Fuel 111 (2013) 430–441

Contents lists available at SciVerse ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Characterising long-term wet-stored fly ash following carbon

and particle size separation
M.J. McCarthy a,⇑, M.R. Jones a, L. Zheng a, T.L. Robl b, J.G. Groppo b
a
Concrete Technology Unit, Division of Civil Engineering, University of Dundee, Dundee DD1 4HN, Scotland, UK
b
Center for Applied Energy Research, University of Kentucky, 2540 Research Park Drive, Lexington, KY 40511-8479, USA

h i g h l i g h t s

" Recovery of coal-derived fly ash from ponds and stockpiles.

" Low-cost wet-processing technology used to homogenize the materials.
" Microstructural, physical nature and chemistry of particle groups investigated.
" Heterogeneity reduction giving particle groups for various construction applications.
" Potential for yield optimisation for holistic use of fly ash.

a r t i c l e i n f o a b s t r a c t

Article history: The paper describes a study carried out to characterise fly ash produced from hard-coal, following recov-
Received 1 May 2012 ery from long-term pond and stockpile storage, and wet process separation. Three power station sites,
Received in revised form 15 February 2013 representative of those in the United Kingdom, were considered and initial work examined the charac-
Accepted 18 February 2013
teristics of recovered material from both types of storage area. Most of this had high loss-on-ignition
Available online 6 March 2013
(LOI) and was relatively coarse, with that in ponds generally having greater variability. Fly ash particles
had reaction products on their surfaces and the presence of agglomerates was also noted. Methods
Keywords:
including, pre-screening, column (hydraulic) separation, froth flotation and lamella hydraulic classifier
Pond and stockpile-stored fly ash
Wet-processing technology
were considered for processing. Three trials were carried out using various combinations, and it was pos-
Physical and chemical characteristics sible to separate carbon and fly ash to give material groups ranging from carbon-rich to low LOI/high fine-
Morphology ness, although fine fly ash particles (<5 lm) were distributed throughout these. Relatively minor changes
End-use potential in chemical composition and mineralogy of the separated materials were noted, except for particle
groups with high LOI (reflecting fly ash dilution). Similar effects were generally obtained with both pond
and stockpile storage. Mass balance calculations were made to quantify material distributions for the
three processing trials carried out. Examples are given of yields achieved for potential end-uses from
these, which indicate that optimisation to target particular fly ash properties should be possible.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Storage periods can range from months to a decade or longer,

It is estimated that 100 Mt of coal-combustion residues, includ-
ing fly ash, were produced from electricity generation in the Euro-
pean Union in 2009 [1]. However, in many markets, supply of fly
ash exceeds demand and as a result much of that remaining un-
used is placed into storage [2]. This is normally by mixing with
water and transporting to designated ponds (slurried with a sol-
ids/water ratio of up to 20%) or stockpiles (moistened up to about
15–20% by mass of solids) within or adjacent to power plants.
⇑ Corresponding author. Tel.: +44 1382 384924; fax: +44 1382 384389.
E-mail address: m.j.mccarthy@dundee.ac.uk (M.J. McCarthy).
depending on demand. As a result, it is estimated that more than
100 Mt of accessible fly ash of this type are available in the United
Kingdom (UK) alone [3]. There is often significant heterogeneity in
this material, reflecting changes at the power plant with time (e.g.
coal type and/or burning conditions), as well as natural separation
occurring during storage. This can be further affected by the pres-
ence of various forms of char from incomplete combustion and
pyrite from coal pulverizer rejects [4].
Research examining the effects of short-term storage under
moist or saturated conditions on low-lime fly ash in UK ponds
and stockpiles suggests that while physical and chemical changes
occur, the material can still be used as a cement component in con-
crete [5,6]. Homogenization of fly ash to meet various property

0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.fuel.2013.02.048
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441 431

Table 1
Characteristics of recovered fly ashes from UK pond and stockpile sites.

