:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 19

Step 3. Using the constants for the Simple Fluid in Table 4.3, calculate Vr from:

(4-36)

where

Denote this calculated value of Vr as Vr (0) .

Step 4 Calculate the compressibility factor of the Simple Fluid, Z (0) , from the relation,

(4-37)

Step 5. Using the constants for the Reference Fluid in Table 4.3, calculate a new value of Vr from
Equation (4-36). Denote this new value of Vr as Vr (R) .

Step 6. Calculate the compressibility factor for the Reference Fluid, Z (R), from the relation,

(4-38)

Step 7. Calculate the compressibility factor, Z , for the fluid of interest from the equation,

(4-39)

where ω(R) = 0.3978


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 20

Figure 4.4 shows the calculated density of a CO2-rich fluid over a range of pressures and tempera-
tures in the dense phase region. These conditions are in fact typical of those that may be encountered in
pipelines associated with a CO2 based enhanced oil recovery project. What is particularly significant about
the dense phase region is the large effect that temperature has on density, compared to the pressure. This is
quite different than for most natural gas pipelines at typical operating conditions. Because of cooling in the
pipeline, and heating during compression(?)/pumping(?), the fluid is typically most dense at the suction side
of the boosting unit and least dense at the discharge side.

The calculated density of this same fluid at 25 °C and 100 °C is shown in Figure 4.5 for the Peng-
Robinson, SRK, and Lee-Kesler equations of state. The behaviour illustrated in this graph (Peng-Robinson
and Lee-Kesler about the same, SRK lower) is typical of that observed in most cases. In general, the SRK
equation under-predicts the density in the liquid-like dense phase region.

Table 4.3

Fluid constants for the Method of Lee and Kesler

Constant Simple Fluid Reference Fluid

b1 0.118193 0.2026579
b2 0.265728 0.331511
b3 0.154790 0.027655
b4 0.030323 0.203488
c1 0.0236744 0.0313385
c2 0.0186984 0.0503618
c3 0.0 0.016901
c4 0.042724 0.041577
d1 x 104 0.155488 0.48736
d2 x 104 0.623689 0.0740336
β 0.65392 1.226
γ 0.060167 0.03754

4.7 Volumetric Gas Flow Rate

It is customary to express gas flow rates in terms of the gas volume at standard conditions (see
Section 2.1). One might thus refer to a gas throughput of 10 MMscf/day, meaning 107 ft3/day when measured
at base (i.e. stock tank) conditions. In SI units, the corresponding rate would be about 2.83 x 105 Sm3/day.

For purposes of the flow calculations however, one must know the actual flowing volume at the
pressure and temperature in the system. The simplest way to compute this is by using the gas formation
volume factor, BG (defined earlier in Section 2.6).

In terms of the density, BG is defined as,


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 21

(4-40)

where ρGo = density of the gas at base conditions P o and T o


ρG = density of the gas at the pressure and temperature of interest.

1000

900
25 °C
800

700
DENSITY (kg/m³)

50 °C
600

500
75 °C
400
100 °C
300 Fluid Composition
CO 2 96%
200 N2 2%
CH 4 2%

100
10 12 14 16 18 20

PRESSURE (MPa)

Figure 4.4

Calculated Densities in the Dense Phase Region for a CO2-Rich Fluid


Using the Peng-Robinson Equation of State

If we substitute for ρGo and ρG using Equation (4-3), we obtain the alternate expression,

(4-41)

In Equation (4-41), it is assumed that Z o = 1.0 , and one must remember that P and T are the
absolute pressure and temperature respectively.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 22

The actual volumetric flow rate of gas at a given pressure and temperature QG can then be obtained
from the simple relation,

(4-42)

where QGo is the flow rate at base conditions.

1000

900

25 °C
800
CALCULATED DENSITY (kg/m³)

700

600 Peng-Robinson
Soave-Redlich-Kwong
Lee-Kesler
500

400

300
100 °C
200

100
10 12 14 16 18 20

PRESSURE (MPa)

Figure 4.5

Comparison of Densities for a CO2-Rich Fluid Calculated by


Various Equations of State

4.8 Enthalpy, Entropy, and Heat Capacity

Depending on the type of flow calculations to be performed, one may require various thermodynamic
properties as functions of pressure and temperature.

When only the gravity or molecular weight of the gas is known, it is necessary to use some very
generalized methods, such as those contained in the GPSA Engineering Data Book.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 23

When a detailed composition is available, it is possible to compute these quantities using one of the
analytical equations of state. It is particularly convenient to actually compute a corresponding quantity called
the Departure Function. In simple terms, this is a measure of the difference between the value of the ther-
modynamic property for the real gas and the value that would be obtained assuming that the gas behaved
ideally. The thermodynamic properties for the ideal gas state can be estimated easily using methods
contained in the API Data Book, GPSA Engineering Data Book, or the method presented by Passut and
Danner (1972).

A detailed derivation and discussion of the various departure functions is beyond the scope of these
Notes and the definitions will simply be presented. The interested reader can pursue this material more
thoroughly in most thermodynamics texts; a particularly good treatment is found in the classic textbook by
Hougen et al (1959).

(i) Enthalpy Departure Function

(4-43)

(ii) Entropy Departure Function

(4-44)

where f is the fugacity, which is related to pressure through the relation,

(4-45)

(iii) Heat Capacity

(4-46)

In the above expressions, the parameters H o, S o , and Cpo refer to values for the ideal gas state.
The integrals and partial derivatives are assumed to be evaluated using the equation of state of interest. The
point that we want to make here is that the equation of state can yield much more than just the P-V-T
behaviour; detailed thermodynamic properties can also be estimated for the fluid(s).

As an example, we will outline the steps involved in estimating the enthalpy of a gas using the Lee
and Kesler (1975) equation of state.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 24

Step 1.

Calculate Vr (0) , Vr (R) , Z (0) , and Z (R) using Steps 1 through 6 , as described in Section 4.6 of these notes.

Step 2.

With Vr = Vr (0) , and Z = Z (0) , and using the constants in Table 4.3 for the Simple Fluid, calculate

from the relation,

(4-47)

where,

Denote this quantity as,

i.e. the enthalpy departure function for the Simple Fluid. In this expression, H o is the enthalpy of the
Simple Fluid at temperature T , but in the ideal gas state.

Step 3.

With Vr = Vr (R) and Z = Z (R) , and using the constants in Table 4.3 for the Reference Fluid, calculate
from Equation (4-47),
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 25

and denote this quantity as

i.e. the enthalpy departure function for the Reference Fluid.

Step 4.

By analogy with Equation (4-39), the enthalpy departure function for the fluid of interest is given by,

(4-48)

where, as before, ω(R) = 0.3978.

Step 5.

Calculate the ideal gas enthalpy, H o , for the fluid of interest. For example, the API correlation has the form,

(4-49)

and where Hio are the ideal gas state enthalpies of the individual pure components and are expressed in the
general form,

(4-50)

Values for ai , bi , ci , di , and ei are tabulated for many pure components in the API Data Book.

Step 6.

Finally, calculate the enthalpy of the fluid of interest, H , from,

(4-51)

A similar procedure is used to evaluate the entropy of the gas. The heat capacity can be determined
simply by evaluating the enthalpy departure function at the pressure of interest, and at two temperatures that
are incrementally above and below the temperature of interest. The numerical derivative with respect to tem-
perature is the heat capacity departure function, as can be seen from Equation (4-46).
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 26

5. P-V-T Behaviour Of Compositional Systems

5.1 Flash Calculations

When the composition of a hydrocarbon mixture is known, it is possible to use a suitable equation
of state (e.g. Peng-Robinson, SRK) to predict,

(i) the fraction of the fluid in each of the gas and liquid phases.
(ii) the composition of each phase
(iii) the density of each phase
(iv) thermodynamic properties of each phase

All of these quantities are functions of pressure, temperature, and the composition of the fluid. A
primary assumption that must be made to be able to perform such calculations is that the mixture exists under
conditions of thermodynamic equilibrium. In most flow cases of interest, this is at least approximately true.
Notable exceptions however include critical flow through chokes and nozzles, and transient flow conditions
resulting, say, from rapid closing of a valve.

When thermodynamic equilibrium exists, then for every component in the mixture,

(5-1)

where ( f i ) V = fugacity of component i in the vapour phase


( fi )L = fugacity of component i in the liquid phase

This corresponds to the minimum Gibbs free energy for the entire system. Since the Equation (5-1)
must apply for each component, flash calculations for multicomponent systems require the simultaneous
solution of numerous equations and can involve some very large matrices (typically, square matrices with
dimension equal to the number of components). A detailed discussion of the complex iterative mathematical
manipulations required for the derivation of the appropriate relationships is beyond the scope of these Notes.
It suffices to say that they are governed by rigorous thermodynamic requirements which must be satisfied by
any solution. Convergence can be slow and/or difficult to achieve, particularly close to the critical point.

It is often observed that different answers are obtained for the same flash calculations using, say, two
different computerized versions of the same equation of state. One of the problems with these calculations
is that the answers can be affected significantly by a number of factors that could be different between the
two programs. These include the particular sets of interaction parameters, the numerical techniques used in
the solution, the magnitude of various convergence criteria, numerical precision of the calculations, etc.

Flash calculations can be performed in several ways, depending on the particular type of information
required. These are summarized briefly below. Note that the term Flash as applied here does not necessarily
imply that a phase change takes place, or that two phases must exist. In many cases, the purpose of the flash
calculation is to actually determine whether or not a two phase mixture should be expected.

(i) Normal Pressure-Temperature Flash

Both the pressure and the temperature of the mixture can be specified. The flash calculation then
determines the composition, densities, and thermodynamic properties at those conditions. This is the
type of calculation that is normally performed to track phase volume and property changes along a
pipeline or through a well tubing.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 27

(ii) Isenthalpic Flash (Constant Enthalpy)

In this case, it is assumed that the enthalpy, and either the pressure or the temperature of the mixture
is known. The flash calculation then determines the unknown temperature or pressure. This is the
type of calculation that would be performed, for example, to determine the temperature change across
a pressure regulator (sudden expansion is assumed to be an isenthalpic process). A normal P-T flash
upstream of the regulator would be performed to determine the enthalpy.

(iii) Isentropic Flash (Constant Entropy)

In this case, it is assumed that the entropy of the fluid is known, and either the pressure or the
temperature is unknown. The flash calculation then determines the unknown temperature or pressure.
This type of calculation would be performed, for example, to determine the temperature rise during
an adiabatic compression process. Again, a normal P-T flash at the initial conditions would be per-
formed to determine entropy.

(iv) Dew Point

In this case, it is assumed that either the pressure or the temperature is known. The purpose of the
flash calculation is to determine the unknown temperature or pressure, above or below which liquid
just starts to appear.

(v) Bubble Point

This is similar to the Dew Point flash calculation, except that the unknown pressure or temperature
corresponds to the first appearance of a vapour phase.

5.2 Pseudo-Components

With the exception of some very dry gases or refined products, a compositional analysis always
contains a residual fraction of some kind. Common examples of this are C 7+ , C 8+ , C 10+ , C 15+ , etc.
fractions that are frequently reported in gas or recombined fluid analysis. One is generally also provided with
at least the average molecular weight, and sometimes, the specific gravity of the residual, but seldom with
much more information. The mole fraction represented by the residual might be very small (e.g. less than
0.5%), as in the case of a relatively dry gas, or very large (e.g. greater than 50%) as is often the case for crude
oil systems.

Before one can perform any flash calculations, the residual component must be characterized as one
or more pseudo-components. Each pseudo-component must then be assigned a molecular weight and various
characterizing properties so that it can be included in the calculation according to the appropriate mixing rules
for the given equation of state. Residual fractions generally represent a much larger proportion of the liquid
phase than of the gas phase. Consequently, the way in which they are dealt with has a significant effect on
the accuracy of the amount of liquid that will be calculated to be present.

At the other extreme, compositional analyses are frequently provided up to C30+ components or
higher, plus assorted aromatic and naphthenic compounds. With N2, CO 2 , and H 2 S, one can thus have
compositional information for 40 or more components. Performing flash calculations (each of which can
involve tens or hundreds of matrix manipulations) for this many components may lead to serious instabilities,
and can take an excessive amount of computer time to perform. The problems in this case are how to lump
the mole fractions of some of the compounds together so as to reduce the total number, and then, how to
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 28

assign characterizing properties to the resulting lumped fractions.

A third type of situation is presented when the analytical data for the fluid (typically a crude oil) are
presented as a true boiling point (TBP) curve, a typical example of which is shown in Figure 5.1. In this case,
the entire composition of the oil will consist of pseudo-components. The actual data from which Figure 5.1
was constructed are shown in Table 5.1. Note that with TBP data, the volume percent reported as the yield
of a given fraction is a relative number. It refers to the volume of the fraction collected as a percentage of
the initial volume of the fluid. Since the volumes of these fractions are not necessarily additive, the relative
molar amounts computed for each fraction must be normalized to complete the composition.

A detailed discussion of how to define pseudo-components in general is beyond the scope of these
Notes. Techniques for splitting up residual fractions and/or recombining the resulting components have been
proposed by numerous authors, including Lohrenz et al (1964), Whitson (1983), Petersen (1982), and
Ahmed (1989). The procedures used by Neotec are discussed in some detail by Gregory (1993a).

In any case, when the mole fractions of the pseudo-components have been determined, it is necessary
to characterize each fraction with an appropriate set of fluid properties. In general, each pseudo-component
will already have a defined value for one or more of the average normal boiling point, the average molecular
weight, and the average API gravity. These data can be used, along with procedures proposed by Cavett
(1962), Hopke and Lin (1974), Bergman (1976), Kesler and Lee (1976), Riazi and Daubert (1980), Brule
et al (1985), and others, to calculate values of Pc , Tc , Vc , and ω for the pseudo-components. Procedures
for characterizing pseudo-components are also discussed in some detail by Gregory (1993a).

In any case, when the mole fractions and characterizing properties of the pseudo-components have
been defined, they are treated exactly the same way as any other component in the composition in terms of
the mixing rules, etc. for any particular procedure that is being used.

The method of Riazi and Daubert is particularly simple, and gives surprisingly good results in many
cases. Their general equation takes the form,

(5-2)

where θ = property to be estimated


Tb = average normal boiling point of residue or fraction (°R)
SG = specific gravity of residue or fraction

Values of the constants α , β , and γ to be used for estimating various properties of interest are given
in Table 5.2.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 29

100

90

80
CUMULATIVE VOLUME PERCENT
70

60

50

40

30

20

10 Ref: Oil & Gas Journal


p. 46, July 8 (1991)
0
0 100 200 300 400 500 600 700 800

AVERAGE BOILING POINT (°C)

Figure 5.1

True Boiling Point Curve for Brent Blend Crude Oil

Table 5.1

True Boiling Point Analysis for Brent Blend Crude Oil *

Boiling Range Yield Gravity


(°C) (Vol.%) (oAPI)
C5 - 149 23.45 64.2
149 - 232 15.0 46.3
232 - 342 20.3 36.2
342 - 369 4.05 30.9
369 -509 18.8 25.8
509 - 550 3.85 21.9
550 + 12.0 12.2

* Source: Oil and Gas J., p. 46, July 8 (1991)

The data in Table 5.1 suggest that each of the boiling range fractions could be treated as a separate
pseudo-component. The molecular weight, critical properties, and acentric factor of each pseudo-component
can be estimated using Equation (5-2) with the average boiling point and the specific gravity (obtained from
the API gravity using Equation (2-3)). The mole fraction of each component is computed as follows:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 30

(i) Compute the relative numbers of moles as the product of the Yield (volume %) and the specific
gravity divided by the molecular weight.

(ii) Divide the individual relative number of moles by the sum of the relative number of moles for
all pseudo-components.

This procedure is illustrated in Table 5.3.

Table 5.4 shows the values of all of the properties assigned to the pseudo-components using Equation
(5-2) with the constants given in Table 5.2. These data, along with the mole fractions computed from Table
5.3 can be treated as a complete compositional description for the crude oil. When used as input data for
INPROP, for example, the API gravity calculated using the Peng-Robinson equation is 39.2 oAPI, will
compare reasonably well with the reported value of 37.5 oAPI.

Table 5.2

Constants for Use with Equation (5-2)*

Property (θ) α β C
MW 4.5673 x 10-5 2.1962 -1.0164
Tc (oR) 24.2787 0.58848 0.3596
Pc (psia) 3.12281 x 109 -2.3125 2.3201
Vc (ft3/lb) 7.5214 x 10-3 0.2896 -0.7666
ω 1.6680 x 10-7 2.1520 -1.1632

* Values for MW, Tc, Pc, and Vc from Riazi and Daubert (1980); values for ω
fitted by the author using data presented by Katz and Firoozabadi (1978)

As a matter of interest the data contained in Table 5.1 can also be used directly as input to the
Technical Utility module HYPOS. The mole fractions and characterizing properties are all computed
automatically. The results obtained using HYPOS are shown in Table 5.5. Note that the characterizing
properties differ somewhat from those in Table 5.4, since HYPOS uses different correlations for estimating
them. In this case, however, the calculated oil gravity is 37.3 oAPI, which is in excellent agreement with the
reported value.

The definition and characterization of pseudo-components is always arbitrary to a degree, regardless


of the particular methods and correlations that are used. When experimental data are available (e.g. liquid
and gas volumes and/or properties from a separator test, bubble point or dew point data) one should always
compare them with predicted values. In some cases, the agreement can be improved by making modest
adjustments to the assigned values of various pseudo-component properties. This must always be done with
caution; if excessively large or unrealistic adjustments are required to make the agreement acceptable, it
probably means that the basic characterization is inadequate (e.g. perhaps more pseudo-components should
have been defined). However, being able to demonstrate good agreement with even very limited data
increases the confidence in values predicted for other conditions.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 31

Table 5.3

Calculation of Mole Fraction of Each Pseudo-Component

Specific Yield Molecular Relative Mole


Fraction Gravity (Vol.%) Weight Moles Fraction
(A) (B) (C) (D) (E)
1 0.7230 23.45 96.4 0.1758 0.4107
2 0.7960 15.00 150.0 0.0796 0.1859
3 0.8440 20.30 214.2 0.0800 0.1869
4 0.8715 4.05 267.1 0.0132 0.0309
5 0.8995 18.80 340.1 0.0497 0.1161
6 0.9225 3.85 431.2 0.0082 0.0192
7 0.9850 12.00 548.4 0.0216 0.0503
Total (S) 0.4281

Table 5.4

Pseudo-Component Characterization for the Brent Blend Crude Oil Using the Correlation of
Riazi and Daubert (1980)

Avg. BP Molecular API Pc Tc Vc


ω
(oC) Weight (oAPI) (kPa) (oC) (m3/mol)
89.5 96.4 64.2 3144.4 271.0 0.35 0.277
190.5 150.0 46.3 2226.5 377.8 0.75 0.421
287.0 214.2 36.2 1646.9 469.9 1.14 0.591
355.5 267.1 30.9 1358.6 531.3 1.42 0.729
439.0 340.1 25.8 1095.6 602.5 1.78 0.920
529.5 431.2 21.9 880.9 674.9 2.17 1.155
650.0* 548.4 12.2 742.1 780.8 2.64 1.446
* Arbitrarily assigned

5.3 Liquid Density

Because of its overall accuracy and relative simplicity, the Peng-Robinson equation of state is a
popular choice for performing flash calculations. However, for well defined compositions, this equation also
tends to under-predict the density of the liquid phase (an even more pronounced phenomenon with the SRK
equation).
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 32

Table 5.5

Pseudo-Component Characterization for the Brent Blend Crude Oil Using the
Technical Utility Module HYPOS 4.1

Avg. BP Molecular API Pc Tc Vc


ω
(°C) Weight (oAPI) (kPa) (°C) (m3/mol)
89.5 93.4 64.2 3178. 271. 0.368 0.278
190.5 150.7 46.3 2243. 378. 0.592 0.422
287.0 226.4 36.2 1612. 470. 0.897 0.592
355.5 292.9 30.9 1298. 531. 1.171 0.730
439.0 389.6 25.8 1016. 602. 1.576 0.920
529.5 519.4 21.9 785. 674. 2.134 1.156
650.0* 663.8 12.2 673. 781. 2.701 1.448
* Arbitrarily assigned

Frequently, the under-prediction, if any, is small, and has relatively little effect on subsequent flow
calculations. When it is apparent however from comparison with available data that the error is significant,
it has become common practice to recompute the liquid density using an alternate equation of state, but with
the liquid composition predicted by the Peng-Robinson equation.

There is no unanimous agreement regarding which alternate equation to use. Varying success has
been reported using the BWRS equation (Starling, 1970), the Schmidt-Wenzel equation (1980), and the
Hankinson and Thomson (1979) corresponding states liquid density (i.e. COSTALD) correlation.

As originally proposed by Hankinson and Thomson, the COSTALD correlation is based on the
assumption that the liquid is saturated and in contact with the vapour phase at equilibrium conditions. This
is, of course, consistent with cases in which flash calculations result in the prediction of a two phase mixture.

In a later study, Thomson et al (1982), presented a modified COSTALD procedure that also applies
to compressed or under-saturated liquids (i.e. single phase liquid at pressures above the bubble point). In
any case, the procedure requires that the composition of the liquid be known, which will always be the case,
following a flash calculation.

In addition to the critical properties, the COSTALD correlation requires two additional parameters
for each component. The first is designated as the characteristic volume; the second is the acentric factor
based on the SRK equation of state. Values of both of these parameters are tabulated in the original paper
for almost 200 components. Mixing rules are provided by the authors to compute appropriate values of the
required parameters for defined mixtures. In general, the COSTALD correlation has been found to give good
results for relatively light liquids (i.e. typical condensates), where the composition is relatively well defined
by known pure components.

For systems where a significant proportion of the composition is represented by pseudo-components,


the situation can be very different. Depending on the characterization of the pseudo-components, the Peng-
Robinson equation can over-estimate the density. Hankinson and Thomson proposed a procedure for
estimating the values of the two characteristic parameters for pseudo-components, but the correlation tends
to underestimate the density, sometimes very substantially

Gregory (1993b) used boiling point assay data for a wide range of crude oils to develop a different
procedure for characterizing the specific volume. This procedure is also tuned to using a standard method
for estimating the acentric factor, rather than having to have an acentric factor that is specific to the SRK
equation of state. Good results have been obtained with matching the reported gravity for crude oils, based
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 33

on a composition that is totally defined by pseudo-components that were derived from the boiling point
fractions.

