processes
Article
The Challenge of Monitoring Impurity Content of CO2 Streams
Bjørn H. Morland *, Gaute Svenningsen






Citation: Morland, B.H.;
Svenningsen, G.; Dugstad, A. The
Challenge of Monitoring Impurity
Content of CO2 Streams. Processes
2021, 9, 570. https://doi.org/
10.3390/pr9040570
Academic Editor: Zhien Zhang
Received: 16 February 2021
Accepted: 18 March 2021
Published: 24 March 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Processes 2021, 9, 570. https://doi.org/10.3390/pr9040570
and Arne Dugstad
Institute for Energy Technology, NO-2027 Kjeller, Norway; gaute.svenningsen@ife.no (G.S.);
arne.dugstad@ife.no (A.D.)
* Correspondence: bjorn.helge.morland@ife.no; Tel.: +47-63-806000
Abstract: Carbon capture and storage has gained increased attention during the last decade, and
several full-scale projects are currently being planned. From economic and public acceptance point of
view it is important to ensure that the transportation system is operated in a safe manner, avoiding
threats such as corrosion or formation of solid matters. Thus, routine chemical analyses are required
to ensure that the CO2 stream complies with the required specifications. The CO2 will usually be
transported in the liquid or supercritical state (high pressure), which makes the practicalities around
chemical analyses difficult. Phase transition from liquid or supercritical state to gaseous state may
also introduce several physiochemical effects that may affect the analyses. This paper discusses
technical and practical challenges with CO2 stream analyses experienced in a joint industry project
that studied corrosion and chemical reactions in a simulated CO2 transport system.
Keywords: CO2 impurities; monitoring; chemical analysis
1. Introduction
Carbon capture and storage (CCS) has gained increased attention as a method to
reduce emissions of anthropogenic CO2 to the atmosphere [1,2]. Separation of CO2 from
gases has been carried out for almost a century and a number of techniques are available.
However, the captured CO2 will always contain some additional components, called
impurities in the present work. The types and concentrations depend on the CO2 source,
the capturing technique, and how many impurities are removed in the liquefaction process.
In most cases, the CO2 streams will need further conditioning/purification before transport
and injection.
It has been shown experimentally that certain combinations of impurities react if they
are present above a critical concentration [3–5]. It was demonstrated that H2 O, H2 S, SO2 ,
NO2 , and O2 at concentrations much less than 100 ppmv reacted and created separate
aqueous phases containing high concentrations (several mol/kg) of sulfuric and nitric acid.
Formation of acid was also observed by Yevtushenko et al. [6] in experiments containing
SO2 , NO2 , and O2 at water saturation. The acidic aqueous phase may introduce corrosion
problems in the CO2 transportation system, a system which most likely will be constructed
of carbon steel due to cost and availability [7,8]. Elemental sulphur is another possible
reaction product [9,10]. If the sulphur remains dissolved in the bulk CO2 phase, it will
probably not cause any problems, but it may precipitate as solids and create particulate
problems in the transportation system and reduced injectivity in the reservoir.
Several CO2 specifications have been proposed to ensure safe operation of the CO2
transportation chain. As the mechanisms for formation of corrosive species and particulate
matter have become better understood, the CO2 specifications have become more and more
strict (lower maximum impurity content). The actual composition of the CO2 stream (or
bulk carrier) needs to be documented by chemical analyses.
In most cases, CO2 will be transported in pipelines or by ship (bulk transport).
Pipelines will be operated at high pressures (>74 bar) and ambient temperature. Bulk
transport will be carried out with CO2 cooled to the liquid state with a small gas cap, for
https://www.mdpi.com/journal/processes
Processes 2021, 9, 570 2 of 12

which the pressure will be in the range of 6–20 bar depending on the temperature [11–13].
Since nearly all CO2 transport specifications [14] are based on the concentration of impu-
rities in the transport system, this paper focuses on the monitoring during transport or
monitoring related to mixing of several CO2 streams.
Measurements of impurities in dense phase CO2 (liquid or supercritical state) is
challenging, particularly since some of the impurities are present at low concentrations,
and in addition some of them may react before analysis can be carried out.
Most of the present work is based on the experience gained during the process of
building a corrosion test system that can be operated under CCS conditions in our lab.
Analysers were used to compare the impurity concentration of inlet and outlet CO2 from
the autoclave (reaction chamber). Work with the test system(s) has been going on for more
than 10 years and the development is still in progress. Even if the practicalities around
such analysis may vary significantly from the lab to the field, many of the general problems
and challenges are still the same and will be discussed in the paper. Currently there are
no standard methods for impurity analyses of dense phase CO2 streams. The present
work highlights the most challenging issues of such analyses. Furthermore, as this work is
intended to focus on the test approaches in general, details about the actual instruments
that were used are intentionally not given. Similarly, chemical reactions, etc., are only
briefly treated here, as more details can be found elsewhere [3,9,15].

2. Analysing the Impurity Content of CO2 Streams

2.1. The Pressure Challenge
Highly accurate gas analysers have been available for a long time and there are nu-
merous instruments and techniques available. However, none of the instruments available
on the open market are able to carry out analyses under actual process conditions (CO2 in
the liquid and supercritical state).
In practice, the pressure must be reduced to near atmospheric before analysis can be
made, and therefore the CO2 must be transformed from the supercritical or liquid state to
the gaseous state. A pressure regulator is needed for this, and usually it will also require a
mass flow controller to maintain a stable (but adjustable) feed of analyte.
With pressure reduction and phase transformation, there are several physiochemical
factors that may affect the chemical analysis. Figure 1 shows the solubility of water in pure
CO2 . When CO2 is transferred from the liquid to the gaseous state, there is a sharp drop
in the water solubility, and hence there is a risk of water precipitation. The problem is
enhanced by the Joule–Thomson effect, which reduces the temperature in the gas reduction
valve. The reduction valves should therefore be operated3 ofwith
Processes 2021, 9, x FOR PEER REVIEW 13
electric heating, as this
prevents precipitation of liquid water. Heating will also prevent/reduce the risk of hydrate
formation, which may occur at temperatures of 11 ◦ C or lower [16,17]. In the gaseous state,
will gradually dissolve again when the pressure is reduced below 50 bar and it will be
the water solubility increases with decreasing pressure (left side of Figure 1).
fully dissolved when the pressure reaches 20 bar.

