GERG-PROJECT 1.

52 COMPARING AND DEFINING A

RELATION BETWEEN EXPERIMENTAL AND CALCULATING
TECHNIQUES FOR HYDROCARBON DEW-POINT

Main author
Angeles YACKOW
GDF SUEZ
FRANCE
IGRC2008@gasunie.nl
Co-author
A. LAUGHTON (Advantica)
U. GRONEMANN (BEB)
A. BENITO (Enagas)
T. LINDGREN (Energinet.dk)
E. KUKOVA (E.ON. Ruhrgas)
H. KAESLER (E.ON. Ruhrgas)
P. VAN CANEGHEM (Fluxys)
S. SOLVANG (Gassco)
H. J. PANNEMAN (Gasunie)
A. PIRON (GDF SUEZ)
D. PAPWORTH (Interconnector)
F. MAGHINI (Snam Rete Gas)
B. VIGLIETTI (Snam Rete Gas)
E. SOLBRAA (StatoilHydro)
C. POSPIECH (Wingas)

Page 2 of 20

1.

ABSTRACT

The natural gas industry has identified a need both for current operation and for future gas
supplies: the accurate prediction and/or measurement of the hydrocarbon dew-point (HCDP).
The HCDP is the temperature and pressure at which heavy components of the stream cause the
formation of a liquid phase. To prevent the condensation phenomena, specifications concerning
the HCDP have been enforced in many countries. In the new context of a deregulated gas
market in Europe, the HCDP might become a key specification. However, it is still very difficult
to determine HCDP unambiguously, since no standard method is available.
Two techniques are actually being used to find the natural gas HCDP:
1. HCDP estimation based on the analysis of gas composition using a gas chromatograph
combined with a calculation method based upon equations of state (EOS).
2. Manual or automatic devices specially designed for HCDP measurements. These devices
use the chilled mirror principle, whereby the sampled gas passes a mirror surface that is
cooled continuously thus causing the hydrocarbon condensation.
The scope of the GERG WG 1.52 was to compare and define a relation between the
experimental and calculating techniques for the determination of HCDP. The project consisted of
3 consecutive phases and each individual phase was defined by a different HCDP determination
method. Measurements and/or calculations were made with real gases (all European gases)
and, only during Phase 1, with binary and synthetic mixtures. The aim of phase 1 was to
investigate and to compare commercial software packages used for the determination of phase
envelopes and thus the HCDP. The objective of the second phase was to compare HCDP
measured with a manual chilled mirror with predicted values calculated from phase envelopes.
Finally, Phase 3 allowed us to compare different HCDP automatic measurement devices.
The obtained results show that all HCDP direct measurement methods are comparable to 2-4C,
while calculating HCDPs using EOS models require a proper characterization of the C6+ fraction.

Copyright 2008 IGRC2008

CONTENTS

1.

Abstract ................................................................................................. 2

2.

Introduction........................................................................................... 4

3.

Phase 1 Study of commercially available HCDP calculation software ... 4

3.1.

Sample definition and characterization..................................................................5

3.2.

Influence of the gas analysis uncertainty on the HCDP calculation .............................5

3.3.

Lumping Methods ..............................................................................................5

3.4.

Phase envelope calculation Property Package Selection .........................................7

3.5.

Phase 1 discussion .............................................................................................7

4.
Phase 2 Manual Chilled mirror Device HCDP Measurements versus
Calculated values ........................................................................................... 8
4.1.

Gas sampling and analysis ..................................................................................8

4.2.

Manual Chilled Mirror Device ...............................................................................9

4.3.

HCDP Calculation Procedure .............................................................................. 10

4.4.

Experimental HCDP data versus calculated values ................................................ 11

4.4.1.
4.4.2.
4.4.3.
4.5.

Phase envelopes with the default lumping method .......................................... 11

Influence of the way of handling isomers ...................................................... 12
GC data treated according to ISO 23874 ....................................................... 12

Phase 2 discussion ........................................................................................... 13

5.
Phase 3 Automatic Device HCDP Measurements versus Calculated
values........................................................................................................... 13
5.1.

Materials ........................................................................................................ 13

5.2.

Natural gas samples and RGM ........................................................................... 14

5.3.

Description of the rig........................................................................................ 14

5.3.1.
5.3.2.
5.4.

Phase 3 measurements..................................................................................... 16

5.4.1.
5.5.

Loading and emptying procedures ................................................................ 15

Measuring procedure .................................................................................. 15
Emerson process GC TCD measurements.................................................... 17

Phase 3 discussion ........................................................................................... 17

6.

GERG 1.52 general discussion .............................................................. 18

7.

References ........................................................................................... 19

14.

List Of Tables ....................................................................................... 19

15.

List Of Figures ...................................................................................... 19

16.

Acronyms ............................................................................................. 20

2.