For example, Ahrabi et al (1987) report measured density data for a North Sea crude oil. The
composition data include 11 pseudo-components, which collectively represent a C7+ mole fraction of almost
53%. This is not at all unusual; in fact, C7+ mole fractions of over 65% are often encountered. Measured
values of the average boiling point, molecular weight, and relative density are given in the paper for each of
the pseudo-components. These data were used with HYPOS to complete the pseudo-component
characterization.

Figures 5.2 and 5.3 show a comparison, at 30 °C and 102 °C respectively, of the measured crude oil
densities, and the densities computed using both the original and the revised COSTALD method, and the
Peng-Robinson EOS. In the “Revised” version calculations, the values of the two COSTALD parameters
(characteristic volume and acentric factor) for the pseudo-components were evaluated using the procedures
suggested by Gregory (1993b). In this particular case, at least, the Peng-Robinson equation gives densities
that are typically 8 - 10% too high. By comparison, the values computed using either version of the
COSTALD method are seen to be in much better agreement with the data. Values predicted by the original
version are, however, almost independent of pressure.

Figures 5.2 and 5.3 provide further illustration the importance of having some data with which to
compare calculated results. Different procedures that are legitimately considered to represent good and
reliable technology can still give significantly different results in a specific case. Even a small amount of
measured data can be used to select the best method for a specific application. It must, of course, also be
understood that the results shown here are for one particular crude oil, and do not necessarily represent a
general trend.

900

Peng-Robinson EOS
880 Original COSTALD
Modified COSTALD
860 Measured
DENSITY (kg/m³)

840

820

800

780

760

740
0 50 100 150 200 250 300 350

PRESSURE (bar)

Figure 5.2

Comparison of Calculated and Measured Densities at 30 oC for the Crude


Oil Composition Reported by Ahrabi et al (1987)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 34

820

Peng-Robinson EOS
Original COSTALD
800
Modified COSTALD
Measured

DENSITY (kg/m³) 780

760

740

720

700
80 90 100 110 120 130 140 150 160 170 180

PRESSURE (bar)

Figure 5.3

Comparison of Calculated and Measured Densities at 102 oC for the Crude


Oil Composition Reported by Ahrabi et al (1987)

6. Phase Behaviour Of Black Oil Systems

6.1 Characteristics of Black Oil Systems

The term Black Oil is used to denote a particular class of fluid models for phase behaviour and
transport properties. All of these models are based on the following assumptions:

(a) The system consists of only two components, an oil and a gas.

(b) The gas is assumed to have a constant composition (effectively a pseudo-pure component), and
therefore has a constant gravity or relative density (compared to air = 1.0), regardless of the
pressure and temperature.

(c) The oil is assumed to have a constant composition (i.e. also a pseudo-pure component), and thus
also has a constant specific gravity (i.e. relative density), API gravity, Watson's K factor, etc.

(d) At pressures greater than 1.0 atmosphere, gas is assumed to go into solution or dissolve in the oil
(e.g. similar to CO2 in a carbonated soft drink). The process is thus viewed as a solubility
phenomenon, rather than as a phase change in the classical thermodynamic sense. Consistent
with this, it is also assumed that there are no inherent heat effects (e.g. heat of solution, heat of
vaporization) accompanying any changes in the amount of dissolved gas.

Before proceeding further, we need a few more definitions.


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 35

Solution Gas/Oil Ratio (Rs)

The solution gas/oil ratio, Rs , is the amount of gas that is in solution, at any given pressure and
temperature, in a specified quantity of oil. At stock tank conditions, it is assumed that all gas has
completely evolved from the oil, and thus, Rs = 0 . Common units for Rs are scf/STB, or m3 at
standard conditions/m3 stock tank oil (i.e Sm3/Sm3).

Produced Gas/Oil Ratio (GOR)

The produced gas/oil ratio, GOR , is the total amount of gas that is produced from the reservoir along
with one stock tank volume of oil. The gas may be produced as either free gas, or gas that is initially
in solution. When the pressure in the system exceeds the bubble point pressure at a given
temperature, all of the produced gas will be in solution, and Rs = GOR. However, if the pressure
in the system is less than the bubble point pressure for the given temperature, Rs < GOR , and a two
phase mixture of oil and gas will exist. Common units for GOR are scf/STB, or m3 at standard
conditions/m3 stock tank oil (i.e. Sm3/Sm3).

Dead Oil

Oil which contains no dissolved gas is referred to as dead oil (i.e. Rs = 0 ). At stock tank
conditions, all gas is assumed to be free gas (i.e. Rs = 0 ), and thus, stock tank oil is always dead
oil. Note however that dead oil can exist at any pressure greater than stock tank pressure, provided
that Rs = 0 . This latter condition can only be satisfied if there is no gas present.

Live Oil

This is the opposite of dead oil, and corresponds to Rs > 0 . Live oil is considered to be saturated
if the oil and gas are in equilibrium. This is always assumed to be the case if the oil is in contact with
a free gas phase. However, if Rs = GOR , the oil could either be at its bubble point, or it could be
undersaturated. That is, the oil has the capacity at the given pressure and temperature to hold more
dissolved gas, and would do so if any more free gas were present. Undersaturated oil therefore
always exists as a single phase liquid system.

Oil Formation Volume Factor (Bo)

The oil formation volume factor, Bo , is the volume occupied at any given pressure and temperature
by one stock tank volume of oil plus its solution gas (if any). Thus, by definition, Bo = 1.0 for
stock tank oil. Typical units for Bo are bbl/STB, or m3/m3 of stock tank oil (i.e. m3/Sm3).

Gas Formation Volume Factor (BG)

As noted earlier in Section 2, the gas formation volume factor, BG , is the ratio of the volume
occupied at a specified pressure and temperature by a given mass of gas to the volume occupied by
the same mass of gas at the base conditions (i.e. stock tank conditions). Typical units for BG are
ft3/scf and m3/m3 at base conditions. While this parameter is commonly used in reservoir studies, it
is seldom used in pipeline or wellbore calculations.

At pressures above 1.0 atmosphere, and any temperature, the amount of free gas is directly related
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 36

to the difference between the produced gas/oil ratio and the solution gas/oil ratio.

At the bubble point pressure for the oil, corresponding to the reservoir temperature, all gas associated
with the oil is assumed to be in solution, and thus Rs = GOR . However, if any free gas remains, the oil is
considered to be saturated, and the system is assumed to be at thermodynamic equilibrium. If all free gas
could be removed without changing either the pressure or the temperature (e.g. at a separator exit), the oil
would still be saturated, and the new oil-gas mixture would be at its bubble point. The value of Rs can thus
be viewed as the value of the GOR for which the particular pressure and temperature would represent the
bubble point. Correlations which are used to describe the relationship between pressure, temperature, gas
and oil properties, and Rs are thus generally referred to as bubble point correlations.

Under the Black Oil model assumptions, the phase behaviour can be determined at any pressure and
temperature using relatively simple correlations which depend only on three basic system parameters:

API, or specific gravity (i.e. relative density) of the oil


Gravity (i.e. relative density) of the gas
Produced gas oil ratio

These data are relatively easy to obtain, compared to a detailed compositional analysis (including the
characteristics of residual fractions), and are thus almost always available. There are thus obvious benefits
to being able to analyze a system as a Black Oil. In most cases, both the bubble point and transport property
correlations can be tuned relatively easily, using even limited measured data, which often results in
significantly improved prediction accuracy.

On the other hand, Black Oil models represent a simple approach to a complex phenomenon, and thus
have some definite limitations. Because of the inherent assumptions, they should not be used with gas
condensate or other highly volatile systems. The PVT behaviour of those systems tends to be quite sensitive
to relatively modest pressure and/or temperature changes, and the assumptions of constant gas and liquid
properties are generally not valid. Compositional methods are definitely preferred for such systems.

For most crude oil-gas systems, however, Black Oil models can often be used to predict PVT
behaviour and transport properties with accuracies that are as good as, or better, than those obtained using
more complex compositional models. This is especially true in those cases where a significant fraction of
the system has been arbitrarily defined as pseudo-components (25% or more, mole basis, is not unusual for
even relatively light crude oil systems). Furthermore, “tuning” or “calibrating” Black Oil models to improve
the calculation accuracy (using minimal measured data) is generally far easier than for compositional
equations of state.

It has become apparent over the years, however, that no single set of Black Oil PVT correlations can
consistently give an adequate description of crude oil systems from all different producing areas of the world.
Consequently, numerous correlations have been developed and reported in the literature, each typically based
on data from a relatively specific region. We now look at a some specific examples of these.

6.2 Standing (1947) Correlations for Rs and Bo

The Standing (1947) correlation for Rs , which is based on data from 22 California crude oil-gas
systems, is shown in Figure 6.1 . This relationship for the bubble point pressure can be approximated by the
expression,

(6-1)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 37

which can also be rearranged, to give an explicit equation for Rs , as,

(6-2)

where

(6-3)

and P = pressure, psia


T = temperature, °F
API = API gravity of the oil, oAPI
Rs = solution gas/oil ratio, scf/STB
C = constant ( = 18 in the original paper)

For the data shown in Figure 6.1, C = 18 , and this is the value that is normally used. However, Standing
observed that a similar plot of bubble point data for other oils may result in a line that is almost parallel to
that shown in Figure 6.1, but slightly displaced from it. Examination of Equation (6-1) suggests that such
a line would simply correspond to a different value of C . This, then, is the basis of a possible calibration
procedure for Equations (6-1) and (6-2). Even a single measured value of Rs , for a given pressure and
temperature (e.g. from a separator test), can be used to compute an appropriate value of C , and thus to force
the equations to match the known data point. With two or more measured points, an even more reliable value
of C could be computed, using say, the well-known method of least squares.

Example

Consider the following crude oil-gas system:

Oil gravity: 32 oAPI


Gas gravity: 0.72
GOR: 610 scf/STB

At the reservoir temperature of 195 °F, the bubble point pressure has been measured to be 2,630
psia.

The measured bubble conditions can be used to determine a revised value of C from a rearranged
version of Equation (6-1), namely,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 38

Figure 6.1

Bubble Point Correlation of Standing (1947)

(6-4)

At the reservoir bubble point pressure and temperature, we know that Rs = GOR = 610 scf/STB.
Using this value with the specified pressure, temperature, and fluid properties gives C = 16.3 .

Figure 6.2 shows a comparison of the values of Rs calculated for this system at the reservoir
temperature using both the default and the adjusted values of the correlation constant. As a matter of interest,
Figure 6.2 also shows the same comparison at 90 °F, which could, say, correspond to the flowing temperature
in a pipeline.

Recall that the amount of free gas in the system is directly related to the quantity (GOR - Rs ) . In this
case, the effect of the calibration is to increase Rs at any given pressure, and thus, to reduce the calculated
amount of free gas. The magnitude of that reduction is summarized in Table 6.1 for some selected pressures.
Whether or not this has a significant effect on the accuracy of the calculations for a given flowing system
must be determined in individual cases. It could, for example, certainly be significant in a flowing well,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 39

where gas-lift mechanisms can be an important factor in the calculation of the overall pressure loss.

700
Oil Gravity = 32.0 °API
Gas Gravity = 0.72
600 GOR = 610 scf/stb
SOLUTION GAS/OIL RATIO (scf/stb)

500
Actual
90 °F Bubble Point
at 195 °F
400
195 °F

300

200 C = 18
C = 16.3

100

0
0 500 1000 1500 2000 2500 3000 3500

PRESSURE (psia)

Figure 6.2

Comparison of Rs Values Calculated Using Calibrated and Uncalibrated Versions


of Standing’s (1947) Bubble Point Correlation

As noted earlier, an adjusted value of the constant C could also have been determined from Equation
(6-4) using a measured value of Rs obtained from a separator test.

Unfortunately, the simple calibration procedure described above can be difficult to apply to other
more complex bubble point correlations. Most of them do not have such an easily identifiable constant that
can be adjusted without significantly altering the behaviour of the correlation with respect to pressure and
temperature changes. Later in these Notes, however, a simple alternative calibration procedure is described
which not only can be applied to any bubble point correlation, but also provides a basis for comparing the
applicability of various correlations in a given case.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 40

Table 6.1

Ratio of the Calculated Free Gas Volume with Calibration to the Calculated Free Gas
Volume without Calibration

Temperature
Pressure
(psia) 90 °F 195 °F

100 0.997 0.998


500 0.977 0.983
1000 0.929 0.952
1500 0.820 0.897
2000 0.376 0.778

Standing’s (1947) correlation for the oil formation volume factor is shown in Figure 6.3 . The curve
in that figure is reasonably well approximated by the expression,

(6-5)

where

(6-6)

where Bo = oil formation volume factor, bbl/STB


SGo = oil specific gravity

6.3 Lasater (1958) Correlation for Rs

The bubble point correlation of Lasater (1958) is shown in Figures 6.4(a) and 6.4(b). It was
developed using 158 data points, which represented 137 crude oils from Canada, western and mid-continent
USA, and South America.

The first step in using this procedure is to determine the effective molecular weight of the oil,
designated as Mo, from Figure 6.4(a), as a function of the oil gravity. The actual molecular weight is seldom
known for most crude oils, and Lasater has based Figure 6.4(a) on crude oils having a Watson's K factor of
11.8 . This is, in fact, a fairly typical value for the crude oils used as the basis of his correlation.

The effective molecular weight is related to the mole fraction of the gas in the overall system, YG,
by the expression,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 41

Figure 6.3

Oil Formation Volume Factor Correlation of Standing (1947)

(6-7)

or,

(6-8)

where YG = mole fraction of the gas


Rs = solution gas/oil ratio, scf/STB
SGo = oil specific gravity
Mo = effective molecular weight of the oil
API = API gravity of the oil, oAPI
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 42

The relationship between YG and the quantity PG/(T + 460) is shown in Figure 6.4(b). The curve
in this figure is reasonably well approximated by two expressions, as follows:

(i) YG > 0.6,

(6-9)

(ii) 0.05 < YG < 0.6,

(6-10)

where
(6-11)

and P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)

Lasater did not present a correlation for the oil formation volume factor, and his bubble point
correlation is usually used in combination with Standing's (1947) correlation for Bo.

6.4 Vazquez and Beggs (1977) Correlations for Rs and Bo

Vazquez and Beggs (1977) used a total of 6,004 data points in the development of their correlations.
Their expression for Rs has the general form,

(6-12)

where P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)
API = API gravity of the oil, oAPI
Rs = solution gas/oil ratio, scf/STB

The correlation is actually defined for two different ranges of the API gravity, as the authors found that this
improved the accuracy significantly. Values of the constants a1, a2, and a 3 in Equation (6-12) are given in
Table 6.2 for the two cases.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 43

6.4

6.0

5.6

55 5.2

T + 460
PG
4.8
GRAVITY OF STOCK TANK OIL (°API)

50
4.4
45

BUBBLE POINT PRESSURE FACTOR,


4.0
40
3.6
35
3.2
30 2.8

25 2.4

20 2.0

15 1.6
100 200 300 400 500
1.2
EFFECTIVE MOLECULAR WEIGHT
OF STOCK TANK OIL 0.8

(a) 0.4

0.0
0.0 0.2 0.4 0.6 0.8 1.0
GAS MOLE FRACTION
(b)

Figure 6.4

Bubble Point Correlation of Lasater (1958)

The correlation for the oil formation volume factor has the general form,

(6-13)

where Bo = oil formation volume factor, bbl/STB

As with the correlation for Rs, values of the constants b1, b2, and b3 in Equation (6-13) depend on the API
gravity of the oil, and they are also given in Table 6-2 .
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 44

Table 6.2

Values of the Constants in Equations (6-12) and (6-13)

Constant API < 30° API > 30°


a1 0.0362 0.0178
a2 1.0937 1.187
a3 25.724 23.931
b1 4.677 x 10-4 4.67 x 10-4
b2 1.751 x 10-5 1.10 x 10-5
b3 -1.8106 x 10-8 1.337 x 10-9

6.5 Glaso/ (1980) Correlations for Rs and Bo

The Glaso/ (1980) PVT correlations were developed using a data base of 45 oil samples taken from
reservoirs located primarily in the North Sea. His bubble point pressure correlation is expressed by the
relation,

(6-14)

where,

(6-15)

where P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)
API = API gravity of the oil, oAPI
Rs = solution gas/oil ratio, scf/STB

To determine Rs at a given pressure and temperature, we must first calculate the value of p* from
Equation (6-14), which is actually a quadratic relation in log10 (p*). The appropriate root is given by:

(6-16)

Finally, Equation (6-15) can be arranged to give,

(6-17)

While the above correlation gave good results for many typical crude oils, Glaso/ also found that the
agreement with data for volatile oils could be significantly improved by using a different exponent for the
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 45

temperature in Equation (6-15). Thus, for volatile oils,

(6-18)

which can be rearranged to give,

(6-19)

Glaso/'s oil formation volume factor correlation is given by the relation,

(6-20)

where

(6-21)

and Bo = oil formation volume factor, bbl/STB

6.6 Al-Marhoun (1985, 1988, 1992) Correlations for Rs and Bo

Al-Marhoun (1985) initially presented correlations for both Rs and Bo that were developed using
about 200 measured data points taken for 75 samples of Saudi crude oils. His equation for Rs was actually
written in terms of the pressure, as follows,

(6-22)

where

(6-23)

and P = pressure of interest (i.e. bubble point pressure corresponding to T), psia
Rs = solution gas-oil ratio, scf/STB
G = gas gravity (relative to air = 1)
SGo = oil specific gravity

Values of the constants a1 through a 7 in Equations (6-22) and (6-23) are given in Table 6.3. The correlation
for the oil formation volume factor is given by,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 46

(6-24)

where

(6-25)

Values of the constants b1 through b7 in Equations (6-24) and (6-25) can also be found in Table 6.3.

Somewhat later, Al-Marhoun (1988) developed new correlations for both Rs and Bo using a data base
that included about 160 measured data point for crude oils from 69 Middle East reservoirs. The bubble point
correlation is again presented as an explicit equation for computing the pressure corresponding to a given
solution gas-oil ratio, i.e.

(6-26)

where P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)
SGo = specific gravity of the oil
Rs = solution gas/oil ratio, scf/STB

Values of the constants c1 through c5 in Equation (6-26) are given in Table 6.3. In this case,
Equation (6-26) can readily be rearranged to give an expression that is explicit in Rs, or

(6-27)

The revised oil formation volume factor correlation is given by

(6-28)

where

(6-29)

and Bo = oil formation volume factor, bbl/STB

Values of the equation constants d1 through d7 are also given in Table 6.3.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 47

Table 6.3

Constants for Use With Equations (6.22) through (6.30)

No. a b c d e
1 -64.13891 0.574095 5.38088 x 10-3 0.497069 1.77342 x 10-4
2 7.02362 x 10-3 7.723532 x 10-4 0.715082 8.62963 x 10-4 2.20163 x 10-4
3 -2.278475 x 10-9 2.454005 x 10-3 -1.877840 1.82594 x 10-3 4.29258 x 10-6
4 0.722569 3.727676 x 10-5 3.143700 3.18099 x 10-6 5.28707 x 10-4
5 -1.879109 0.501538 1.326570 0.742390 -
6 3.046590 -0.145526 - 0.323294 -
7 1.302347 -5.220726 - -1.202040 -

Still later, Al-Marhoun (1992) presented yet another correlation for the oil formation volume factor,
this time based on about 4,000 data points, representing some 700 crude oil samples taken from all over the
world, but primarily from the Middle East and North America. This relationship is given by,

(6-30)

Values of the constants e1 through e4 are again given in Table 6.3 .

6.7 Abdul-Majeed and Salman (1988) Correlation for Bo

This correlation is based on 420 measured data points for 119 crude oil-gas systems, which primarily
represent Middle East reservoirs. The correlation is given by the relationship,

(6-31)

where

(6-32)

and Bo = oil formation volume factor, bbl/STB


T = temperature, °F
G = gas gravity (relative to air = 1)
SGo = specific gravity of the oil
Rs = solution gas/oil ratio, scf/STB

Values of the constants a1 through a4, and b1 through b3 , are given in Table 6.4.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 48

Table 6.4

Constants for Use with Equation (6-31) and (6-32)

a b
1 0.9657876 1.2
2 4.8141 x 105 -0.147
3 -6.8987 x 10-10 -5.222
4 7.73 x 10-4 -

6.8 Asgarpour et al (1989) Correlations for Rs and Bo

These authors based their correlations on data taken from over 310 crude oil samples taken from
western Canadian crudes. It was found that the best results were obtained when a unique correlation was used
for each of three particular reservoir formation types (Cardium-Viking, Nisku, and Leduc). They noted that
the importance of these particular formations is such that they account for more than 50% of the original oil
in place in western Canada.

The solution gas/oil ratio correlations for the various formations are expressed as follows:

Cardium-Viking

(6-33)

Nisku

(6-34)

Leduc

(6-35)

where

(6-36)

and P = pressure, kPa


T = temperature, °C
G = gas gravity (relative to air = 1)
API = API gravity of the oil, oAPI
Rs = solution gas/oil ratio, m3/m3

Values of the constants a1 through a5 in Equation (6-36) are given in Table 6.5 . It is not noted in
the paper whether the pressure is gauge or absolute, although SI convention generally considers it to be
absolute.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 49

In either case, there are two limitations of the above relationships that should be noted. Firstly, the
temperature appears in Equation (6-36) with units of °C, but it is raised to an exponent. This inherently limits
the use of these correlations to temperatures greater than 0 °C (i.e. greater than 32 °F ), since a negative
number raised to a non-integer exponent is undefined. Secondly, the polynomial form chosen for Equations
(6-33) and (6-35) means that they do not satisfy the normal Black Oil model requirement that Rs = 0 at
stock tank conditions. The deviation is small however, and should not cause any problems in a practical
sense.

The correlations for the oil formation volume factor are similar in nature, and are written as:

Cardium-Viking

(6-37)

Nisku

(6-38)

Leduc

(6-39)

where

(6-40)

and Bo = oil formation volume factor, bbl/STB

Values of the constants b1 through b 6 for Equation (6-40) are given in Table 6.5 . As for the
solution gas/oil ratio correlations, the polynomial form used for these relationships means that they do not
satisfy the requirement that Bo = 1.0 at stock tank conditions. At typical operating pressures, however, the
calculated results are reasonable.