7000

6000
H2O solubility (ppmv)

5000

4000

3000

2000

1000

0
0 20 40 60 80 100 120 140 160
Pressure (bar)
0°C 10°C 20°C 30°C 40°C 50°C

Figure 1. Water
Figure 1. Water solubility
solubility (ppmv)
(ppmv) in CO in CO
2 as a function
2 as a function
of pressure of pressure
and temperature. and temperature. The data were
The data were
calculated with the OLI software.
calculated with the OLI software.
3000

2500
)
 6000

H2O solubility (ppmv)

5000

4000
Processes 2021, 9, 570 3 of 12
3000

2000

An example that shows the effect of water precipitation in the gas regulator is given
1000
in Figure 2. The initial water analysis was very stable (500 ppmv), but it started to fluctuate
0
somewhat 0 when20 the water
40 content
60 was increased
80 100 to 1200
120 140ppmv. 160Shortly after injection of
fully water-saturated CO2 (in Pressure this example
(bar) it started at 54 h), the water signal started to
0°C
fluctuate significantly. 10°CThis is20°C
believed 30°C
to be the 40°Cresult50°C
of water precipitation/dissolution
dynamics in the heated gas regulator. This behaviour is so common that it is used by the
Figure 1. Water solubility (ppmv) in CO2 as a function of pressure and temperature. The data were
authors as an
calculated with theindication
OLI software. of water saturation.

3000

2500

2000
H2O (ppmv)

1500

1000
H2O
500

0
0 20 40 60 80 100
Time (h)
Figure 2. Water analysis from an experiment where the water injection was increased from 500 to
Figure 2. Water analysis from an experiment where the water injection was increased from 500 to
1250 ppmv (28 h) and from 1250 to 2200 ppmv (54 h).
1250 ppmv (28 h) and from 1250 to 2200 ppmv (54 h).

Figure 3 shows the predicted water solubility at 25 ◦ C assuming full equilibrium

and no temperature changes. If the water content is low, precipitation of water is not
expected when the pressure is reduced from 100 to 1 bar (example with 250 ppmv shown
by the green line/circles). The black lines/circles show the same process for a higher water
content of 2000 ppmv. The water solubility is clearly exceeded during the phase transition
from liquid to gaseous CO2 , where about 800 ppmv of water will precipitate as liquid
water. It will gradually dissolve again when the pressure is reduced below450
Processes 2021, 9, x FOR PEER REVIEW bar and it
of 13
will be fully dissolved when the pressure reaches 20 bar.

4000

3500
H2O solubility (ppm-mol)

3000

2500

2000
precipitation
Water

1500

1000
Dissolution of liquid water
500

0
0 20 40 60 80 100 120
Pressure (bar)

Figure
Figure3.3.The
Theblue lineline
blue shows the water
shows solubility
the water (ppmv)(ppmv)
solubility in CO2 as
inatCO
252°C at 25 ◦ C (calculated
as(calculated with the with the
OLI software). The green and black lines represent two CO 2 streams with 2000 ppmv and 250
OLI software). The green and black lines represent two CO2 streams with 2000 ppmv and 250 ppmv
ppmv water, respectively.
water, respectively.
A sharp drop in solubility with decreasing pressure is also observed for other species
such as nitric acid, sulfuric acid, and elemental sulphur. If the precipitation is fast, species
may accumulate in the heated gas regulator. This will of course result in erroneous chem-
ical analysis, but it may also introduce clogging of the regulator, the mass flow controller,
Processes 2021, 9, 570 4 of 12

A sharp drop in solubility with decreasing pressure is also observed for other species
such as nitric acid, sulfuric acid, and elemental sulphur. If the precipitation is fast, species
may accumulate in the heated gas regulator. This will of course result in erroneous chemical
analysis, but it may also introduce clogging of the regulator, the mass flow controller, and
the analysers (see the example in Section 3.2). Furthermore, if certain species precipitate,
they could introduce chemical reactions in the feeding lines from the gas regulator to the
analysers. If an aqueous phase with acids precipitates inside the heated gas regulator,
the water measurements may also fluctuate even in the low parts per million by volume
(ppmv) range, as shown in Section 3.3.

2.2. The Calibration Challenge

Some instruments, such as gas chromatographs, as well as UV and IR photometers,
need regular calibration. Even if this process is automated, it can be a somewhat tedious
process, particularly in the field. It also means that bottles with calibration gas (one for
each concentration) must be available at the site and be refilled in due time. Thus, there is
a technological and logistic issue that has to be dealt with. This should be manageable at a
CCS facility, which nevertheless will require a certain amount of qualified work force. The
situation may be different on a remote site.

2.3. Saturating the Sampling System with Analyte

Several impurities may adsorb on internal surfaces of the sampling loop, such as
phase transition regulators, flow meters, and analysing lines. Normally, the volume in
the sampling system should be fully replaced with CO2 feed (analyte) three to five times
before a representative sample can be collected. This is not a problem for a continuous
analysing system, but the lag time could be long when considering both the volume
exchange and adsorption/desorption effects. This is commonly referred to as “saturating”
the analysing system. If the analyte composition changes, these impurities will adsorb
or desorb according to the surface equilibrium. It has been observed that H2 O, H2 S, and
NH3 require long saturation times, while SO2 , O2 , and CO are much faster. The actual
response time depend on the length of the analysis line and the volume and surface area of
additional equipment.
One of the setups in our lab consists of 20 m 1/1600 stainless steel tubing, three valves,
two filters, a heated gas regulator, and a 300 mL autoclave. It takes about 16 h to saturate
this setup when changing the water content from 5 to 1500 ppmv with 500 mL/min total
gas flow. If the water content is reduced from 1500 to 1000 ppmv, stable measurements are
achieved after about 2 h. The dry-up time for such a system is very long. The experience is
that it takes about 2 weeks to reduce the water content from 1000 to 1 ppmv, and the last
50 ppmv are the most time consuming. It is possible to purchase tubing that has a special
surface treatment (“inert tubing”) that is claimed to reduce the problem. This effect has not
yet been tested by the authors.
The problem can be reduced by designing the sampling system with a high-volume
exchange rate. In practice, this means low diameter tubing or a large flow of CO2 . With
smaller diameter tubing, there is an increased risk of clogging by solids and therefore some
optimisation has to be made.
The phase transition regulator should be made of inert material. Usually, a fine tuning
of the heat supply would also be necessary. More heat would at least give shorter adsorp-
tion/desorption time, but it could increase the risk of reaction between the impurities.
Thus, there is a trade-off between response time and accuracy.

