PROCEEDINGS OF ECOS 2012 - THE 25TH INTERNATIONAL CONFERENCE ON

EFFICIENCY, COST, OPTIMIZATION, SIMULATION AND ENVIRONMENTAL IMPACT OF ENERGY SYSTEMS

JUNE 26-29, 2012, PERUGIA, ITALY

Process efficiency and optimisation of

precipitated calcium carbonate (PCC) production
from steel converter slag
Hannu-Petteri Mattilaa, Inga Grigalinaitb, Arshe Saidc, Sami Filppulad,
Carl-Johan Fogelholme and Ron Zevenhovenf
a

Thermal and Flow Engineering Laboratory, bo Akademi University, Turku, Finland,

[email protected], CA
b
Thermal and Flow Engineering Laboratory, bo Akademi University, Turku, Finland,
currently at Feyecon, Weesp, The Netherlands, [email protected]
c
Department of Energy Technology, School of Engineering, Aalto University, Espoo, Finland,
[email protected]
d
Thermal and Flow Engineering Laboratory, bo Akademi University, Turku, Finland, [email protected]
e
Department of Energy Technology, School of Engineering, Aalto University, Espoo, Finland,
[email protected]
f
Thermal and Flow Engineering Laboratory, bo Akademi University, Turku, Finland, [email protected]

Abstract:
CO2 emissions could be reduced with capture and storage (CCS) methods. CCS causes costs for the
industry, creating an extra barrier for implementation of these techniques. A recently developed mineral
carbonation process producing a valuable product covering process implementation costs is advanced
towards commercial scale. In this two-step process, calcium-rich steel converter slag is treated with an
aqueous ammonium salt solution to selectively extract the calcium. Then the dissolved calcium is removed
from the process liquid as pure precipitated calcium carbonate (PCC) by introducing gaseous CO2 to the
system. This concept would utilise waste material from steel industry, and spare the natural resources used
for conventional PCC production. Experimental results have shown that papermaking grade PCC can be
produced with ammonium nitrate, chloride or acetate solvents. The used solvent has also been regenerated
and recycled between the process stages, which reduces the need for fresh solvent and lowers the process
costs. To maximise the PCC production in a continuous process, a thermodynamic process model in Aspen
Plus is now constructed. The modelling results are validated with a series of semi-continuous experiments. It
is found that by using staged carbonate precipitation the yield of pure product is increased. Ammonium
nitrate is observed to enable the highest conversion of calcium from steel slag to calcium carbonate also in
continuous operation. At ambient conditions the process steps generate heat but this low temperature heat
cannot be utilised. The mixing and pumping energies for process operation are small in comparison to the
heat duties in chemical reactors. Preliminary results on solid-liquid separation studies, on washing of the
solid outlet streams, as well as on recovery and make-up needs of process chemicals are discussed. The
nd
work is a continuation of work presented by Said et al. at 22 ECOS in Brazil 2009.

Keywords:
Ammonium salt solution, Chemical thermodynamics, Mineral carbonation, pH swing process, Steel
converter slag.

1. Introduction
It is a widely accepted fact that the climate on Earth is changing because of human activities. One
major factor causing global warming is the high emission rate of so-called greenhouse gases. Of
these, carbon dioxide is an especially problematic one, since modern economies are based on
combustion of fossil, carbon-based fuels. Thus the amount of CO2 emissions is also high compared
to other greenhouse gas emissions like N2O that are possibly more harmful to the climate calculated
per mass only [1].
Various methods have been proposed to diminish CO2 emissions. Apart from improving the
efficiencies of existing combustion processes, choosing CO2 neutral fuels and decreasing the need
114-1

of power and electricity on global scale, carbon capture and storage (CCS) technologies are seen as
one means of reducing carbon dioxide emissions. In CCS, carbon is removed from the gases at
different stages in combustion or gasification processes depending on the chosen technology. The
capture techniques are already on a relatively mature level, especially for oxygen-free gases, since
they have during several decades been applied for industrial processes to produce pure CO2 [2].
Thus, one major threshold in CCS applications is currently the lack of a permanent, leakage-free
storage [1].
Mineral carbonation, where the carbon dioxide gas is reacted directly with some magnesium or
calcium containing compounds, is one option for safe storage. Simultaneously it offers the
possibility of removing the separate capture stage. However, the mineralization reactions with
natural materials are very slow by their nature, and thus development is still needed before
industrial applications [3].
Apart from this, there are industrial waste materials, in which calcium or magnesium is present in
more reactive forms [3-7]. For example, steel slags from steelmaking plants have been found to
react with suitable solvents in such a way that calcium is selectively extracted from the slag matrix.
This selective reaction enables the utilization of the produced carbonates in other industrial
processes. Thus, two waste streams, CO2 in flue gases and calcium in steel slags can be used for
generating a marketable calcium carbonate product [3, 6, 7].
In this paper, a two-step pH swing process (Fig. 1) operating at ambient temperature and pressure is
discussed. In this concept, calcium from steel converter slag is first leached out with ammonium
chloride, nitrate or acetate solution under slightly acidic conditions, and then, after separating the
slag residue, the dissolved calcium is carbonated in a separate reaction vessel with a feed of gaseous
CO2. This CO2 gas could be absorbed to the process liquid directly from the flue gas stream from
e.g. a lime kiln or a steel plant, and thus there would be no need for a separate capture unit.
Naturally, the concentration of CO2 in the flue gas would be a crucial factor for the process kinetics.
In the carbonation step the ammonium salt solvent is regenerated and can be recycled back to the
extraction reactor after the precipitated carbonate product has been separated.