Fly ash characteristicsa Power plant site

1 2 3
b b
Storage type (no. of samples) Pond (34) Processing Stockpile/processing Stockpile (45) Processingb
LOI, % 3.8–20.2 (8.2) 14.7 10.0 11.2–25.4 (15.3) 14.4
Particle size distribution, lm D10 3.2–14.3 (6.2) 7.3 12.1 3.7–7.1 (4.9) 4.5
D50 13.4–100.5 (43.7) 42.9 55.3 27.4–43.6 (34.8) 31.2
D90 54.1–521.5 (153.0) 186.6 164.7 96.5–208.5 (127.5) 113.6
45 lm ret, % 12.0–61.9 (39.8) 58.6 53.4 31.1–42.3 (36.7) 39.2
Chemical composition, % by mass
CaO 2.0–4.3 (2.7) 4.3 4.6 1.2–3.5 (2.2) 2.7
SiO2 40.0–52.5 (47.4) 45.7 41.7 35.8–45.8 (40.4) 40.5
Al2O3 24.3–31.4 (28.5) 26.1 25.9 19.5–23.6 (21.5) 21.8
Fe2O3 3.3–6.8 (4.9) 5.8 9.0 5.3–8.5 (6.7) 7.5
MgO 1.1–2.8 (1.6) 2.8 1.6 0.7–1.5 (1.2) 1.2
TiO2 1.2–1.6 (1.4) 1.5 1.2 1.0–1.7 (1.1) 1.1
K2O 0.7–1.6 (1.1) 1.4 2.1 1.7–3.1 (2.6) 2.8
Na2O 0.3–3.8 (1.1) 0.5 0.9 0.3–2.0 (0.6) 0.5
P2O5 0.5–1.5 (0.9) 1.5 0.6 0.5–0.8 (0.6) 0.6
Cl 0.4–6.4 (1.8) 1.0 0.0 0.0–3.0 (0.3) 0.0
SO3 0.2–1.2 (0.6) 0.3 2.0 0.2–1.8 (0.7) 0.7
Main mineral compositionc, % by mass
Quartz 3.0–13.9 (8.3) 6.3 5.3 2.4–4.7 (3.2) 2.6
Mullite 13.9–32.3 (22.3) 17.3 12.9 5.0–7.9 (6.5) 6.2
Non-crystalline contentd 47.3–72.3 (59.1) 60.7 68.9 60.7–72.6 (67.2) 71.4
a
Where multiple samples were taken, results are given as ranges (means).
b
Typical properties of samples used during processing trials.
c
For power plant site 3 stockpile, results are from 9 measurements.
d
By difference. Note: this does not necessarily equal the pozzolanically reactive ‘glass’ content.