6.9 Dokla and Osman (1992) Correlations for Rs and Bo

The data used to develop these correlations represented 51 bottom-hole samples taken from UAE
reservoirs in the Middle East. The solution gas/oil ratio correlation is given by,

(6-41)

where P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)
SGo = specific gravity of the oil
Rs = solution gas/oil ratio, scf/STB

Values of the constants a1 through a5 in Equation (6-41) are given in Table 6.6.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 50

The oil formation volume factor correlation has the form,

(6-42)

and

(6-43)

where Bo = oil formation volume factor, bbl/STB

Values of the constants b1 through b4 in Equation (6-42) and c1 through c3 in Equation (6-43) are also
given in Table 6.6.

Table 6.5

Values of Constants in Equations (6-36) and (6-40)

Constant Cardium/Viking Nisku Leduc


a1 4.891 x 10-5 1.462 x 10-3 2.08 x 10-4
a2 0.165 0.4013 0.0752
a3 1.185 0.2368 1.1232
a4 -0.223 -0.1167 -0.3572
a5 0.3614 0.2898 0.2499

b1 0.0899 0.1851 0.179


b2 0.0567 0.1149 -0.445
b3 0.297 0.05903 0.2490
b4 0.0753 0.1473 5.39 x 10-4
b5 0.0955 0.0831 0.0799
b6 0.0928 0.0806 0.0815

Table 6.6

Values of the Constants in Equations (6-41), (6-42), and (6-43)

Constant No. a b c
1 3.825276 x 10-6 4.31935 x 10-2 0.773572
2 1.381126 1.56667 x 10-3 0.404020
3 1.395614 1.39775 x 10-3 -0.882605
4 -1.49492 x 10-1 3.80525 x 10-6 -
5 1.315638 - -
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 51

6.10 Petrosky and Farshad (1993) Correlations for Rs and Bo

These authors present correlations which they developed based on data from 81 oil samples taken
from reservoirs in the Gulf of Mexico. Their correlation for the solution gas/oil ratio is expressed as,

(6-44)

where

(6-45)

and P = pressure, psia


T = temperature, °F
G = gas gravity (relative to air = 1)
API = API gravity of the oil, oAPI
Rs = solution gas/oil ratio, scf/STB

Values of the constants a1 through a8 in Equations (6-44) and .6-45) are given in Table 6.7 .

The oil formation volume factor correlation is given by,

(6-46)

where Bo = oil formation volume factor, bbl/STB


SGo = specific gravity of the oil

Values of the constants b1 through b8 in Equation (6-46) are also given in Table 6.7

6.11 Kartoatmodjo and Schmidt (1994) Correlations for Rs and Bo

These authors developed their correlations using a relatively large data base consisting of 5,392 data
points, representing 740 different crude oil samples. Data represent oils produced from Southeast Asia
(mostly Indonesia), North America, the Middle East, and Latin America. A second data base consisting of
998 data sets was not used in the development of the correlations, but was reserved for subsequent testing
purposes. Their correlation for Rs has the following general form:

(6-47)

here Rs = solution gas/oil ratio, scf/STB


G100 = gas gravity at the separator temperature and a pressure of 100 psig
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 52

API = gravity of the oil, oAPI


P = pressure of interest, psia
T = temperature of interest, °F

Table 6.-7

Constants for Use in Equations (6-44) through (6-46)

Constant No. a b
1 112.727 1.0113
2 12.340 7.2046 x 10-5
3 0.8439 0.3738
4 1.73184 0.2914
5 7.916 x 10-4 0.6265
6 1.5410 0.25626
7 -4.561 x 10-5 0.5371
8 1.3911 3.0936

The values of the constants a1 , a2, a3, and a4 depend on the API gravity of the oil in a similar way to those
of the Vazquez and Beggs (1980) correlations, and are given in Table 6.8 .

Table 6.8

Constants for Use in Equation (6-47)

Constant API < 30 API > 30


a1 0.05958 0.03150
a2 0.7972 0.7587
a3 1.0014 1.0937
a4 13.1405 11.2895

The gas gravity is often known only for some pressure other than 100 psig (e.g. at some arbitrary
separator pressure). If no pressure is specified, the known value should simply be used in the correlation in
place of G100 . However, when the actual pressure is known (e.g. from a separator test), the following
expression, which is based on the functional form used by Vazquez and Beggs (1977), can be used to adjust
the known gas gravity to correspond to the required pressure of 100 psig. The constants in this equation have
been re-evaluated using regression analysis with the available data base:

(6-48)

where Gsep = known gas gravity at separator conditions Psep and Tsep
Psep = separator pressure (psia)
Tsep = separator temperature (°F)

Their correlation for Bo is expressed by the relationship,


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 53

(6-49)

where

(6-50)

and SGo = specific gravity of the crude oil

6.12 Velarde et al (1999) Correlations for Rs and Bo

These authors have taken a somewhat different approach to formulating their correlations in that they
are defined in terms of what they refer to as reduced functions. The definition of these reduced functions is
analogous to the familiar definitions of the reduced pressure and temperature, but the normalizing factors are
not the critical properties.

The reduced pressure variable is defined in terms of the bubble point pressure, as

(6-51)

where Ppr = reduced pressure variable


P = the pressure of interest, psia
Pb = bubble point pressure at the current temperature, psia

A comparable expression is used to define the reduced solution gas-oil ratio, i.e.

(6-52)

where Rsr = reduced solution gas-oil ratio


Rs = solution gas-oil ratio at the pressure and temperature of interest
Rsb = solution gas-oil ratio at the bubble point pressure for the temperature of interest

The reduced variables defined by Equations (6-51) and (6-52) were then correlated according to the
following expression,

(6-53)

In Equation (6-53), the parameters a1 , a2 , and a3 are evaluated according to the following
expressions as functions of the temperature and the parameters that define the fluid system:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 54

(6-54)

(6-55)

(6-56)

where G = gas gravity (relative to air = 1)


API = gravity of the oil, °API
T = temperature, °F
Pb = bubble point pressure corresponding to T, psia

The values of the coefficient Ai , Bi , and Ci in Equations (6-54), (6-55), and (6-66) respectively, are given
in Table 6.9 below.

Table 6.9

Values of the Constants for Equations (6-54), (6-55), and (6-56)

Constant No. A B C
0 9.73 x 10-7 0.022339 0.725167
1 1.672608 -1.004750 -1.485480
2 0.929870 0.337711 -0.164741
3 0.247235 0.132795 -0.091330
4 1.056052 0.302065 0.047094

6.13. Brief Comparison of Various Bubble Point Correlations

For comparison purposes, the solution gas/oil ratio has been computed over a range of pressures at
a given reservoir temperature (210 oF) for one particular crude oil-gas system, using the various bubble point
correlations which have been presented. The results of these calculations are summarized graphically in
Figures 6-5(a), 6-5(b), and 6-5(c). The Standing (1947) correlation has been shown on all three plots to give
a convenient reference.

These data, of course, pertain only to one specific oil/gas system and should not therefore be viewed
as necessarily representing any general trends. They have been included here, however, to illustrate the large
variability that can result from using different correlations. Table 6-9 shows the bubble point pressure at 210
o
F that is predicted by each of the methods.

It must also be noted, however, that each correlation represented in Figure 6-5 was derived from
actual measured data, and is therefore a legitimate procedure that can be used to predict the phase behaviour
for, at the very least, some limited or well-defined set of crude oil-gas systems. When one goes back to the
original papers, it is apparent that the data tended to come mostly from some particular producing area, and
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 55

that there is thus a significant geographic effect. A correlation that is based on data from another region, in
which the crudes are significantly different, should not then be expected to show the same accuracy.

The variability is thus not an inherently bad thing. The correlations are all trying to predict the same
properties, but not necessarily for the same oils. However, if the authors of a particular correlation actually
based their work on a truly representative sample of their regional crude oils, their procedure might be
expected to yield more accurate results than some other procedure.

There are times, however, when one is not certain of the origin of the oil for which some type of
design calculations are required. The variability that is observed in this particular comparison study
reinforces the importance of being able to use even limited measured data to either help select or to calibrate
a particular correlation for any given crude oil-gas system.

Table 6-9

Comparison of Calculated Bubble Point Pressures at 210 °F

Pbub
Correlation
(psia)
Standing (1947) 3940
Vazquez and Beggs (1977) 4395
Lasater (1958) 3975
Glaso/ (1980) [normal crude] 4284
Glaso/ (1980) [volatile oil] 3486
Al-Marhoun (1988) 4354
Asgarpour et al (1989) [Cardium-Viking] 3487
Asgarpour et al (1989) [Nisku] 3033
Asgarpour et al (1989) [Leduc] 3334
Petrosky and Farshad (1993) 4559
Dokla and Osman (1992) 3103
Kartoatmodjo and Schmidt (1994) 4541
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 56

1000

900

800

SOLUTION GAS/OIL RATIO (scf/stb) 700

600

500

400
Standing (1947)
300
Vazquez & Beggs (1977)
Lasater (1958)
200
G lasø (1980) [normal crude]
G lasø (1980) [volatile oil]
100
Al-Marhoun (1988)

0
0 1000 2000 3000 4000 5000

PRESSURE (psia)

Figure 6-5 (a)

Comparison of Solution Gas-Oil Ratios Calculated by Various Correlations


(gas gravity = 0.73, oil gravity = 32.5 oAPI, GOR = 875 scf/STB, T = 210 oF)

1000

900
SOLUTION GAS/OIL RATIO (scf/stb)

800

700

600

500

400

300
Standing (1947)
200 Asgarpour et al (1989) [Cardium-Viking]
Asgarpour et al (1989) [Nisku]
100 Asgarpour et al (1989) [Leduc]

0
0 1000 2000 3000 4000 5000

PRESSURE (psia)

Figure 6-5 (b)

Comparison of Solution Gas-Oil Ratios Calculated by Various Correlations


(gas gravity = 0.73, oil gravity = 32.5 oAPI, GOR = 875 scf/STB, T = 210 oF)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 57

1000

900

SOLUTION GAS/OIL RATIO (scf/stb)


800

700

600

500

400

300
Standing (1947)
200 Petrosky and Farshad (1993)
Dokla and Osman (1992)
100 Kartoatmodjo and Schmidt (1994)

0
0 1000 2000 3000 4000 5000

PRESSURE (psia)

Figure 6-5 (c)

Comparison of Solution Gas-Oil Ratios Calculated by Various Correlations


(gas gravity = 0.73, oil gravity = 32.5 oAPI, GOR = 875 scf/STB, T = 210 °F)

6.14 Recommended Calibration Procedure for Rs and Bo Correlations

The author has devised a simple way to apply a systematic shift to the calculated value of Rs , without
changing either the temperature or fluid property dependencies inherent in a given correlation. It consists
of simply replacing R s in the correlation with the term CRS Rs . The parameter CRS is a calibration factor, and
it has the nominal default value of 1.0 (i.e. if the correlation is uncalibrated). One of the main advantages
of this procedure is that it can be used with any correlation, as it does not depend on the equation inherently
having an arbitrary constant that can be adjusted. Other positive characteristics are that the actual value of
Rs remains preserved, and the extent to which CRS differs from 1.0 gives a good indication of how well the
original form of the correlation agrees with the known data point.

For example, if we use this procedure to perform the calibration of the Standing (1947) correlation
described earlier in Section 6.2, Equation (6-1) becomes,

(6-57)

which can be rearranged to give,


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 58

(6-58)

With Pb = 2630 psia, Rs = 610 scf/STB, API = 32o, G = 0.72, and T = 195 oF, we obtain the
result from Equation (6-58) that CRS = 0.888 . A value of 1.0 would have indicated that the correlation is
in perfect agreement with the data. The actual value indicates that the calculated solution gas/oil ratio would
only be 88.8% of the actual value at the specified bubble point, using the original Standing correlation.

As a matter of interest, these same data were used to compute values of the calibration factor for the
other solution gas/oil ratio correlations discussed above. The results are summarized in Table 6-10. Using
the calibration factors given in that table, all of the correlations will exactly match the known bubble point
condition. For pressures that are in the same general order of magnitude as the bubble point pressure, it is
therefore reasonable to expect that the accuracy of those that were in otherwise poor agreement would be
significantly improved. It is also apparent from Table 6-10 that several of the correlations considered (e.g.
Glasø [volatile oil], Asgarpour et al [Cardium/Viking and Leduc], and Dokla-Osman) are inherently in better
agreement with the data than others (i.e. because the values of CRS are close to 1.0). However, as is apparent
in Figure 6-5, recall that the behaviour of any given correlation with pressure can be quite different than that
of some other correlation. At pressures that differ significantly from that of the calibration point, there is no
assurance that the accuracy of a given correlation will be improved, and in fact, exactly the opposite may
occur! Ideally, one will have more than just a single measured point to use for validation purposes.

Because the most important parameter affecting the oil formation volume factor is also the solution
gas oil ratio, an almost identical procedure is used to calibrate the correlations for Bo. In this case, however,
the parameter Rs is replaced with C Bo R s everywhere that it occurs in the correlation. The latter can thus be
viewed as the effective solution gas/oil ratio that is required to match the data. Again, this method has the
advantage that the functional dependence of the correlation on temperature or other fluid properties is not
altered. As with the bubble point correlation calibration factor, values of CBo that are close to 1.0 indicate
inherently good agreement between the original form of the correlation and the data.

These calibration procedures can easily be used with any solution gas/oil ratio or oil formation
volume factor correlation, and are, in fact, the methods used in Neotec’s fluid property and PVT module,
INPROP. Values of the calibration factors are reported, when appropriate, in the .MSG output report from
that module.

6.15 Correction to Bo for Undersaturation

Figure 6-6 shows the typical behaviour of the oil formation volume factor that is observed as the
system pressure is increased at a constant temperature. From the initial pressure up to the bubble point
pressure (i.e. the point at which GOR = Rs, which happens to be 3,073 psia in this case), the oil is assumed
to be saturated, and Bo continues to increase, as more and more gas goes into solution. The effect of this
increasing solution gas is always much greater than the corresponding shrinkage of the oil due to pure
compression effects. At the bubble point, there is no more gas to go into solution, and the oil then becomes
progressively more undersaturated with increasing pressure. With the solution gas-oil ratio being constant,
the portion of the curve in Figure 6-6 labelled “Compressibility Ignored” shows the behaviour that would be
predicted by the correlations for Bo that we have looked at to this point. In actual fact, however, at pressures
greater than the bubble point pressure, Bo is decreasing, due totally to the compressibility of the oil. The
actual behaviour that is observed is thus indicated in Figure 6-6 by the portion of the curve labelled
“Compressibility Included”. In general, the compressibility of liquids tends to be relatively low, and the
pressure effect on Bo is thus not large. In this particular case, B o decreases from 1.417 at the bubble point
pressure to 1.389 at a pressure of 6,000 psia, which represents a volume decrease of only about 2% for a
pressure increase of almost 50%. For some fluid systems, however, particularly lighter oils with relatively
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 59

high GOR values, the effect can be significantly larger.

In most pipeline and wellbore applications, the effect of this phenomenon on design or performance
analysis calculation results will be negligible and could be ignored without significant error. In some
reservoir studies, however, ignoring the undersaturation effect could lead to an under-estimation of the oil
in place.

A number of correlations have been presented in the literature for adjusting the value of the oil
formation volume factor at the bubble point to account for undersaturation, and, for the sake of completeness,
we list some of those here. Almost all of the published correlations are presented in the form of equations
for computing the parameter Co in the following general expression for P > Pb,

(6-59)

Table 6-10

Comparison of Calculated Calibration Factors for


Various Solution Gas/Oil Ratio Correlations

Correlation CRs
Standing (1947) 0.888
Vazquez and Beggs (1977) 0.776
Lasater (1958) 0.783
Glaso/ (1980) [non-volatile] 0.754
Glaso/ (1980) [volatile] 0.989
Al-Marhoun (1988) 0.704
Asgarpour et al (1989) [Cardium-Viking] 0.982
Asgarpour et al (1989) [Nisku] 1.122
Asgarpour et al (1989) [Leduc] 1.082
Petrosky and Farshad (1993) 0.724
Dokla and Osman (1992) 1.087
Kartoatmodjo and Schmidt (1994) 0.788

where Co = isothermal oil compressiblity, psia-1


Bo = oil formation volume factor at pressure P, bbl/STB
Bob = oil formation volume factor at the bubble point pressure, bbl/STB
P = pressure, psia
Pb = bubble point pressure, psia

One of the earliest methods for estimating the value of Co was presented by Calhoun (1947) in the
form of a graphical correlation . However, his curve is reasonably well represented by the expression,

(6-60)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 60

where SGob is the effective oil specific gravity at the bubble point (includes solution gas), and is given by,

(6-61)

1.5

Bubble Point
OIL FORMATION VOLUME FACTOR (bbl/STB)

Compressibility Ignored

1.4 Compressibilit
y Included

1.3

1.2

1.1
Saturated Undersaturated

1
0 1000 2000 3000 4000 5000 6000

PRESSURE (psia)

Figure 6-6

Effect of Pressure on the Oil Formation Volume Factor (Bo) at Constant


Temperature

The undersaturation correlation developed by Vazquez and Beggs (1977) is based on almost 4,500
data points that covered a wide range of conditions, including pressures up to 9,515 psia (65.6 MPa). Their
equation for Co is given by,

(6-62)

where T = temperature, oF
G = gas gravity (relative to air = 1)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 61

API = API gravity of the oil, oAPI


Rs = solution gas/oil ratio, scf/STB
P = pressure, psia

Another correlation, with a somewhat different form, was proposed by Al-Marhoun (1992) based
on extensive data collected from all over the world. In this case, the relation used to compute Bo in the
undersaturated region is given by,

(6-63)

where the exponent m is defined by the relation,

(6-64)

Values of the constants a1 through a4 in Equation (6-64) are given in Table 6-11.

Petrosky and Farshad (1993) also developed a correlation for the effect of undersaturation using
the data they collected for Gulf of Mexico crude oils. They assume that Bo is given by Equation (6-59), but
with the isothermal compressibility of the oil computed using the relation,

(6-65)

Values of the constants b1 through b6 in Equation (6-65) are also given in Table 6-11 .

More recently, Kartoatmodjo and Schmidt (1994) proposed a new correlation, which they derived
using their data bank containing 2,545 measured observations. It is again based on Equation (6-59), with Co
computed from the expression,

(6-66)

Values of c1 through c6 in Equation (6-66) are also given in Table 6-11.

6.16 Density of the Gas and Oil Phases

Calculation of the density for single phase gas characterized only by its gravity (with acid gas
component mole fractions perhaps also being known) was discussed earlier in Sections 4.1 through 4.4 of
these Notes. The procedures outlined there are also applicable to the gas phase in a Black Oil model system
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 62

Table 6-11

Values of the Constants in Equations (6-64), (6-65), and (6-66)

Constant No. a b c
1 -0.0136680 x 10-3 1.705 x 10-7 6.8257 x 10-6
2 -0.0195682 x 10-6 0.69357 0.5002
3 0.02408026 0.1885 0.3613
4 -0.0926019 x 10-6 0.3272 0.76606
5 - 0.6729 0.35505
6 - -0.5906 -1.0

The density of the oil phase is computed from the simple material balance relation,

(6-67)

where ρo = density of the oil (lb/ft3, or kg/m3)


Bo = oil formation volume factor (bbl/STB, or m3/m3)
ρoo = density of the oil at stock tank conditions (lb/ft3, or kg/m3)
ρGo = density of the gas at standard conditions (lb/ft3, or kg/m3)
Rs = solution gas/oil ratio (scf/STB, or m3/m3)
α = volume conversion factor,
= 5.615 ft3/bbl [field units]
= 1.0 [SI units]

The density of the stock tank oil is computed from the simple relation,

(6-68)

where SGo = specific gravity of the oil

6.17 In Situ Flow Rates of Gas, Oil and Total Liquid

It is customary in the oil and gas industry to express throughput and production flow rates in terms
of the volumes that would be measured at the base (i.e. stock tank) conditions. However, it is the in situ
volume (i.e. actual flowing volume) for each phase that we must know for performing the flow calculations,
since that is what determines the fluid velocities.

For Black Oil models, the total throughput or production rate is always defined in terms of the flow
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 63

rate of stock tank oil. The overall gas flow rate is specified in terms of the produced gas-oil ratio, and the free
gas flow rate is determined by difference between that quantity and the solution gas-oil ratio. Similarly, the
overall water flow rate is expressed in terms of either a water-oil ratio, or a water cut. The water cut is simply
the fraction (usually expressed as a percent), by volume, of the total stock tank liquid flow that is water.

The following equations briefly summarize the relationships between the various parameters that are
commonly used in Black Oil models. Equations are given for both Field units and SI units; the corresponding
units for individual parameters and values of the constants in the equations are given in Table 6-12.

(i) Gas Flow Rate

(6-69)

where

(6-70)

(ii) Oil Flow Rate

(6-71)

(iii) Total Liquid Flow Rate (oil + water)

(6-72)

where QGo = gas flow rate at stock tank conditions


QG = gas flow rate at system pressure and temperature
Qoo = oil flow rate at stock tank conditions
Qo = oil flow rate at system pressure and temperature
QL = total liquid flow rate at system pressure and temperature
Rs = solution gas/oil ratio at system pressure and temperature
GOR = produced gas/oil ratio at stock tank conditions
WOR = produced water/oil ratio at stock tank conditions
WCUT = produced water cut at stock tank conditions
Bo = oil formation volume factor at system pressure and temperature
Bw = water formation volume factor at system pressure and temperature
Po = base (i.e. stock tank) pressure
To = base (i.e. stock tank) temperature
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 64

P = system pressure
T = system temperature
Z = gas compressibility factor at system pressure and temperature

Table 6-12

Consistent Units and Values of the Constants in Equations (6-69) through (6-72)

Field Units SI Units


QGo MMscf/day Sm3/day
Qoo STB/day Sm3/day
Bo, Bw bbl/STB m3/Sm3
GOR, Rs scf/STB Sm3/Sm3
WOR STB/STB Sm3/Sm3
WCUT STB/STB Sm3/Sm3
Po, P psia kPa
To, T o
F o
C
3 3
QG ft /sec m /sec
Qo, QL ft3/sec m3/sec
a1 10-6 1.0
a2 1.1574 x 10-5 1.1574 x 10-5
a3 460 273
a4 6.499 x 10-5 1.1574 x 10-5

7. Gas Viscosity

7.1 Introduction

The viscosity of a fluid is an important parameter in the calculation of irreversible pressure losses
resulting from what we traditionally refer to as friction effects. The determination of the various contributions
to overall pressure losses in general is dealt with in considerable detail elsewhere and will not be belaboured
here. For the moment, we simply note that the relative importance of the gas viscosity to flow calculations
depends on the nature of the system under consideration. In single phase pipelines and gas wells, it can be
a dominant factor in the calculated pressure drop. In multiphase oil-gas systems, it becomes a relatively
insignificant factor, particularly in oil wells, where friction effects are minimal. In gas-condensate systems,
whether or not it is important depends on the overall pressure and the range of flow rates under consideration.
In any case, there are enough situations in which gas viscosity is important that we must have suitable
methods for computing it under a wide variety of conditions.