Fig. 1. A principal process scheme of the two-step pH swing process.

In this text especially the efficiencies and losses of different process steps are evaluated based on
both thermodynamic modelling and experimental work. The development and scale-up of the
process from laboratory scale batch tests towards industrial applications is discussed too [8-9].
The general target of the current study is to produce calcium carbonate that would meet the
demands of papermaking applications; if the product could be utilised as filler or a coating pigment,
Precipitated Calcium Carbonate (PCC), the profits of selling this material would according to
preliminary calculations cover the costs of the carbonation process. This in turn would remove or
lower the financial threshold of the utilization of CCS technologies, which usually bring large costs
to CO2 generating industries or power plants. Also, the financial incentive would enable a larger
scale demonstration of this mineral carbonation concept [3, 6, 9]. To ensure a high quality product,
the CO2 feed should be free of sulphur and particulate matter.

114-2

2. Process development
2.1. Background
During earlier studies [6, 10, 11], several batch experiments have been performed to verify the
effects of different process changes such as temperature and CO2 pressure on the process chemistry.
It was found that the reaction steps, which both generate some heat (see Section 4), could be
performed at room temperature (2030 C). An increase in temperature to 70 C had hardly any
effect on the extraction stage kinetics. In carbonation the shape of the precipitated particles changes
with the temperature. This restricts the applicable carbonation temperatures below 3040 C. The
effect of carbon dioxide pressure was limited to changes in process kinetics, the precipitation rate
being slower at lower partial pressures of CO2. Thus, 2030 C was chosen as a suitable process
operation temperature interval. At these temperatures the kinetic reaction rates are acceptable [10]
and solubility and volatility of gaseous components such as NH3 and CO2 are beneficial for the
process.
Steel converter slag was reported to contain calcium as free lime (CaO), larnite (Ca2SiO4) and
various calcium-iron compounds that seem not to react with ammonium salt solvents. The
dissolution reactions of lime and larnite are presented as (R1) and (R2). Carbonate precipitation
chemistry can be summarised as (R3) and (R4). X in the reaction equations represents Cl-, NO3- or
CH3COO- (acetate), depending on the chosen salt.

CaO(s) + 2NH 4 X (aq) + H 2O(l ) CaX 2 (aq) + 2NH 4OH (aq)

(R1)

2CaO SiO2 (s) + 2NH 4 X (aq) + H 2O(l ) CaX 2 (aq) + CaO SiO2 (s) + 2NH 4OH (aq) (R2)
2 NH 4 OH (aq ) + CO2 ( g ) (NH 4 )2 CO3 (aq ) + H 2 O(l )

(R3)

(NH 4 )2 CO3 (aq ) + CaX 2 (aq ) CaCO3 (s) + 2 NH 4 X (aq )

(R4)

In the batch experiments it was also observed that if solvents with molarities higher than 1.0 mol/L
or solid-to-liquid ratios higher than 100 g/L were used, also some iron and manganese was extracted
from the steel slag, decreasing the purity and whiteness of the process solution and the produced
carbonates [10-12]. Thus, these specifications were used in the current work. Mainly ammonium
chloride solutions were used, both in modelling work and in experiments, since the available data
were most complete for this solvent, and also because it is cheaper than the other two ammonium
salts. Some observations of experimental work with ammonium nitrate will be presented in later
sections.
In larger scale, the process should preferably be operated on continuous basis to achieve
presumably lower operational costs and better adaptability to changes in feedstock quality.
Problems arising especially from the continuous operation as identified already by [9, 13] are the
losses of solvent components (NH3 and water vapour, ammonium and calcium salt precipitates) and
dissolution of excess carbon dioxide as bicarbonate and carbonate ions in the carbonation step. The
solvent losses cause an unnecessary increase in process costs, both as a need of a solvent make-up
but also as a need for purification of precipitates and purged gases. On the other hand, if excess
dissolved carbon species are recycled from carbonation to the extraction unit, solid calcium
carbonate is precipitated on the slag particles, lowering the overall production rate of pure carbonate
product (PCC). These problems exist to some extent also in a batch type process.