requirements, [e.g. 7–9], may then enable adoption in a range of
applications, beyond bulk filler, whilst also reducing pressure on
storage sites. Furthermore, this approach can contribute to the
increasing requirements for sustainable materials in twenty-first
century design and construction [10].
Several systems have been developed recently towards enabling
fly ash separation and their potential is becoming recognised [11].
The research described was, therefore, concerned with investigat-
ing the characteristics of long-term stored, low-lime fly ash follow-
ing recovery and processing, with a view to identifying suitability
for various end-uses.
2. Sourcing of test materials
In order to examine different power plant storage conditions, fly
ash from both ponds and stockpiles was considered. A survey of
storage areas at UK power plants [3], enabled three representative
sites for material recovery to be identified.
At Power Plant 1, material (Fly Ash 1) was stored in a 0.4 Mt
pond and ranged in age up to 30 years. At the time of testing, the
pond had been drained of water and some fly ash excavated, en-
abling access for sampling in exposed vertical faces. A total of
33 kg samples were taken at horizontal distances of approximately
75 m and depths of 1–2, 2–3 and 3–4 m, by hand auger. In addition,
a 2 tonne bulk sample was obtained at a later stage for use during
the processing work.
At Power Plant 2, there was a 1 Mt stockpile, where material
(Fly Ash 2) had been produced in the previous few years prior to
recovery. A bulk sample of approximately 3.5 tonnes was taken
from this site for processing.
At Power Plant 3, the material (Fly Ash 3) was in a 16 Mt stock-
pile, covering about 50 hectares and of height around 50 m. This
was divided into three areas according to when fly ash had been
placed there, i.e. (i) recently produced material and that stored
for (ii) 5–10, and (iii) 15–20 years. Samples were taken at 5 posi-
tions in these areas, with horizontal distances of about 75 m and
depths of 1–2, 4–5 and 9–10 m, giving a total of 45. Two large sam-
ples of about 1 tonne were obtained from (iii) above and combined
for processing.
3. Test procedures
The loss-on-ignition (LOI) of recovered and separated materials
was determined to BS EN 450-1 [7]. The fineness of the fly ashes
was measured in terms of (i) the quantity of sample retained on
a 45 lm sieve (following wet sieving; 45 lm ret) as described in
BS EN 451-2 [12] and (ii) particle size distribution (PSD) by LASER
diffraction.
Pressed powder pellets of milled fly ash were analysed for oxide
composition using a sequential X-ray fluorescence spectrometer
with Rh Ka source. The mineralogy was examined on powder sam-
ples by X-ray diffractometer with Cu Ka source. An internal stan-
dard was used (5.0% corundum) to enable estimation of the
crystalline/non-crystalline (by difference) constituents.
4. Characteristics of recovered material
The results from the tests on recovered fly ash samples are sum-
marised in Table 1. The LOI data highlight the variability of the
materials, as mentioned above, in the storage areas, which for
the test sources was slightly greater in the pond than stockpile.
The results indicate that the fineness (45 lm ret) and range of par-
ticle sizes were also wider in the former. In addition, the values for
many of the samples were greater than the limits set for the mate-
rial’s use as a cement component in concrete [7].
As expected for bituminous coal sources, all fly ashes had low
lime contents, with the maximum level 4.6%. The chemical compo-
sition generally corresponded to the typical ranges for UK fly ashes
[13]. Differences to these, found in some cases, e.g. Fe2O3, alkalis
and sulfate, may relate to the wet storage conditions [5]. It was also
noted that some samples contained relatively high levels of chlo-
ride, which would be an issue for their application in reinforced
concrete [14]. This reflects either the use of salt water to moist-
en/slurry fly ash, or proximity of the storage areas to coastal/estu-
432 M.J. McCarthy et al. / Fuel 111 (2013) 430–441

Fig. 1. Typical morphologies of recovered fly ash samples from the three power plants.

1. Pre-Screening 4a. Column Flotation 5. Lamella Hydraulic

(5 mm sieve) Classifier
Carbon-rich Froth
(CARE US Patent)
6.0 Floated
Feed (Raw)
l/min Cenospheres
10-15% 1.5-3.0
Solids l/min
2.5-5.0
l/min
Air Injection
Fly Ash-
rich Tails U1 U2 U3 U4 U5
2. Slurrying Underflows Ultrafine
(800 l slurry tank) 0.6 0.3 0.3 0.3 4.5
l/min l/min l/min l/min l/min

Fig. 2. Schematic illustration of the elements of the processing system for Trial 1. Note: stage 3 not used (see Fig. 4).
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441 433

Table 2
Main characteristics of fly ash at the various processing points (Trial 1, Fly Ash 3).

