SPE 122827

Inaccurate Gas Viscosity at HPHT Conditions and Its Effect on

Unconventional Gas Reserves Estimation
Ehsan Davani, SPE, Kegang Ling, SPE, Catalin Teodoriu, SPE, William D. McCain Jr, SPE, Gioia Falcone, SPE,
Texas A&M University

Copyright 2009, Society of Petroleum Engineers

This paper was prepared for presentation at the 2009 SPE Latin American and Caribbean Petroleum Engineering Conference held in Cartagena, Colombia, 31 May–3 June 2009.

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

Abstract

Viscosity is one of the gas properties that is always uncertain at high-pressure and high-temperature (HPHT) conditions
because no reliable correlation has been developed for HPHT and accurate measurement of gas viscosities at HPHT is
difficult. However, the industry currently does not pay sufficient attention to the need for accurate gas viscosity values in
HPHT gas fields’ development.
We show that small errors in gas viscosity affect the inflow performance relationship (IPR) curves and eventually change
the reserves estimate for HPHT conditions in ways that can drastically influence production forecasting. A sensitivity analysis
for a synthetic case at these conditions shows that a 1% error in gas viscosity measurement results in a 1% error in gas
flowrate, which in a large scale of gas production may severely underestimate or overestimate recovery from an HPHT field.
This linearity between gas viscosity and gas flowrate continues to ±10% errors.
An evaluation of gas viscosity correlations commonly used in the industry shows that none of them is reliable at HPHT
conditions, mainly for two reasons: measured data are not available to confirm their consistency and they do not account for
impurities in the gas.
Clearly, the industry needs accurate gas viscosity measurements and needs a new procedure to develop correlations based
upon accurate HPHT gas viscosity data.

Introduction

Gas production worldwide has been steadily increasing over the past 15 years, but the world’s natural gas reserves-to-
production ratio declined to 60.3 years in 2007, even though reserves increased by more than 1 trillion cubic meters (BP,
2008). This growing demand for natural gas is driving the E&P industry to look for new resources in previously unexplored
and deeper areas, where HPHT reservoirs may be encountered. These are reservoirs that exhibit pressures greater than 10,000
psia and temperatures over 300ºF. Thus, more attention must be paid to gas properties at HPHT conditions.
Among all gas properties, viscosity still remains uncertain and is usually characterized by some available correlation.
Values of gas viscosities are used to model gas mobility in the reservoir and this has a significant impact on reserves
estimation during field development planning. At these tremendous pressures and temperatures, the reservoir fluids will be
very lean gases, typically methane with some degree of impurity, and therefore the gas properties may behave totally
differently in comparison with gases of lower pressures and temperatures.
A large database was reviewed for this research, which revealed that most of the published data on the viscosity of pure
methane or impure methane are extremely limited in terms of both conditions and quantity, and in some cases their accuracy is
unknown. All of the currently available viscosity correlations were developed from measured data at low to moderate
pressures and temperatures, and their applicability at HPHT conditions is doubtful. However, it is important to note that any
correlation is only as good as the data on which it is based, but because of the simplicity of current correlations and
unavailability of accurately measured data for HPHT conditions, the petroleum industry applies them with some degree of
confidence. This leads to erroneous estimates of gas viscosities and to major errors in reservoir engineering calculations.
Generally, the problem can be overcome in two ways: by improving the available gas viscosity measurement equipment
and the procedures for use at HPHT conditions, and by developing new correlations for HPHT supported by accurately
measured data.
2 SPE 122827

Among the viscosity correlations that are most commonly used in the industry, the following are discussed here: the Carr,
Kobayashi and Burrrow correlation, and the Lee, Gonzalez and Eakin correlation. In addition, the National Institute of
Standards and Technology (NIST) have developed a computer program to predict thermodynamic and transport properties of
hydrocarbon fluids. We used viscosity values of methane and nitrogen to evaluate the reliability of current correlations at
HPHT and also the capability of these equations to get accurate results in gas reservoirs with high quantities of impurities.
Then, we conducted a sensitivity analysis to show how small errors in values of gas viscosity can affect the estimated
cumulative recovery from a reservoir.