Many correlations have been proposed for calculating the viscosity of the gas phase. This section
presents a number of both non-compositional and compositional methods that are considered by the author
to be among the most reliable. The former can be used for both compositional and non-compositional
systems.

7.2 Carr et al (1954) Correlation

The original correlation, as proposed by Carr et al (1954) is is a non-compositional graphical method,


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 65

but it does contain special correction terms to account for the presence of the non-hydrocarbon components,
N2, H2S, and CO2. It is also a corresponding states procedure, in that the first step is to determine µG1 (the
viscosity of the gas at 1.0 atm), from a plot of µG1 as a function of the gravity and the temperature of the gas.
A second plot was then used to determine the ratio µG/µG1 as a function of the pseudo-reduced pressure and
pseudo-reduced temperature of the gas. Since µG1 has already been determined, µG can thus be computed
directly.

Since this method inherently contains a correction for acid gas components, the Wichert and Aziz
(1971) correction, discussed earlier in Section 4.4, should not be used when estimating the values of the
pseudo-reduced pressure and temperature.

7.3 Dempsey (1965) / Standing (1977) Approximation for the Carr et al (1954) Correlation

As a graphical procedure, the Carr et al (1954) correlation is not suited for computer usage.
Dempsey (1965) presented an analytical version that has been used quite extensively. He used two complex
polynomials to fit the surfaces represented by the two plots in the original Carr et al (1954) paper.

Standing (1977) proposed that the following set of relationships be used in place of Dempsey’s
polynomial expression for the viscosity at atmospheric pressure,

(7-1)

where

(7-2)

(7-3)

(7-4)

(7-5)

and G = gas gravity (relative to air = 1)


YN2 = mole fraction of N2 in the gas
YCO2 = mole fraction of CO2 in the gas
YH2S = mole fraction of H2S in the gas

Values of the constants a1 through a4 and b1 through b6 in the above equations are given in Table 7-1.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 66

Table 7-1

Values of the Constants in Equations (7-2) through (7-5)

Constant No. a b
1 1.709 x 10-5 8.48 x 10-3
2 -2.062 x 10-6 9.59 x 10-3
3 8.188 x 10-3 9.08 x 10-3
4 -6.150 x 10-3 6.24 x 10-3
5 - 8.49 x 10-3
6 - 3.73 x 10-3

The ratio of the gas viscosity at the pressure and temperature of interest to the low pressure value is
determined from Dempsey’s equation for fitting the second plot,

(7-6)

where

(7-7)

and µ = gas viscosity at pressure and temperature of interest, cP


µ1 = gas viscosity at a pressure of one atmosphere and the temperature of interest, cP
Pp r = pseudo-reduced pressure [ = P/Pp c]
Tp r = pseudo-reduced temperature [ = ( T + 460)/Tp c]
P = pressure, psia
T = temperature, oF
Pp c = pseudo-critical pressure, psia
Tp c = pseudo-critical temperature, oR

Values of α i , j are given in Table 7-2.

Table 7-2

Values of the Constants α i , j in Equation (7-7)

i α i, 0 α i, 1 α i, 2 α i, 3

-2.462118 2.808609 -7.933857 x 10-1 8.393872 x 10-2


2.970547 -3.498033 1.396433 -1.864088 x 10-1
123
-2.862641 x 10-1 3.603730 x 10-1 -1.491449 x 10-1 2.033679 x 10-2
8.054205 x 10-3 -1.044324 x 10-2 4.410155 x 10-3 -6.095793 x 10-4
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 67

7.4 Dranchuk et al (1986) / Standing (1977) Approximation for the Carr et al (1954) Correlation

Somewhat later, Dranchuk et al (1986) published another analytical version which they claim to be
applicable over a wider range of conditions than the polynomial approximation proposed by Dempsey.

They retain the use of Equations (7-1) through (7-5) for computing the gas viscosity at atmospheric
pressure, but proposed a new procedure for computing the viscosity ratio term. This is somewhat more
complex than the Dempsey version in that the overall range is mapped into three different surfaces, with
comparable sets of polynomial equations for each. The revised procedure also includes a number of
provisions for ensuring continuity at the boundaries of the different regions.

Despite their common heritage, the two numerical versions often give rather different results and
effectively represent two different correlations.

7.5 Lee et al (1966) Correlation

The viscosity of the gas at the pressure and temperature of interest is given by,

(7-8)

where

(7-9)

(7-10)

(7-11)

and µG = gas viscosity, cP


ρG = gas density, lb/ft3
G = gas gravity
T = temperature, oF

7.6 Dean and Stiel (1965) Correlation

This is a fully compositional model, based on the corresponding states principle. While it can be
used for gases at any pressure, it has been found to be particularly good in the dense phase region. The
overall correlation and its associated mixing rules are expressed by the relations,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 68

(7-12)

(7-13)

(7-14)

(7-15)

(7-16)

(7-17)

(7-18)

(7-19)

For Tr m < 1.5

(7-20)

For Tr m > 1.5

(7-21)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 69

(7-22)

where

µG = gas viscosity, cP
µGo = low pressure gas viscosity, cP
ρG = gas density, lb/ft3
yi = mole fraction of component i
Mi = molecular weight of component i
R = gas law constant (i.e. 10.73 psi - ft3/lbmole oR )
Tci = critical temperature of component i, oR
Pci = critical pressure of component i, psia
Zci = critical compressibility factor for component i
Vci = critical molar volume of component i, ft3/lb-mole
T = temperature, oF

Figure 7-1 shows a comparison of observed and calculated values of the left-hand side of Equation
(7-12) plotted against the reduced density, ρrm. The agreement is seen to be excellent, even for relatively large
values of ρrm. Correlations of this type have also been used with reasonable success for light condensate
liquids, although this is not generally recommended.

Figure 7-1

Gas Viscosity Correlation of Dean and Stiel (1965)


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 70

7.7 Pedersen and Fredenslund (1986) Correlation

These authors presented a corresponding states correlation that was actually developed to predict the
viscosities of crude oils for which a full compositional analysis was available. They chose to use methane
as the reference fluid, partly because it is a simple molecule, it is a primary component of most naturally
produced hydrocarbon fluids, its properties are well known, and Hanley et al (1975) had earlier presented a
complex mathematical model of its viscosity behaviour. Because of the reference fluid choice, they claimed
that the overall procedure should also be applicable to gas and condensate liquids, although most of the data
they presented to support the method were for crude oils. The values predicted for gases by this correlation
are generally somewhat higher than those predicted by other available procedures, but they appear to be
reasonable.

Additional details of this correlation are presented in Section 12.3 of these Notes.

7.8 A Brief Comparison of Gas Viscosity Correlations

For comparison purposes, values of gas viscosity have been calculated at a number of temperatures
and two pressures (15 psia and 1,000 psia) which cover a typical range of pipeline operating conditions. The
composition of the gas, which has a gravity of 0.65, is given in Table 7-3. The correlations included in the
comparison are:

(i) Lee et al (1966)


(ii) Carr et al (1954), Dempsey (1965) / Standing (1977) version
(iii) Carr et al (1954), Dranchuk et al (1986) / Standing (1977) version
(iv) Dean and Stiel (1965)
(v) Jossi et al (1962)
(vi) Pedersen and Fredenslund (1986)

The Jossi et al (1962) correlation has not previously been discussed in these Notes. It is claimed
to be applicable to non-polar gases. It is similar in many ways to the Dean and Stiel correlation, and can also
be used for light condensate liquids.

The results of these calculations for 15 psia and 1,000 psia are summarized graphically in Figures
7-2(a) and 7-2(b) respectively. There are no measured values and this exercise has been performed only to
give a brief overview of how gas viscosity varies with pressure and temperature, and to provide a qualitative
comparison of the different methods. It is apparent that the different methods can give significantly different
results.

It can also be seen from these Figures that, unlike typical behaviour for a liquid, the viscosity of a
gas increases with increasing temperature. As the temperature increases, the energy level (and hence, the
velocity) of individual gas molecules increases. This causes proportionately more particle-particle collisions,
which manifests itself as an increase in the viscosity of the fluid. The viscosity-temperature relationship is
typically almost linear for a gas, which, as we shall see shortly, is also quite different from typical liquid
behaviour.

Comparison of Figures 7-2(a) and 7-2(b) shows that the viscosity also increases with increasing
pressure. Again, the effect of a pressure increase is to increase the number of particle-particle collisions by
decreasing the volume occupied by a given amount of gas.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 71

Table 7-3

Composition of Gas Used in Comparison of Calculated Viscosities

Component Mole Fraction


N2 .00074
CO2 .01301
C1 .89248
C2 .04864
C3 .02509
i-C4 .00553
n-C4 .00637
i-C5 .00292
n-C5 .00228
C6 .00198
C7+ .00096
C7+ Mole Wt. 124

0.016

Dempsey / Standing
0.015 Dranchuk et al / Standing
Lee et al
Dean & Stiel
CALCULATED VISCOSITY (cP)

0.014 Jossi et al
Pedersen & Fredenslund
0.013

0.012

0.011

0.010

0.009
Pressure = 15 psia

0.008
40 60 80 100 120 140 160 180 200 220 240 260

TEMPERATURE (°F)

Figure 7-2 (a)

Comparison of Calculated Gas Viscosities


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 72

0.016
Pressure = 1,000 psia
0.015

0.014
CALCULATED VISCOSITY (cP)

0.013

0.012

0.011
D em psey / S tan ding
D ranchu k e t a l / S tand in g
0.010
Lee et al
D ean & S tiel
0.009 Jossi et al
Pedersen & Fred en slun d

0.008
40 60 80 100 120 140 160 180 200 220 240 260

T EMPER AT U R E (°F)

Figure 7-2 (b)

Comparison of Calculated Gas Viscosities

8. Dead Oil Viscosity (Black Oil Models)

8.1 Introduction

Unless otherwise specified, single phase crude oil is generally assumed to be dead oil and its viscosity
is thus independent of any gas phase properties. However, as we shall shortly see, even in two/three phase
systems, methods for computing the viscosity of the saturated oil phase generally require the dead oil to be
known. Viscosity can have a very significant effect on both the pressure drop and heat transfer behaviour
of flowing liquids and gas-liquid mixtures. It is thus especially important to be able to calculate the dead oil
viscosity with reasonable accuracy. It is also a property that can vary widely in a given flow system, as it can
be quite sensitive to temperature, the amount of solution gas in the oil, and the basic composition of the oil
itself. Two oils that appear to be quite similar (i.e. comparable API gravities) can have quite different
viscosities at a given temperature because of, say, differing paraffin content.

A number of the more useful methods for calculating this important quantity are presented in this
Section.

8.1 General Relationships for the Effect of Temperature on Dead Oil Viscosity

A number of general relationships have been proposed for describing how dead oil viscosity changes
with temperature. One of the simpler methods is given by,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 73

(8-1)

where µdo = dead oil dynamic viscosity, cP


C, S = constants for a given oil
T = oil temperature, oF

If we take the logarithm of both sides of Equation (8-1), we obtain,

(8-2)

Thus, a plot of log10 µdo vs log10 (100/T) should result in a straight line on a log-log scale. The slope of
that line will be the parameter S, and the intercept is C (i.e. C is the value of the dead oil viscosity at 100 oF).
Figure 8-1 shows such a plot for three different crude oils from Kuwait.

This log-linear relationship is of course the basis for the frequently encountered practice of writing
Equation (8-1) as,

(8-1a)

In any case, if µdo is known at two temperatures, the two parameters can be calculated algebraically.
If µdo is known at three or more temperatures, one can estimate the values of the parameters either using the
method of least squares, or from a log-log plot of the data. Equation (8-1) can then be used to calculate µdo
at any other temperatures of interest. Values of the parameters C and S shown in Figure 8-1 were evaluated
from the measured data using the method of least squares.

A second, and somewhat more complex, relation was originally proposed by Walther (1931). It was
subsequently used by Wright (1969) as the basis for what is now generally known as the ASTM D-341
viscosity chart (1981), and is therefore also commonly known as the “ASTM equation”. It is is given by the
expression,

(8-3)

where

Z = νdo + 0.7
νdo = dead oil kinematic viscosity (cS)
A, B = constants for a given oil
T = oil temperature, oF
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 74

4.0

measured
3.5
C = 241.8, S = 3.194 14.79 °API
C = 33.53, S = 2.122
3.0
C = 5.073, S = 1.227
LOG10 µdo (cP)

2.5
24.62 °API
2.0

1.5

35.96 °API
1.0

0.5

0.0
-0.2 -0.1 0 0.1 0.2 0.3 0.4

LOG10 (100 / T )

Figure 8-1

Typical Plot Used for Evaluating the Parameters C and S in Equation (8-1)
(Data taken from Al-Besharah et al, (1989))

The kinematic viscosity, νdo is given by,

(8-4)

where ρo is the density of the oil at the temperature of interest, expressed in g/cm3.

When the density of the oil is not otherwise known, SGT, the specific gravity of the oil at the tem-
perature of interest can be estimated from Figure 8-2, based on the known value of SGo or API.

In any case, values of µdo or νdo that are known for two or more temperatures can be used to estimate
the parameters A and B, as described above for Equation (8-1). Equation (8-3) can then be used to estimate
νdo (and hence, µdo) at any temperature of interest.

Equations (8-3), as written above, is applicable only for νdo > 2.0 centistokes. An extension to that,
which permits values of νdo to be as low as 0.21 centistokes, has been proposed by Manning (1974), who
suggested that when νdo < 2.0, Z should be defined as,

(8-5)

which can also be arranged as,


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 75

(8-6)

where X = Z - 0.7

A third general expression that is often used is the Eyring (1936) equation, which is given by,

(8-7)

Again, the parameters A and B can be evaluated from data, just as for Equations (8-1) and (8-3),
and the absolute viscosity can be computed from the results of Equation (8-7) using Equation (8-4).

Whenever possible, the viscosities used to evaluate the equation constants should be measured at
temperatures close to the range of interest. The fitted equations can give significantly different results as one
moves away from the range covered by the measured values. An example of this can be seen in Figure 8-3,
which shows the predicted values based on three observations from a particular North Sea Reservoir. All
three equations predict essentially the same viscosity at temperatures above 100 oF. Below 100 oF, the
predictions begin to differ significantly. Note that as the singularity in Equation (8-1) (which occurs at T =
0 °F) is approached, the calculated viscosity begins to increase very rapidly.

When there are no measured data are available, it is still possible to make use of Equations (8-1),
(8-3) and (8-7) if one uses some other procedure to estimate the viscosity of the particular oil at two or more
temperatures. The estimated values can then be used in place of actual measured data to compute values for
the unknown constants in the equations. There are a number of methods, based on the API gravity of the oil,
which can be used for that purpose. A few of these are described below, as well as a number of different
procedures that are complete in themselves (i.e. “stand-alone” correlations for dead oil viscosity as a function
of temperature). As with the bubble point correlations, some of these are distinctly regional in their
derivation.

8.3 Beal (1946) Correlation

The Beal (1946) correlation relates dead oil viscosity to temperature and the API gravity of the oil.
This is a graphical correlation and it is shown in Figure 8-4 (the dashed lines represent extrapolations of the
original version). Because it did not extend to temperatures below 100 oF, its direct use as a prediction
method was somewhat limited. However, when no measured data are available, it can readily be used to
estimate the viscosity at two temperatures (e.g. 100 oF and 220 oF), which can then be used to compute the
constants for either of Equations (8-1) or (8-3).

When only a single measured value is available, the following procedure is recommended,

(i) Determine C and S in Equation (8-1) using values of viscosity estimated from Figure 8-4 at,
say, 100 °F and 220 °F. Denote the calculated values of C and S as C* and S*.

(ii) Substitute the measured viscosity and the corresponding temperature, along with S* into
Equation (8-1) and compute a new value of the parameter C, say C**.

C** and S* can then be used as the fitted values of C and S in the equation to estimate the viscosity of the
oil at other temperatures of interest.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 76

Figure 8-2

Effect of Temperature on Apparent Specific Gravity of Liquid Petroleum Fractions


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 77

40
Oil Gravity = 32.5 °API
35
Equation (2.8-1) with C = 4.907, S = 1.182
Equation (2.8-3) with A = 6.712, B = 2.479
30
Equation (2.8-5) with A = .05155, B = 1.446

VISCOSITY (cp)
Measured
25

20

15

10

0
0 20 40 60 80 100 120 140 160 180 200 220 240

TEMPERATURE (°F)

Figure 8-3

Comparison of Calculated Viscosities for Various General Equations


(constants evaluated using all three measured data points)

Figure 8-4

Beal (1946) Correlation for Dead Oil Viscosity


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 78

Values of C and S determined for three cases are given in Table 8-1, and these results are also shown
graphically in Figure 8-5.

As can be seen from Figure 8-5, in this case the results obtained using the Beal correlation only are
very similar to those obtained with Beal plus a single observation. This will not always be the case of course.
When three measured values are used, the constants are significantly different, but the greatest effect is seen
at the lower temperatures. Clearly, having a measured value at 50-60 °F would be highly desirable if the
actual temperatures of interest went down to that range.

Table 8-1

Fitted Values of C and S for Equation (8-1)

Basis C S

Beal only 7.141 2.121


Beal + 1 point 6.728 2.121
3 points 4.907 1.182

40
Oil Gravity = 32.5 °API
35
All 3 data points
Beal (1946) + 140 °F data point
30
Beal (1946) correlation
Measured
VISCOSITY (cp)

25

20

15

10

0
0 20 40 60 80 100 120 140 160 180 200 220 240

TEMPERATURE (°F)

Figure 8-5

Comparison of Calculated Viscosities Using Equation (8-1) With Parameters


Determined in Various Ways
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 79

8.4 Abbot et al (1970) Correlation

This correlation is in fact an analytical representation of a nomograph presented by the API (1966).
It actually consists of two equations, which are used to compute values of the kinematic viscosity at 100 °F
and 210 °F respectively. These two values can then be used to evaluate the parameters A and B in Equation
(8-3), (or, along with Figure 8-2, to evaluate the parameters C and S in Equation (8-1)), and thus, a
relationship for the dead oil viscosity at any temperature of interest.

The two equations are as follows:

(8-8)

(8-9)

and

(8-10)

(8-11)

where ν = kinematic viscosity, cst


K = Watson's K factor
API = oil gravity, oAPI

The constants a0 through a10 and b0 through b8 in Equations (8-8) through (8-11) are listed in Table 8-2.

8.5 Twu (1986) Correlation

This is another analytical representation of the API (1966) oil viscosity nomograph. It is claimed to
be a significant improvement over the correlation of Abbot et al (1970), as the latter has some singularities,
and its agreement with the API “data” (i.e. the chart) is better at low and medium viscosities than at higher
values. As with the Abbot et al (1970) procedure, Twu’s correlation consists of equations for computing the
kinematic viscosity at two temperatures (100 °F and 210 °F) , which can then be used to compute the values
of the parameters in any of Equations (8-1), (8-3), or (8-5).

Twu based his correlation on two standard Reference Fluids, and it is thus a form of corresponding
states model. The kinematic viscosity of a given reference fluid at a give temperature is computed using the
expression,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 80

Table 8-2

Values of Constants for Equations (8-8) through (8-11)

Number a b
0 4.39371 -0.463634
1 -1.94733 -0.166532
2 0.12769 5.13447 x 10-4
3 3.2629 x 10-4 -8.48995 x 10-3
4 -1.18246 x 10-2 8.0325 x 10-2
5 0.171617 1.24899
6 10.9943 0.19768
7 9.50663 x 10-2 26.786
8 -0.860218 -2.6296
9 50.3642 -
10 -4.78231 -

(8-12)

where ν(R) = kinematic viscosity of a reference fluid at a specific temperature, cst


TB = normal boiling point for the oil in question, oR

and C1 through C6 are constants that are defined for each reference fluid at each of two specific
temperatures. The values of these constants are given in Table 8-3 .

The kinematic viscosity of the fluid of interest at a given temperature is computed from the relation,

(8-13)

where K = Watson’s K factor for the oil of interest


ν (Ri) = kinematic viscosity of the ith reference fluid

As with the method of Abbot et al (1970), both the Watson’s K factor and the API gravity (or specific
gravity) must be known (or have been estimated) for the oil. These parameters can then be used with
Equation (2-4) to calculate TB.

The procedure for using the Twu (1986) correlation is as follows:

Step 1: Compute the value of TB for the oil of interest from Equation (2-4), using the known
value of the Watson’s K factor and the API gravity of the oil.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 81

Step 2: Compute a value of the reference fluid viscosity using Equation (8-12) with the
value of TB determined in Step 1, and values of the constants given in Table 8-3 for
Reference Fluid 1 at 100 oF. Denote this value as ν (R1) .

Step 3: Repeat Step 2, this time using the values of the constants given in Table 8-3 for
Reference Fluid 2 at 100 oF. Denote this value as ν (R2).

Step 4: Compute the kinematic viscosity for the oil of interest at 100 oF using Equation (8-
13).

Step 5: Repeat Steps 2 through 4 to compute a value of the kinematic viscosity of the oil at
210 oF.