Fly ash separation characteristicsa Processing point (see Fig. 2)

4a. Column flotation 5. Lamella hydraulic classifier
Froth Tails U1 U2 U3 U4 U5
Mass proportion, % of feed 13.6 86.4 43.3 17.4 10.4 8.0 7.3
LOI proportion, % of feed 32.9 67.1 47.1 9.5 4.7 3.2 2.6
LOI, % 39.4 13.4 17.8 8.8 7.4 6.7 5.9
<5 lm proportion, % of feed 9.1 90.9 28.0 14.4 9.3 7.7 31.5
45 lm ret, % 36.7 19.7 42.4 10.9 5.4 3.4 0.0
Chemical composition, % by mass
CaO 2.9 2.0 2.1 1.9 1.9 1.9 1.8
SiO2 32.7 41.9 40.8 42.5 42.3 43.3 43.7
Al2O3 18.5 22.9 22.0 23.0 23.3 23.7 24.4
Fe2O3 7.4 7.4 7.3 7.3 7.4 7.0 6.6
MgO 1.3 1.2 1.2 1.2 1.2 1.2 1.2
TiO2 1.0 1.0 1.0 1.1 1.1 1.1 1.1
K2O 2.7 3.2 3.0 3.2 3.3 3.3 3.4
Na2O 0.4 0.5 0.5 0.5 0.5 0.5 0.5
P2O5 0.7 0.5 0.6 0.5 0.5 0.5 0.5
Cl 0.0 0.0 0.0 0.0 0.0 0.0 0.0
SO3 1.0 0.4 0.5 0.2 0.2 0.1 0.1
Main mineral composition, % by mass
Quartz 3.0 4.5 7.9 7.3 6.5 7.1 3.5
Mullite 4.3 7.4 10.1 10.4 10.6 11.8 9.3
Non-crystalline contentb 51.8 73.7 62.0 71.4 74.4 71.3 80.6
a
For recovered fly ash characteristics, see Table 1.
b
By difference. See note under Table 1.

arine conditions. The mineralogy results indicate that Fly Ash 1 had
higher quartz and mullite levels (mean and processing values) than
Fly Ashes 2 and 3, with the corresponding non-crystalline contents
higher in the latter two materials.
Fig. 1 shows typical morphologies (by scanning electron micros-
copy) of the recovered samples including, fly ash particles and
residual unburned components. The presence of reaction products
on particle surfaces, giving them a rough appearance, and agglom-
erates, corresponding to the wet storage conditions [5,6], were also
noted.
5. Approach to carbon and particle size separation
The research was concerned with characterising UK fly ashes
following carbon and particle size separation. This considered var-
ious wet processing techniques (given the fly ash storage condi-
tions) used in the mineral recovery industry and which have
found application with fly ash previously. The main trials with
these considered Fly Ash 3, with confirmatory tests made using
Fly Ashes 1 and 2.
A stepwise approach was followed, with different techniques
combined and refinements made, as required, to target material
of low LOI/high fineness. Physical and chemical (selectively) char-
acterisation of particle groups was made at the various stages of
separation. Similarly mass balance calculations were also carried
out, based on the solids contents and flow rates on entry to and
exit from these. This enabled material distributions throughout
the processing systems to be determined.
6. Main carbon and particle size separation techniques
6.1. Hydraulic classification
Initially, hydraulic classification, with separation by differences
in particle velocity passage through water (influenced by density,
size, and shape) [15] was considered. The particular system
adopted was a lamella hydraulic classifier (LHC), which has oblique
plates, enabling rapid particle settlement and reduced tank area
space requirements [15,16]. The LHC was developed for use with
fly ash at the University of Kentucky and offered several advanta-
ges including an ability to (i) operate at high solids contents, (ii)
achieve good recoveries (yields), and (iii) enable ultrafine fly ash
particles (<10 lm) to be obtained [16–18].
The LHC used was of 2.4 m length (laboratory-scale) with a la-
mella spacing of 50 mm. The feed slurry had a solids/water ratio
of 10–15%, with a dispersant added in-line. The LHC had five outlet
(processing) points (four underflows, U1–U4; and an overflow, U5)
along its length, with progressively finer material obtained at
greater distances from the inlet, reflecting increased settling times
for smaller particles (or continued suspension over the tank length
for the finest, collected in U5). Initial tests with the LHC on fly
ashes of LOI >10.0%, following pre-screening (to remove coarse
particles/vegetation) through a 5 mm mesh, gave a minimum value
(LOI) for separated materials of >7.0%, (i.e. the UK limit for use as a
cement component in concrete). This suggested that carbon be-
haved similarly to fine fly ash during separation, and that addi-
tional processes were required to reduce the LOI to lower levels.
6.2. Froth flotation
Given the wet processing approach adopted during the study,
froth flotation was considered for carbon removal. The technique
achieves this through, (i) physico-chemical property differences
between suspended particles, (ii) the use of appropriate re-agents,
(iii) passage of air through the flotation unit, and (iv) formation of a
stable carbon froth [15,19]. The methods used during the trials
were column flotation, with air introduced by compressor and
sparger; and mechanical flotation, which achieved this by impeller.
In order to determine the potential for carbon removal from the
test fly ashes, release analyses were carried out with a batch flota-
tion machine, using the collector and frother re-agents adopted for
the main trials [4,18]. The process defines the limits of separation
achievable with the technique and was carried out on 200 g fly ash
samples after removal of particles >150 lm by sieving. Results, fol-
lowing repeat flotations for the three fly ashes, indicate that the
LOI of each material could be reduced. This was most readily
achieved with Fly Ash 3, where a LOI of 3.0% was obtained with
434 M.J. McCarthy et al. / Fuel 111 (2013) 430–441