Laboratory Measurement of Gas Viscosity

Viswanathan (2007) showed that many techniques are available to measure gas viscosities, however just a few of them are
applicable at HPHT conditions. The following viscometers have been used to measure gas viscosities:

• Rolling ball viscometer

• Capillary tube viscometer
• Falling body viscometer
• Vibrating wire viscometer

All of these are also used to measure liquid viscosities, but because of the low densities and very low viscosities of gases,
they need to be adjusted to be applicable. Rich natural gases usually contain some heavier components, making the viscosity
measurement of these gases relatively easier than for lean gases due to lubricant effect of those heavy components. At HPHT
conditions, most gas reservoirs contain lean gases, essentially methane, and the dry nature of lean gas creates erratic friction
through the measurement process that obstructs the experiment. Low viscosities make the effect of friction significant, and
most of the basic formulations for each instrument should be modified to account for this complication.

Correlations

Several well-known correlations are used in the petroleum industry to determine values of gas viscosity. Here, we discuss the
most commonly adopted correlations, namely the Carr, Kobayashi and Burrrow correlation (1954), and the Lee, Gonzalez and
Eakin correlation (1966). In addition, NIST (2008) developed a computer program that predicts thermodynamic and transport
properties of hydrocarbon fluids. Comparing the values from correlations with those by NIST gives an insight into the current
understanding of gas viscosity correlations.

Carr, Kobayashi, and Burrrow correlation

The Carr, Kobayashi and Burrrow correlation includes two steps to estimate natural gas viscosity. First, a value of gas
viscosity at atmospheric pressure is determined and some corrections have to be applied to account for the effect of impurities
such as N2, CO2 and H2S on natural gas viscosity (Fig. 1). Dempsey (1965) developed a fit for Fig 1 and 3, but it does not
precisely represent the data given by the correlation. The value of viscosity at atmospheric conditions is adjusted to reservoir
conditions by corrections based on the charts presented in Figs. 2 and 3. A value of gas viscosity ratio (µg/µ1atm) is estimated
from either of these figures and used to adjust the value of gas viscosity at atmospheric pressure to reservoir conditions.

µ1atm = ( µ1atm uncorrected) + ( N 2 correction) + (CO2 correction) + ( H 2 S correction)

............................................................... (1)
−5 −6 −3 −3
µ1atm uncorrected = (1.709X10 − 2.062X10 γ g )T + 8.188X10 − 6.15X10 log γ g
........................................................ (2)
 − 3
N 2 correction = y N (8.48X10 ) log γ + 9.59X10 − 3 
2
 g  .................................................................................................................... (3)

CO2 correction = yCO (9.08X10 − 3 ) log γ + 6.24 X10 − 3 

2
 g  .................................................................................................................... (4)
H 2 S correction = y H S (8.49X10 − 3 ) log γ + 3.73X10 − 3 
2  g  ............................................................................................................ (5)

The advantage of the Carr et al. correlation is that it can correct for the effect of nonhydrocarbons that significantly affect
the gas viscosity. According to Carr et al., it also gives a 1.5% absolute error in estimating gas viscosity.
SPE 122827 3

Fig. 1—Viscosity of hydrocarbon gas mixture at atmospheric pressure and correction for N2, CO2 and H2S, from Carr et al (1954).

Fig. 2—Viscosity ratio vs. pseudo-reduced pressure for Fig. 3—Viscosity ratio vs. pseudo-reduced temperature for
hydrocarbon gases, from Carr et al (1954). hydrocarbon gases, from Carr et al (1954).
4 SPE 122827

One limitation of the Carr et al. correlation is that it is based on a limited (30) number of data points. These data points are
related to measurements of viscosities of pure gases, such as nitrogen, carbon dioxide, methane, ethane, ethylene, propane, and
a few mixture samples. Thus, the universal accuracy of this correlation for samples from real hydrocarbon reservoirs still
needs to be proved. Carr et al. claimed that their correlation is appropriate for gas specific gravities from 0.55 to 1.55 and
temperatures from 100 to 300oF.
Another inconvenience of the Carr et al. correlation comes from the use of charts to get gas viscosity at the conditions of
interest. To date, there is no satisfactory curve fit for Figs. 2 , which has not been digitized for use in computers over theirs
entire range.