2.4. Sampling Location

Corrosion and impurity reactions can form solid and liquid phases [3–5]. These
products (solids, liquids) may separate from the CO2 bulk phase due to density differences
and either stick to the wall or follow the CO2 stream. Depending on the pressure and
temperature (density of CO2 ), the products may be lighter or heavier than the bulk CO2
Processes 2021, 9, 570 5 of 12

phase. This is further complicated by the large variation of CO2 density (varies significantly
with pressure and temperature), meaning that certain components can be both lighter or
heavier than the bulk CO2 depending on the conditions. Flow and turbulence may also
have an effect, particularly in pipelines. Products that are expected to be separated by
gravity could be kept in the bulk phase as emulsions or small particles, but they may
also accumulate either at the top or bottom of the transportation system. The practical
sampling locations will have to be evaluated for each system, but one approach could be to
sample from the top position (light components), middle position (bulk phase), and bottom
position (heavy components), as indicated in Figure 4. Applying a large-diameter flange at
the bottom would allow for accumulation of heavy components if the turbulence is6not
Processes 2021, 9, x FOR PEER REVIEW of 13
too high.

Figure4.4.Schematic
Figure Schematicdrawing
drawingindicating
indicatingpossible
possiblesample
samplelocations
locationsinina apipeline
pipelineororCO
CO 2 bulk carrier.
2 bulk carrier.

Partitioning
3. Analysis of impurities
and the in two-phase
Need for Phase SeparationCO2 is the case for a number of impurities,
and sampling
The CO2 from
streamthecan
gasinorprinciple
liquid phase willboth
contain therefore giveliquid,
gaseous, different
andresults [18–20].
solid products.
Experiments have shown that H 2 O, SO 2 , and H 2 S partition preferentially
Each of these components will usually require different types of analysing techniques into liquid CO 2
and
phase, while O2 partition preferentially into the gas phase. This means that
will therefore have to be separated. With gaseous components, it is here meant compo- it is important
tonents
carrythat
outare
sampling in a manner
in gaseous that prevents
state around a two-phase
room temperature system,
and e.g., by
at pressure introducing
near a
1 atm (after
large pressure drop by fast sampling. A piston cylinder with back pressure
pressure reduction). The analysing approach will probably vary from site to site, but onecould be used
toexample
avoid two
of aphase formation
possible solution during analysis
is shown of a batch
in Figure 5. CO2 sample.

2.5. Impurity Range

High pressure side Low-pressure side (1 – 2 bara)
The impurity levels are expected to be low (typically < 10 ppmv) during normal
operation,Heated
and gas
consequently low detection limits are required. Upset conditions, on the
To waste
regulator Gas
other hand, may introduce relatively high impurity levels.GasThus, there is a trade-off
controller
Gas flow

analyser analyser
Separator

between
Filter wide range and accuracy of the lowest concentrations.
High speed
sampling loop 3. Analysis and the Need for Phase Separation
The CO2 stream can in principle contain both gaseous, liquid, and solid products. Each
of these components will usually requireLiquid different types of analysing techniques and will
Large diametertherefore have tubing
to be separated. Withsampling
gaseous components, it is here meant components that
Small diameter
tubing are in gaseous state around room temperature and at pressure near 1 atm (after pressure
reduction). The analysing approach will probably vary from site to site, but one example
Figure
of a possible 5. Schematic
solution flowchart
is shown for analysis
in Figure 5. system.

3.1. Gasseous Species

After the heated gas regulator, the gas can be analysed directly. The choice of analys-
ing method is usually based on a number of factors, of which economy, number of ana-
lysed species, detection limit, reproducibility, need for calibration, ease of repair/mainte-
 3. Analysis and the Need for Phase Separation
The CO2 stream can in principle contain both gaseous, liquid, and solid products.
Each of these components will usually require different types of analysing techniques and
will therefore have to be separated. With gaseous components, it is here meant compo-
Processes 2021, 9, 570 6 of 12
nents that are in gaseous state around room temperature and at pressure near 1 atm (after
pressure reduction). The analysing approach will probably vary from site to site, but one
example of a possible solution is shown in Figure 5.

High pressure side Low-pressure side (1 – 2 bara)

Heated gas To waste

regulator Gas Gas

controller
Gas flow
analyser analyser

Separator
Filter
High speed
sampling loop

Liquid
Large diameter sampling
Small diameter tubing
tubing

Figure 5. Schematic flowchart for analysis system.

Figure 5. Schematic flowchart for analysis system.

3.1. Gasseous3.1. Gasseous Species

Species
After the heated gas regulator, the gas can be analysed directly. The choice of analys-
After theing
heated gas regulator, the gas can be analysed directly. The choice of analysing
method is usually based on a number of factors, of which economy, number of ana-
method is usually based on a number
lysed species, detection of factors, of which
limit, reproducibility, need foreconomy,
calibration,number of analysed
ease of repair/mainte-
species, detection
nance, limit, reproducibility,
and equipment lifetime areneed for calibration,
the most ease of repair/maintenance,
important. Analysers based on laser absorp-
and equipment lifetime arewere
tion spectroscopy the selected
most important. Analysers
in the authors’ based
lab as the best ondue
choice laser absorption
to high analysis
spectroscopyfrequency and longin
were selected calibration
the authors’intervals.
labOne
as disadvantage
the best choice with due
the current setup
to high is that
analysis
frequency and the number of possible analysed
long calibration intervals. species
Oneis disadvantage
fixed. Furthermore,withnotthe
all species
current cansetup
be ana-is
lysed for (there are several reasons for this). Although this probably can be improved in
that the number of possible analysed species is fixed. Furthermore, not all species can be
the future, it means that these laser-based techniques must be combined with other ana-
analysed for lytic
(there are several
methods. reasons
Several for this).
laser-based Although
analysers may alsothis
be probably can beFTIR
applied in series. improved
spectros-in
the future, it copy
means that these laser-based techniques must be combined with other
is another analytical technique that probably could be used, but at present it has notanalytic
methods. Several laser-based analysers may also be applied in series. FTIR spectroscopy is
another analytical technique that probably could be used, but at present it has not been
tested by the authors. The technique is promising since it is capable of multi component
analysis, and new components in principle can be added to the analysis by only changing
the software.
Gas chromatography (GC) needs regular calibration, and therefore we did not use
it for long-term analysis (with dense phase CO2 ). It is sometimes used for infrequent
analysis. However, GC has the advantage that it relatively easy can be modified to analyse
for new species.
Regardless of the analysis technique, it is difficult to find one technique and instrument
that can handle all species, and thus several analysers have to be used in combination.