INTRODUCTION

The natural gas industry has identified a need both for current operation and for future gas
supplies: the accurate prediction and/or measurement of the hydrocarbon dew-point (HCDP).
The HCDP is the temperature and pressure at which heavy components of the gas stream cause
the formation of a liquid phase. Two different sources of hydrocarbon liquids can be
distinguished. The first source of hydrocarbon liquids can be lubricants and/or seal-oil
originating from the compressors used in the transmission system. The other source is the
condensation of heavy hydrocarbons from the natural gas itself. To prevent the condensation
phenomena, specifications concerning the HCDP have been enforced in many countries. In the
new context of a deregulated gas market in Europe, the HCDP might become a key
specification. However, it is still very difficult to determine HCDP unambiguously, since no
standard method is available.
The HCDP can be determined either by direct measurement or by calculation from gas
composition data. Measurement methods are based on the chilled mirror principle, in which the
sampled gas passes by a mirror surface that is cooled continuously. The HCDP is reached when
the very first amount of hydrocarbon liquid is detected either visually (manual devices) or
optically (automatic devices) on the mirror surface. However, the HCDP determined by such a
method is only available at a fixed measuring pressure and, when the manual devices are used,
we must also take into account the operators effect. Calculation methods enable a complete
phase envelope to be drawn, thus specifying all the existing HCDP at different pressures, but
are more complex. First, it is necessary to have a detailed analysis of the natural gas available
in which the heavy hydrocarbons (C11, C12) are given at a sub-ppm level. Furthermore, the
way of handling isomers of heavy hydrocarbons is crucial for the outcome of the calculation
procedure. Finally, the phase envelope calculation is also influenced by the physical model and
the thermodynamic parameters used, e.g. different software packages and different equations
of state (EOS) within them, different methods used for the calculation of densities and the
binary interaction coefficients, etc.
The GERG working group 1.52 was created to conduct a study to compare the results of
experimental and calculating techniques for HCDP and, if possible, to define a relation between
both techniques. GERG 1.52 was made up of 13 European gas companies: Advantica (UK), BEB
(Germany), Enagas (Spain), E.ON Ruhrgas (Germany), Fluxys (Belgium), Gassco (Norway),
Energinet.dk (Denmark), Gasunie (Netherlands), GDF SUEZ (France), Interconnector (UK),
Snam Rete Gas (Italy), StatoilHydro (Norway), Wingas (Germany). GDF SUEZ was the convenor
of this working group.
GERG 1.52 project consisted of 3 consecutive phases. The aim of Phase 1 was to investigate
and to compare commercial software packages used for the determination of phase envelopes
and thus the HCDP. The objective of the second phase was to compare HCDP, measured with a
manual chilled mirror, with calculated values obtained from phase envelopes. Finally, Phase 3
allowed us to compare different HCDP automatic measurement devices.
This paper presents the main results of the GERG 1.52 project.

3.

PHASE 1 STUDY OF COMMERCIALLY AVAILABLE HCDP

CALCULATION SOFTWARE

The aim of Phase 1 was to define how to choose the best software(s) for calculating HCDP and
the following software packages were tested:

GasVLe 5.0 Advantica

Multiflash 3.3.57 Infochem

PVTSim 13.1 (for pure gases, binary and synthetic mixtures) & 14.0 (for real gases)
Calsep

Gaspack 1.5 GDF SUEZ


PVTp 6.5 Petroleum Experts

Aspen HYSYS 3.2 Hyprotech

AGA Program Snam Rete Gas

PRO/II 7.2 Sim. Sci. Inc.

At the beginning, a comprehensive listing allowed us to define all the current difficulties
encountered when calculating phase envelopes from given natural gas compositions. An
ordinary software user does not usually know the optimum way to perform such a calculation.
Some questions readily arise, e.g. "Which equation of state is the best suited to predict the
HCDP correctly?" or "How the various isomers in the gas analyses should be grouped?".
Since no widely accepted recipe is available to perform these calculations, very different results
can be obtained. Several commercially available software programs were examined and a test
methodology was applied: full technical details of the software packages, calculations were
performed with different types of mixtures, lumping methods were used and the influence of
several specific parameters was studied in detail. The most important aspects of this analysis
are described in the following paragraphs.

3.1.

Sample definition and characterization

Calculations were completed with three different kinds of mixtures binaries, synthetic mixtures
and real gases besides pure gases. For validating these calculations, available experimental
HCDP data was used: literature data for pure compounds[1], some of the binary mixtures[10] and
manual chilled mirror measurements performed by StatoilHydro for the multi-component
mixtures and real gases.
As for the synthetic mixtures the compositions were known, the first step of this study was to
characterize the real gases. The 6 real gases used during all GERG 1.52 were sampled by some
of the participating companies mentioned above. These gases represent the spectrum of the
natural gases transmitted in Europe and the maximum hydrocarbon dew points are in the range
30C and +20C.

3.2.

Influence of the gas analysis uncertainty on the HCDP calculation

The uncertainty values used in this exercise are based on the experience of the participating
companies. The uncertainties in Table 1 have been obtained by considering the maximum
uncertainty that could be encountered in the given concentration range. The uncertainty
estimates presented here are worst-case values. However, it should be noted that by using a
non-suitable sampling technique, it is quite easy to introduce much larger deviations in the
analysis.
Table 1: Estimated uncertainty for
each concentration range

3.3.

Concentration range
(ppm mol)

Estimated
uncertainty

<1

100%

1 to 10

20%

10 to 40

10%

40 to 100

5%

100 to 1000

2%

Lumping Methods

It is common practice in hydrocarbon gas chromatographic analysis to group isomers according

to their carbon number. In this work, isomers of the normal alkane with i carbons or nCi
represent all the eluted compounds by gas chromatography (with a non-polar column) between
the fractions nCi-1 and nCi.
Before comparing different lumping methods, some assumptions had to be made to establish a
common ground :

O2, H2, He and Ar have no or little influence on the HCDP determination. If their
concentrations were available, they have been intentionally added to N2 fraction.
N2, C1, CO2, C2, C3, i- and n-C4,i-, n- and neo-C5 are considered separately.
Other well-known isomers have been treated separately, if these components were
included in the detailed analysis: cyclopentane, cyclohexane, methylcyclohexane,
benzene and toluene.
The normal alkane nCi and its isomers (for carbon number i > 5) were lumped into the
normal alkane fraction or nCi