Lee, Gonzalez, and Eakin correlation

Lee et al. (1966) derived correlations that are simpler than that by Carr et al. to predict gas viscosities at reservoir conditions.
The input data required in the Lee et al. correlation are gas density, gas molecular weight and temperature. Values of these
input data can be readily obtained if reservoir temperature and pressure and gas specific gravity are known. Eqs. 6 through 9
are used to get gas viscosity, provided that the molecular weight and density at the relevant conditions are known.

−4 Y
µ g = 10 K exp( X ρ g )
............................................................................................................................................................ (6)
1.5
(9.379 + 0.01607 M w )T
K=
209.2 + 19.26 M w + T
..................................................................................................................................................... (7)
where
 986.4  + 0.01009M
X = 3.448 +
 T  w
................................................................................................................................................... (8)
Y = 2.447 − 0.2224 X ................................................................................................................................................................ (9)

The correlation by Lee et al. fit their 234 data points to within a 2.0% average absolute error at low pressure and 4% at
high pressure. It provides better results if the gas specific gravity is less than unity. The correlation can be used to forecast gas
viscosities at temperatures from 100 to 340oF and pressures from 100 to 8,000 psi.
However, this correlation does not consider natural gases that contain high quantities of nonhydrocarbon components. We
used viscosity data of pure nitrogen and also our lab data to test the validity of the Lee et al. correlation equations in predicting
nonhydrocarbon viscosities.
The viscometer used for the measurements is a modified falling ball viscometer (the Cambridge Viscosity SPL440
viscometer). It is a piston-style viscometer designed exclusively for measuring the viscosity of petroleum fluids, oils and
gases. The measurable range of the gas viscosity is from 0.02 to 0.2 cp.
As expected, the correlation by Lee et al. is unable to forecast nonhydrocarbon viscosities. Laboratory experiments of
nitrogen viscosities and NIST values of the same illustrate that this correlation is inappropriate for gases with high nitrogen
content (Fig. 5).
Jeje and Mattar (2004) compared the Lee et al. correlation to the Carr et al. correlation for both sweet and sour gases. They
found that, for sweet gases, both give almost identical results. However, for sour gases, the differences between the Lee et al.
correlation and Carr et al. correlation can be significant and the former, which was derived for sweet gases only, should not be
applied to sour gases.
Viswanathan (2007) used NIST Value of pure methane at pressures from 5,000 to 30,000 psi and temperatures from 100 to
400oF. Based on them, Viswanathan modified the correlation by Lee et al. as follows:

µ g = K exp Xρ g Y ( ) .............................................................................................................................................................. (10)

1.832
0.0001 * (5.0512 − 0.2888 M w )T
K = ........................................................................................................................... (11)
− 443.8 + 12.9 M w + T

where

X = −6.1166 +
 3084.9437  + 0.3938 M ............................................................................................................................. (12)
 T  w

Y = 0.5893 − 0.1563 X ............................................................................................................................................................. (13)

SPE 122827 5

Viswanathan then compared the modified Lee et al. correlation to the NIST values for pure methane. The results showed a
very good match with the NIST values compared to the performance of the original correlation by Lee et al. (Fig.4). However,
Viswanathan’s findings cannot be directly extrapolated to situations where impurities are present in the gas.

Fig. 4— As modified by Viswanathan (2007), the Lee, Gonzalez, and Eakin correlation accurately reproduces the NIST viscosity values
at 300 oF.

Nonhydrocarbons such as nitrogen, carbon dioxide, and hydrogen sulfide are often found in natural gases. As deeper wells
are drilled, more nonhydrocarbons are likely to be found in the gas, with some reservoirs reaching up to 50 or 60% in
nonhydrocarbon content. Gas reservoirs in the South China Sea contain 55 to 60 vol% of carbon dioxide and hydrogen sulfide,
while sour gas reservoirs in East China Sea have been discovered with 20 to 30 vol% carbon dioxide. The development and
production of sour gas reservoirs requires a gas viscosity correlation that can take into account the effect of these
nonhydrocarbon compounds.
The data upon which the Lee et al. correlation is based cover a relatively limited pressure range. The application of this
correlation for gas reservoir pressures greater than 8000 psi is uncertain. As deeper reservoirs are discovered, under HPHT
conditions, the major component found in the gas fluids is methane. Thus, a model that predicts methane viscosities at HPHT
conditions is desirable.