2.2. Process modelling

Thermodynamic modelling and simulation software Aspen Plus 7.2 were utilised to study the
possibilities to decrease losses of both carbonate product and solvent components. The software
114-3

also provided information on sizes and compositions of different process streams. The model design
is shown in Fig. 2. It consists of one extraction step EXTRACTO, and two carbonation steps,
CARBONAT and SETTLER. All these reactors are so called RGIBBS units, which calculate
the output by minimizing Gibbs free energy of the system. Carbonation is divided in two stages to
enhance the precipitation rate of calcium (see Section 3.1.).
25 ton/h dry steel converter slag, containing 5%-wt CaO and 59%-wt Ca2SiO4, the remaining 36%wt consisting of inert compounds, is fed to the extraction reactor together with an ammonium salt
solution (~1 mol/L NH4Cl). The solid-to-liquid ratio used in modelling is approximately 100 g slag
in one litre of solvent [10-11].
In the first carbonate precipitation unit, CARBONAT, 85-100% of the calcium-rich solution from
extraction unit is put in contact with 25 ton/h flue gas containing 20% CO2 and 80% N2, being
approximately the composition of lime kiln flue gases. The 300 C flue gas is fed to the process via
a cooling unit to estimate the released heat, when the gas is brought down to room temperature
(20C). The flue gas feed amount is adjusted so that approximately 5% of the CO2 gas leaves the
system unreacted. After the first step the solution is flashed to 0.5 bar in VAPORSEP, thus
removing the non-reacted gases. Re-pressurisation to 1.0 bar is done by a separate PUMP unit.
The SETTLER unit is used to increase the pH of the once carbonated solution, so that the
chemical equilibrium can be shifted to favour additional precipitation of calcium carbonate. This is
done by introducing 0-15% of the calcium-rich solution coming from the extraction unit directly to
SETTLER as a bypass stream of CARBONAT. At the same also some calcium is added to the
SETTLER reactor, shifting the equilibrium even further towards carbonate. After this second
precipitation unit the solution is again flashed to 0.5 bar in VAPORSE2 and pressurised to 1.0 bar
with PUMP2 to remove dissolved gases.

Fig. 2. A detailed process scheme of the pH swing process as simulated with Aspen Plus.
Solids are separated from the process solution with RESIDSEP for slag residue and PRODSEP
and PRODSEP2 for carbonate product. These units are based plainly on percentages of separation
efficiency specified for each component. They can be thickeners, filters, hydrocyclones and
combinations of these. Later in this paper some studies on gravitational separation of steel slag are
presented. Both the slag residue (RESIWASH) and produced PCC (PRODWASH) are washed
114-4

to dissolve the chloride salts precipitated on these particles. To demonstrate the recovery of
vaporised ammonia from the carbonation steps, an additional RGIBBS unit, NH3SCRUB is
introduced to the model. In this unit an HCl solution is reacted with NH3 vapour to produce aqueous
NH4Cl. These aspects are discussed in more detail in Section 5.
The solvent liquid is recycled in the process. Also make-up streams for ammonia and water are
included in the model to maintain the balance with vaporization losses. The CO2SEPAR unit is
not active in this model, although it could be used to artificially decrease the concentrations of
carbon species in the recycled stream.

2.3. Experimental work

A set of five experiments was performed to provide information that could not be obtained from the
modelling work, but also to confirm that the modelling results can be applied for predicting the
behaviour of the process in practice. Thus, the product yields and conversion rates from the Gibbs
energy minimization performed in Aspen Plus software and experimental work could be compared
with each other, but also the purity and crystal shape of PCC product, properties that are difficult to
model, were studied in experiments. In future, the exact experimental result values may be used in
modelling work to get better estimates for the large scale stream sizes and energy needs of the
process.
The tests were done at ambient conditions (20C, 1 bar), with a mechanical stirrer (170 rpm) in the
extraction reactor, using 1 mol/L ammonium chloride or ammonium nitrate solution, and
maintaining the slag-to-liquid ratio at approximately 100 g/L during extraction. After each
experiment the process equipment was modified according to the obtained results before the next
experiment (Tables 1 and 2).
Table 1. Experimental parameters
Experiment
Reaction time, min
Ammonium salt
Volume of ammonium salt, ml
System filled in the beginning
Initial slag amount, g
Total slag amount used, g

1
75
NH4Cl
750
No
50
50

2
135
NH4Cl
1500
No
50
70

3
180
NH4Cl
2500
Yes
50
90

4
255
NH4Cl
3500
Yes
50
110

5
255
NH4NO3
2500
Yes
50
110

Table 2. Amount of equipment used in experimental set-ups

Experiment
Filters
Settlers
Pumps
H2O lock
pH adjustment unit
CO2 removal unit

1
2
0
2
0
0
0

2
4
1
4
1
0
0

3
4
0
5
1
1
1

4
4
1
4
1
1
1

5
4
0
4
1
1
1

Table 3. Volumes, volume flows and residence times in experiments 4 and 5

Unit
Extraction
Carbonation
pH adjustment
CO2 removal
Filter equipment

Volume, ml
500
300
500
500
600

Volume flow, ml/min

20
15
20
20
N/A
114-5

Residence time, min

25
20
25
25
N/A

Fig. 3. Process scheme for experiments 4 and 5. 1 Extraction step; 2 Carbonation step; 3
water lock; 4 pH adjustment unit; 5 CO2 removal unit; 6 Settler (not used in experiment 5); 7
Filtration step; 8 Pump.

Fig. 4. Process equipment used in experiments 4 and 5.