When choosing a lumping method, several issues must be addressed, for example :
1. Influence of aromatic and cyclic compounds: For each sample containing aromatic
compounds, two calculations were performed: either the aromatics are treated
separately; or these components are lumped into the nC7 (for benzene) and the nC8
(toluene) fractions. It seems that there is no significant influence on the outcome if their
concentration is below 70 ppm, since the observed difference in the cricondentherm
values is less than 0.5C. Cyclic compounds seem to have a larger influence compared
to the aromatics, regardless of their concentrations. However, these observations
should be taken as a partial conclusion since results depend on the complete gas
composition (contents of C6, C7, and heavy hydrocarbons: C10, C11). Given the possible
influence of aromatic and cyclic components on the final cricondentherm value, it is
advised to choose software packages that allow benzene, toluene, cyclohexane and
methylcyclohexane to be treated separately, within their own specific component
fraction, or as a users defined fraction an hypothetical component fraction created by
the user.
2. Influence of heavy hydrocarbons: To study the influence of the heavy hydrocarbons,
various artificial compositions were devised from the original real gas composition as
follows: truncated basic composition up to Cn and heavy fraction Cn+ with n taking
values from 7 up to the highest hydrocarbon analysed. The different trends showing the
change of the cricondentherm according to the heaviest hydrocarbon group analysed
are given in Figure 1. All software packages deliver similar results: the results are given
for only one of them. Although the results depend heavily on the composition of each
real gas, it is not sufficient to perform gas analysis only up to C8 when calculating the
HCDP. It is recommended to extend the gas analysis up to the sub-ppm level, e.g. when
the concentrations of two consecutive heavy hydrocarbons (probably C11 and C12) are
both below 1 ppm.

40

30

20

T of cricondentherm (C)

10

0
5

10

11

12

13

14

-10

Gas 1

Gas 2

-20

Gas 3

Gas 4

-30

Gas 5

Gas 6
-40

-50
No. of C atoms

Figure 1: Cricondentherm as a function of the highest carbon group used for the HCDP
calculation (PR EOS for one software)

3.4.

Phase envelope calculation Property Package Selection

All the selected software packages in this study use the same kind of property models for
calculating phase envelopes : an equation of state (EOS) based method.
In this work, calculations were made with the two most popular EOS (if available in the selected
software) : the Peng-Robinson EOS (PR EOS) and the Redlich-Kwong Soave (RKS EOS). For real
gases, the difference between the calculated cricondentherms values with the RKS EOS and with
the PR EOS varies between 2.5C and 4C. Other factors, besides the EOS choice, such as the
exact formulation of the EOS used algorithm, internal parameters, component parameters,
etc., the physical properties of input compounds, the additional equations used to calculate
densities, etc also influence the results of phase envelopes calculations.
Using the RKS EOS of state always results in higher cricondentherm and cricondenbar values.
The temperature difference depends on the gas composition and seems to be higher for gases
containing heavy hydrocarbons. The RKS EOS seems to be the more suitable EOS for very lean
gases, as already mentioned in literature[12]. In this paper, only the results obtained with the PR
EOS will be presented.

3.5.

Phase 1 discussion

The key parameters that influence the phase envelope calculation of natural gases have been
identified. The main conclusions of the GERG 1.52 Phase 1 are:

It is important to be able to quantify the heaviest hydrocarbons in the natural gas at a

level of 0.1-0.2 ppm mol for HCDP calculations. Components should be taken into
account up to the last two consecutive heavy hydrocarbons (probably C11 and C12) with
concentrations below 1 ppm.
The uncertainty in the gas composition resulting from the chromatographic analysis has
only a minor influence on the outcome of HCDP calculations.
The
aromatic
and
cyclic
compounds
(benzene,
toluene,
cyclohexane,
methylcyclohexane) might have significant influence on the HCDP calculation.

The most popular physical models for calculating phase envelopes, the PR EOS and the
RKS EOS, show a significant and rather constant difference : cricondentherm values
calculated with the PR EOS are normally 2 - 3 C lower than the RKS EOS estimations.
Results are similar for cricondenbars (2-3 bars differences). PR EOS has been chosen to
calculate HCDPs in phase 2 and 3 of this project.
All software packages evaluated during Phase 1 show the same average behavior when
all the discussed key parameters are adjusted in a similar way. Consequently, only the
software packages that permit aromatic and cyclic compounds to be treated separately,
implement both PR and RKS EOS and include heavy hydrocarbons were retained for
Phases 2 and 3 of the GERG WG 1.52 project.

4.

PHASE 2 MANUAL CHILLED MIRROR DEVICE HCDP

MEASUREMENTS VERSUS CALCULATED VALUES

Phase 2 of the GERG 1.52 Project comprised sampling and analyzing the same 6 natural gases
used during Phase 1, with the purpose of comparing and qualifying the agreement between
HCDP measurements and predictions.

4.1.

Gas sampling and analysis

The gas sampling was mostly performed according to ISO 10715[7]. After sampling, both
StatoilHydro and the sampling gas company performed GC analyses. These two analyses
allowed us to define: (1) the influence of transportation on the gas composition and (2) the
influence of gas analysis uncertainty on the HCDP prediction.
To avoid having transportation effects reflected in the gas analysis, precautions were taken
and the samples were allow to warm up, and agitated bottles were rolled before being
analysed.
All compositions were determined by direct analysis, and no pre-concentration methods were
used. Three companies (StatoilHydro, GDF SUEZ, and E.ON. Ruhrgas) have developed their own
methods to perform detailed analysis dedicated to HCDP calculations up to C12 or higher
hydrocarbons, whereas Fluxys follow the standard ISO 6975[6]. The basic analysis
requirements were:

Analysis of N2, CO2, methane, ethane, propane, i-butane, n-butane, i-pentane and npentane, each of them separately.