(a) Feed (Raw) (b) Froth

(c) Tails (d) U1

(e) U2 (f) U3

(g) U4 (h) U5

Fig. 3. Typical morphologies of feed (raw) and material at different processing points (Trial 1, Fly Ash 3).
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441 435

1. Pre-Screening
(5 mm sieve) 3. Column
Separation

2.8-4.3
l/min
10-15% Feed (Raw) Overflow
5. Lamella Hydraulic Classifier
Solids
3.0-4.5 (CARE US Patent)
l/min 6.0 Floated
l/min Cenospheres
Clean 0.5-0.7
Water l/min
2. Slurrying 0-0.5
(800 l slurry tank) l/min
Underflow

4b. Mechanical Flotation (1 cycle) U1 U2 U3 U4 U5

Underflows Ultrafine
0.6 0.3 0.3 0.3 4.5
Froth
Air l/min l/min l/min l/min l/min

Processing Test Combinations:

TRIAL 2: 1 2 3 4b 5
Or TRIAL 3: 1 2 3 4c 5
Tails
2nd cycle
4c. Mechanical Flotation (2 cycles)

Fig. 4. Schematic illustration of the elements of the processing system for Trials 2 and 3.

a yield of about 65%. Corresponding yields at this LOI with Fly compared to the recovered material. The process gave similar fine-
Ashes 1 and 2 were about 30% and 50%. ness between this and the froth, with an increase in fineness
(approximately 50% reduction in 45 lm sieve retention) in the
tails.
7. Results and discussion The LHC results show that 43.3% of the material settled in U1,
with progressively less from U2 to U5. U1 had an increased LOI,
7.1. Trial 1: column (froth) flotation and LHC was coarser than the column flotation tails and contained 28.0%
of the particles <5 lm. At subsequent outlet points, there were
For Trial 1, the processing system shown in Fig. 2 was used and gradual reductions in LOI, which was <7.0% by U4 and increases
combined pre-screening (1), slurrying (2), and then column (froth) in fineness, which was <10.0% by U3, with all material passing
flotation using a 4.5 m  U100 mm column (4a), followed by the the 45 lm sieve at U5, which contained 31.5% of particles <5 lm.
LHC (5). Table 2 gives the main characteristics of the separated The presence of cenospheres was also noted with Fly Ash 3 during
materials for Fly Ash 3. These indicate that 13.6% of the feed separation.
(raw) material (i.e. that following pre-screening and slurrying) The chemical composition data (Table 2) indicate noticeable
was present in the froth, with 86.4% in the tails after column flota- changes in certain compounds between the froth and tails after
tion. Hence, the tails had only a small reduction in LOI of 1.0%, column flotation (and compared to the recovered material), with
relatively minor differences following separation in the LHC. These
may reflect (i) variations in fly ash/carbon levels in the materials,
Table 3
and (ii) the tendency for certain compounds to concentrate in fly
Main characteristics of fly ash at the various processing points (Trial 2, Fly Ash 3).
ash particles of particular sizes. The effect of (i) is most likely be-
Processing Fly ash separation characteristicsa tween the froth and tails of column flotation, and to a lesser extent
point (see
Mass LOI LOI, <5 lm particle 45 lm at the outlet points of the LHC. In the case of (ii), it has been found
Fig. 4)
proportion, proportion, % proportion, % of ret, % that SiO2 and Al2O3 tend to increase in finer material [20,21]. Sim-
% of feed % of feed feed
ilarly variations in Fe2O3 may reflect its density and presence in
3. Column separation particles >20 lm [22]. Levels of CaO and SO3 increased slightly in
Underflow 20.3 25.9 17.5 13.6 69.5
the froth compared to the recovered material, which suggests re-
Overflow 79.7 74.1 14.3 86.4 25.5