Figs. 3 and 4 represent the process equipment of experiments 4 and 5, for which the most successful
and advanced setup was constructed. Volume flows and residence times for these tests are shown in
Table 3. In experiment 4 the whole system was filled with ammonium chloride solution, of which
0.5 L with 50 g steel converter slag (size fraction 125-250 m) was fed to the extraction reactor. 5 g
slag was added to the extractor with 15 minutes interval during the whole experiment. To allow for
the calcium concentration in the extractor to increase towards a constant level, the solution was
allowed to react 5 minutes before the pumps were switched on. Fine particles were continuously
removed from extraction reactor via an overflow of 20 ml/min. The overflow was filtered to prevent
solid particles from entering the carbonation stages. 75% of the filtrate (15 ml/min) was pumped to
the 1st carbonation reactor, where a small overpressure (0.02 bar) of carbon dioxide (99.9996%, Oy
Aga Ab) was maintained with a water lock. CO2 was introduced through a Waters 10 micron Hplc
solvent inlet filter (WAT025531) to provide an even gas distribution. Pure CO2 gas was used
because of the impracticality of actual flue gas, but as mentioned above, this has been found to
mainly affect the process kinetics, if the flue gas is sulphur-free and contains no solid particles.
114-6

The remaining 25% (5 ml/min) of the filtrate from extraction was fed to the 2nd precipitation stage
to adjust the pH of this step to approximately 7. The neutralised solution was transferred by
gravimetric flow into a vessel, where 99.9999% nitrogen (Oy Aga Ab, 0.5 L/min) was bubbled
through the liquid to remove the unreacted CO2. Finally, the slurry was allowed to settle in another
vessel, after which the overflow was filtered with 0.45m/47mm membrane (Pall Life Sciences,
Supor-450) and the particle-free solution (20 ml/min) was recycled back to the extraction reactor.
Since the product was not continuously removed from the system, this process can be described as
semi-continuous. For comparison, the experiment was repeated with ammonium nitrate solvent,
other parameters unchanged. The settler, however, was not used in the nitrate experiment, since it
was observed to have no effect in experiment 4.
Samples were taken of both the slag residue and the produced PCC. They were dried overnight at
105C and analysed with SEM/EDX to get an approximate elemental composition. The
concentrations of dissolved calcium ions were measured after each step with an ion selective
electrode (NICO 2000).

3. Conversion efficiencies of calcium and CO2 to PCC

3.1. Modelling results
Theoretically, based on presented assumptions on extraction chemistry and slag composition, 44.7%
of the calcium present in the feed can be extracted. However, if only one carbonation unit is used,
almost 7% of this extracted calcium will precipitate as a mixture with slag residue in the extraction
reactor. Table 4 represents the effect of different bypass fractions on the efficiencies of both
calcium and carbon dioxide conversions. As can be seen, according to the model already a 5%
bypass stream is enough to reduce the losses by 40%. With a 15% bypass practically all extracted
calcium could be utilised as marketable product. On the other hand, the amount of sequestered CO2
remains constant, since the total amount of CaCO3 precipitate is not changing.
Table 4. Modelled conversion efficiency calculations of calcium and CO2 to PCC with various
process configurations
Bypass fraction
%
0
5
10
15
Reactive calcium feed
kmol/h (ton/h)
108 (4.3)
PCC product
kmol/h (ton/h)
101(10.1) 104(10.4) 107(10.7) 108(10.8)
PCC in residue
kmol/h (ton/h)
7(0.7)
4(0.4)
1(0.1)
0(0)
Loss of marketable calcium
% of reactive input
6.5
3.7
0.9
0.0
Conversion of total Ca to PCC
%
42
43
44
45
CO2 feed
kmol/h (ton/h)
114 (5.0)
CO2 not sequestered
kmol/h (ton/h)
6 (0.3)
%
5.3
Table 5. Modelled changes in solution pH values at different stages of the process
Bypass fraction
pH after extraction
pH after 1st carbonation
pH after 2nd carbonation

114-7

0
10.61
7.68
7.71

5
10.70
7.64
7.87

10
10.83
7.61
8.18

15
11.13
7.61
8.68

1
log[Ca2+] (mol/l)

0,5
0

pCO2
(bar)

-0,5
-1

0.10

-1,5

0.25

-2

1.0

-2,5
-3
5

5,5

6,5

7,5

pH

Fig. 5. Theoretical dependency of dissolved Ca2+ equilibrium concentration at 20C on pH and

CO2 pressure modified from [14].
The chemical explanation for this behaviour can be summarised in Table 5 and Fig. 5. When the pH
of solution is increased, the concentration of dissolved calcium at thermodynamic equilibrium is
decreased. Consequently, the precipitation of CaCO3 is favoured at higher pH values. In the Aspen
model, the increase of pH after both carbonation steps at different bypass rates is almost one pH
unit from 0 to 15% bypass, resulting in the above described shift in marketable carbonate
production efficiency. Thus, pH measurement and adjustment can be used for process control
purposes.