Analysis of aromatics and cyclics: benzene, cyclohexane, methylcyclohexane and

toluene.

Analysis of heavy hydrocarbons (at least normal alkanes higher than nC6 and if possible
isomers with a limit of detection of 1 ppm or lower).
Generally, the analyses performed by the different sampling companies and StatoilHydro
analysis were in agreement and gas compositions were not affected by transport. In Table 2 we
show the StatoilHydro gas analyses for the GERG 1.52 Phase 2.

Table 2: GERG 1.52 real gases used during Phase 2 (StatoilHydro data)
Gas 1 mol % Gas 2 mol % Gas 3 mol % Gas 4 mol % Gas 5 mol % Gas 6 mol %
He

H2

N2

3.987

1.48

14.21

1.899

0.76

1.46

CO2

0.428

0.729

0.965

1.032

0.043

0.787

C1

86.67

92.39

81.40

90.49

98.07

92.70

C2

6.83

4.144

2.811

4.824

0.750

4.179

C3

1.459

0.686

0.381

1.185

0.256

0.432

iC4

0.179

0.297

0.0609

0.159

0.046

0.236

nC4

0.277

0.0940

0.0715

0.2167

0.0485

0.0608

neoC5

iC5

0.0541

0.0525

0.0167

0.0534

0.0095

0.0421

nC5

0.0587

0.0213

0.0170

0.0472

0.0073

0.0164

C6

0.0353

0.0472

0.0195

0.0383

0.0035

0.0373

Benzene

0.00452

0.00076

0.0139

0.0140

0.00009

0.00024

CycloC6

0.00215

0.01685

0.00329

0.0065

0.00061

0.01341

MethylCycloC6

0.00242

0.01038

0.00287

0.0060

0.00064

0.00801

Toluene

0.00075

0.00081

0.00368

0.0030

0.00010

0.00039

C7

0.01356

0.03041

0.0108

0.0173

0.00151

0.02399

C8

0.00464

0.00346

0.00543

0.0051

0.00031

0.00245

C9

0.00197

0.00103

0.00535

0.0022

0.00014

0.00064

C10

0.000386

0.00008

0.00265

0.00035

0.00008

0.00003

C11

0.000039

0.000001

0.000684

0.000015

0.00003

0.000014

C12

0.000003

0.000026

0.000038

0.000008

4.2.

Manual Chilled Mirror Device

The HCDP measurements were made by StatoilHydro. Experiments were conducted with a
custom-made with the help of Chandler Engineering (dew point mirror and cooling system)
manual chilled mirror device to determine the HCDP temperature of natural gas samples at
different pressures. Figure 2 shows the experimental setup used[11].
The system is filled with the gas sample to the highest possible pressure (depending on the gas
bottle pressure). The pressure sample within the rig was reduced in small steps by venting
portions of the gas, and once the target pressure is reached, the system is closed and
stabilized.
The gas is then circulated back and forth between two equally large chambers propelled by a
piston. The sample circulates in a closed loop, passing by a mirror whose temperature is
controlled by fitting a cooled copper rod to the back of it. The copper rod is cooled by a
controlled flow of liquid CO2 (Figure 3). All external piping is heat traced to a temperature of
60oC, and the rest of the experimental apparatus (including the gas cylinders) is held at 40oC.

Gas Chamber
Piston

Figure 2: Schematic set up of the experimental equipment

The cooled mirror temperature and the gas pressure are constantly logged. When the first liquid
precipitates on the mirror, the temperature is registered, giving the dew point temperature at
the given pressure.

Light

Figure 3: Cooling of the mirror

This dew point mirror and cooling system allow HCDP detection down to -40oC at pressures up
to 400 bars. As the device is a manual one, measurements were made by two operators, to
minimize the human factor. At each pressure, at least two temperature measurements were
made. Pressure incremental step varied from 1 bar near cricondentherm to 5 bars.

4.3.

HCDP Calculation Procedure

Calculations were made with the software packages that were selected during Phase 1 plus the
the GERG2004 EOS that was included during Phase 2. Calculations were made with both PR and
RKS EOS and rules were adopted concerning the way of lumping isomers, normalization of
compositions, etc.
Isomers of Ci will represent all the eluted compounds by gas chromatography (with a nonpolar column like PONA) between nCi-1 and nCi. All compositions were normalized to 100% mol.
The following rules were adopted regarding the isomer lumping method within each software
package:

Isomers of Ci were either lumped into nCi (default option if not otherwise indicated);

or they were lumped into nCi-1 (for specific applications)

or they were lumped into a default-fraction available in some software packages.

When using GERG2004 EOS the following rules were retained:

GERG2004 does not take into account benzene, toluene, cycloC6 and methylcycloC6.
Benzene and cycloC6 were then added into nC7 and toluene and methylcycloC6 into
nC8.

Higher hydrocarbons than C8 were lumped into the nC8 fraction.

In this paper, only PR EOS results will be shown. As it was explained, a constant discrepancy of
2.5 to 4C has been found between PR EOS and RKS EOS calculated HCDP RKS EOS
calculations give higher cricondentherms and cricondenbars.

4.4.

Experimental HCDP data versus calculated values

4.4.1. Phase envelopes with the default lumping method

Figure 4 summarizes the cricondentherm differences between experimental and calculated data
PR EOS. As a reminder, all isomers of Ci were lumped into nCi, except when the
calculations were made with GERG2004 (see paragraph 4.3).