3.2. Solids
Solids can be separated from the bulk CO2 phase with filters. Since the solubility of
certain components vary with pressure, it is best to apply the filter before the phase transi-
tion regulator. This will also prevent clogging problems of the gas regulator. Nevertheless,
the filter housing must be able to handle the CO2 pressure, and it must withstand frequent
depressurization in a controlled manner (particularly soft materials such as rubber gaskets
or Teflon-coated diaphragm can be damaged due to rapid depressurisation). The filters
could be analysed by standard methods, e.g., by X-ray powder diffraction (XRD), scanning
electron microscopy (SEM), and energy-dispersive X-ray spectroscopy (EDS). Dissolving
the products in a suitable solvent could be another solution.
A filter can also be installed on the low-pressure side for protection of the analysers
and the mass flow controllers, etc. Comparison of filters from up- and downstream
the reduction valve may provide useful information, since the pressure drop over the
heated gas regulator may lead to precipitation of certain species (see Figure 6). Previous
research [3,10,15] has shown that interaction between impurities may occur when certain
types are present, and the concentrations exceed a critical limit. The analysis system is
vulnerable to precipitation of solids if they settle inside the analytical instruments. The
green/yellow solids seen in Figure 6 (diaphragm to the right) was identified with XRD
to be elemental sulphur. The sulphur was initially fully dissolved in the CO2 stream not
transported to the gas regulator as solid but rather as dissolved species [10] in the CO2 and
precipitated when the pressure dropped over the regulator. This is commonly observed,
 (XRD), scanning electron microscopy (SEM), and energy-dispersive X-ray spectroscopy
(EDS). Dissolving the products in a suitable solvent could be another solution.
A filter can also be installed on the low-pressure side for protection of the analysers
and the mass flow controllers, etc. Comparison of filters from up- and downstream the
reduction valve may provide useful information, since the pressure drop over the heated
gas regulator may lead to precipitation of certain species (see Figure 6). Previous research
Processes 2021, 9, 570 [3,10,15] has shown that interaction between impurities may occur when certain types are 7 of 12
present, and the concentrations exceed a critical limit. The analysis system is vulnerable
to precipitation of solids if they settle inside the analytical instruments. The green/yellow
solids seen in Figure 6 (diaphragm to the right) was identified with XRD to be elemental
sulphur. The sulphur was initially fully dissolved in the CO2 stream not transported to
and
the gasmost of the
regulator impurity
as solid but rather solubilities shows
as dissolved species [10] inathe
minimum solubility right before the phase
CO2 and precipitated
when the pressurefrom
transmission dropped overto
gas theliquid
regulator.(see
This is
ancommonly
example observed, and most1).
in Figure of As long as such species
the impurity solubilities shows a minimum solubility right before the phase transmission
remain
from gas to dissolved inexample
liquid (see an the CO in 2 stream,
Figure 1). As they
long asmight notremain
such species be a problem
dissolved for the transport system,
but
in theitCO
may introduce
2 stream, they mightproblems for for
not be a problem thetheanalysing loop.
transport system, but it may intro-
duce problems for the analysing loop.

Figure 6. Pictures showing products that have precipitated and accumulated inside the heated gas
Figure
regulator6. Pictures
body (left) andshowing products
on the regulator’s that
diaphragm have precipitated and accumulated inside the heated gas
(right).
regulator body (left) and on the regulator’s diaphragm (right).

3.3. Liquids
Processes 2021, 9, x FOR PEER REVIEW Liquids can be difficult to separate from dense phase8 ofCO 13
2 since they may have al-
most the same density. The density differences are much higher downstream the pressure
regulator
3.3. Liquids (where CO2 is in the gaseous state), and gaseous and liquid species can then be
separated
Liquids canbybe top and
difficult bottom
to separate fromstreams
dense phase inCOa 2small
since theyseparator.
may have almostThe separator could even be
the same density. The density differences are much higher downstream the pressure
equipped with a small window for in situ observation of liquids. The liquid could be anal- reg-
ulator (where CO2 is in the gaseous state), and gaseous and liquid species can then be
ysed using
separated by topconventional
and bottom streams method,
in a smallsuch as ion
separator. Thechromatography,
separator could even beliquid chromatography, etc.
equippedEvenwith if there
a small is noforseparate
window liquidof phase
in situ observation in liquid
liquids. The the CO could stream,
2 be an- liquids may condense
alysed using conventional method, such as ion chromatography, liquid chromatography,
inside
etc.
the heated regulator due to the pressure drop, as shown in Figure 6. An ion chro-
matograph
Even if thereanalysis of the
is no separate liquidliquid
phase inintheFigure 6 showed
CO2 stream, liquids may that it consisted of both nitric and
condense
inside the heated
sulphuric regulator
acid. due to the pressure
Installation drop, as
of a drain onshown in Figure 6. An
the regulator ion chro-allow for sampling of such
would
matograph analysis of the liquid in Figure 6 showed that it consisted of both nitric and
liquids while at the same time providing a simple method
sulphuric acid. Installation of a drain on the regulator would allow for sampling of such for purging the regulator for
liquids and particles. It is expected that routine purging would
liquids while at the same time providing a simple method for purging the regulator for decrease the need for
liquids and particles. It is expected that routine purging would decrease the need for
maintenance due to corrosion of the regulator material
maintenance due to corrosion of the regulator material and solid accumulation. It may
and solid accumulation. It may also
improve
also improvethe chemical
the chemical analysis.
analysis. Typical
Typical signs signs
of liquid of liquid
condensation condensation
inside or down- inside or downstream
stream
the the regulator
regulator arefluctuations
are fluctuations in the inwater analysis,analysis,
the water as shown inas theshown
example inin the example in Figure 7.
Figure 7.