National Institute of Standards and Technology

NIST (2008) provides a program to predict thermodynamic and transport properties of hydrocarbons. Not only pure
components, but also mixtures containing up to 20 components can be used with the program to obtain thermodynamic
properties. The ranges of pressures and temperatures are up to 44,100 psi and 1,340o F. However, the program predicts the fluid
properties using correlations developed at low to moderate pressures and temperatures. Unfortunately, the validity of
extrapolating correlations that are derived from low to moderate pressure and temperature data to predict gas viscosities for
HPHT fields is doubtful
6 SPE 122827

Nitrogen viscosity as a function of pressure at T=116°oF

F.

0.065

0.055
Viscosity,cp

0.045

0.035

Lab Data
0.025
Lee et al. Correlation

NIST Values
0.015
0

3000

6000

9000

12000

15000

18000

21000
Pressure, psi

Nitrogen viscosity as a function of pressure at T=152 °

oF
F.

0.065

0.055
Viscosity, cp

0.045

0.035

Lab Data
0.025
Lee et al. Correlation

NIST Values
0.015
0

3000

6000

9000

12000

15000

18000

21000

24000

Pressure, psi

Fig. 5— The Lee et al. correlation does not accurately predict the viscosity of nitrogen gas.
SPE 122827 7

Viscosity effect on the IPR: sensitivity analysis

A Cartesian plot of bottomhole pressure versus surface flow rate is used to construct an IPR curve, which reflects the capacity
of a gas reservoir to deliver fluid to the wellbore. The IPR curve is a diagnostic tool used in the industry to determine the
performance of a flowing system.
The exact solution to the differential form of Darcy’s law for compressible fluids under pseudo steady-state flow conditions is
(Ahmed and McKinney, 2005):

qg =
(
kh ψ r − ψ wf )
 r  
1422T  Ln e  − 0.75 + s 
  rw   ..................................................................................................................................................... (14)
where

pr p
ψ r = 2∫ dp
0 µZ ........................................................................................................................................................................... (15)
p wf p
ψ wf = 2 ∫ dp
0 µZ ...................................................................................................................................................................... (16)

This research used the trapezoidal method to solve the integrals in Eqs 10 and 11 for a synthetic case. Because the case is
representative of HPHT conditions, it is not unrealistic to assume that the hydrocarbon reservoir fluid is pure methane. Since
an accurate correlation to predict the viscosity of mixed gases at such conditions does not exist, we constrained the model to
pure methane. Obviously, the problem becomes more significant when the reservoir gases are gas mixtures, such as methane
with some quantities of ethane, propane, carbon dioxide, nitrogen and/or hydrogen sulfide.

IPR curves were calculated using the viscosity and density of pure methane from the NIST tables for the selected pressure
and temperature conditions. The original NIST gas viscosity values were then perturbated by ± 1% to ± 10%, and the IPR
curves re-calculated.
The reservoir parameters used for the synthetic case are summarized in Table 1. It will be seen from Fig. 6 that small
changes in gas viscosity can have a dramatic impact on the IPR’s.

Table 1— Reservoir parameters

o
Temperature 400 F
Initial Pressure 15000 psi
Permeability 65 md
Reservoir radius 1000 ft
Wellbore radius 0.25 ft
Thickness 15 ft
Skin -0.4

Fig. 7 shows the changes in calculated gas flow rates as the values of gas viscosities vary. A 10% error in the estimated
gas viscosity causes a change of approximately 10% in the predicted value of gas flow rate. An interesting result is that
underestimating the gas viscosity gives slightly worse results than overestimating the gas viscosity.
As the predicted gas flowrate would then be used in production forecasting, an error in viscosity estimation would be
reflected in the accuracy of ultimate field recovery calculations, with a potential for under /over estimation of the reserves.
Hernandez et al. (2002) highlighted the impact of hydrocarbon viscosity prediction on the accuracy of reservoir
engineering calculations. They focused on oil viscosity, not gas viscosity, and concluded that, for light oils, a ±10% error in
viscosity leads to a ±1-2% error in the prediction of cumulative production; for heavy oils, the same viscosity error results in a
±10% error in cumulative production.
8 SPE 122827

Pi=15000, T=400 °F.