Step 6: Use the values computed in Steps 4 and 5 to algebraically compute values for the
parameters in any of Equations (8-1), (8-3), or (8-5), as appropriate. Note that if the
viscosities computed by the Twu correlation are to be used to evaluate the
parameters in Equation (8-1), they must first be converted to values of the absolute
viscosity, using Equation (8-4) .

The Twu (1986) correlation is recommended for general use with all crude oil systems for which dead
oil viscosity data are not known. In the absence of a known value for the Watson’s K factor, it is suggested
that one use a general default value of 11.9 . Note however that the accuracy of the calculated viscosity may
be reduced if the actual Watson’s K factor is not known.

Table 8-3

Values of the Constants for Equation (8-12)

100 °F 210 °F
Constant Reference Fluid 1 Reference Fluid 2 Reference Fluid 1 Reference Fluid 2
C1 -234.362 -509.138 -258.131 -243.268
C2 7,831.91 22,615.7 11,124.9 11,787.9
C3 38.3044 80.4388 40.4918 37.3558
C4 -0.0430100 -0.0763095 -0.0356499 -0.0283454
C5 0.55514 x 10-8 0.819108 x 10-8 0.281192 x 10-8 0.162873 x 10-8
C6 0.841988 x 10-18 0.244956 x 10-18 0.468001 x 10-18 0.406341 x 10-18

8.6 Beggs and Robinson (1975) Correlation

This is a stand-alone correlation that was developed using a data base that was compiled primarily
from North American sources. The equations are as follows,

(8-14)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 82

where

(8-15)

(8-16)

It is worth noting that all of the data that were used to develop this correlation were measured at
temperatures of 70 °F or greater. At lower temperatures, it should be considered to be unreliable, unless one
has measured data which demonstrate otherwise, as it tends to predict a very rapid increase in viscosity as
the temperature drops below that value. For example, at 30 °F , this method predicts a viscosity of almost
11,000 cp for a 35 oAPI oil, compared with predicted values in the order of 25-40 cp for most other methods.
On the other hand, it is characteristic of some heavy and/or waxy crudes for the viscosity to start increasing
very rapidly as the pour point is approached. The author has been involved in a study of one such oil for
which this correlation matched the observed data very well, unlike most of the other viscosity correlations
that were investigated.

8.7 Glaso/ (1980) Correlation

Glaso/ proposed the following expression for dead oil viscosity, based on viscosity data for a number
of crude oils from North Sea reservoirs,

(8-17)

where
(8-18)

and T = temperature, °F
API = oil gravity, oAPI
C = constant (3.131 x 1010 in the original paper)

It is suggested that the constant C could in fact be viewed as an adjustable parameter for calibration
purposes, as described earlier with the bubble point correlation. A single observation would be all that was
required to determine an appropriate value for matching purposes.

8.8 Ng and Egbogah (1983) Correlation

These authors essentially present a revised version of the Beggs and Robinson (1975) correlation.
They concluded that the general form of that correlation was reasonable, but re-evaluated the constants using
a different data base which incorporated data for a greater temperature range. Their correlation is given by,

(8-19)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 83

where µdo = dynamic viscosity, cP


API = oil gravity, oAPI
T = temperature, oF

They also proposed a second version of their correlation that takes into account the pour point temperature
of the oil. However, the improvement in accuracy was considered to be only marginal, and did not justify
the extra complexity.

8.9 Amin and Maddox (1980) Correlation

This correlation is used to compute the values of A and B in Equation (8-5) directly. It requires either
the mean boiling point or the Watson’s K factor to be known (Equation (2-4) can be used to compute one
from the other), and one must also know either the specific or API gravity of the crude oil . The equations
are as follows,

(8-20)

(8-21)

where Tb = mean boiling temperature, oF


K = Watson’s K factor

8.10 Beg et al (1988) Correlation

This is very similar to the Amin and Maddox (1980) correlation in that it computes the parameters
for Equation (8-5) directly from the mean boiling point and API gravity of the oil. Their equations are,

(8-22)

(8-23)

where Tb = mean boiling point of the oil, °F


API = gravity of the oil, oAPI

8.11 Puttagunta et al (1992) Correlation

These authors presented a correlation for the kinematic viscosity of a dead oil for cases in which a
single measured value is known. Their correlation is expressed by the following equations,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 84

(8-24)

(8-25)

(8-26)

where νdo = kinematic viscosity at any temperature T, cSt


νdo* = kinematic viscosity at 37.78 °C (i.e. 100 oF), cSt
T = temperature, °C

One must thus know, or be able to estimate, the kinematic viscosity of the dead oil at a fixed reference
temperature of 37.78 °C (i.e. 100 °F ). In fact, it is very common to have such data. However, in cases
where a measured value is available at some other temperature, the authors suggest that one can perform trial
and error calculations to determine an appropriate value. One would thus set νdo and T equal to the known
values and then try different values of νdo* until the equations were satisfied.

8.12 Kartoatmodjo and Schmidt (1994) Correlation

In this correlation, the dead oil viscosity is computed using equations that are the same form as those
proposed by Glasø (1980). The constants have all been re-evaluated, however, using an expanded data base.
The revised relationships are as follows:

(8-27)

(8-28)

where API = gravity of the oil, oAPI


T = temperature, °F

8.13 Brief Comparison of Dead Oil Viscosity Correlations

Figure 8-6(a) shows a comparison of calculated dead oil viscosity values and measured data for a
wide range of API gravities at a temperature of 104 °F (40 °C), while Figure 8-6(b) shows a similar
comparison for data measured at 212 °F (100 °C). The general behaviour is similar for all of the correlations,
but significant differences are observed in the absolute values predicted. On the other hand, there is also wide
scatter among the data for any given API gravity oil. It is worth noting that most of the major oil producing
areas in the world are represented in the data that were used in these plots. This once again underscores the
value of having even one or two observations to use as a guide in selecting the most appropriate correlation
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 85

in a given case.

Figure 8-7(a) shows the measured dead oil viscosity behaviour of a 25.42 oAPI ( K = 11.82) oil as
a function of temperature compared with the behaviour predicted by the various generalized methods. In each
case, the measured value at 40 °C was used to “tune” the procedure. In general, the agreement is reasonable,
although the procedure of Amin and Maddox (1980) predicts rather low values at higher temperatures.

Finally, Figure 8-7(b) shows a similar comparison of the same data with values predicted by the
various “stand-alone” correlations. The curved designated as “Beal” was generated using an analytical
approximation of his original graphical correlation that was presented by Sutton and Farshad (1990). Again,
in each case, the measured value at 40 °C was used to “tune” the procedure

As previously noted, these should not be viewed as general results, since the data shown in the plots
have been computed for an oil having specific properties. It is clear from the data scatter in Figures 8-6(a)
and 8-6(b) that the API gravity alone is not sufficient information to define the viscosity at a given
temperature. On the other hand, the behaviour of dead oil viscosity with temperature change shown by the
data in Figures 8-7(a) and 8-7(b) is qualitatively, but not necessarily quantitatively, typical for an oil. That
is, it will follow a smooth curve. The observed differences between the methods in the latter Figures is due
to differences in the predicted temperature sensitivity, since all procedures were forced to match the data at
the same point. Since different oils will also exhibit different temperature sensitivities, it can be quite useful
to have a number of different procedures from which to choose.

1000
Beal (1946)
Beggs and Robinson (1975)
Ng and Egbogah (1983)
Glasø (1980)
100 Measured
VISCOSITY (cp)

10

0.1
10 20 30 40 50 60 70

OIL GRAVITY (°API)

Figure 8-6 (a)

Comparison of Predicted vs. Measured Dead Oil Viscosity at 104 °F


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 86

1000
Beal (1946)
Beggs and Robinson (1975)
Ng and Egbogah (1983)
100 Glasø (1980)
Measured
VISCOSITY (cp)

10

0.1
0 10 20 30 40 50 60 70

OIL GRAVITY (°API)

Figure 8-6 (b)

Comparison of Predicted vs. Measured Dead Oil Viscosity at 212 °F

100
ASTM - Twu (1986)
General - Beal (1946)
ASTM - Abbot et al (1970)
Eyring - Amin & Maddox (1970)
Eyring - Beg et al (1988)
10
VISCOSITY (cp)

Measured

0.1
20 40 60 80 100 120 140 160 180 200

TEMPERATURE (°C)

Figure 8-7 (a)

Comparison of Predicted vs. Measured Dead Oil Viscosity for a 25.42 oAPI Oil
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 87

100
Beal (1946)
Beggs & Robinson (1975)
Ng & Egbogah (1983)
Glasø (1980)
Puttagunta et al (1992)
10 Measured
VISCOSITY (cp)

0.1
20 40 60 80 100 120 140 160 180 200

TEMPERATURE (°C)

Figure 8-7 (b)

Comparison of Predicted vs. Measured Dead Oil Viscosity for a 25.42 oAPI Oil

9. Saturated Oil Viscosity

9.1 Introduction

It is always observed in gas/oil systems that, at any given temperature the viscosity of the saturated
oil is less than viscosity of the dead oil value at pressures greater than one atmosphere. This is due to the fact
that dissolved gas tends to reduce the average molecular weight and density of the liquid, which, in turn, leads
to a lower viscosity. It is thus not surprising to find that almost all correlations for live oil viscosity show a
strong dependence on the solution gas/oil ratio. Several methods for computing the viscosity of saturated oil
are presented below.

9.2 Chew and Connally (1959) Correlation

The correlation of Chew and Connally has the general form,

(9-1)

where µ = viscosity of the saturated oil, cp


µdo = viscosity of the dead oil, cp

The constants A and B are functions of the solution gas/oil ratio, Rs and are computed from the following
expressions:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 88

(9-2)

(9-3)

where Rs = solution gas/oil ratio, scf/STB

This correlation is recommended for general use in combination with any of the dead oil viscosity
correlations discussed in the previous sections of these Notes.

9.3 Beggs and Robinson (1975) Correlation

The Beggs and Robinson (1975) correlation for saturated oil viscosity is also based on the use of
Equation (9-1). However, the expressions for evaluating A and B are somewhat different, as follows:

(9-4)

(9-5)

This correlation is specifically recommended for use with the Beggs and Robinson (1975) dead oil
viscosity correlation, but there is no particular reason why it should not be used with any other dead oil
viscosity correlation.

Figure 9-1 shows the values of A and B for both the Chew and Connally (1959) and the Beggs and
Robinson (1975) correlations. It is clear that somewhat different values of the saturated oil viscosity will be
obtained for a given dead oil viscosity. Thus, calibrating a dead oil viscosity correlation with available data
does not in itself ensure that the saturated oil viscosity relationship will also be optimal. As a calibration
procedure for Equations (9-2) through (9-5), it is suggested that a procedure similar to that described in
Section 6.14 be followed. That is, replace the parameter Rs with, say, αRs. The tuning parameter α will thus
have a nominal value of 1.0, but could be adjusted if empirical data were available for saturated oil viscosity
at known solution gas/oil ratios.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 89

1.0
Chew & Connally (1959)
0.9 Beggs and Robinson (1975)

0.8

0.7
Parameters A and B

0.6 B
B
0.5

0.4
A
A
0.3

0.2

0.1

0.0
0 200 400 600 800 1000 1200 1400 1600 1800 2000

SOLUTION GAS/OIL RATIO (scf/STB)

Figure 9-1

Comparison of Saturated Oil Viscosity Correlation Parameters

9.4 Khan et al (1987) Correlation

The correlation of Khan et al (1987) is of significantly different form than the two previous methods.
The first step is to compute the viscosity at the bubble point pressure, Pb , which corresponds to a given
temperature T from the relations,

(9-6)

(9-7)

where µob = viscosity of the oil at temperature T and pressure corresponding to the bubble point
where Rs = GOR, cp
Rs = solution gas/oil ratio, scf/STB
G = gas gravity (relative to air = 1)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 90

GOR = produced gas/oil ratio, scf/STB


SGo = specific gravity of the stock tank oil
T = temperature, °F

The viscosity at the pressure of interest is then computed from the relation,

(9-8)

where µ = saturated oil viscosity at pressure and temperature of interest, cp


P = pressure, psia
Pb = bubble point pressure for Rs = GOR and temperature T, psia

9.5 Kartoatmodjo and Schmidt (1994) Correlation

According to this correlation, the viscosity of a saturated oil is computed using the polynomial
expression,
(9-9)

where

(9-10)

(9-11)

(9-12)

(9-13)

and µdo = dead oil viscosity at the temperature of interest, cp


Rs = solution gas oil ratio at the pressure and temperature of interest, scf/STB

9.6 A Brief Comparison of Some Saturated Oil Viscosity Correlations

Figure 9-2 shows a comparison between the saturated viscosities predicted at 175 °F (79.4 °C) by
some of the correlations discussed above with some corresponding measured data for the fluids from a North
Sea reservoir. This is a 32.5 oAPI crude oil with a relatively low GOR of about 254 scf/STB (45.0 Sm3/Sm3).
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 91

Because of the low GOR, the viscosity only decreases from a dead oil value of 2.535 cp to 1.035 cp at the
measured bubble point pressure of 1,082 psia (7,460 kPa). With higher GOR oils, the viscosity reduction can
be very much larger.

4.0
Chew and Connally (1959)
Beggs and Robinson (1975)
3.5
Khan et al (1979)
Kartoatmodjo and Schmidt (1994)
3.0 Measured
VISCOSITY (cp)

2.5

2.0

1.5

1.0

0.5

0.0
0 100 200 300 400 500 600 700 800 900 1000 1100

PRESSURE (psia)

Figure 9-2

Comparison of Predicted vs. Measured Saturated Oil Viscosities at 175 °F


for a 32.5 oAPI North Sea Crude Oil-Gas System (GOR = 254 scf/STB)

10. Undersaturated Oil Viscosity

10.1 Introduction

For a given temperature, an oil is said to be undersaturated at any pressure above the bubble point
pressure. Increasing the pressure would force more gas to go into solution if there was any, but above the
bubble point pressure, there is no more free gas. With no more gas going into solution above the bubble
point, the viscosity of the oil actually begins to increase with increasing pressure due to the compressibility
of the oil. Since liquid compressibility is typically small, the effect of pressure on viscosity is much smaller
above the bubble point than below.

A number of correlations have been proposed for computing the viscosity of undersaturated oils, and
a few of these are described below. All of these procedures assume that the bubble point pressure PB is
known at the temperature of interest, as well as µob, the saturated oil viscosity corresponding to the bubble
point pressure. PB can be evaluated using any of the methods discussed earlier in Section 6, while µob can be
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 92

estimated using any of the methods discussed under Sections 9.

10.2 Beal (1946) Correlation

The Beal (1946) correlation for the viscosity of an undersaturated oil at a given temperature is given
by the equation,

(10-1)

where µo = viscosity of undersaturated oil, cp


µob = viscosity of saturated oil at the bubble point pressure, cp
P = pressure, psia
PB = bubble point pressure at the temperature of interest, psia

10.3 Vazquez and Beggs (1977) Correlation

The effect of pressure on the viscosity of an undersaturated oil can be estimated from the saturated
oil viscosity using a correction proposed by Vasquez and Beggs (1977). This is given by the relation,

(10-2)

where

(10-3)

and µo = viscosity of the undersaturated oil, cp


µob = viscosity of the saturated oil at the bubble point, cp
P = pressure of interest, psia
PB = bubble point pressure, psia

Values of the constants C1 through C4 are given in Table 9-1.

Table 10-1

Values of the constants in Equation (10-3)

Parameter Value
C1 2.6
C2 1.187
C3 -11.513
C4 -8.98 x 10-5
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 93

10.4 Khan et al (1987) Correlation

This correlation is given by the simple relation,

(10-4)

where µo = viscosity of undersaturated oil, cp


µob = viscosity of saturated oil at the bubble point pressure, cp
P = pressure, psia
PB = bubble point pressure at the temperature of interest, psia

10.5 Abdul-Majeed et al (1990) Correlation

These authors used a data base derived from 254 PVT tests on 41 bottom-hole oil samples, primarily
representing reservoirs in North Africa and the Middle East, to develop their correlation. It is given by the
following equations,

(10-5)

(10-6)

(10-7)

where µo = viscosity of undersaturated oil, cp


µob = viscosity of saturated oil at the bubble point pressure, cp
P = pressure, psia
PB = bubble point pressure at the temperature of interest, psia
API = gravity of the oil, oAPI
Rs = solution gas/oil ratio, scf/STB

10.6 Kartoatmodjo and Schmidt (1994) Correlation

This correlation follows the same general form as Equation (10-1) for the Beal (1946) correlation,
but the constants were re-evaluated using the authors’ expanded data base. The revised expression is as
follows:

(10-8)

where µo = viscosity of undersaturated oil, cp


µob = viscosity of saturated oil at the bubble point pressure, cp
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 94

P = pressure, psia
PB = bubble point pressure at the temperature of interest, psia

10.7 A Brief Comparison of Undersaturated Oil Viscosity Correlations

Figure 10-1 shows the predicted undersaturated viscosity over a range of pressures computed using
the various correlations compared to some measured data for an 18 oAPI oil at 100 °F (37.8 °C). The
measured bubble point pressure for this system (a Mid-East crude) at that temperature is 2,450 psia (16.892
MPa) and Rs = 218 scf/STB (38.83 Sm3/Sm3). In this particular case, most of the correlations over-estimate
the effect of pressure, and again, we see substantial differences between the various predictions. This is a
rather heavy oil, however, and heavy oils tend to have low compressibilities. Lighter oils with higher GOR’s
will generally tend to show a somewhat stronger pressure effect.

From a practical perspective, unless P/PB > 1.1, which is relatively unlikely in most pipeline
applications, the magnitude of the pressure effect is small, and the viscosity change will typically be in the
order of 1 - 2 % or less. The correction can thus be ignored in most practical cases for flow in pipelines and
wells. It may, however, be significant in reservoir modelling or other cases where higher pressures occur,
especially if laminar flow is expected.

36
Beal (1946)
34 Vazquez and Beggs (1977)
Khan et al (1987)
32 Abdul-Majeed et al (1990)
Kartoatmodjo and Schmidt (1994)
30 Measured
VISCOSITY (cp)

28

26

24

22

20

18
2000 3000 4000 5000 6000 7000

PRESSURE (psia)

Figure 10-1

Comparison of Predicted vs. Measured Viscosity for an Undersaturated


Oil (oil gravity = 18 oAPI, Rs = 218 scf/STB, T = 100 °F, PB = 2,450 psia)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 95

11. Viscosity Of Blended Hydrocarbon Liquids

It is very common for two or more hydrocarbon liquid streams to be blended together at some point
in a pipeline system. This may occur in a gathering system, where the mixing occurs at the junction with a
lateral flowline, say, from a battery or some other facility. In fact, it is not unusual to have 20 - 30 or more
such junctions in a moderately long crude oil pipeline. The oils may be similar nature with respect to their
viscosity-temperature behaviour, or they may be quite different. For example, a heavy oil is often mixed with
a condensate liquid or some other lighter oil to reduce its viscosity and thus, to reduce the pumping costs for
the pipeline.

In any case, if we are to perform any flow calculations for the resulting blended stream, we must be
able to compute its viscosity under any temperature and pressure conditions that might be encountered.
Several procedures for doing this have been discussed in some detail by Gregory (1985) in Technical Note
No. 6. Procedures have also been proposed by Twu and Bulls (1981), Al-Besharah et al (1989a, 1989b, and
1989c), and Anhorn and Badakhshan (1994).

12. Liquid Viscosity (Compositional Methods)

12.1 Introduction

The most frequently encountered compositional liquids tend to be condensate fluids, although it is
becoming more common to obtain a full compositional analysis for crude oils. Because condensate liquids
are generally much lighter than crude oils, both the quantity and the composition of the liquid in a multiphase
system can change significantly with temperature and/or pressure. Normal “black oil” model behaviour can
not, therefore, be assumed to apply, and considerably more complex compositional models are required.

On the other hand, using a compositional model for a crude oil, even with a very detailed analysis,
does not always yield greater accuracy than the simpler black oil models. More often, a large fraction of the
composition is a simply a lumped residual fraction (e.g. C7+, C10+, etc.) and the characterization of one or more
pseudo-components becomes a significant factor in the predicted viscosity. Compositional methods are also
very difficult to tune to match existing data, compared to black oil models.

In any case, several of the more reliable compositional procedures are discussed below.

12.2 Dean and Stiel (1965) Correlation

This method was described in detail earlier in these Notes in Section 7.5 as a recommended procedure
for computing the viscosity of gases. It can, however, also be used with relatively light condensate fluids,
and it has been shown to give reasonable accuracy throughout the dense phase region, including the
conditions where the fluid tends to behave as a liquid. In general, however, it is not recommended for use
with condensate liquids as there are better methods available, some of which are discussed below.

This correlation should never be used for compositional crude oils as it will generally predict
viscosities that are much too low.

12.3 Van Velzen et al (1972) / Letsou and Stiel (1973) Correlation

This procedure is actually a composite method based on a combination of two correlations. Which
one to use in a given case depends on the value of the reduced temperature. These correlations are functions
of temperature and composition only, since, as we have seen in Section 10 of these Notes, pressure has a
very small direct effect on liquid viscosity. In any case, for a two phase system, there is a pressure effect that
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 96

is implicit in the equilibrium composition of the liquid.

The first step is to compute the viscosity of each of the individual components in the mixture as
follows, based on the reduced temperature for that component (i.e. Tri = T / Tci):

(i) For Tri < 0.76, use the correlation of Van Velzen et al (1972),

(12-1)

where µi = viscosity of component i, cP


T = temperature, °F
Bi, ti = parameters that depend on the molecular structure of component i

(ii) For 0.76 < Tri < 0.98, use the correlation of Letsou and Stiel (1973)

(12-2)

where,

(12-3)

(12-4)

(12-5)

(12-6)

and Tci = critical temperature of component i, oR


Pci = critical pressure of component i, psia
Mi = molecular weight of component i
αi = acentric factor of component i

Values of the parameters Bi and ti are tabulated in the book by Reid et al (1977) for most components
likely to be encountered in gas-condensate systems.