Figure 7. Water analysis from two experiments. The blue line is from an experiment where the
Figure
exhaust CO Water
7.2 was analysis
saturated from
with sulfuric acidtwo experiments.
(after 11 Theinblue
h), which precipitated linegasis from an experiment where the
the heated
regulator and introduced significant fluctuations in the water
exhaust CO2 was saturated with sulfuric acid (after 11 h), which content. The black line is from an- precipitated in the heated gas
other experiment without sulfuric acid.
regulator and introduced significant fluctuations in the water content. The black line is from another
4. Monitoring Multiple
experiment withoutCO 2 Streams
sulfuric acid.
For single CO2 streams, it should be relatively straightforward to document that the
quality is within a given specification by using the earlier mention techniques. If no reac-
tion or corrosion occurs, the content of impurities should be the same along the whole
transport system. Changes in impurity concentration would indicate that chemical reac-
tions or even corrosion is taking place. This will obviously require multiple analysing
points.
For a large CO2 transportation network, there will be CO2 streams from several CO2
sources gathered in large transport lines or temporary stored for ship transport (see the
Processes 2021, 9, 570 8 of 12

4. Monitoring Multiple CO2 Streams

For single CO2 streams, it should be relatively straightforward to document that the
quality is within a given specification by using the earlier mention techniques. If no reaction
or corrosion occurs, the content of impurities should be the same along the whole transport
system. Changes in impurity concentration would indicate that chemical reactions or even
corrosion is taking place. This will obviously require multiple analysing points.
For a large CO2 transportation network, there will be CO2 streams from several CO2
sources gathered in large transport lines or temporary stored for ship transport (see the
Processes 2021, 9, x FOR PEER REVIEW 9 of 13
illustration in Figure 8). The different CO2 streams might have different impurities in
both type and concentration. Even if these streams are stable individually (no chemical
reactions), type and concentration.
chemical reactions Even if theseoccur
could streamswhen
are stable
theindividually
streams(noare
chemical
mixed. reac-Thus, there is a
tions), chemical reactions could occur when the streams are mixed. Thus, there is a need
need to documentto documentthe
theimpurity content
impurity content before before
and after and after mixing.
mixing.

Figure 8. Illustration of several CO2 streams mixed in a hub for then to follow one main flow to storage or utilization.
Figure 8. Illustration of several CO2 streams mixed in a hub for then to follow one main flow to
storage or utilization.
4.1. Reactions
Due to the large number of possible impurities, the number of possible chemical re-
4.1. Reactions
actions is high. The most important reactions that were identified in the Kjeller dense
phase CO2 project (KDC) [3,4,9,15,21] were
Due to the large numberH S of possible
3NO → SO H impurities,
O 3NO the number of (1)
possible chemical
reactions is high. The most important
2NO O reactions
→ 2NO that were identified(2)
in the Kjeller dense
SO H O NOwere
phase CO2 project (KDC) [3,4,9,15,21] → NO H SO (3)
2NO H O → HNO HONO (4)
Nitrogen dioxide is a strong oxidation agent and will readily react with hydrogen
sulphide to sulphur dioxide + 3NO
H2 S and → SO(1)).
water 2(reaction 2 +AsHlong +oxygen
2 Oas 3NOis present, nitro- (1)
gen dioxide will be regenerated according to reaction (2). It should be noted that due to
this regeneration, only trace amount of NO
+ 2 isOrequired to oxidise H2S as long as oxygen
2NO 2 → 2NO2 (2)
is present. The only NO2 sink will be formation of nitric acid or corrosion.
Sulphuric acid is formed according to reaction (3), but to form liquid acid inside the
SO2 +observed
transport system, it has been
H2 O + NO → NO + H2 SO
that the2SO2 content needs to 4 50–60 ppmv
exceed
(3)
before H2SO4 forms and precipitates as a liquid phase (25 °C and 100 bar).
2NO + H2 Oguidance
Reactions (1) to (4) could2be used for
→ HNO + HONO
when3interpreting the results from the
(4)
analyses of the CO2 streams. This will be discussed further in the following chapters.
Nitrogen dioxide is a strong oxidation agent and will readily react with hydrogen
sulphide to4.2.
sulphur dioxide
False Accordance withand water (reaction (1)). As long as oxygen is present, nitrogen
the Specification
dioxide will beMonitoring
regenerated according
the composition of a COto2 stream
reaction where(2). It should
the impurities maybe noted
react before that due to this
the analysis introduces the need for special evaluations in addition to the chemical analy-
regeneration, only
sis. For tracemonitoring
example, amountparameters
of NO2such is required to oxidise
as flow, pressure, H2 S ascould
and temperature long as oxygen is
present. The only
reveal NO2risk
possible sink will be formation
of precipitation of nitric
(see, for example, Figure acid or corrosion.
3). Comparison of analyses
at different positions can also provide valuable information. All of this should be com-
Sulphuric acid is formed according to reaction (3), but to form liquid acid inside the
bined and compared with the known chemical reactions and limits (see Section 4.1) and
transport system, it has
also compared been observed
to thermodynamic models that the SO2 content needs to exceed 50–60 ppmv
(if available).
before H2 SO4 forms
Examplesand precipitates
of how misleading theas a liquid
analyses can bephase (25in◦Figures
are shown C and9 and
10010.bar).
The
injected impurity concentrations were the same in both experiments [9] but the injections
Reactions
of the impurities were started at different times. In Figure 9, it is quite clear that no chem-results from the
(1)–(4) could be used for guidance when interpreting the
analyses ofical
thereactions
CO2 streams. ThisHwill
took place when 2O, SObe2, Odiscussed
2, and H2S werefurther
injectedin
as the
they following
all reached chapters.
stable target values. Once the NO2 injection started, reactions took place, and an aqueous
phase of H2SO4 and HNO3 precipitated. In the other experiment (Figure 10), where all the
4.2. False Accordance with the Specification
impurities were injected simultaneously from start-up, the analysis showed practically
zero content
Monitoring theofcomposition
H2S, and SO2 andofO2 were
a CO much lower than the injected content. The read-
2 stream where the impurities may react before
ings in Figure 9 could therefore erroneously lead to the conclusion that the CO2 stream
the analysiswas
introduces the
in accordance need
with for special
the specification, evaluations
even inpercent
if more than 70 addition
of theto the chemical analysis.
impurities
For example, monitoring parameters such as flow, pressure, and temperature could reveal
possible risk of precipitation (see, for example, Figure 3). Comparison of analyses at
different positions can also provide valuable information. All of this should be combined
and compared with the known chemical reactions and limits (see Section 4.1) and also
compared to thermodynamic models (if available).
Processes 2021, 9, 570 9 of 12