moval of clay residue fragments containing these, or CaSO4-based
4b. Mechanical flotation
compounds (during this process) that may have formed with wet
Froth 23.4 46.4 33.1 20.0 37.2
Tails 56.3 27.7 8.0 66.4 21.3 storage (in the stockpile) [6]. Given the fly ashes were wet initially,
it is likely that chemical effects due to water contact, referred to
5. Lamella hydraulic classifier
U1 17.8 11.1 10.1 8.2 58.3 above, would have mainly occurred prior to recovery.
U2 12.0 5.9 7.9 8.4 23.3 The main mineral composition results (Table 2) show that
U3 10.2 4.7 7.4 8.6 11.7 quartz, mullite and non-crystalline contents were similar or re-
U4 9.2 3.7 6.6 8.0 6.3 duced in the froth following column flotation compared to the
U5 6.2 2.3 6.0 33.2 0.0
recovered material, which is likely to reflect fly ash/carbon levels,
a
For recovered fly ash characteristics, see Table 1. with small increases noted for these between the latter and the
436 M.J. McCarthy et al. / Fuel 111 (2013) 430–441
(a) Feed (Raw)
(c) Overflow
(e) Tails
Fig. 5. Typical morphologies of feed (raw) and material following column separation and mechanical flotation processes (Trial 2, Fly Ash 3).
tails. In the LHC, quartz tended to reduce from U1 to U5, with cor-
responding increases in the non-crystalline content. In addition to
fly ash/carbon level influences, this may correspond to differences
in SiO2 and Al2O3, which are linked to the amorphous content of fly
ash [23]. These effects also agree with other research, which indi-
cates increased amorphous nature of and reduced quartz in fine fly
ash particles after a similar type separation [24]. The presence of
mullite has also been found to depend on fly ash particle size
[24], but only minor differences in this were found at the outlet
points of the LHC.
The typical morphology of material groups from Trial 1 before
and following the different separating stages is shown in Fig. 3
and generally agrees with the effects noted above. Indeed, the feed
(raw) (Fig. 3a) had a range of particle types and sizes. After column
flotation, the tails (Fig. 3c) were finer than the froth (Fig. 3b), with
increasing levels of carbon particles found in the latter. Fine fly ash
particles were also present in both materials. Fig. 3d–h shows
material following separation with the LHC and indicates coarsen-
ing between the tails and U1, with more irregular shaped particles
(b) Underflow
(d) Froth
in the latter, and then a gradual increase in fineness and quantity of
spherical particles from U2 to U5 (but with fine particles noted in
all LHC-separated materials).
7.2. Trial 2: column (hydraulic) separation, mechanical (froth)
flotation and LHC
In Trial 2, column (hydraulic) separation (3), mechanical (froth)
flotation (4b) and LHC (5) were used (following pre-screening (1)
and slurrying (2)), as shown in Fig. 4. Thus, coarser particles
(approximately >150 lm) were initially removed and then carbon
by mechanical flotation (with a bank of two  7 l cells), to try and
improve the separation of this compared to Trial 1. The results for
Fly Ash 3, Table 3, indicate that about 20% of material was removed
during column separation, giving slight increases in fineness, with
little change in LOI of the overflow compared to the recovered
material. Following mechanical flotation, 56.3% of the material
was in the tails, which had a LOI of 8.0% and fineness of 21.3% (with
corresponding values of 33.1% and 37.2% in the froth). At this stage,
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441 437