3.2. Experimental results

Experimentally obtained conversion rates (Table 6) are remarkably lower than modelling results
(Table 4). This can be partially explained by experimental inaccuracies, the main reason being the
lower extraction efficiency from the used steel converter slag. As discussed in [10], this might be
due to a long storage time in laboratory conditions. However, in general it can be observed that by
introducing the pH adjustment unit with a 25% bypass stream passing the first carbonation unit, the
total conversion of calcium from slag to precipitated product can be enhanced. Compared to a case
with only two reactors, one extractor and one carbonator (Experiment 1), the setup described in
Figs. 3 and 4 (Experiment 4) was able to increase the conversion efficiency by 13 %-units. By
changing the ammonium chloride solvent to ammonium nitrate an additional increase of 7 %-units
was measured (Experiment 5). When carbon dioxide concentration was decreased before the pH
adjustment unit (Experiment 3), the calcium conversion was 5 %-units lower than in experiment 4.
Nevertheless, results from experiment 2 show that only by introducing an additional settler unit,
compared to experiment 1, a similar conversion efficiency as with the pH adjustment system can be
obtained. From the measured calcium concentrations (Table 7) it can be seen that this is actually
due to a more efficient extraction conversion, depending on the variation of the input slag
composition and on the different operation of pumps during the experiments, which resulted in
longer residence times during extraction.
Fig. 6 shows the recorded pH values from experiments 3-5, where the pH adjustment unit was used.
It can be seen that the levels are stabilising quite well to constant values both in extraction and in
carbonation, especially in experiments 4 and 5. Still, the extraction pH is slightly decreasing with
time, indicating carbonate precipitation also in this reactor. The stable carbonation pH at
approximately 6.3 with a calcium concentration of 0.004-0.007 mol/L is quite well corresponding to
the theoretical values of Fig. 5 (pH 6.3 gives 0.002 mol/L), when the experimental inaccuracy is
taken into account as well. Also, the pH values obtained from Aspen Plus are systematically 1.3-2
units too high compared to the experimental results.
114-8

Table 6. Experimental conversions of steel slag into PCC

Experiment
Reactive calcium feed
PCC product (carbonator)
PCC product (pH adjustment)
CaCO3 (other system)
CaCO3 (total)
Conversion of reactive Ca
Conversion of total Ca

g/h
g/h(% of total)
g/h(% of total)
g/h(% of total)
g/h
%
%

1
5.8
2.5(100)
2.5
17
8

2
4.5
4.2(75)
1.4(25)
5.6
49
22

3
4.3
1.9(51)
0.8(20)
1.1(29)
3.8
35
16

4
3.7
1.4(31)
2.3(51)
0.8(18)
4.5
48
21

5
3.7
2.0(35)
3.5(60)
0.3(5)
5.8
62
28

Table 7. Experimental Ca2+ concentrations after and before extraction reactor

1
0.026
0.007

mol/L
mol/L

2
0.041
0.019

4
0.015
0.007

5
0.018
0.004

10

Experiment 4

Experiment 5

Experiment 3
Experiment 4

Experiment 5

50

100

150

200

250

Time (min)

50

100

150

200

250

Time (min)

10

Experiment 3

Experiment 4

Fig. 6. pH during experiments 3-5 in extraction

(up left), carbonation (right) and pH adjustment
(down left) units.

Experiment 5

Experiment 3

6
0

pH

3
0.012
0.004

10

pH

pH

Experiment
Ca2+ after extraction reactor
Ca2+ before extraction reactor

7
6
5
0

50

100

150

200

250

Time (min)

A remark that can be made based on Fig. 6 is that the pH in the pH adjustment unit has been quite
unstable during the experiments. Thus, applying a process control device to steer the bypass rate as
proposed in the previous section would improve the performance of the experimental setup.

4. Energetic properties of the process

Based on the presented Aspen model, heat duties of different process steps can be estimated. In
Table 8 values are shown for the case with a 15% bypass, which was in above discussion found to
result in highest production rate of pure PCC. It can be observed that all process steps, except for
the flashing units, are exothermic, i.e. release heat. However, as the process is operated at low
114-9

temperature, this heat can in practice not be utilised. Only exception is the cooling unit for flue gas,
where the gas is cooled from 300C to 20C. The maximum amount of reversible work from these
process steps, when the ambient temperature is assumed to be 20C, is shown as exergy in Table 8.
Calculations were made using (1) for heat Q [15].
(1)
T0

Exergy(Q) = (1

) *Q

Table 8. Heat duties, inlet temperatures and pressures as well as exergies of heat in different
process steps according to Aspen Plus model (Fig. 2) with 25 ton/h slag feed and a 15% bypass
Process unit
EXTRACTO
CARBONAT
COOLER
VAPORSEP
SETTLER
VAPORSE2
SUM