Examples of how misleading the analyses can be are shown in Figures 9 and 10. The
injected impurity concentrations were the same in both experiments [9] but the injections of
the impurities were started at different times. In Figure 9, it is quite clear that no chemical
reactions took place when H2 O, SO2 , O2 , and H2 S were injected as they all reached stable
target values. Once the NO2 injection started, reactions took place, and an aqueous phase
of H2 SO4 and HNO3 precipitated. In the other experiment (Figure 10), where all the
impurities were injected simultaneously from start-up, the analysis showed practically
zero content of H2 S, and SO2 and O2 were much lower than the injected content. The
readings in Figure 9 could therefore erroneously lead to the conclusion that the CO2 stream
was in accordance with the specification, even if more than 70 percent of the impurities
were missing due to reactions. The reason for the missing impurities
Processes 2021, 9, x FOR PEER REVIEW 10 of 13 was reaction to acids
Processes 2021, 9, x FOR PEER REVIEW 10 of 13
(reactions (3) and (4)) and solid formation. If all the sulphur species (SO2 + H2 S) react
to sulphuric acid (reaction (1) + reaction (3)), about 500 g would be produced per ton
were missing due to reactions. The reason for the missing impurities was reaction to acids
CO
were . A system
2missing
(reactions (3) due
and to
transporting
(4))reactions.
and solidThe
1 megaton
reason
formation. forallthe
If
per
themissing
yearspecies
would
sulphurimpurities (SO
in this
was reaction case
2 + H2S) react to
produce 500 tonnes of
to acids
acid per(3)
(reactions
sulphuric year.
acidand This
(4))
(reaction and large
(1) +solid amount
formation.
reaction of
(3)), about acid
If all
500the would
sulphur
g would most
specieslikely
be produced (SO +H
per2 ton threaten
2S) 2react
CO . A to the integrity of the
sulphuric acid (reaction
transportation
system transporting system (1) +due
1 megaton reaction
perto (3)), about
corrosion,
year would 500
and
in this g would
caseroutinesbe produced
produce must
500 beper
tonnes tonper CO2. A
ofimplemented
acid to prevent (and
system transporting
year. This large amount1 megaton per year would
of acid would likelyin this case produce
integrity500 tonnes of acid per
detect) such a situation. This most
emphasises threaten
thetheimportance of the
oftransporta-
several measuring points to
year. This large
tion system dueamount of acid
to corrosion, andwould most
routines likely
must threaten the integrity
be implemented to preventof(and
the transporta-
detect)
ensure
suchsystem
tion sufficient
a situation.
due toThis control
emphasises
corrosion, over
and the the transport
importance
routines must of system,
beseveral and
measuring
implemented the
points
to preventdesign
to(and
ensure of the analysing system
detect)
must
such take
sufficient intoThis
control
a situation. consideration
over the transport the
emphasises that
system, newtheproducts
and
importance design from
of the
of several chemical
analysing
measuring system reactions
points must
to ensure could appear in
take into consideration
sufficient
addition control thattransport
over impurities
to those the new products
system,from
included andchemical
inthethe reactions
design could
of the appearsystem
analysing
specification. in addi-must
tioninto
take to those impurities included
consideration that newinproducts
the specification.
from chemical reactions could appear in addi-
tion to those impurities included in the specification.

Figure 9. The result of first injecting of 300 ppmv H2O, 350 ppmv O2, 100 ppmv of SO2, and H2S
Figure
until the 9. The result
concentration of first
stabilized injecting
in 100 of 300
bar CO2 and 25 °C.ppmv H2 O,
After about 350
50 h, the ppmv
injection O , 100 ppmv of SO2 , and H2 S
of2100
ppmv NO 2 was started and immediately a reaction between the impurities ◦
until the concentration stabilized in 100 bar CO2 and 25 C. After about 50 h, the injection of 100
caused a deviation
Figure 9. The
from the result[9].
setpoints of first injecting of 300 ppmv H2O, 350 ppmv O2, 100 ppmv of SO2, and H2S
ppmv
until the NO 2 was started
concentration andinimmediately
stabilized 100 bar CO2 anda25 reaction
°C. Afterbetween
about 50 h,the
the impurities caused a deviation from
injection of 100
ppmv NO2 was started
the setpoints [9]. and immediately a reaction between the impurities caused a deviation
from the setpoints [9].

Figure 10. The result of simultaneous injection of 300 ppmv H2O, 350 ppmv O2, 100 ppmv of NO2,
SO2, and H2S in 100 bar CO2 and 25 °C. None of the impurities reached their setpoint due to reac-
tions [9]. The O2 content was not measured due to experimental problems.