The final step is to estimate the viscosity of the liquid mixture from the viscosities µi and mole
fractions xi of the individual components using the relation,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 97

(12-7)

12.4 Pedersen and Fredenslund (1986) Correlation

This correlation is actually a revised form of a corresponding states model presented earlier by
Pedersen et al (1984). It is a very complex model, and the interested reader should consult the original
references, as the equations will not be reproduced here. As a reference substance, the authors chose to use
methane, which was also used in another procedure proposed by Ely and Hanley (1981). The latter authors
presented a 32 constant equation of state, specific to methane, that is also used as the basis of the Pedersen
and Fredenslund correlation. An advantage of this approach is that the method is not limited to the liquid
state, and is in fact claimed by the authors to be applicable to the gas phase as well. On the basis of
comparisons using over 100 viscosity data points for crude oils (primarily from North Sea reservoirs), the
proposed correlation showed an average error of less than 3% and an average absolute error of about 5%.

Experience with this procedure is mixed to date, and it has been observed to underestimate the
viscosity of medium to heavy oils. However, it appears to be very powerful and certainly is representative
of the direction in which viscosity models should be developed. At this time however, it is recommended that
this model be used only in cases where measured data are available for comparison purposes (even if only
a few points) to ensure that the proper order of magnitude is being predicted. As always with compositional
models, a major area of uncertainty is how to handle pseudo-components. If the characterization is good and
the compositional method is robust, one need not worry about data range limits, as for wholly empirical
correlations. However, one must also remember that for crude oils, the pseudo-component (i.e. arbitrarily
assigned) fraction of the composition is typically about 30-60%. Furthermore, the characterization that is
optimal for phase behaviour calculations may not necessarily be optimal for viscosity calculations.

12.5 A Brief Comparison of Compositional Oil Viscosity Correlations

Figure 12-1 shows a comparison between the viscosity vs pressure behaviour computed by the
Pedersen and Fredenslund (1986) correlation and some measured data for a North Sea reservoir at 175 °F .
The composition of the reservoir fluid includes a C7+ residual fraction that represented about 55% (mole
basis) of the total fluid. This was characterized as six pseudo-components, using HYPOS, and the resulting
composition was then tuned using “Shift Factors” by attempting to match known bubble point pressures at
140 °F , 175 °F , and 200 °F . Table 12-1 shows the results of those calculations, compared to the measured
data. It is clear from those data that the overall agreement is excellent from a PVT perspective.

In Figure 12-1, the qualitative agreement is seen to be very good, in that the compositional model
predicts both the viscosity decrease with increasing pressure below the bubble point and the increase with
increasing pressure above the bubble point. Quantitatively, the agreement is not as good. However,
considering that more than 50% of the composition consists of an undefined residual fraction, it is not bad.
At least in this case, it is also consistent with conservative design practice; calculated pressure losses based
on these viscosities would be somewhat higher than might actually be expected.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 98

4 .0
P ed ers en a n d F red en s lu n d (1 9 8 6 )
3 .5 B lac k O il m od ellin g
M eas u red

VISCOSITY (cp) 3 .0

2 .5

2 .0

1 .5

1 .0

0 .5

0 .0
0 1000 2000 3000 4000 5000

P R E S S U R E ( p s ia )

Figure 12-1

Comparison of Predicted vs. Measured Viscosity for a North Sea Crude Oil

Table 12-1

Comparison of Dead Oil Viscosity Predictions by the Pedersen and Fredenslund


Correlation With Measured Values for a North Sea Crude Oil

Property Measured Calculated % Error

PB at 140 °F (psia) 1004 977 -2.7


PB at 175 °F (psia) 1082 1082 -0.0
PB at 200 °F (psia) 1137 1155 1.6
Oil Gravity (oAPI) 32.5 32.38 -0.3
GOR (scf/STB) 254 26.1 2.8

For comparison, Figure 12-1 also shows the results of some Black Oil viscosity model calculations.
In this case, the dead oil viscosity was computed using Equation (8-1), with C and S having been evaluated
from data measured at 140 °F, 175 °F, and 200 °F. Live oil viscosity was computed using the Beggs and
Robinson (1975) correlation, as expressed by Equations (9-4) and (9-5), and calibrated using the known
viscosity at the 175 °F bubble point. The undersaturated viscosity was computed using the Beal (1946)
correlation, as expressed by Equation (10-1). Clearly, the agreement is very good indeed. This illustrates
the value of being able to make use of known data for tuning the Black Oil models. Fully compositional
models are not nearly so easy to adjust. We could, for example, base the tuning of the psuedo-component
characterization on matching the viscosities, but that might well reduce the accuracy of the PVT calculations.

The other compositional correlations that were discussed are intended primarily for condensate-type
liquids (i.e. light, low gravity), and should not be used with crude oil systems. They tend to significantly
underestimate crude oil viscosities, which can lead to very optimistic estimates of pressure losses in flowing
systems.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 99

13. Viscosity Of Oil-Water Emulsions

An emulsion can is a stable dispersion of one liquid phase in another with which it is immiscible.
Emulsions are most commonly encountered in pipelines and flowlines because of simultaneous production
of oil and water from a well. This could result from water coning because of high production rates, or as a
result of a waterflood to enhance oil recovery. In any case, the water cut (i.e. fraction of total liquid that is
water) tends to increase over the producing life of a reservoir, and may reach as high as 90% - 95% before
the well is abandoned.

The rheology of emulsions can be extremely complex, and a detailed treatment of that subject is
beyond the scope of these Notes. The effective viscosity of an emulsion generally depends on the relative
amounts of the hydrocarbon liquid and water, on the properties of the hydrocarbon liquid, and to a lesser
extent, on the properties of the water phase. In a water-in-oil emulsion (i.e. the oil is the continuous phase),
it is higher than the viscosity of the oil itself. In an oil-in-water emulsion (i.e. the water is the continuous
phase), the viscosity will always be higher than that of the water phase alone.

At some particular water cut, an emulsion will typically switch very abruptly from a water-in-oil to
an oil-in-water dispersion. This is known as the inversion point and it typically occurs at water cuts of 60%
to 80% although values outside this range are not uncommon. The rheological behaviour of the emulsion is
normally very different from one side of the inversion point to the other.

The simplest generalized model is based on the concept of a viscosity multiplier, which is defined
as,

(13-1)

where µe = effective viscosity of the emulsion, cp


µo = viscosity of the oil, cp
Kemul = oil viscosity multiplier

Probably the best known method for estimating Kemul is the graphical correlation of Woelflin (1942)
which simply relates this parameter to the water cut. Woelflin noted that, as the inversion point is
approached, the viscosity of the emulsion can be 40 times or more the viscosity of the oil alone.

More recently, in Chapter 19 of the Petroleum Engineering Handbook, Smith and Arnold (1987)
recommended the use of the simple quadratic relationship shown below.

(13-2)

where Cw = water cut, expressed as a decimal fraction

While Equation (13-2) appears to be a simple empirical relationship, it actually has a rather illustrious
history. It dates back to the early 1930's and in fact has a much broader applicability than is readily apparent.
As written above, it was presented by Guth and Simha (1936), who based it on an earlier and purely
theoretical equation that was derived by Albert Einstein! This pedigree, and the actual versatility of the
equation, are discussed in greater detail by Gregory (1990) in Technical Note No. 11.

Figure 13-1 shows a comparison between the Woelflin (1942) and the Guth and Simha (1936)
correlations. As can be seen from this Figure, the two correlations give virtually identical values of the
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 100

multiplier for Cw < 0.4, but diverge rapidly at higher water cuts. As is often the case with hydrocarbon fluid
systems, neither method gives consistently better results than the other.

For example, Figure 13-2 shows the comparison of emulsion viscosities calculated using both
methods with values measured for a 17.5 oAPI oil at 10 °C. Predicted values are in good agreement for both
correlations at the lower water fractions. For a water fraction of 0.5 and greater, the Woelflin correlation is
clearly in better agreement with the data.

Figure 13-3 shows a similar comparison for the same oil-water emulsion, but this time the
temperature is 80 °C.. Once again, both methods are in good agreement with the data at the lower water
fractions, but now the Guth and Simha correlation is clearly superior at higher values.

For some emulsions, there is a serious problem caused by the basic form of both of these correlations.
Since the multiplier is determined solely from the water fractions, it is inherently assumed that the effect of
temperature on the viscosity of the emulsion is adequately taken into account by the effect of temperature on
the viscosity of the oil. The same value of the multiplier is assumed to apply at all temperatures for a fixed
water cut. The complexity of emulsion behaviour is such that this is not always the case, and the multiplier
itself can also be a function of temperature in a way that is independent of the temperature effect on the oil
alone.

50

45 W o elf lin (1 9 4 2 )
G u th a n d S im h a (1 9 3 6 )
40
OIL VISCOSITY MULTIPLIER

35

30

25

20

15

10

0
0 10 20 30 40 50 60 70 80

W AT E R C U T ( % )

Figure 13-1

Comparison of the Woelflin (1942) and Guth and Simha (1936)


Oil-Water Emulsion Viscosity Correlations

There is little or no technology available to predict the conditions under which a stable emulsion will
occur. While a relatively high water fraction is normally required, that in itself is by no means a sufficient
condition. The stability of an emulsion is strongly influenced by surface tension, and even trace amounts of
some compounds, either on their own, or in combination with other trace compounds, can strongly affect
whether or not the system will form a stable emulsion. Because the effect on viscosity, and therefore on
pressure losses and/or pumping costs, can be very dramatic, it is generally a major concern for anyone
designing pipeline facilities for a high water cut system. Fortunately, however, it is also generally true that
there are many more people worrying about getting stable emulsions than are actually experiencing them.

In summary, there really is no substitute for a laboratory investigation in cases where emulsions
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 101

represent a potential concern. Such studies must always be carried out with samples of the actual produced
liquids to ensure that the results will be reliable.

24 Woelflin (1942)
Guth and Simha (1936)
Measured
20
VISCOSITY /1000 (cp)

16

12

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

WATER FRACTION

Figure 13-2

Comparison of Predicted vs. Measured Emulsion Viscosities


at 10 °C for a 17.5 oAPI Crude Oil

320
Woelflin (1942)
280 Guth and Simha (1936)
Measured
240
VISCOSITY (cP)

200

160

120

80

40

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
WATER FRACTION

Figure 13-3

Comparison of Predicted vs. Measured Emulsion Viscosities


at 80 °C for a 17.5 oAPI Crude Oil
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 102

14. Gas Phase Enthalpy

14.1 Introduction

Rigorous algorithms that are used for simultaneously calculating pressure and temperature profiles
in a pipeline and/or well require the specific enthalpy of each phase at any pressure and temperature. We thus
require procedures for computing these thermodynamic parameters with reasonable accuracy.

In particular, for the gas phase, knowing the enthalpy as a function of both pressure and temperature
allows us to account for the Joule-Thompson effect implicitly (i.e. expansion cooling), without making any
simplifying assumptions (e.g. constant Joule-Thompson coefficient), when performing our flow calculations.
In this Section, we look at how to compute the specific enthalpy for the gas phase.

14.2 Compositional Gases

When the composition of the gas is known, the specific enthalpy is almost invariably reported as one
of the results of a flash calculation. For single phase gas (i.e. dry gases), the specific enthalpy can be
computed using the procedures outlined earlier in Section 4.8 of these Notes.

14.3 Undefined Gases

By undefined, we mean that the actual composition of the gas is not known, and all we have to work
with is the gas gravity. A simple procedure for dealing with such cases is to use an equation of state to
compute the specific enthalpy for a binary mixture of gases, chosen such that the mixture has the same gravity
as the gas in question. As the components of the mixture, it is usual to choose the two simplest gases with
gravities that bracket that of the gas in question.

For example, a natural gas with a gravity of 0.68 could be viewed as a binary mixture of methane
(gravity = 0.554) and ethane (gravity = 1.038). The mole fraction of methane in the mixture, y, can be
computed from the simple balance relationship,

which results in y = 0.740 . One would thus compute the enthalpy for a binary mixture consisting of 74%
methane and 26% ethane (mole basis) and use that as an approximation for the actual gas in question.

For gases with a gravity greater than 1.038, one would base the calculations on a binary mixture of
ethane (gravity = 1.038) and propane (gravity = 1.522), and so on. In actual fact, most naturally produced
gases of practical interest can be treated as a methane-ethane mixtures.

This procedure works reasonably well for sweet gases (i.e. those with negligible amounts of N2, CO2,
or H2S). As the overall fraction of non-hydrocarbon components in the gas increases, the accuracy of the
approximation can be expected to decrease. Usually, however, the accuracy is sufficient for most purposes.

15. Liquid Phase Enthalpy

As noted above, calculations which involve thermal effects (e.g. flowing temperature profile,
heater/cooler sizing) can be performed quite accurately if the enthalpy of the fluids involved are known at
the temperatures and pressures of interest.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 103

At any given pressure, the specific enthalpy Hi, of a liquid at some absolute temperature T1* is given
by,

(15-1)

where Tref* is an arbitrarily selected absolute temperature at which the enthalpy is defined to be zero.

In most cases, we actually need to know the change of enthalpy associated with a temperature change,
say, from T1* to T2*, rather than the actual enthalpy. If we assume there is no phase change involved, this
is given by,

(15-2)

Thus, the reference temperature has no effect on the enthalpy change, provided that the same value
of Tref is used in both cases. In cases where CP is essentially constant over the temperature range of interest,

(15-3)

where ∆T is the temperature change.

15.2 Compositional Hydrocarbon Liquids

As in the case of gases, the specific enthalpy of a hydrocarbon liquid phase (e.g. oil, condensate
liquid) is almost invariably reported, along with the PVT behaviour, as part of the output results from a flash
calculation. For real liquids, the specific enthalpy is a function of both pressure and temperature, and this
will be reflected in the values that are reported from most equation of state computations.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 104

15.3 Crude Oil and Other Undefined Hydrocarbon Liquids

Rather than computing the specific enthalpy directly, Watson and Nelson (1933) proposed the
following equation for estimating the specific heat capacity of a crude oil,

(15-4)

where
(15-5)

(15-6)

(15-7)

and Cp = specific heat capacity, BTU/lb-°F


T = temperature, °F
Sgo = specific gravity of the oil
K = Watson K factor for the oil

To compute the absolute specific enthalpy, Equation (15-4) must be re-written in terms of absolute
temperature, T* , i.e.,

(15-8)

where
(15-9)

(15-10)

and T* = absolute temperature, oR

Equation (15-8) can now be substituted into Equation (15-1) and the integration performed, to obtain,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 105

after some re-arranging,

(15-11)

where T1 = temperature, °F
Tref = reference temperature, °F
H1 = specific enthalpy at temperature T1, BTU/lb - °F

Similarly, the change in enthalpy associated with a temperature change from T1 to T2 can be shown
to be,

(15-12)

In the Watson and Nelson (1933) correlation, it is inherently assumed that CP is independent of
pressure, and therefore, the calculated enthalpies are also independent of pressure. In actual fact, the pressure
effect is generally very small for liquids and, for the most part, this assumption is reasonable.

One important consequence, however, of ignoring the effect of pressure on the heat capacity is that
enthalpy changes computed using Equation (15-12) cannot be used to estimate the magnitude of frictional
heating effects in a crude oil pipeline. However, in some real cases of interest, this heating can be significant
and can influence the specification of insulation or other heat transfer related considerations.

An approximate procedure for introducing a pressure effect has been developed by the author.
Specific enthalpies were computed using the Peng-Robinson (1976) equation of state, over a relatively wide
range of pressures and temperatures, for a variety of hydrocarbon liquids. In all cases, the composition of
the liquid was known, and the API gravities ranged from about 15 to 65 oAPI. The effect of pressure was
surprising constant in all cases, and is well represented by the simple relationship,

(15-13)

where HP = specific enthalpy at any pressure P, BTU/lb-°F


Ho = specific enthalpy at one atmosphere, BTU/lb-°F
P = absolute pressure, psia

It is thus a simple matter to show that for any pressure and temperature change, to a good approximation,

(15-14)

where ∆H = overall enthalpy change for the hydrocarbon liquid, BTU/lb-°F


∆Ho = enthalpy change computed using Equation (15-12), BTU/lb-°F
P1 = initial pressure, psia
P2 = final pressure, psia

This modification does permits us to account for the frictional heating that can occur in heavy oil pipelines
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 106

with large pressure gradients.

16. Properties Of Produced Water

16.1 Introduction

Although it is seldom desirable to produce water along with hydrocarbon fluids, production of water
from the reservoir is often unavoidable. The quantity of water, relative to the produced hydrocarbon fluids
may be very small, and result simply because the gas phase in the reservoir is saturated with water vapour
at the reservoir temperature (due, say, to an underlying aquifer). On the other hand, the amount of produced
water can be very substantial, and represent a significant fraction (up to 100%) of the total liquids produced.
This could occur as a result of coning in a water-drive reservoir, break-through in a waterflood system,
condensed water from a conventional steam flood (break-through) or from a "huff'n puff" enhanced oil
recovery scheme.

As noted earlier in Section 13, the volume fraction of the total liquid phase produced in an oil well
that is represented by water, expressed as a percent, is referred to as the water cut.

Because of the potential influence of produced water on the pressure drop and general flow behaviour
of multiphase flow systems, one must be able to accurately estimate its properties. In general, these will not
simply be the same as the properties of pure water, since formation water normally contains significant quan-
tities of dissolved salts and other minerals.

The correlations that are described below should yield property data for produced water that
are sufficiently accurate for most practical purposes.

16.2 Density of Produced Water

The density of produced water, ρw, is given by the relation,

(16-1)

where SGw = specific gravity or relative density of the produced water


Bw = formation volume factor of pure water (bbl/STB -or- m3/m3 at stock tank conditions)
ρwo = density of pure water at stock tank conditions

The specific gravity of produced water, SGw, depends on both the nature and concentration of
dissolved salts. In most cases, SGw will be reported as part of any commercial analysis that has been
requested. Occasionally however, only the overall salt concentration, typically reported in milligrams salt
per litre (i.e. parts per million, ppm) is known. The specific gravity must thus be estimated in such cases.

Frick (1970) reported specific gravity and salt concentration data for produced water from a variety
of North American oil and gas fields. These data are shown plotted in Figure 16-1, and suggest a simple
correlation of the form,

(16-2)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 107

where α = a constant
Cs = overall salt concentration in parts per million (i.e. mg/litre)

Values of α have been determined by the author for a number of geographic regions and are given
in Table 16-1, along with an overall average value.

Table 16-1

Value of α in Equation (16-2) for Various Geographical Regions

Geographical Region No. of Data Points α Correlation Coefficient

Appalachian 26 0.728 x 10-6 0.980


Kansas 24 0.582 x 10-6 0.855
Oklahoma 42 0.690 x 10-6 0.968
Texas 10 0.757 x 10-6 0.957
Canada 39 0.654 x 10-6 0.935

Overall Average 141 0.682 x 10-6 0.941

Figure 16-1 shows a comparison of the data reported by Frick(1970) with the corresponding values
computed using Equation (16-2) and the overall average value of the constant. The agreement is seen to be
quite reasonable over the whole range.

As a matter of interest, seawater typically has an overall salt concentration of 30,000 ppm or less.
Produced waters may have salt concentrations ranging from a few hundred to more than 300,000 ppm; this
range is in fact represented by the data in Table 16-1.

An approximation of the formation volume factor for pure gas-free water can be obtained from
saturated steam tables, which give the density of pure water in equilibrium with saturated steam at any
temperature or pressure up to the critical point for water, i.e.

(16-3)

where ρws = density of saturated pure water at some pressure and temperature, lb/ft3 or kg/m3
ρwo = density of pure water at stock tank conditions, 62.37 lb/ft3 or 999.0 kg/m3.

A more complex correlation, which explicitly takes into account the effect of both pressure and
temperature has been proposed by Numbere et al (1977), as follows:
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 108

1.40

Calculated
Measured
Specific Gravity 1.30

1.20

1.10

1.00
0 100 200 300 400

Salt Concentration (ppm/1,000)

Figure 16-1

Specific Gravity vs. Dissolved Salts Concentration for Produced Water


(Calculated using Equation (16-2) with α = 0.682 x 10-6 )

(16-4)

ere

(16-5)

(16-6)

(16-7)

and T = temperature, °F
P = pressure, psia

Values of the constants a00 through a22 are given in Table 16-1.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 109

Table 16-1

Values of Constants for Equations (16-5) through (16-7)

Constant Value

a00 0.9947
a01 5.8 x 10-6
a02 1.02 x 10-6
a10 -4.228 x 10-6
a11 1.8376 x 10-8
a12 -6.77 x 10-11
a20 1.3 x 10-10
a21 -1.3853 x 10-12
a22 4.285 x 10-15

The density of gas-free water, calculated using Equation (16-1), with SGw = 1.0, is shown in Figure
16-2 over a range of pressures and temperatures covering most conditions likely to be encountered in
pipelines and wells. While not shown in Figure 16-2, the density obtained from the steam tables as described
above corresponds roughly to the values shown in Figure 16-2 for atmospheric pressure.

The solubility of natural gas in water is relatively low, but it may have a significant effect on the
density of water produced from deep high pressure wells. To account for this, the density obtained from
Figure 16-2 should be multiplied by the appropriate value of the factor estimated from Figure 16-3. Note that
the solubility of gas in formation water is also affected to some degree by the salinity of the water.

65

64
60 °F
63
120 °F
62
180 °F
DENSITY (lb/ft³)

61
240 °F
60

59
300 °F
58

57

56

55
0 1 2 3 4 5 6 7 8 9 10

P R E S S U R E (p s ia /1 0 0 0 )

Figure 16-2

Density of Gas-Free Water


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 110

16.3 Volumetric Flow Rate of Produced Water

For crude oil systems, the volumetric flow rate of produced water may be computed from the
relations,

(16-8)

where Qw = water flow rate, ft3/sec


Qo = oil flow rate, STB/day
WOR = water/oil ratio, STB/STB
Bw = water formation volume factor, bbl/STB

1 .0 0 0
0 .9 9 8
0 .9 9 6
0 .9 9 4
DENSITY MULTIPLIER

0 .9 9 2
0 .9 9 0
6 0 °F
0 .9 8 8
0 .9 8 6 1 8 0 °F

0 .9 8 4 3 0 0 °F
0 .9 8 2
0 .9 8 0
0 .9 7 8
0 .9 7 6
0 .9 7 4
0 1 2 3 4 5 6 7 8 9 10

P R E S S U R E (p sia /1 0 0 0 )

Figure 16-3

Density Correction Factor to Account for Dissolved Gas in Water

or, in SI units,

(16-9)

where Qw = water flow rate, m3/sec


Qo = oil flow rate, Sm3/day
WOR = water/oil ratio, Sm3/Sm3
Bw = water formation volume factor, m3/Sm3

When the flowing system is primarily gas, the volumetric water flow rate is calculated as follows,

(16-10)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 111

where Qw = water flow rate, ft3/sec


QGo = gas flow rate, MMscf/day
WGR = water/gas ratio, STB/MMscf
Bw = water formation volume factor, bbl/STB

or, in SI units,

(16-11)

where Qw = water flow rate, m3/sec


QGo = gas flow rate, Sm3/day
WGR = water/gas ratio, Sm3/Sm3
Bw = water formation volume factor, m3/Sm3

16.4 Viscosity of Produced Water

The viscosity of produced water, as a function of pressure, temperature, and concentration of


dissolved salts, can be computed using the following equations that were presented by Meehan (1980), which
are claimed to cover the whole range of conditions expected in both pipelines and wells.