Q(kW)
-1638
-2614
-2010
258
-62
215
-5851

in(C)
20
20
300
20
20
20

p(bar)
1
1
1
0.5
1
0.5

Exergy (Q)@T0=293.15K(kW)
0
0
982
0
982

It must also be noted that the heat duties for process units where gas absorption into process
solution, or vaporization of solvent liquid is occurring, include also these values, besides of the
actual heats of chemical reactions. If the results are presented as specific values, the result becomes
66 W/kg slag for the extraction step. During carbonation, 390 W/kg PCC is generated. This value is
a sum of all units listed in Table 8, excluding the extractor.
Two energy-related topics not yet discussed are the energy inputs needed for pumping the process
liquids and mixing the contents of the reactor units, especially during extraction, where a large
amount of solids is blended with a liquid stream. From the presented Aspen model the power
consumption of the two pumps to pressurise the liquid streams after flashing is 2.8 kW for PUMP
and 3.4 kW for PUMP2, calculated according to P=p*V , where p is the pressure change (0.5
bar) and V is the volume flow. Compared to heat duties in reactors, these are quite small numbers.
In general, also other pumping equipment will be needed for the process, but since the pressure
losses can be assumed to remain low with small distances and height differences, even for flows of
200-250 m3/h.
For mixing, the energy dissipation from the stirrer can be calculated with (2) [16], where Po is
Power number, n is the frequency of rotation for the mixer (1/s), d is the diameter of the mixer (m),
and V is the total volume (m3).

Po n 3 d 5
=
V

(2)

Power number can be estimated from Reynolds number of mixing according to (3) [16], where is
the density of the solution and is the viscosity of slag-ammonium salt solution mixture, calculated
from (4) [17] with p as the volume fraction of slag (m3/m3) and p the viscosity of the dispersed
material. Also is assumed that the viscosity of ammonium salt solution will not differ noticeably
from viscosity of pure water.

Re v =

nd 2
mixture

(3)

114-10

1 + 1.5

mixture =

p p
p +H O
2

1 p

H O
2

1 + 1.5 p
1 p

H O , using p >> H O
2

(4)

With input values H2O = 1002.7*10-6 Pa*s, slag=2604 kg/m3and solvent =1008 kg/m3, (4) yields
mixture= 1102*10-6 Pa*s. Together with =1074 kg/m3, and with experimental values n=5 1/s and
d=0.09 m, Rev =39471. In other words, the mixing situation is clearly turbulent. Power number in a
vessel without baffles for a turbine stirrer is then estimated to 1.25, resulting finally to energy
dissipation of 0.076 W/kgsolution. In a fully baffled reactor, resulting to better mixing, the power
number would be approximately 6, yielding to 0.369 W/kgsolution [16].
If it is further assumed that the ratio of mixer diameter and vessel volume is maintained constant in
scale-up, this means that plainly mixing of slag and ammonium salt solvent in extraction requires
20.8 kW/25 ton slag, or 0.8 kW/ton slag. Related to PCC production, this equals to 1.9 kW/ton
PCC. If the experimental residence times (Table 3) would be used for the larger scale, this would
mean 0.8 kW/ton slag * (25/60) h = 0.33 kWh/ton slag. However, these numbers do not include the
electrical losses of the system (90% efficiency would yield to 23.1 kW/ 25 ton slag), nor they are
optimised considering the mixer dimensions and reactor size relation. When compared to heat
duties of the different reactors (Table 8), it can be seen that also the mixing energy has a small, yet
significant role for the overall energy balance of the process.
For the carbonation reactor, the energy dissipation in mixing would be lower since the high gas
flow through the reactor lowers the solution density and viscosity. Also, depending on the gas inlet
arrangement, the mixing caused by the gas flow itself could be sufficient to obtain the required
dispersion of solution components.

5. Studies on additional process units

5.1. Separation equipment
As mentioned in Section 2.2, various options exist for continuous separation of solids from the
aqueous streams. For steel slag separation, a gravitational settling tube (Fig. 7) was tested
experimentally [18]. The utilised plastic tube with a circular cross-section had a volume of 8.64 L
(length 110 cm, diameter 10 cm), and it was used with a feed flow of 40 L/h. The underflow was
maintained at ~7 L/h, a value high enough to guarantee a steady stream for the concentrated particle
suspension, resulting thus to an overflow of 33 L/h.
The feasibility of using the settling tube was tested with spherical Ballotini glass beads (p = 2634
kg/m3, d p = 163 m) in water. Tests were performed with the settling tube positioned horizontally
(0) and at an angle of 10. The beads settled within 0-22 cm from the feed orifice on the horizontal
tube and approximately within the same range in the tube with 10 angle. The distances where the
settling occurred were observed visually during the experiments. Because of the large size and high
density of the glass beads the separation efficiency was ~100%.
To obtain an estimate with separation efficiencies with the actual steel slag, tests were performed
with the steel converter slag (p = 2604 kg/m3, d p = 96 m) that was used also as an input material
for the modelling work described in Section 2. Angles of 30 and 45 were used with water as the
liquid phase, but 45 was tested also with 1 mol/L NH4Cl. Sedimentation was observed between 083 cm (0-80 cm for 45), although the determination was quite challenging due to a very turbid
suspension.
After the process had reached a steady-state, a sample was taken from the outflow to determine the
concentration of particles. The samples were filtrated, dried in an oven at 105 C and weighed.
Calculated from (5), the separation efficiencies were 99.9% for both 30 and 45 experiments.