Gas 6

Gas 4

Gas 3
GERG2004
Gas 2

Software D
Software C
Software B

Gas 1

Software A
0

10

12

14

16

18

T (C)

Figure 4 : Cricondentherm differences (T = Texperimental - Tcalculated) in C. Isomers are

lumped in nCi fractions, except for GERG2004.
All software packages mostly give very similar results when using the same way of lumping
isomers and the same equation of state. However, deviations between calculated and
experimental values regarding cricondentherm temperature are quite high.
Results of GERG EOS allow a better match with experimental data for some gases. The
differences between this model and the other software results can be explained by how the
isomers are lumped:

since aromatics and cyclics are put into nC7 or nC8, these higher hydrocarbons could
lead to an increase in the HCDP.

all heavy isomers (C9, C10) are put into a lower hydrocarbon group (nC8): in this case,
a decrease in the HCDP is likely.
These two opposite effects combine and either one can be predominant depending on the gas
composition.
There are no experimental measurements for Gas 5: the StatoilHydro custom-made device uses
CO2 as cooling fluid allowing the temperatures as low as 40C, and no HCDP detection was
made for this gas. Two possible reasons for this phenomena are: Gas 5 HCDP is lower than
40C or the device sensitivity (not published) does not detect the condensate quantity deposited
on the chilled mirror surface. Finally, a special remark should be made about the experimental
phase envelopes of Gases 2 and 3: the measured dew points around the cricondentherm did not
form a smooth curve.

4.4.2. Influence of the way of handling isomers

Figure 5 summarizes cricondentherm differences between experimental and the different
lumping methods calculated by PR EOS. The aim of this calculation is to find if using software
default-fractions allows a better match between predicted and experimental HCDP data.
Calculations were performed with two different software packages using both PR and RKS EOS,
and the following lumping rules:

Ci isomers with nCi lumped into a specific new fraction or default-fraction (group
characterized by average physical properties).

Ci isomers lumped with nCi (group characterized by nCi physical properties)

Ci isomers lumped with nCi-1 (group characterized by nCi-1 physical properties)

Gas 6

Gas 4

Gas 3

Gas 2
Ci isomers in nCi-1
Ci isomers in nCi

Gas 1

Default Fraction
0

10

12

14

16

T (C)

Figure 5 : Cricondentherm differences (T = Texperimental - Tcalculated) in C between

experimental and a specific lumping method calculated data using the Software A.
The following conclusions arise from this analysis:

Different lumping methods result in different HCDP temperatures: differences between

5C and 15C on the cricondentherm values were found.

In most cases, the real calculated value that would be obtained by using the defaultfraction is between both extreme values obtained by lumping isomers into nCi or nCi-1.

The best matches are obtained using the default-fractions lumping method combined
with PR EOS and nCi-1 lumping method combined with the RKS EOS. Nevertheless, the
PR EOS should be preferred because the results show a better match with the
experimental data.

Gas 3 is unusual since all the predicted cricondentherm values (meaning using all three
lumping methods) are always higher than the experimental ones.

4.4.3. GC data treated according to ISO 23874

ISO 23874[9] describes the performance requirements for natural gas analysis in sufficient detail
so that the HCDP temperature can be calculated using an appropriate EOS. This procedure
covers the measurement of hydrocarbons in the range C5 to C12. Major components are
measured using ISO 6974[5] and nC5 acts as bridge-component.
Chromatographic data supplied by StatoilHydro for the C5+ analysis using a FID (retention times
RT and peak areas) was treated in a modified version of the Software A to calculate the
phase envelopes using a boiling point generating tools for creating new default-fractions.
Table 3 summarises the results.
Table 3. Comparing Software A calculations using the ISO 23874 characterization
method with experimental data and the other software packages results

Gas 1
Gas 2
Gas 3

Gas 4
Gas 6

Using PR EOS
Very good fit (cricondentherm
differences between 0.5 and 1C)*
Very good fit (cricondentherm
difference 1C)*
Same behavior as other software
packages (except GERG2004, better fit)
(cricondentherm difference 11.7C)*
Better fit than other software packages
(3.8C difference)*
Very good fit (1.7C difference)*

Using RKS EOS

Better fit than other software packages
(3C difference)*
Very good fit (cricondentherm difference
1.3C)*
Same behavior as other software
packages (except GERG2004, better fit)
(cricondentherm difference 14.5C)*
Better fit than other software packages
(6.5C difference)*
Better fit than other software packages
(5C difference)*

* Differences between experimental and calculated cricondentherm values (T = Texperimental - Tcalculated)

The following conclusions arise from this analysis:

Differences between PR and RKS EOS results are still between 2-3C and PR EOS
predictions better fit the experimental data.

Results are still not satisfactory for Gases 3 and 4 : the fit is not as good as expected,
although there is a slight improvement compared to the other software predictions.

Gas 3 still gives higher predicted cricondentherm values compared to the experimental
data. These results cannot be easily explained and a combination of several factors
could be responsible, i.e. Gas 3 shows an important tail of heavy hydrocarbons.

ISO 23874 has only been tested with Software A. But, it is clearly obvious that for this
specific software, default fractions or fractions defined with the ISO standard have a
better fit to the experimental data than any other lumping method.

4.5.