(16-12)

where

(16-13)

and P = pressure, psia


T = temperature, °F

(16-14)

(16-15)

Cs = concentration of salts, ppm

For convenience in using these relations, Equation (16-2) can be rearranged as,

(16-16)

and, upon substituting the overall average value for α, we obtain the relation,
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 112

(16-17)

16.5 Enthalpy of Produced Water

For most practical purposes, it can be assumed that dissolved salts have no effect on the enthalpy of
the water phase, and the latter may be assumed to be that of pure water.

This may be obtained from steam tables, or estimated simply (assuming a value of 1.0 BTU/lb-°F
for the heat capacity of water) as:

(i) T is in °F ,

(16-13)

(ii) T is in °C ,

(16-14)

In either case, the reference temperature (i.e. the temperature at which HW = 0 ) is taken as the
freezing point of water at a pressure of one atmosphere.

16.6 Equilibrium Water Content of Natural Gas

When gas is produced from a reservoir with an underlying aquifer, it is generally assumed to be
saturated with water vapour at the reservoir conditions. Once the gas has been produced through the well and
enters the gathering system, conditions will most likely have changed to the point where some of the water
vapour will have dropped out as liquid. If this mechanism is the only source of liquid water, the contribution
by the condensed water to the total amount of liquid present is not significant in most cases and can safely
be ignored. The exceptions usually occur when bottom-hole temperatures and pressures are high, and in such
cases, it can be significant. Of course, if hydrate formation is likely to occur, the presence of liquid water in
any amount is a concern.

Figure 16-5 shows a generalized plot that gives the equilibrium water content of natural gas as a
function of temperature and pressure. Strictly speaking, this Figure should only be used for sweet gases (i.e.
those with no H2S or CO2), but it should be sufficiently accurate for most flow calculation purposes.

Figure 16-5 shows the saturation value in units of lb water/MMscf of total gas. The amount of water
that will condense or drop out in the well or pipeline can be determined by looking at the initial and final
conditions of the system.

For example, in a well with a bottom-hole pressure and temperature of 2,500 psia and 200 °F, Figure
16-5 gives the equilibrium water content to be about 340 lb/MMscf of total gas. At the wellhead, we will
assume that the pressure has dropped to 1,000 psia and the temperature has decreased to about 80 °F. Under
these conditions, Figure 16-5 gives a water saturation of about 35 lb/MMscf of total gas. The difference
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 113

between these two values, 305 lb/MMscf of total gas represents the amount of liquid water that will drop out
in the tubing, and thus be produced at the wellhead.

For the record, one barrel of water weighs approximately 350 lb and at stock tank conditions, would
only represent approximately 0.007 MMscf of the total gas volume. In more familiar terms then, the total
liquid water in the above case would be less than 0.9 bbl/MMscf (i.e. less than 5 x 10-6 m3/Sm3). From a
multiphase flow point of view, this is very low, and would normally be ignored unless the gas flow rate was
very low. As noted above, however, it would be an important factor if hydrate formation conditions were
approached anywhere in the system.

Figure 16-5

Equilibrium Water Vapour Content in Sweet Natural Gas

17. Surface Tension


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 114

17.1 Introduction

In porous medium systems (i.e. oil reservoirs), surface tension is an important parameter in the
estimation of recoverable reserves because of its effect on residual saturations. On the other hand, most
correlations and models for predicting two phase flow phenomena in pipelines are relatively insensitive to
surface tension, and one can generally use an average value for calculation purposes. Calculations for wells
have a somewhat stronger dependence on surface tension, in that this property can be important in predicting
bubble and droplet sizes (maximum stable droplet size increases as surface tension increases), which in turn,
can significantly influence the calculated pressure drop. Even then, however, surface tension typically
appears in the equations raised to only about the ¼ power.

This relatively weak dependence is fortunate, since the accurate calculation of surface tension is very
complex. Physically, surface tension can also be greatly affected by even trace amounts of surfactants or
other chemical contamination. Thus, even values calculated using a very accurate procedure can be subject
to large error if such contamination exists.

In this section, we will look at some relatively simple procedures that yield calculated values that are
satisfactory for most pipeline and wellbore flow modelling purposes.

17.2 Baker and Swerdloff (1956) Correlation

This correlation is used for estimating the surface tension for gas-hydrocarbon liquid systems, and
it is a two part procedure. The first step is to determine the surface tension at atmospheric pressure, as a
function of the API gravity of the oil, from Figure 17-2. Note that there are only two temperature lines on
this graph. When the temperature of interest is greater than 100 °F, the value should be read directly from
the 100 °F line. Similarly, when the temperature is less than 68 °F, the value should be read directly from
the 68 °F line. The lines can thus be viewed as limiting boundaries. At temperatures between these values,
interpolation should be performed between the two lines.

In the second step, a multiplying factor which adjusts this value to the pressure of interest is read from
Figure 17-2. The oil-gas surface tension at any given pressure is thus obtained from the relation,

(17-1)

where σo = oil-gas surface tension at any pressure, dyne/cm


σoo = oil-gas surface tension at one atmosphere, dyne/cm
K = multiplying factor obtained from Figure (2.17-1)

It should be obvious from the scatter of the data shown in these Figures that the correlation must be
considered as very general and approximate. In fact, simply using a value of 20-30 dyne/cm for crude oils
and 5-10 dyne/cm for condensate liquids will usually give results that are just as satisfactory.
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 115

Figure 17-1

Baker and Swerdloff (1956) Correlation for Surface Tension


of Oil-Gas Mixtures at 1.0 Atmosphere

Figure 17-2

Baker and Swerdloff (1956) Multiplying Factor for Computing Oil-Gas


Surface Tension at a Given Pressure from the Surface Tension at One
Atmosphere
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 116

17.3 Gould (1972) Correlation

Gould proposed a simple correlation which expresses surface tension solely as a polynomial in
pressure. It is claimed to be applicable for both crude oil and gas condensate systems.

(17-2)

where P = pressure, psia


σo = gas-oil surface tension, dyne/cm

Values of the constants a0 through a3 are given in Table 17-1.

At pressures above about 3,500 psia, Equation (17-2) predicts negative values and that pressure must
therefore be considered to be an upper limit for this procedure.

Gould also presented a second equation which can be used to estimate the surface tension in a gas-
water system. This relation, which includes the effect of both pressure and temperature is given by,

(17-3)

where
(17-4)

and T = temperature °F
P = pressure, psia
σw = gas-water surface tension, dyne/cm

Values of the constants b0 through b5 and c0 through c5 are given in Table 17-1. As with Gould's gas/oil
correlation, Equation (17-3) should only be considered valid up to a pressure of about 3,500 psia.

Both of these correlations are shown graphically in Figure 17-3.

Table 17-1

Constants in Equations (17-2) and (17-4)

Number a b c

0 2.5698 x 10-2 -0.0984 81.71


1 -2.27803 x 10-5 6.9 x 10-6 -3.527 x 10-2
2 7.5762 x 10-9 6.4 x 10-8 2.304 x 10-5
3 -8.9147 x 10-12 -6.09 x 10-11 -9.94 x 10-9
4 - 1.94 x 10-14 2.3 x 10-12
5 - -2.025 x 10-18 -2.1665 x 10-16
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 117

80

60 °F
70
120 °F
SURFACE TENSION (dyne/cm)
60
180 °F
50
240 °F
40

30 Gas-Oil
Gas-Water
20

10

0
0 1000 2000 3000 4000 5000

PRESSURE (psia)

Figure 17-3

Gould (1972) Correlations for Gas-Oil and Gas-Water Surface Tension

17.4 Gray (1978) Correlation

Gray presented a useful correlation for gas-water surface tension that accounts for all the effects of
both pressure and temperature, as follows:

(17-5)

where σw = gas-water surface tension, dyne/cm


P = pressure, psia
T = temperature, °F

Values of σw calculated using Equation (17-5) are shown in Figure 17-4 for a range of temperatures and
pressures that cover most practical cases of interest. Comparison of Figure 17-3 with Figure 17-4 shows that
the surface tension values predicted by the Gray (1978) correlation are comparable to those predicted by the
Gould (1972) equation for pressures below about 2,500 psia. Because of both its simplicity and its stability
over the full range of interest, it is recommended that the Gray correlation be used.

. When both water and a hydrocarbon liquid phase are present (i.e. two immiscible liquids), a
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 118

question naturally arises regarding the effective value of the “gas-liquid” surface tension. Gray also presented
a mixing rule for such cases, proposing that an overall average value can be computed using the relation,

(17-6)

where sm = average surface tension for mixture


so = surface tension for gas-oil system
sw = surface tension for gas-water system
WOR = water-oil ratio, STB/STB or Sm3/Sm3

The actual system units used for the parameters in Equation (17-6) does not matter, as long as they are
consistent. The form of this expression is such that the effect of the water-gas surface tension is smaller than
if the weighting was simply based on the relative flow rates.

80

70
SURFACE TENSION (dyne/cm)

60 °F

60 120 °F
180 °F
240 °F
50
300 °F

40

30

20

10
0 1 2 3 4 5 6 7 8 9 10

PRESSURE (psia/1000)

Figure 17-4

Gray (1978) Correlation for Gas-Water Surface Tension

17.5 Gomez (1987) Correlation


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 119

Another correlation for the surface tension of gas-oil systems was presented by Gomez (1987). An
unfortunate printing error occurred in the original publication; the corrected version of the correlation is as
follows:

(17-7)

where so = gas-oil surface tension, dyne/cm


K = Watson's K factor for the oil
T = temperature, °F
SGo = specific gravity of the oil

Since Equation (17-7) does not include a pressure effect, it is presumed that the calculated values would
correspond to normal atmospheric pressure. A comparison of values computed using Equation (17-7) with
the Baker and Swerdloff (1956) correlation for atmospheric pressure (i.e. Figure 17-1) also suggests this is
the case. It is thus recommended that values computed using Equation (17-7) be multiplied by the factor Fp,
defined as,

(17-8)

where P = pressure, psia

Equation (17-8) in fact represents an approximation of the Baker and Swerdloff (1956) pressure
factor shown in Figure 17-2 and using it should improve the accuracy of the Gomez correlation sufficiently
for most purposes.

Figure 17-5 shows the calculated surface tension as a function of pressure for a 35 oAPI oil with a
Watson K factor of 11.9 . The calculations were performed using Equation (17-7) for a temperature of 120
°F and then adjusted for pressure using Equation (17-8). Also shown in this Figure is the calculated surface
tension for water at 120 °F taken from Figure 17-4, and the curve corresponding to oil-water mixtures having
20% and 50% water cuts. The effect of the reduced weighting of the water surface tension is clearly evident
in the general magnitude of the values for the mixtures.

17.6 Compositional Systems

For hydrocarbon mixtures of known composition. the surface tension can be calculated from the
relation,

(17-9)

where s = surface tension, dyne/cm


Pi = Parachor for component i
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 120

80
100 % Oil (Gomez)
70 100% Water (Gray)
80 % Oil - 20% Water
SURFACE TENSION (dyne/cm)
60 50 % Oil - 50% Water

50

40

30

20

10

0
0 500 1000 1500 2000 2500 3000 3000

PRESSURE (psia)

Figure 17-5

Calculated Surface Tension vs. Pressure at 120 °F for a 35 oAPI


Crude Oil, Water, and Two Crude Oil-Water Mixtures

rL = density of the liquid phase, g/cm


rG = density of the gas phase, g/cm
ML = molecular weight of the liquid phase
MG = molecular weight of the gas phase
xi = mole fraction of component i in the liquid phase
yi = mole fraction of component i in the gas phase

The parachor for a given component is a property that can be estimated using group contribution
values, as given in the GPSA Engineering Data Book, or simply estimated from the approximate relation.

(17-10)

where Mi = molecular weight of component i

Note that Equation (17-10) should only be used for paraffinic hydrocarbons.

Fully compositional calculations always encounter the same problems when it comes to residual
fractions and the more rigorous surface tension models are no exception. For example, how does one estimate
the parachor for a pseudo-component, and how does one test the calculated results? Measured surface
tensions are seldom available unless required for some very specific purpose.

In actual fact, it is seldom necessary to perform fully compositional surface tension calculations if
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 121

the surface tension value is required only for pipeline or tubing flow calculations. The methods discussed
earlier in this Section will normally be entirely adequate for such purposes.

18. Hydrates

18.1 Introduction

A hydrate consists of ice-like crystals that can be formed in a pipeline or well under particular
conditions of pressure and temperature when a hydrocarbon gas is flowing in the presence of liquid water.
The gas molecules fit into the spaces in the water molecule and cause a solid to form at temperatures well
above the normal freezing point for pure water. These crystals can accumulate, with a resulting decrease in
the available cross-section flow area, and may even block the pipeline or tubing completely (i.e. a hydrate
plug). Hydrates can occur almost anywhere, but they are of special concern for Arctic and offshore
installations where pipelines are exposed to low ambient temperatures for long distances, and where the cost
of insulation and/or heating is high.

A number of different crystal structures have been identified, and a detailed discussion of these is
beyond the scope of these Notes. However, because the gas molecules must fit into available spaces in the
crystal structures, hydrates can only form with relatively small gas molecules (i.e. relatively low molecular
weight components). For example, hydrate forming molecules for structures I and II include methane, ethane,
propane, i-butane, nitrogen, carbon dioxide, hydrogen sulphide, oxygen, argon, Kr, Xe, and C2-C3-cyclo-C3.
N-butane requires the presence of at least one of these compounds in order to participate in forming a hydrate.
Unfortunately, some or all of these components are typically present to some degree in most natural gas
production.

Figure 18.1 shows a phase envelope with three curves, as generated by the D.B. Robinson software,
HYDRATE. These curves show the temperature at which hydrates are predicted to form in the presence of
liquid water. The hydrate formation curve for the original fluid is labeled “Locus 1"; hydrates are predicted
to form at conditions to the left of the hydrate formation curve. Note that as the pressure increases, the
maximum temperature at which hydrates form also increases.

Figure 18.1

Hydrate Formation Envelope With and Without Inhibitors


Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 122

Hydrate inhibitors, such as methanol, ethylene glycol and triethylene glycol act by destroying the
cavity structure of the water molecules so that there is no room for the hydrate forming molecules. The curve
labeled “Locus 2" in Figure 18.1 shows the predicted hydrate formation temperature when a methanol molar
concentration of 20% has been added (i.e. moles of methanol divided by the moles of methanol plus moles
of liquid water) . The curve labeled as “Locus 3" shows the hydrate formation envelope when the methanol
concentration has been increased to 40 mole percent. Clearly, the addition of inhibitors has decreased the
maximum temperature at which hydrates form.

Another way to prevent blockage of the flow by hydrates is to use a class of chemicals known as anti-
agglomerants that prevent the hydrates from conglomerating into the solid (or, nearly so) masses that actually
form the plugs (Frostman, 2000). Yet another class of chemicals, known as kinetic inhibitors, can delay the
nucleation and growth of the hydrate crystals themselves (Paez et al, 2001).

There are a number of methods for computing the maximum temperature at which hydrates will form.
These range from relatively simple (and, less accurate) procedures that are based solely on the gas gravity,
to more complicated methods that are based on the actual gas composition and vapour-solid equilibrium
constants, to more recently developed complex methods based on equations of state and include the
interaction between components.(e.g. Ng-Robinson method).

For example, the GPSA Engineering Data Book (1972, 1980, 1998) presents a graph of the predicted
hydrate formation temperature versus pressure for gases of several different gravities, as shown in Figure18.2
Hydrates are predicted to form under conditions to the left of the curves. Note that the GPSA also issues the
caution that these curves should only be used for first approximation estimates.One reason for this is that the
hydrate formation temperature increases with increasing concentrations of CO2 and H2S. However, such
graphs do not differentiate between gases with the same gravity yet with differing sour gas content.

Figure 18-2

Pressure-Temperature Curve for


Predicting Hydrate Formation
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 123

18.2 Katz et al (1959) Correlation

Katz et al (1959) presented a method for predicting the hydrate temperature based on the composition
of the gas and some empirically derived vapour-solid equilibrium constants. The latter are defined for each
component of the gas as,

(18-1)

where K v - s = vapour-solid equilibrium constant for a given hydrocarbon component


xg = mole fraction of the component in the gas phase on a water-free basis
xs = mole fraction of the component in the solid phase on a water-free basis

For hydrates to form, the following equation must be satisfied,

(18-2)

The equilibrium constants are presented in charts as functions of temperature; these charts can be
found in the GPSA Engineering Data Book (1972, 1980, 1998). Evaluation of the hydrate temperature
involves a trial and error procedure. For the known composition and an initial guess of the temperature, one
determines values of the equilibrium constants from the charts. These are used to compute the value of the
left hand side of Equation (18-2), which is then compared with the target value of 1.0 . If the sum is not
acceptably close to 1.0, a new temperature must be selected and the procedure is repeated. The charts can
be used for mole concentrations of H2S up to about 15 - 20%. When the H2S concentration is about 30% or
higher, hydrates will form at about the same temperatures as for pure H2S .

18.3 Hammerschmidt (1969) Correlation

A commonly used relationship for estimating the extent to which the gas hydrate formation
temperature can be reduced by adding an inhibitor is known as the Hammerschmidt (1969) equation, which
is as follows:

(18-3)

where ∆Td = the number of degrees that the hydrate formation temperature has been lowered by
the addition of the inhibitor (°F or °C)
M = molecular weight of the inhibitor
w = mass percent of the inhibitor in the aqueous phase
α = a constant (2,335 for ∆Td in °F, or 1,297 for ∆Td in °C)

Molecular weights of some common inhibitors are given in Table 18.1 below. Note that the
Hammerschmidt (1969) equation is based only on the concentration of the inhibitor is the aqueous phase.
In practice, however, the amount of inhibitor required for a given hydrate temperature suppression must also
take into account the inhibitor (if any) that is lost into the vapour and hydrocarbon liquid phases. This can
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 124

be especially significant for methanol, but, unfortunately, such losses are not estimated by this procedure and
the required methanol injection can be underestimated. In some circumstances, Equation (18-3) will also
substantially over-estimate the amount of inhibitor required, and it must thus be viewed as a relatively
approximate method.

Table 18.1

Molecular Weight and Specific Gravity for Common Hydrate Inhibitors

Molecular Specific
Inhibitor
Weight Gravity

methanol 32.0 0.797

ethylene glycol 62.1 1.109

diethylene glycol 106.1 1.120

triethylene glycol 150.2 1.127

tetraethylene glycol 194.2 1.129

18.4 Ng and Robinson Method for Hydrate Prediction and Suppression

A considerably more reliable (and, more complex) method is that of Ng and Robinson (1976, 1977,
1980, 1994), which has also been studied and reported on by Ng et al (1987), Parrish and Prausnitz (1972),
Robinson and Ng (1986), Robinson et al (1987), Ng (1992, 1999), Ng and Borman (1999), Skovborg et al
(1993), and Zuo (2000).

This is an iterative procedure which predicts the maximum temperature at which hydrates will form
for both the case where the water is present as liquid and the case when the water exists totally as vapour in
the gas phase. In the latter case, the actual amount of water is taken into account when computing the
maximum temperature at which hydrates will form; a higher water vapour concentration results in a higher
hydrate formation temperature. In either case, the amount of inhibitor present in each phase is also taken into
account.

When the inhibitor is to be included in the overall fluid composition, it may be necessary to express
the amount of inhibitor that is being added as an “equivalent gas flow rate”. This can be computed using the
following simple relation, which is derived from the Engineering gas law,

(18-4)

3
where Qi, eq = equivalent gas flow of inhibitor (MMscf/d -or- Sm /d)
K = constant ( = 0.00914 for Field units; = 567.71 for SI units)
wi = mass flow rate of inhibitor (lb/hr -or- kg/hr)
Mi = molecular weight of the inhibitor

For Field units, the value of K given above is for an assumed base pressure of 14.65 psia. For a different base
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 125
o o
pressure, this value of K (i.e. 0.00914) should be multiplied by 14.65/P , where P is the actual base
pressurein psia.

19. Some Concluding Comments

Fluid property and PVT behaviour calculations almost always have to be made as part of any fluid
flow modelling study for pipelines and wells, since one seldom has enough measured data available to cover
all of the cases and conditions that may be of interest. The technology contained in these Notes includes a
wide variety of methods that can be used with almost any amount of information about the fluids. In
preliminary studies, fluid data tend to be very sparse and simplified procedures are required. Later in the life
of a project, detailed fluid compositions, results of PVT studies, measurements of transport properties, etc.
may be available, and fully compositional procedures become more important.

Correlations based on data from specific producing areas can be especially useful if you are looking
at systems from the same area. The properties of produced hydrocarbons can vary widely, and no single
correlation or model can realistically be expected to be universally applicable.