Figure 10. The result of simultaneous injection of 300 ppmv H2O, 350 ppmv O2, 100 ppmv of NO2,
Figure 10. The result of simultaneous injection of 300 ppmv H2 O, 350 ppmv O2 , 100 ppmv of
SO2, and H2S in 100 bar CO2 and 25 °C. None of the impurities reached their setpoint due to reac-
NO2[9].
, SO ◦ C. None of the impurities reached their setpoint due to
tions The2,O
and H2 Swas
2 content in 100 bar CO2due
not measured andto 25
experimental problems.
reactions [9]. The O2 content was not measured due to experimental problems.
 tored, it is possible to identify upsets. If no reactions take place, a simple mass balance o
the small streams should give the composition of the mixed stream.
However, even if most of the chemical reactions are known, a full simulation of th
mixing of multiple CO2 streams with occurring reactions is quite a complex process. Pa
Processes 2021, 9, 570 rameters such as chemical kinetics and competition between reactions, surface adsorption
10 of 12
and catalysing effects are not fully understood at present. Another aspect that would add
complexity to the simulation occurs if a separate aqueous phase forms and accumulate i
the system. The presence of an aqueous phase might not only change the kinetical param
4.3. Simulationeters
and Predicting
but could also change the ongoing reactions or introduce new reactions, includin
By using reactions reactions,
corrosion thatwere
(1) to (4), we would further
able complicate
to make simple the simulation.
predictions of the results of
mixing CO2 streams. TheIfresults
all sidefrom the and
streams KDC-project
the mainare
COcurrently being implemented
2 stream (Figure in a thermody
11) are monitored,
namic
it is possible to model
identify developed
upsets. If no by OLI Systems
reactions [22,23].
take place, The model
a simple massis balance
still in the developmen
of the
small streamsstage
shouldforgive
use intheCCS systems. of the mixed stream.
composition

Figure11.
Figure Themixing
11.The mixingofofdifferent
different CO
CO 2 streams
2 streams with
with known
known impurities
impurities and and concentration
concentration in a hub.
in a hub.

However, even if most

Even thoughof the chemical
an accurate reactionsmight
simulation are known, a fullthe
be complex, simulation
predictionoffrom
the the reac
mixing of multiple CO streams with occurring reactions is quite a complex process.
tion alone2would give valuable input to the outcome of mixing streams or building Pa- spec
rameters suchifications.
as chemical kinetics and competition between reactions, surface adsorption,
Table 1 shows an example of such prediction, where no aqueous phase wi
and catalysingform
effects arethe
since notsum
fully
of understood
H2S and SO2atis present.
well belowAnother aspect that
the threshold would
value add (3) (se
for reaction
complexity toSection
the simulation occurs if a outcome
4.1). An interesting separate of
aqueous phaseisforms
the exercises and accumulate
that reaction 1 leads toinhigher con
the system. Thecentration
presenceofofSO 2 and
an H2O, phase
aqueous which might
in certain
notsituations
only changecould
theexceed theparam-
kinetical threshold for re
eters but couldaction
also (3), eventhe
change if the original
ongoing SO2 content
reactions was well below
or introduce this level.including
new reactions, Further increase o
impurity
corrosion reactions, thatcontent
wouldcould eventually
further result
complicate theinsimulation.
exceeding the accepted specification. Yet, re
search
The results from has
theshown that the concentrations
KDC-project are currently in Table
being 1 are safe [3].in a thermody-
implemented
namic model developed by OLI Systems [22,23]. The model is still in the development
The
stage for use in understanding
CCS systems. of ongoing processes in the transported CO2 might ease the simulations but it
might also untangle the complex analysis after mixing several CO2 streams loaded with even mor
Even though an accurate simulation might be complex, the prediction from the reaction
types of impurities.
alone would give valuable input to the outcome of mixing streams or building specifications.
Table 1 shows an example of such prediction, where no aqueous phase will form since the
sum of H2 S and SO2 is well below the threshold value for reaction (3) (see Section 4.1). An
interesting outcome of the exercises is that reaction 1 leads to higher concentration of SO2
and H2 O, which in certain situations could exceed the threshold for reaction (3), even if the
original SO2 content was well below this level. Further increase of impurity content could
eventually result in exceeding the accepted specification. Yet, research has shown that the
concentrations in Table 1 are safe [3].

Table 1. Prediction based only on the reactions; the input impurities are the sum of all streams while
the output concentration are the results of the reactions.

Concentration Concentration
Impurity before Reaction after Reaction
(ppmv) (ppmv)
NO2 10 3
O2 10 0
H2 O 30 39
SO2 10 19
H2 S 9 0
NO 0 7
Processes 2021, 9, 570 11 of 12

The understanding of ongoing processes in the transported CO2 might ease the
simulations but it might also untangle the complex analysis after mixing several CO2
streams loaded with even more types of impurities.

5. Conclusions
Analysis of impurities in dense phase CO2 streams requires pressure reduction and
phase transformation to gaseous CO2 and is therefore more complicated than analysis of
gaseous CO2 alone. Possible reactions of impurities make such analysis even more chal-
lenging, particularly since even 99.95% pure CO2 (food grade) has the potential to produce
acids and solids, which may have negative effects on the analysis system. Impurities may
partition between phases, and this may be another complicating factor during analysis.
Precipitation of solids and liquids on the low-pressure side of the analysis line may also
introduce challenges due to particle accumulation, corrosion, and fluctuation of certain
impurities that interact with the liquids.
Multiple analysing points along the CO2 transportation system (e.g., inlet and outlet)
will increase the possibility to reveal ongoing processes such as chemical reactions and/or
corrosion. If multiple CO2 streams are being merged in a network, it might be necessary
to analyse all CO2 streams before and after mixing in order to ensure that the specifica-
tion is fulfilled. Such analysis could also be assisted by predictions based on identified
chemical reactions.

Author Contributions: Conceptualization, B.H.M., G.S., and A.D.; methodology, B.H.M.; formal
analysis, B.H.M. and G.S.; investigation, B.H.M.; data curation, G.S.; writing—original draft prepara-
tion, B.H.M. and G.S.; writing—review and editing, B.H.M., G.S., and A.D.; visualization, B.H.M.;
supervision, G.S. and A.D.; project administration, G.S. and A.D.; funding acquisition, G.S. and A.D.
All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by CLIMIT, project number 618094. Shell, Total, ArcelorMittal,
Equinor, Vallourec, and Gassco are acknowledged for financial support to the KDC-III project.
Institutional Review Board Statement: Not applicable (no animals were used in this work).
Informed Consent Statement: Not applicable.
Data Availability Statement: This present paper describes a method/approach and do therefore
only show a few examples, of which most already has been published.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

atm atmosphere (pressure unit)

CCS Carbon capture and storage
GC Gas chromatography
FTIR Fourier transform infrared spectroscopy
IR Infrared
ppmv parts per million by volume (or by mole)
SEM Scanning electron microscopy
UV Ultraviolet
XRD X-ray powder diffraction
EDS Energy-dispersive X-ray spectroscopy