16000

14000

12000
Pressure, psi

10000

8000

6000

4000

2000

0
0.E+00

1.E+05

2.E+05

3.E+05

4.E+05

5.E+05

6.E+05

7.E+05

True Viscosity
True Viscosity -4% error
True Viscosity -8% error
True Viscosity +2% error
True Viscosity +6% error
True Viscosity +10% error
Gas production rate,q, scf/day
True Viscosity -1% error
True Viscosity -5% error
True Viscosity -9% error
True Viscosity +3% error
True Viscosity +7% error
True Viscosity -2% error
True Viscosity -6% error
True Viscosity -10% error
True Viscosity +4% error
True Viscosity +8% error
8.E+05
True Viscosity -3% error
True Viscosity -7% error
True Viscosity +1% error
True Viscosity +5% error
True Viscosity +9% error

Fig. 6— Small errors in gas viscosity considerably shift the IPR curve to the left or right.

15

10
Gas production rate Change, %

0
-15 -10 -5 0 5 10 15

-5

-10

-15
Viscosity calculation error, %

Fig. 7— Flowrate change relates directly to uncertainties in gas viscosity.

SPE 122827 9

Conclusions

Accurate measurements of natural gas viscosity under HPHT conditions are yet to be obtained.

Gas viscosity correlations derived from data obtained at low to moderate pressures and temperatures cannot be confidently
extrapolated to HPHT conditions.

The gas viscosity correlations that are currently available to the industry were derived from data obtained with gases with
limited impurities, and so their accuracy for use with gases containing large quantities of impurities is unknown.

A 1% error in the estimate of gas viscosity can cause a 1% error in predicted gas flowrate; this relationship is consistently
linear up to ±10%. If the predicted gas flowrates is then used in production forecasting to estimate the ultimate field recovery,
this will add uncertainty to the gas reserves estimation.

There is a need to elaborate a methodology for providing accurate measurements of gas viscosity at HPHT conditions, and also
to develop a new viscosity correlation based on HPHT measurements.

Nomenclature

h= thickness, ft
k= permeability, md
MW= molecular weight
qg = gas flow rate, Mscf/day
re = drainage radius, ft
rw = wellbore radius, ft
s= skin factor
T= temperature, ˚R
y N 2 ,CO2 , H 2 S = mole fraction of the non-hydrocarbon component
γg = gas specific gravity (air=1)
µ1atm= gas viscosity at 1 atm, cp
µg = gas viscosity, cp
ρg= gas density, g/cc
ψr= average reservoir real-gas pseudo-pressure, psi2/cp

References

BP, 2008. Statistical Review of World Energy. June 2008. www.bp.com. Downloaded 15 February 2009.

Carr, N.L., Kobayashi, R., and Burrows, D.B. 1954. Viscosity of Hydrocarbon Gases Under Pressure. Trans., AIME 201: 264-272

Dempsey, J.R.1965. Computer Routine Treats Gas Viscosity as a Variable. Oil and Gas J. (16 August 1965) 141-143.

Jeje, O., and Mattar, L. 2004. Comparison of Correlations for Viscosity of Sour Natural Gas. Paper 2004-2004-214 presented at Petroleum
Society’s 5th Canadian International Petroleum Conference (55th Annual Technical Meeting), Calgary, Alberta, The Canada, 8 – 10. June.

Hernandez, J.C., Vesovic, V., Carter, J.N., and Lopez. E. 2002. Sensitivity of Reservoir Simulations to Uncertainties in Viscosities. Paper
SPE 75227 presented at the SPE/DOE IOR Symposium, Tulsa, Oklahoma, 13–17. April. SPE-75227.

NIST. Web-Book of Thermodynamic Properties of Fluids. National Institute of Science and Technology (NIST).
http://webbook.nist.gov,chemisty/fluid. Downloaded 11 November 2008.

Lee, A.L., Gonzalez, M.H., and Eakin, B.E. 1966. The Viscosity of Natural Gases. Paper SPE 1340 presented at the SPE Annual Technical
Conference and Exhibition, Shreveport, Louisiana, 11–12. November. SPE-1340-MS.

Ahmed, T., and McKinney, P. 2005. Advanced Reservoir Engineering, 188. Amsterdam, The Netherlands: Gulf Professional Publishing.

Viswanathan, A. 2007. Viscosities of Natural Gases at High Pressures and High Temperatures. MS thesis, Texas A&M University, College
Station, Texas.