Finally, at the risk of excessive repetition, we again stress the value of making use of whatever data
you may have available to help select appropriate methods and then, to verify the accuracy of fluid property
and PVT calculations. It pays to remember that if the fluid properties used in your flow calculations do not
match those that will actually occur in your system, it is simply not your system that you are modelling!
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 126

LITERATURE REFERENCES

Abbot, M. M., Kaufmann, T. G., and Domash, L. Amin, M. B., and Maddox, R. N.
“A Correlation for Predicting Liquid Viscosities of Petro- “Estimate Viscosity vs. Temp.”, Hydrocarbon Processing,
leum Fractions”, Can. J. Chem. Eng., Vol. 49, p. 379, June p. 131, Dec. (1980)
(1971)
Asgarpour, S., McLauchlin, L., Wong, D., and Cheung, V.
Abdul-Majeed, G. H., and Salman, N. H. “Pressure-Volume-Temperature Correlations for Wes-tern
“An Empirical Correlation for Oil FVF Prediction”, J. Can. Canadian Gases and Oils”, J. Can. Petrol. Technol., Vol. 28,
Petrol. Technol., Vol. 27, No. 6, p. 118, Nov.-Dec. (1988) No. 4, p. 103, Jul-Aug (1989)

Abdul-Majeed, G. H., Kattan, R. R., and Salman, N. H. Baker, O., and Swerdloff, W.
“New Correlation for Estimating the Viscosity of Under- “Finding Surface Tension of Hydrocarbon Liquids”, Oil and
saturated Crude Oils”, J. Can. Petrol.Technol., Vol. 29, No. Gas J., p. 125, January 2 (1956)
3, p. 80, May-June (1990)
Beal, C.
Ahrabi, F., Ashcroft, S. J., and Shearn, R. R. “The Viscosity of Air, Water, Natural Gas, Crude Oil and its
“High Pressure Volumetric, Phase Composition, and Associated Gases at Oil Field Temperatures and Pressures”,
Viscosity Data for a North Sea Crude Oil and NGL”, Chem. Trans. AIME, Vol. 165, p. 94 (1946)
Eng. Res. Des., Vol . 65, p. 63, Jan. (1987)
Beg, S. A., Amin, M. B., and Hussain, I.
Al-Besharah, J. M., Akashah, S. A., and Mumford, C. J. “Generalized Kinematic Viscosity-Temperature Correlation
“The Effect of Temperature and Pressure on the Viscosities for Undefined Petroleum Fractions”, The Chem. Eng. J., Vol.
of Crude Oils and Their Mixtures”, Ind. Eng. Chem. Res., 38, p. 123 (1988)
Vol. 28, p. 213 (1989a)
Beggs, H. D., and Robinson, J. R.
Al-Besharah, J. M., Akashah, S. A., and Mumford, C. J. “Estimating the Viscosity of Crude Oil Systems”, J. Petrol.
“Viscosities of Binary Crude-Oil Mixtures Correlation”, Oil Technol., p. 1140, September (1975)
& Gas J., p. 35, Feb. 20 (1989b)
Benedict, M., Webb, G. B., and Rubin, L. C.
Al-Besharah, J. M., Akashah, S. A., and Mumford, C. J. “An Empirical Equation for Thermodynamic Properties of
“Binary Mixtures of Petroleum Products and Mixtures of Light Hydrocarbons and Their Mixtures - Part 1: Methane,
Base Oils Examined”, Oil & Gas J., p. 50, March 6 (1989c) Ethane, Propane and N-Butane”, J. Chem. Physics, Vol. 8,
April (1940)
Al-Marhoun, M. A.
“Pressure-Volume-Temperature Correlations for Saudi Crude Bergman, D.
Oils”, paper No. SPE 13718, presented at the Middle East Oil “Predicting the Phase Behaviour of Natural Gas Pipelines”,
Tech. Conf. and Exhib., Bahrain (1985) Vol. 1 and 2, Ph.D Thesis, University of Michigan (1976)

Al-Marhoun, M. A. Brule, M. R., Kumar, K. H., and Watanasiri, S.


“PVT Correlations for Middle East Crude Oils”, J. Petrol. “Characterization Methods Improve Phase Behaviour Pre-
Technol., p. 660, May (1988) dictions”, Oil and Gas J., p. 87, February 11 (1985)

Al-Marhoun, M. A. Calhoun, J. C.
“New Correlations for Formation Volume Factors of Oil and Fundamentals of Reservoir Engineering, Univ. of Oklahoma
Gas Mixtures”, J. Can. Petrol. Technol., Vol. 31, No. 3, p. 22 Press, Norman, OK (1947)
(1992)
Carr, N. L., Kobayashi, R., and Burrows, D. B.
American Gas Association “Viscosity of Hydrocarbon Gases Under Pressure”, Trans.
“Compressibility and Supercompressibility for Natural Gas AIME, Vol. 201, p. 264 (1954)
and Other Hydrocarbon Gases”, Transmission Measurement
Committee Report No. 8, December 15 (1985) Cavett, R. H.
“Physical Data for Distillation Calculations - Vapor-Liquid
American Petroleum Institute Equilibria”, Trans. API, San Francisco, CA, Vol. 42, No. 3,
API 44 Tables: Selected Values of Properties of Hydro- p. 351, May (1962)
carbons and Related Compounds, (1975)
Chew, J., and Connally, C. A.
American Society for Testing Materials “A Viscosity Correlation for Gas Saturated Crude Oils”,
1981 Annual Book of ASTM Standards - Part 23, Petroleum Trans. AIME, Vol. 216, p. 23 (1959)
Products and Lubricants, ASTM, Philadelphia, PA, p. 205
(1981) Dean, D. E., and Stiel, L. I.
“The Viscosity of Nonpolar Gas Mixtures at Moderate and
High Pressures”, AIChE J., Vol. 11, p. 526 (1965)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 127

Dempsey, J. R. Gregory, G. A.
“Computer Routine Treats Gas Viscosity as a Variable”, Oil “Viscosity of Heavy Oil/Condensate Blends”, Technical Note
and Gas J., p. 141, August 16 (1965) No. 6, Neotechnology Consultants Ltd., Calgary, Canada,
July (1985)
Dokla, M. E., and Osman, M. E.
“Correlation of PVT Properties for UAE Crudes”, SPE Form. Gregory, G. A.
Eval., p. 41, Mar. (1992) “Pipeline Calculations for Foaming Crude Oils and Crude
Oil-Water Emulsions”, Technical Note No. 11,
Dranchuk, P.M., Purvis, R.A., and Robinson, D.B. Neotechnology Consultants Ltd., Calgary, Canada, January
“Computer Calculations of Natural Gas Compressibility (1990)
Factors Using the Standing and Katz Correlations”, Inst. of
Petrol. Technical Series, No. IP74-008, p. 1 (1974) Gregory, G. A.
“Characterizing Residual Fractions and Crude Oil Assay Data
Dranchuk, P. M., and Abou-Kassem, J. H. as Pseudo-Components Using HYPOS”, Technical Note No.
“Calculations of Z Factors for Natural Gases Using Equa- 1, Neotechnology Consultants Ltd., Calgary, Canada,
tions of State”, J. Can. Petrol. Technol., p. 34, July-Sept. November (1993a)
(1975)
Gregory, G. A.
Dranchuk, P. M., Islam, R. M. , and Bentsen, R. G. “Improvements to the Corresponding States Liquid Density
“A Mathematical Representation of the Carr, Kobayashi, and (COSTALD) Correlation for Handling Pseudo-components”,
Burrows Natural Gas Viscosity Cor-relations”, J. Can. Petrol. Technical Note No. 18, Neotechnology Consultants Ltd.,
Technol., p. 51, January (1986) Calgary, Canada, November (1993b)

Elsharkawy, A. M., Hashem, Y. S., and Alikan, A. A. Gregory, G. A.


Compressibility Factor for Gas-Condensates”, Paper SPE “Calculate the Density of Non-hydrocarbon Gases
59702, presented at the SPE Permian Basin Oil and Gas Correctly”, Technical Note No. 24, Neotechnology
Recovery Conf., Midland, TX, March (2000) Consultants Ltd., Calgary, Canada, November (2000)

Eyring, H. Hall, K. R., and Yarborough, L. A.


“Viscosity, Plasticity and Diffusion as Examples of Absolute “A New Equation of State for Z-Factor Calculations”, Oil and
Reaction Rates”, J. Chem. Phys., Vol. 4, p. 283 (1936) Gas J., p. 82, June 18 (1973)

Fricke, T. C., Editor-in-Chief Hankinson, R. W., and Thomson, G. H.


Petroleum Production Handbook, Vol. 2, Society of “A New Correlation for Saturated Densities of Liquids and
Petroleum Engineers, Richardson, Texas (1962) Their Mixtures”, AIChE J., Vol. 25, p. 653 (1979)

Frostman, L.M. Hanley, H. J. M., McCarty, R. D., and Haynes, W. M.


"Antiagglomerant Hydrate Inhibitors for Prevention of Cryogenics, Vol. 15, p. 413 (1975)
Hydrate Plugs in Deepwater Systems", Paper No. SPE 63122
presented at the 2000 SPE Annual Tech. Conf. and Exhib, Hopke, S. W., and Lin, C. J.
Dallas, TX, Oct. (2000) “Improve Absorber Predictions”, Hydrocarbon Proc., p. 136,
June (1974)
Gas Processors Association (GPSA)
Engineering Data Book, Tulsa, Oklahoma, 9th Edition Hougen, O. A., Watson, K. M., and Ragatz, R. A.
(1977), SI Edition (1980), 10th Edition (1987) Chemical Process Principles, Vol. 2, p. 593, John Wiley &
Sons, Inc., New York, N.Y. (1959)
Glasø, Ø.
“Generalized Pressure-Volume-Temperature Correlations”, Jossi, J. A., Stiel, L. I., and Thodos, G.
J. Petrol. Technol., p. 785, May (1980) “The Viscosity of Pure Substances in the Dense, Gaseous,
and Liquid Phases”, AIChE J., Vol. 8, p. 59 (1962)
Gomez, J. V.
“Method Predicts Surface Tension of Petroleum Fractions”, Katz, D. L., and Firoozabadi, A.
Oil and Gas J., p. 68, December 7 (1987) “Predicting Phase Behaviour of Condensate/Crude Oil
Systems Using Methane Interaction Coefficients”, J. Petrol.
Gould, T. Technol., p. 1649, November (1978)
“Vertical Two-Phase Flow in Oil and Gas Wells”, Ph.D.
Thesis, The Univ. of Michigan, Ann Arbor, Michigan Kay, W. B.
(1972) “Density of Hydrocarbon Gases and Vapor at High
Temperature and Pressure”, Ind. Eng. Chem., p. 1014,
Gray, H. E. September (1936)
“Vertical Flow Correlation - Gas Wells”, API Manual 14
BM, Second Edition, Appendix B, p. 38, American Petroleum Kesler, M. G., and Lee, B. I.
Institute, Dallas, Texas, January (1978) “Improve Prediction of Enthalpy of Fractions”, Hydrocarbon
Processing, p. 153, March (1976)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 128

Khan, S. A., Al-Marhoun, M. A., Duffuaa, S. O., and Abu- Ng, H.-J., and Borman, C.
Khamsin, S. A. "Hydrate Phase Composition for Multicomponent Gases",
“Viscosity Correlations for Saudi Arabian Crude Oils”, paper GPA Research Report RR-168 (1999)
No. SPE 15720, presented at the 5th SPE Middle East Oil
Show, Manama, Bahrain, March (1987) Ng, H.-J., Chen, C.-J., and Saeterstad, T.
"Hydrate Formation and Inhibition in Gas Condensate and
Lasater, J. A. Hydrocarbon Liquid Systems", Fluid Phase Equilibria, 36, p.
“Bubble Point Pressure Correlation”, Trans. AIME, Vol. 213, 99 (1987)
p. 379, (1958)
Ng, H.-J., and Robinson D.B.
Lee, A. L., Gonzalez, M. H., and Eakin, B. E. "The Measurement and Prediction of Hydrate Formation in
“The Viscosity of Natural Gases”, J. Petrol. Technol., Vol. Liquid Hydrocarbon-Water Systems", Ind. Eng. Chem. Fund.,
18, p. 997 (1966) v. 15, p. 293 (1976)

Lee, B. I., and Kesler, M. G. Ng, H.-J., and Robinson D.B.


“A Generalized Thermodynamic Correlation Based on Three "The Prediction of Hydrate Formation in Condensed
Parameter Corresponding States”, AIChE J., Vol. 21, p. 510 Systems", AIChE J., v. 23, p. 477 (1977)
(1975)
Ng, H.-J., and Robinson D.B.
Lohrenz, J., Bray, B. G., and Clark, C. R. "A Method for Predicting the Equilibrium Gas Phase Water
“Calculating Viscosities of Reservoir Fluids from Their Content in Gas-Hydrate Equilibrium", Ind. Eng. Chem.
Compositions”, J. Petrol. Technol., p. 1171, October (1964) Fund., v. 19, p. 33 (1980)

Letsou, A., and Stiel, L. I. Ng, H.-J., and Robinson D.B.


“Viscosity of Saturated Nonpolar Liquids at Elevated "The Influence of Methanol on Hydrate Formation at Low
Pressures”, AIChE J., Vol. 19, p. 409 (1973) Temperatures", GPA Research Report RR-74 (1984) and RR-
74 (Revised) (1986)
Manning, R. E.
“Computation Aids for Kinematic Viscosity Conversions Ng, H.-J., and Robinson, D.B.
from 100 and 210 oF to 40 and 100 oC”, J. of Testing and "New Developments in the Measurement and Prediction of
Evaluations (JVETA), Vol. 2, p. 522, November (1974) Hydrate Formation for Processing Needs", Int. Conf. on
Natural Gas Hydrates, Annals of the New York Academy of
Meehan, D. N. Sciences Vol. 715, p. 450-462 (1994)
“A Correlation for Water Viscosity”, Petrol. Eng. Int., July
(1980) Ng, J. T. H., and Egbogah, E. O.,
“An Improved Temperature-Viscosity Correlation for Crude
McCain, W. D. Oil Systems”, Paper No. 83-34-32, presented at the 34th
“Black Oils and Volatile Oils - What’s the Difference?”, Pet. Ann. Tech. Mtg. of The Petrol. Soc. of CIM, Banff, Alta,
Eng. Intl., p. 24, November (1993) May (1983)

McCain, W. D. Numbere, D. T., Brigham, W. E., and Standing, M. B.


“Volatile Oils and Retrograde Gases - What’s the Proc. Petroleum Research Inst., p. 8, Stanford University,
Difference?”, Pet. Eng. Int., p. 35, January (1994a) November (1977)

McCain, W. D. Parrish, W.R., and Prausnitz, J.M.


“Heavy Components Control Reservoir Fluid Behaviour”, J. "Dissociation Pressure of Gas Hydrates Formed by Gas
Petrol. Technol., p. 764, September (1994) Mixtures", Ind. Eng. Chem. Process Des. Develop., 11, p. 26
(1972)
Moses, P. L.
“Engineering Applications of Phase Behaviour of Crude Oil Passut, C. A., and Danner, R. P.
and Condensate Systems”, J. Petrol. Technol., p. 715, July “Correlation of Ideal Gas Enthalpy, Heat Capacity, and
(1986) Entropy”, Ind. Eng. Chem. Proc. Des. Dev., Vol. 11, No. 4,
p. 543 (1972)
Ng, H.-J.
"Solid Deposition in Hydrocarbon Systems: Hydrate Pedersen, K. S., Thomassen, P., and Fredenslund, A.
Inhibitor Loss Studies", Annual Report to Gas Research “Phase Equilibria and Separation Processes”, Report SEP
Institute (1992) and (1993) 8207, Inst. for Kemiteknik, Denmark Teckniske Hojskole,
July (1982)
Ng, H.-J.
"Hydrate Phase Composition for Multicomponent Gas Pedersen, K. S., and Fredenslund, A.
Mixtures", Presented at 3rd Int. Conf. on Gas Hydrates, Salt “An Improved Corresponding States Model for the Prediction
Lake City, USA, July (1999) of Oil and Gas Viscosities and Thermal Conductivities”,
Preprint No. SEP 8602, Danish Tech. Univ., February (1986)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 129

Pedersen, K. S., Fredenslund, A., Christensen, P. L., and Society of Petroleum Engineers
Thomassen, P. Petroleum Engineering Handbook, H.B. Bradley, Editor-in
“Viscosity of Crude Oils”, Chem. Eng. Sci., Vol. 39, No. 6, Chief, Richardson, Texas (1987)
p. 1011 (1984)
Standing, M. B.
Peng, D. Y., and Robinson, D. B. “A Pressure-Volume-Temperature Correlations for Mixtures
“A New Two Constant Equation of State”, Ind. Eng. Chem. of California Oils and Gases”, Drill. Prod. Practice, API, p.
Fund., Vol. 15, p. 59 (1976) 247 (1947)

Peng, D. Y., and Robinson, D. B. Standing, M. B.


“Two and Three Phase Equilibrium Calculations for Systems Volumetric and Phase Behaviour of Oil Field Hydrocarbon
Containing Water”, Can. J. Chem. Eng., Vol. 54, p. 595, Systems, Society of Petroleum Engineers of AIME, Dallas,
December (1976) Texas, 8th Printing (1977)

Petrosky, G. E., and Farshad, F. F. Standing, M. B., and Katz, D. L.


“Pressure-Volume-Temperature Correlations for Gulf of “Density of Natural Gases”, Trans. AIME, Vol. 146, p. 140
Mexico Crude Oils”, Paper No. SPE 26644, presented at the (1942)
68th Ann. Tech. Conf. & Exhib. of the SPE, Dallas, TX,
Sept. (1987) Starling, K.
“Applications of Multiproperty Analysis in Equation of State
Puttagunta, V. R., Miadonye, A., and Singh, B. Development and Thermodynamic Property Prediction”,
“Viscosity-Temperature Correlation for Prediction of paper presented at the 49th NGPA Annual Convention,
Kinematic Viscosity of Conventional Petroleum Liquid”, Denver, Col., March (1970)
Trans. I. Chem. E., Vol. 70, Part A, p. 627, Nov. (1992)
Sutton, R. P.
Redlich, O., and Kwong, J. N. S. “Compressibility Factor for High Molecular Weight
“The Thermodynamics of Solutions. V. An Equation of Reservoir Gases”, Paper SPE 14265, presented at the Ann.
State. Fugacities of Gaseous Solutions”, Chem. Rev., v. 44, Tech. Mtg. and Exhib. of the SPE, Las Vegas, September
p. 233 (1949) (1985)

Reid, R. C., Prausnitz, J. M., and Sherwood, T. K. Sutton, R. P., and Farshad, F.
The Properties of Gases and Liquids, 3rd Edition, McGraw- “Evaluation of Empirically Derived PVT Properties for Gulf
Hill Book Co., New York (1977) of Mexico Crude Oils”, SPE Res. Eng., p. 79, Feb. (1990)

Riazi, M. R., and Daubert, T. E. Thomas, L. K., Hankinson, R. W., and Phillips, K. A.
“Simplify Property Predictions”, Hydrocarbon Processing, p. “Determination of Acoustic Velocity for Natural Gases”, J.
115, March (1980) Petrol. Technol., v. 22, p. 889 (1970)

Robinson, D.B., and Ng, H.-J. Thomson, G. H., Brobst, K. R., and Hankinson, R. W.
"Hydrate Formation and Inhibition in Gas or Gas Condensate “An Improved Correlation for Densities of Compressed
Streams", JCPT, 25, No. 4, p. 26-30 (1986) Liquids and Liquid Mixtures”, AIChE J., Vol. 28, p. 671
(1982)
Robinson, D.B., Ng, H.-J., and Chen, C.-J.
"The Measurement and Prediction of the Formation and Twu, C. H.
Inhibition of Hydrates in Hydrocarbon Systems", Proceedings “Generalized Method for Predicting Viscosities of Petroleum
of the 66th GPA Annual Convention, March (1987) Fractions”, AIChE J., Vol. 32, No. 12, p. 2091 (1986)

Schmidt, G., and Wenzel, H. Twu, C. H., and Bulls, J. W.


“A Modified Van Der Waals Type Equation of State”, Chem. “Viscosity Blending Tested”, Hydrocarbon Proc., p. 217,
Eng. Sci., Vol. 35, p. 1503 (1980) April (1981)

Skovborg, P., Ng, H.-J., Rasmussen, P., and Mohn, U. Van Velzen, D., Cardozo, R. L. and Langenkamp, H.
"Measurement of Induction Times for the Formation of “A Liquid Viscosity-Temperature-Chemical Constitution
Methane and Ethane Gas Hydrates", Chem.Eng. Science Relation for Organic Compounds”, Ind. Eng. Chem. Fund.,
48(3), p. 445-453 (1993) Vol. 11, p. 20 (1972)

Soave, G. Vasquez, M., and Beggs, H. D.


“Equilibrium Constants from a Modified Redlich-Kwong “Correlations for Fluid Physical Property Prediction”, Paper
Equation of State”, Chem. Eng. Sci., Vol. 27, p. 1197 (1972) SPE 6719, presented at the 52nd Annual Technical
Conference and Exhibition, Denver, Col. (1977), Published
Society of Petroleum Engineers in J. Petrol. Technol., p. 968 (1980)
Petroleum Engineering Handbook, Chapter 19, “Crude Oil
Emulsions”, by Smith, H.V., and Arnold, K.E., p. 19-6, Walther, C.
Richardson, Texas (1987) “The Evaluation of Viscosity Data”, Erdöl und Teer, Vol. 7,
p. 382 (1931)
Hydrocarbon Fluid Properties and PVT Behaviour (Revised 05-02) Page 130

Watson, K. M., and Nelson, E. F.


“Improved Methods for Approximating Critical and Thermal
Properties of Petroleum Fractions”, Ind. Eng. Chem., Vol. 25,
p. 880, August (1933)

Whitson, C. H.
“Characterizing Hydrocarbon Plus Fractions”. SPE J., p. 683,
August (1983)

Wichert, E., and Aziz, K.


Compressibility Factor of Sour Natural Gases”, Can. J.
Chem. Eng., Vol. 49, p. 267, April (1971)

Wichert, E., and Aziz, K.


“Calculated Z’s for Sour Gases”, Hydrocarbons Processing,
p. 119, May (1972)

Woelflin, W.
“Viscosity of Crude Oil Emulsions”, Oil and Gas J., Vol. 40,
No. 45, p. 35, March 19 (1942)

Wright, W. A.
J. Materials, Vol. 4, No. 1, p.19 (1969)

Zuo, J., Zhang, D., and Ng, H.-J.


"Representation of Hydrate Phase Equilibria in Aqueous
Solutions of Methanol and Electrolytes Using an Equation of
State", Energy & Fuels, 14, p. 19-24 (2000)

You might also like