References
1. Towards Zero Emissions CCS from Power Stations Using Higher Capture Rates or Biomass. Available online: https://
ieaghg.org/publications/technical-reports/reports-list/9-technical-reports/951-2019-02-towards-zero-emissions (accessed on
23 March 2021).
2. Bui, M.; Adjiman, C.S.; Bardow, A.; Anthony, E.J.; Boston, A.; Brown, S.; Fennell, P.S.; Fuss, S.; Galindo, A.; Hackett, L.A.; et al.
Carbon capture and storage (CCS): The way forward. Energy Environ. Sci. 2018, 11, 1062–1176. [CrossRef]
Processes 2021, 9, 570 12 of 12

3. Morland, B.H.; Tjelta, M.; Norby, T.; Svenningsen, G. Acid reactions in hub systems consisting of separate non-reactive CO2
transport lines. Int. J. Greenh. Gas Control 2019, 87, 246–255. [CrossRef]
4. Morland, B.H.; Norby, T.; Tjelta, M.; Svenningsen, G. Effect of SO2 , O2 , NO2 , and H2 O Concentrations on Chemical Reactions and
Corrosion of Carbon Steel in Dense Phase CO2 . Corrosion 2019, 75, 1327–1338. [CrossRef]
5. Morland, B.H.; Dugstad, A. Corrosion of carbon steel in water equilibrated with liquid and supercritical CO2 . In Proceedings of
the CORROSION/2016, Vancouver, BC, Canada, 6–10 March 2016; NACE International: Houston, TX, USA, 2016; p. 7740.
6. Yevtushenko, O.; Bettge, D.; Bäßler, R.; Bohraus, S. Corrosion of CO2 transport and injection pipeline steels due to the condensation
effects caused by SO2 and NO2 impurities. Mater. Corros. 2015, 66, 334–341. [CrossRef]
7. Paul, S.; Shepherd, R.; Bahrami, A.; Woollin, P. Selection of materials for high pressure CO2 transport. In Proceedings of the Third
International Forum on the Transportation of CO2 by Pipeline, Newcastle, UK, 18–19 June 2012.
8. Adi Jatmoko, F.; Kusrini, E. Analysis of CO2 transmission pipelines for CO2 enhanced oil recovery networks: Gas field X to oil
field Y. E3s Web Conf. 2018, 67, 04009. [CrossRef]
9. Dugstad, A.; Halseid, M.; Morland, B. Testing of CO2 specifications with respect to corrosion and bulk phase reactions. Energy
Procedia 2014, 63, 2547–2556. [CrossRef]
10. Morland, B.H.; Halseid, M.; Dugstad, A.; Tjelta, M.; Svenningsen, G. Sulphur formation in CCS streams. In Proceedings of the
TCCS-9, Trondheim, Norway, 12–14 June 2017.
11. Platform, Z.E. The Costs of CO2 Capture: Post-Demonstration CCS in the EU; Zero Emission Platform: Brussels, Belgium, 2011.
12. Suzuki, T.; Toriumi, M.; Sakemi, T.; Masui, N.; Yano, S.; Fujita, H.; Furukawa, H. Conceptual Design of CO2 Transportation
System for CCS. Energy Procedia 2013, 37, 2989–2996. [CrossRef]
13. Mulighetsstudier Av Fullskala CO2-Håndtering i Norge. Available online: https://www.regjeringen.no/globalassets/
departementene/oed/pdf/mulighetsstudien.pdf (accessed on 23 March 2021).
14. Herron, S.; Myles, P. Quality Guidelines for Energy System Studies—CO2 Impurity Design Parameters; Report no. DOE/N/NETL-
341/011212; National Energy Technology Laboratory: Pittsburg, PA, USA, 2013.
15. Morland, B.H.; Tjelta, M.; Dugstad, A.; Svenningsen, G. Corrosion in CO2 Systems with Impurities Creating Strong Acids.
Corrosion 2019, 75, 1307–1314. [CrossRef]
16. Chapoy, A.; Burgass, R.; Tohidi, B.; Austell, J.M.; Eickhoff, C. Effect of Common Impurities on the Phase Behavior of Carbon-
Dioxide-Rich Systems: Minimizing the Risk of Hydrate Formation and Two-Phase Flow. SPE J. 2011, 16, 921–930. [CrossRef]
17. Onyebuchi, V.E.; Kolios, A.; Hanak, D.P.; Biliyok, C.; Manovic, V. A systematic review of key challenges of CO2 transport via
pipelines. Renew. Sustain. Energy Rev. 2018, 81, 2563–2583. [CrossRef]
18. Dugstad, A.; Halseid, M.; Morland, B. Corrosion in dense phase CO2 pipelines—Consequences of upset conditions. In Proceedings
of the 4th International Forum on Transportation of CO2 by Pipeline, Newcastle, UK, 17–19 June 2013.
19. Dugstad, A.; Halseid, M.; Morland, B.; Sivertsen, A.O. Dense phase CO2 corrosion and the impact of depressurization and
accumulation of impurities. In Proceedings of the CORROSION/2013, Orlando, FL, USA, 17–21 March 2013; NACE International:
Houston, TX, USA, 2013; p. 2785.
20. Dugstad, A.; Halseid, M.; Morland, B.; Sivertsen, A.O. Corrosion in dense phase CO2 —The impact of depressurisation and
accumulation of impurities. Energy Procedia 2013, 37, 3057–3067. [CrossRef]
21. Dugstad, A.; Halseid, M.; Morland, B. Experimental techniques used for corrosion testing in dense phase CO2 with flue gas
impurities. In Proceedings of the CORROSION/2014, San Antonio, TX, USA, 9–13 March 2014; NACE International: Houston,
TX, USA, 2014; p. 4383.
22. Wang, P.M.; Anderko, A.; Young, R.D. A speciation-based model for mixed-solvent electrolyte systems. Fluid Phase Equilibria
2002, 203, 141–176. [CrossRef]
23. Springer, R.D.; Wang, Z.; Anderko, A.; Wang, P.; Felmy, A.R. A thermodynamic model for predicting mineral reactivity in
supercritical carbon dioxide: I. Phase behavior of carbon dioxide—water—chloride salt systems across the H2 O-rich to the
CO2 -rich regions. Chem. Geol. 2012, 322, 151–171. [CrossRef]