Phase 2 discussion

Phase 1 conclusions were confirmed: the development of a specific HCDP chromatographic

method (i.e. quantification at 0.1 ppm up to C12) reduces the gas composition influence in the
prediction of cricondentherm values. It is also important to analyse and to account for benzene,
toluene, cyclohexane and methylcyclohexane.
However, it was not possible to choose a standard method for lumping isomers: deviations
between the two common solutions (add isomers of Ci in nCi or in nCi-1) are not negligible. The
use of defined software default-fractions might be a good compromise. The ISO 23874 Natural
GasChromatographic requirements for hydrocarbons dew point calculation was applied and
results were obtained using a modified version of Software A : results show that the use of ISO
23874 does improve the predictions, but not significantly more than the default-fractions
method already available in the software package.

5.

PHASE 3 AUTOMATIC DEVICE HCDP MEASUREMENTS

VERSUS CALCULATED VALUES

Phase 3 allowed us to compare the performance of different HCDP automatic devices. The
experimental data was then compared to new manual device measurements and with calculated
values.

5.1.

Materials

Three automatic HCDP analysers, one manual chilled mirror device, one Process GC and a
Varian GC micro gas chromatographer were used during Phase 3. The manual chilled mirror
device was described in paragraph 4.2. The main characteristics of the automatic HCDP meters
are detailed in Table 4.
Table 4. Automatic HCDP analyzers specifications

HCDP Analyzer

Condumax II

Ametek Model 241CE II

Cong-Prima-10

Manufacturer
Measuring Method

Michel Instruments
Ametek
Vympel
A gas sample is locked A gas sample is locked HCDP measurements
into the sensor cell
into the sensor cell
under continuous
during the cooling of
during the cooling of the
flow
the mirror
two surface mirror
Operating Pressure
Up to 100 barg
Up to 137 barg
Up to 100 barg
Temperature range
-30C to ambient
Up to 60C below
-30 to 30 C
temperature
ambient temperature
Flow rate
0.5 Nl/min
1 to 5 Nl/min
1 to 2 Nl/min
Accuracy
+/- 0.5 C
+/- 1 C
+/- 0.5 C
*
Measurement cycle
10 min.
30 min.
10 to 30 min.
*
The measurement cycle comprises the cooling an recovery purge and heating of the sensor cell phases.
The Process GC is an Emerson Danalyzer 700 GC/TCD, already used as an on-line gas
chromatographic analyzer for:

Separation, determination and quantification of inert compounds (CO2/N2) and

hydrocarbons from C1 up to C9/C10+ in natural gas.

Calculation of the calorific value (CV) of natural gas method based on the quantified
compounds and their physical properties.

Calculation of natural gas HCDP method based on the quantified compounds, the HCDP
is calculated from EOS (RKS and PR).
The Emerson Danalyzer 700 GC/TCD is equipped with two TCD detectors. TCD detector 1 is
set up for standard C6+ analysis while TCD detector 2 takes from C6 up to C9+. The gas analysis
is later combined in post-analysis calculations to calculate CVs and HCDPs.
A VARIAN Micro-GC/TCD model CP4900 was installed to control gas compositions within the test
rig during Phase 3 measurements. The CP4900 used in this work is comprised of four micro GC
modules, each of them with a separate injector, column and detector and all of them with
separate flow and temperature controls. Sample transfer and GSV-sampling were done in
accordance with ISO 10715[7].
Two additional analysers were also used during Phase 3 : a MCM DewLuxe (UK) for determining
moisture in the gas matrix, and a Perkin Elmer TurboMass GC/MS where ATD tubes were
analyzed for glycol quantification.

5.2.

Natural gas samples and RGM

The natural gas samples are the same ones used during Phase 2. These samples have been
stored on high pressure stainless steel (SS316) cylinders and the storage temperature has been
room temperature (> 15C).
Calibration gases[2-4,8] synthetic mixtures by Yara Industrial were used to calibrate the Varian
Micro-GC/TCD and to check and track any adsorption effects from heavier hydrocarbons inside
the test rig.

5.3.

Description of the rig

The main components in the test rig are two pumps and a valve arrangement (EV1 to EV4)
giving a continuous flow through a DP instrument. The 4 HCDP instruments and the Varian
Micro-GC/TCD are all connected to the rig but it is only possible to carry out measurements at
one of them at a time. The rig is operated at room temperature (20 C) and up to 200 bar. In
Figure 6 the operation of the manual device is shown as an example. The gas flows from Pump

2 through the inlet manifold to the Chandler DPM and then back via the outlet manifold to Pump
1.
Most of the rig material of construction is stainless steel (SS 316) and some polymer material
for piston rings and gaskets. Any possible adsorption/desorption effect (ADE) could be caused
by the affinity heavier HC have to different material of construction (MOC). Variation of
temperature and pressure conditions could also lead to increased adsorption/desorption effects.
Even if no hydrocarbon condensation takes place inside the rig, adsorption/desorption effects
could happen with the test rig material of construction during the gas contact. Temperature and
pressure variations could also promote these phenomena.
A study of the adsorption/desorption effects, both outside and inside the rig was performed
prior to any measurements using RGM mixtures: components up to C10 (including aromatics
and cyclics) were used to quantify the adsorbed/desorbed hydrocarbons. It seems that mostly
heavy hydrocarbons are adsorbed in the test rig material of construction.
For minimizing this phenomena during measurements, a loading and emptying procedure was
performed every time a new gas was tested.

5.3.1. Loading and emptying procedures

The rig is vacuumed to about 1 mbar prior to each filling.

Then it is filled with the current gas to a pressure of 30 bar and the gas is circulated
through all the HCDP instruments for at least 2 hours.
Then the rig is emptied and refilled with the same gas to the measurement pressure.