114-11

(5)

cin cout
100%
cin

However, a SEM picture taken of an overflow sample revealed that the particle size in this fraction
was approximately 1-2 m, indicating thus that for the actual slag particles the separation efficiency
had been 100%, in accordance with the glass particle experiments. Only a fraction of small particles
released from the slag was leaving the separation unit through the overflow. Thus, according to
these tests, a gravitational settler could be utilised for steel slag separation in a continuous process,
but it should be accompanied with a filter to remove also the fraction of dissolved, micrometre scale
particles. At the same, this filter could also be used to protect the pump needed to transport the
process liquid from extraction to carbonation. In a larger scale process, the height of the settler tube
would directly affect the needed pumping effect. Optimisation of the required tube dimensions and
angle for particles of a known size and density with a known flow is on-going.

Fig. 7. A 30 inclined settling tube 0 minutes (left) and 2 minutes (right) from the start of the test.

5.2. Washing steps

As discussed above, losses of ammonia and water vapour as well as ammonium chloride precipitate,
mixed with slag residue or the PCC product, should be minimised to increase the profitability of the
process. During the described experiments it has been measured that residual slag exiting the
process had moisture content of approximately 10%. Moreover, the chlorine level in dry residue
was 5.5%-wt according to SEM/EDX measurements. For the modelled process, this would mean
that 1670 kg/h ammonium chloride (~0.15 kg per kg PCC product) would be lost with the residue,
assuming that all chlorine is bound with ammonia.
However, from the experimental results of applying one washing stage for the residue, with
washing water amount of 0.50 m3/t residue, the reduction in calcium content was found to be 0.65
mol, while approximately 1.5 mol chlorine was dissolved. Thus, it would mean that only 14% of the
chlorine would have been bound as NH4Cl, the rest being CaCl2. In other words, 230 kg/h
ammonium chloride and 1485 kg/h calcium chloride would be lost with the slag residue.
According to experimental results the one-stage wash would lower the chlorine level in the residue
to 0.7%-wt, corresponding to 28 kg/h NH4Cl and 177 kg/h CaCl2, if calculated with the same
distribution as above. The used washing liquid of 10 m3/h would then contain 0.003 mol/L Cl-,
0.001 mol/L Ca2+ and 0.0004 mol/L NH4+, that could be re-utilised in the process by evaporating
the water in such extent that the need for make-up water for the process would be satisfied at the
same. This, however, would bring an extra energy penalty to the process.
Regarding the produced PCC, washing tests will be done as a part of future work. It can be
mentioned, that for unwashed PCC produced with 1 mol/L ammonium chloride, the experimental
chlorine levels have been below 1 %-wt, being still too high for a marketable product. Using
ammonium nitrate solvent instead of ammonium chloride would remove the problem of chlorine
precipitates.
114-12

Calculating from the Aspen Plus model with 15% bypass of the 1st carbonation unit, the
vaporization losses of ammonia would be 18 kg/h (1.1 kmol/h) with PCC production of 10795 kg/h.
This amount, 0.4% of the total ammonium species present, is of the same order of magnitude what
has been obtained experimentally by FTIR measurements [13]. However, also this ammonia could
be captured, e.g. by using an HCl-based scrubber for the flue gas stream leaving the system (Fig. 2).
The need for make-up water to replace the moisture lost with streams of solid matter and water
vapour flashed from the separation units depends on many variables, and has not yet been
quantitavely studied. However, assuming a 20% moisture content for the PCC product and 10% for
slag residue, the need for make-up water for the above discussed process would be 5830 kg/h. This
amount could easily be taken from washing units, but it requires further optimisation. In general, all
the values presented in this section are based on preliminary studies, and thus need to be verified by
additional experimental studies and analysis as a central part of future work. Only then can the
usage of fresh make-up chemicals be optimised too.

6. Conclusions
The studied pH swing process possesses potential for commercial application. Based on both
modelling and experimental work, precipitated calcium carbonate can be produced with a
continuously operating system instead of batch reactors. By choosing the solvent with highest
conversion efficiency (ammonium nitrate) and by dividing the precipitation step in two stages, the
conversion of calcium in the input material to a valuable product can be enhanced. Energetically the
process will require some mixing and pumping power, but the chemical reaction steps are
exothermic. The recovery and separation units for process chemicals and for solids need to be
studied in more detail in future work, but it seems to be possible to use a gravitational settler to
separate slag residues from the process liquid. Also, washing the output streams will aid the
recovery of chemicals, and thus decrease the need for make-up material. However, additional
studies on particle quality, shape, purity and whiteness of the produced PCC must be performed to
guarantee that a marketable product is manufactured.

Acknowledgements
The authors want to acknowledge Cleen Ltd. and Tekes (the Finnish Funding Agency for
Technology and Innovation) for their financial support for the research via the Cleen CCSP project
(2011-2015). H.-P. Mattila also acknowledges the Graduate School in Chemical Engineering for
support for his work.