Manual
Device

Chandler DPM

Ametek DPM 2
Analyzer

N2
V71

V33

V72

V31

V23

V21

V7
V32

V22

Vent

V41

Gas
V5

V43

EV1

EV2

Vacuum
V6

V12
P
P4

T
T2

V11

EV4

Pump 2

EV3

P
P2

Michell DPM 1
Analyzer

P3

P
P1

V13

BP3 Inlet Manifold

T
T1

Pump 1

V42

BP3 Outlet Manifold

Vympel DPM
Analyzer
3

Varian GC

V51
V52

Figure 6 : StatoilHydros HCDP test rig

5.3.2. Measuring procedure

The measurement started at the highest pressure level. Mostly 15 to 18 pressure levels were
measured for each gas, starting from just below the cricondenbar down to 5 bar. For each
pressure level, at least 5 measurements were carried out with each HCDP instrument in the rig.
The sample gas was analyzed with the Varian MicroGC/TCD after the loading procedure and
once all measurements were achieved for a certain pressure level.

5.4.

Phase 3 measurements

The results, measurements made with the 3 automatic and the manual HCDP devices, are
shown in Figure 7 and summarized in Table 5. The automatic devices are named Analyzers 1 to
3, and the manual device is StatoilHydro custom-made Chandler analyzer used during Phase 2
and described in paragraph 4.2.
The calculated values PR EOS Software A are the cricondentherm and the cricondenbar
values calculated for an average gas composition, using the Software A and the PR EOS. This
average gas composition was calculated for each gas sample, from all the Varian MicroGC/TCD
gas analysis obtained during each measurement cycle. In general calculated cricondentherm
values are slightly higher than measured values (except for Gas 4). These discrepancies can be
explained by measurements requiring a finite not infinitesimal amount of liquid for detection,
and they are not the consequence of the observed ADE.
Table 5. Phase 3 HCDP measurements
HCDP Data
Gas 1
Gas 2
Gas 4
Gas 5
Gas 6

Analyzer 1
P (barg) T (C)
35
-10.4
35
-14.4
30
-8.6
31
-18.5

Analyzer 2
P (barg) T (C)
31
-9.3
41
-12.4
35
-8.6
36
-16.3

Analyzer 3
Manual Device
P (barg) T (C) P (barg) T (C)
35
-10
30
-10.2
35
-13.6
31
-13.4
35
-9.5
30
-9.6
31
-17.7
31
-17.3

Emerson GC/TCD
P (barg)
T (C)
30
-7.8
32
-4.2
30
-3
18
-38.3
30
-9

PR EOS - Software A
P (barg)
T (C)
30
-9.8
30
-10.5
31
-13.1
20
-39.7
29
-17.7

Gas 6

Gas 4

Gas 2

Gas 1

-20

-15

-10

-5

Cricondentherm (C)
Analyzer 1

Analyzer 2

Analyzer 3

Manual Device

PR EOS - Softw are A

Figure 7 : Phase 3 HCDP measurements

Again, no experimental HCDP measurements could be made with Gas 5, only a calculated HCDP
value is given in Table 5. This is probably because Gas 5 is a very dry gas with a HCDP below
the temperature ranges of the analyzers.
Gas 3 measurements revealed the presence of moisture as it can be seen in Figure 8 : the
automatic analyzers 1 and 3, and the manual device data indicate the presence of water.
Further measurements were made with Gas 3 after it was loaded into the test rig through a
molecular sieve. In Figure 9, we can observe that after removing the moisture the analyzers 1
and 3, and the manual device could detect some HCDP points. However, Gas 3 composition was
altered as it shown by the phase envelope calculated with the average gas composition in the
test rig : heavy hydrocarbons were also removed with the molecular sieve because the
calculated cricondentherm decreased by 20C.

90

80

80

70

70

60

60

50

50

P (barg)

P (barg)

90

40

40

30

30

20

20

10

10

0
-100

-80

-60

-40

-20

0
-100

-80

-60

T (C)

Analyser 1
Manual Device

-40

-20

20

T (C)

Analyzer 3
PR EOS - Software A

Figure 8 : Gas 3 HCDP measurements

Analyzer 1
Analyser 3
PR EOS - Software A

Analyzer 2
Manual Device

Figure 9 : Gas 3 HCDP measurements

gas loaded into the rig through a
molecular sieve.

Gas 3 was also analyzed with MCM Dewluxe water analyzer : 375 ppm of moisture were
detected at 10 bars and 550 ppm at 1 bar. The same gas was analyzed after passing through
the molecular sieve, and the water content was 3 ppm. It is believed that moisture was
originally in the gas cylinder.

5.4.1. Emerson process GC TCD measurements

The Emerson Danalyzer 700 GC/TCD was installed after all the final set up of the test rig was
completed. Hence, this analyzer only made direct measurements from gas sample cylinders and
was not integrated in the test rig.
As the set up of this instrument was delayed and due to the problems encountered during the
installation, not enough experimental HCDP data could be retrieved from measurements. It has
also been noticed that there was a stability problem with the TCD detector probably due to a
short stable equilibration period between installation and calibration of the unit : a drift in the
TCD signal was detected during measurements.
As measurements were made from gas cylinders, we cannot directly compare Emersons
calculated data to Phase 3 HCDP measurements or calculated values. However, some calculated
data was added to Table 5, showing extremely high cricondentherms. However, Emersons
calculated cricondenbars actually agree with the PR EOS Software A calculated values. HCDP
calculations were made with the PR EOS.

5.5.