Nomenclature
c
d

concentration, g/L
diameter of the mixer, m

d
n
P
p
Po
Q
Rev
T
T0
V
V

mean diameter, m
frequency of rotation, 1/s
power, kW
pressure, bar
power number,
heat duty, kW
Reynolds number of mixing
temperature, K
ambient temperature, K
volume, m3
volume flow, m3/h
114-13

Abbreviations
CCS
FTIR
PCC
SEM/EDX

Carbon Capture and Storage

Fourier Transform Infrared Spectroscopy
Precipitated Calcium Carbonate
Scanning Electron Microscopy Energy Dispersive X-ray

Greek symbols

difference

energy dissipation, W/kgsolution

viscosity, Pas; efficiency, %

temperature, C

density, kg/m3
Subscripts and superscripts
in
at inlet
p
particle

References
[1]

IPCC, IPCC Special Report on Carbon Dioxide Capture and Storage, Cambridge University
Press, New York, 2005.
[2] Herzog, H., An Introduction to CO2 Separation and Capture Technologies, MIT Energy
Laboratory, 1999. Available at:
<http://sequestration.mit.edu/pdf/introduction_to_capture.pdf> [accessed 27.1.2012].
[3] Teir, S., Fixation of carbon dioxide by producing carbonates from minerals and steelmaking
slags. [Dissertation]. Espoo, Finland: Helsinki University of Technology; 2008.
[4] Baciocchi, R., Costa, G., Polettini, A., Pomi, R., Prigiobbe, V., Comparison of different
reaction routes for carbonation of APC residues, Energy Procedia 2009;1: 4851-4858.
[5] Velts, O., Uibu, M., Kallas, J., Kuusik, R., CO2 mineral trapping: Modeling of calcium
carbonate precipitation in a semi-batch reactor. Energy Procedia 2011;4: 771-778.
[6] Eloneva, S., Reduction of CO2 Emissions by Mineral Carbonation: Steelmaking Slags as Raw
Material with a Pure Calcium Carbonate End Product. [Dissertation]. Espoo, Finland: Aalto
University, School of Science and Technology; 2010.
[7] Kodama, S., Nishimoto, T., Yamamoto, N., Yogo, K., Yamada, K., Development of a new
pH-swing CO2 mineralization process with a recyclable reaction solution. Energy 2008;33:
776-784.
[8] Eloneva, S., Teir, S., Savolahti, J., Fogelholm, C.-J., Zevenhoven, R., Co-utilization of CO2
and calcium silicate-rich slags for precipitated calcium carbonate production (Part II). In:
Mirandola, A., Arnas, O., Lazzaretto, A., editors. ECOS 2007: Proceedings of the 20th
International Conference on Efficiency, Cost, Optimization, Simulation and Environmental
Impact of Energy Systems; 2007 Jun 25-28; Padova, Italy. Volume II:1389-1396.
[9] Said, A., Eloneva, S., Fogelholm, C.-J., Mattila, H.-P., Zevenhoven, R., Process simulation of
utilization CO2 and steelmaking slags to form Precipitated Calcium Carbonate (PCC). In:
ECOS 2009: Proceedings of the 22nd International Conference on Efficiency, Cost,
Optimization, Simulation and Environmental Impact of Energy Systems; 2009 Aug 31- Sep 3;
Foz du Iguacu, Brazil. pp. 1261-1270.
[10] Mattila, H.-P., Grigalinait, I., Zevenhoven, R., Chemical kinetics modeling and process
parameter sensitivity for precipitated calcium carbonate production from steelmaking slags.
Chem. Eng. J. 2012;192: 77-89.
114-14

[11] Mattila, H.-P., Experimental studies and process modeling of aqueous two-stage steel slag
carbonation. M.Sc. Thesis. Turku, Finland: bo Akademi University; 2009.
[12] Mattila, H.-P., Grigalinait, I., Said, A., Fogelholm, C.-J., Zevenhoven, R., Production of
papermaking grade calcium carbonate from steelmaking slag product quality and
development of a larger scale process. Accepted to be presented at SCANMET IV conference,
Lule, Sweden, Jun 2012.
[13] Eloneva, S., Mannisto, P., Said, A., Fogelholm, C.-J., Zevenhoven, R., Ammonium salt-based
steelmaking slag carbonation: Precipitation of CaCO3 and ammonia losses assessment.
Greenhouse Gases: Science and Technology 2011;1(4): 305-311.
[14] Wiklund, A., Sipil, J., Eloneva, S., Zevenhoven, R., Assessment of Shell work on
carbonation of calcium based materials. bo Akademi University, Heat Engineering
Laboratory; 2008. Report.
[15] Szargut, J., Exergy Method: Technical and Ecological Applications. Southampton, UK: WIT
Press; 2005.
[16] Beek, W. J., Muttzall, K. M. K., van Heuven, J. W., Transport Phenomena, 2nd edition.
Guildford, UK: Biddles Ltd; 2000.
[17] Kunitz, M., An empirical formula for the relation between viscosity of solution and volume of
solute. J.Gen.Physiol. 1926, 9:715-25.
[18] Davis, R. H., Gecol, H., Classification of concentrated suspensions using inclined settlers.
International Journal of Multiphase Flow 1996; 22(3): 563-574.

114-15