Phase 3 discussion

All the automatic HCDP analyzers give similar results, and it should not be surprising because all
three of them share the same technology : a cooled mirror cell where the natural gas forms a
heavy hydrocarbon rich liquid. Of course, each of them have their own particular specifications,
set-points the detection threshold can be adjusted for some of the analysers, and operation
procedures.
The aim of Phase 3 was to compare their performance when analyzing real gases, and some
results must be highlighted :

The adsorption/desorption effects on the test rig material of construction were rather
important, and a specific procedure of loading and emptying the test rig had to be
established.

The gas quality should be defined before measuring HCDP. As it was shown for Gas 3, a
high water content can affect the HCDP measurement. The use of molecular sieves is
recommended. Previous studies have shown that gas compositions are not significantly
altered specifically the loss of heavy hydrocarbons when molecular sieves are

6.

installed. Gas 3 results should only be taken as an isolate case of a contaminated gas
cylinder.
We could not evaluate the performance of Process GC for the HCDP determination : the
results obtained from the Emerson Danalyzer 700 GC/TCD are not enough and several
technical problems arose during the installation and operation of this analyzer.

GERG 1.52 GENERAL DISCUSSION

The aim of this study was not to choose a specific HCDP prediction method and making a HCDP
determination standard out of it, but to try to understand and quantify the deviations between
experimental and predictive HCDP values.
The comparison between experimental and calculational methods is complicated: each method
has its own limitations and in order to give a correct basis of comparison the maximum detail
available should always be given (i.e. if HCDPs are calculated using an EOS model, details such
as type of equation, adjustable parameters, gas composition detail, and isomer lumping method
should be known).
When calculating HCDP values, it is important to be able to quantify the heaviest hydrocarbons
in the natural gas at a level of 0.1-0.2 ppm mol. Components should be taken into account up
to the last two consecutive heavy hydrocarbons (probably C11 and C12) with concentrations
below 1 ppm. Moreover, the development of a specific HCDP chromatographic method reduces
the gas composition influence in the prediction of cricondentherm values. It is also important to
analyse and to account for aromatic and cyclic compounds. However, it was not possible to
choose a standard method for lumping isomers: deviations between the two common solutions
(add isomers of Ci in nCi or in nCi-1) are not negligible. The use of defined software defaultfractions might be a good compromise. Finally, results show that using ISO 23874 does not
improve significantly predictions compared to the default-fractions method already available in
the software packages.
Automatic and manual devices both have their uses. While the use of automatic devices is
practical and todays technology allow their continuous operation, measurements with manual
devices are still useful to periodically check the measurements or if problems arise. Besides, the
gas quality has an important influence in the maintenance schedules of automatic devices (e.g.
water content, compressors oil contamination, etc.).

7.

REFERENCES

1.

FRIEND D.G., INGHAM,H. and ELY J.F., Thermophysical Properties of Ethane,

J.Phys.Chem. Ref. Data, (1991), 20(2):275-347.
2. ISO 6141: Gas analysis Requirements for certificates for calibration gases and gas
mixtures, (2006).
3. ISO 6142: Gas Analysis Preparation of calibration gas mixtures Gravimetric method,
(2006).
4. ISO 6143: Gas analysis Comparison methods for determining and checking the
composition of calibration gas mixtures, (2006).
5. ISO 6974-(1-6), Natural gas Determination of composition with defined uncertainty by
gas chromatography, (Part 1, 3, 4 and 5: 2003, part 2: 2004, part 6: 2005).
6. ISO 6975 Natural gas Extended analysis Gas-chromatographic method, (2005).
7. ISO 10715: Natural gas sampling guidelines, (2001).
8. ISO 14167: Gas analysis General quality assurance in the use of reference gas mixtures
guidelines, (2003).
9. ISO 23874: Natural gas Gas chromatographic requirements for hydrocarbon dewpoint
calculation, (2006).
10. KAY, W.B., "Vapor-liquid equilibrium relationships of binary systems", J. Chem. Eng. Data,
(1971), 16(2).
11. MORCH, O et al. Measurement and modeling of hydrocarbon dew points for five synthetic
natural gas mixtures, Fluid Phase Equilibria, (2006), 239 : 138-145.
12. NASRIFAR, K., BOLLAND, O. and MOSHFEGHIAN, M, Predicting natural gas dew points
from 15 equations of states, Energy and Fuels, (2005), 19:561-572.

13.LIST OF TABLES
Table 1: Estimated uncertainty for each concentration range
Table 2: GERG 1.52 real gases used during Phase 2 (StatoilHydro data)
Table 3: Comparing Software A calculations using ISO 23784 characterization method with
experimental data and the other package results
Table 4: Automatic HCDP analyzers specifications
Table 5. Phase 3 HCDP measurements

14.LIST OF FIGURES
Figure 1: Cricondentherm as a function of the highest carbon group taken into account for the
HCDP calculation (PR EOS for one software).
Figure 2: Schematic set up of the experimental equipment
Figure 3: Cooling of the mirror
Figure 4: Cricondentherm differences (in C) between experimental and calculated data
(isomers are lumped in nCi fractions, except for GERG2004)
Figure 5: Cricondentherm differences (in C) between experimental and a specific lumping
method calculated data using the Software A.
Figure 6: StatoilHydros HCDP test rig
Figure 7: Phase 3 HCDP measurements
Figure 8: Gas 3 HCDP measurements
Figure 9: Gas 3 HCDP measurements gas loaded into the rig through a molecular sieve

15.ACRONYMS
AED : adsorption/desorption effects
EOS: Equation of state
GC : Gas Chromatography
HCDP : Hydrocarbon Dew Point
MOC : material of construction
PR : Peng-Robinson
RGM: Reference Gas Mixtures
RKS: Redlich-Kwong-Soave