80

70 3. Column 4b. Mechanical 5. Lamella Hydraulic Fly Ash 3

Mass Proportion, % of feed

Separation Flotation (1 cycle) Classification Fly Ash 1
60

50

40

30

20

10

35 Fly Ash 3
Fly Ash 1
30

25
LOI, %

20

15

10

70 Fly Ash 3
Fly Ash 1
60

50
45 µm ret, %

40

30

20

10

0
Under- Over- Froth Tails U1 U2
flow flow

Fig. 6. Distribution and properties of Fly Ashes 1 and 3 during Trial 2.

Table 4
Main characteristics of fly ash at the various processing points (Trial 3, Fly Ash 3).

Processing point (see Fig. 4) Fly ash separation characteristicsa

Mass proportion, % of feed LOI proportion, % of feed LOI, % <5 lm particle proportion, % of feed 45 lm ret, %
3. Column separation
Underflow 24.9 27.4 16.8 18.7 55.8
Overflow 75.1 72.6 15.0 81.3 23.1
4c. Mechanical flotation, 2 cycles
Froth-Cycle-1 23.2 48.9 34.6 21.5 30.1
Tails-Cycle-1 52.0 23.7 7.5 59.8 24.1
Froth-Cycle-2 6.6 11.7 30.2 7.8 35.6
Tails-Cycle-2 45.4 12.0 4.2 52.1 22.0
5. Lamella hydraulic classifier
U1 14.0 5.7 5.8 8.1 56.5
U2 10.5 2.2 3.0 7.8 19.1
U3 8.7 1.6 2.6 8.2 8.2
U4 8.1 1.4 2.4 9.0 4.4
U5 4.1 1.0 3.6 19.0 0.0
a
For recovered fly ash characteristics, see Table 1.
438 M.J. McCarthy et al. / Fuel 111 (2013) 430–441

80
3. Column 4c. Mechanical 5. Lamella Hydraulic
70 Separati on Flotation, (2 cycles) Classification Fly Ash 3

Mass Proportion, % of feed

Fly Ash 2
60

50

40

30

20

10

35 Fly Ash 3
Fly Ash 2
30

25
LOI, %

20

15

10

70 Fly Ash 3
Fly Ash 2
60
45 µm ret, %

50

40

30

20

10

0
Under- Over- Froth-1 Tails-1 Froth-2 Tails-2 U1 U2 U3 U4 U5
flow flow

Fig. 7. Distribution and properties of Fly Ashes 2 and 3 during Trial 3.

a third of the <5 lm particles had been taken out. These processes
gave a lower LOI for material entering the LHC than Trial 1 (by
5.4%), but this was slightly coarser (by 1.6%).
The typical morphology of these material groups is shown in
Fig. 5 and indicates that the feed (raw) and underflow (from
column separation) generally had greater quantities of coarse
particles compared to the overflow (Fig. 5a–c). The flotation tails
(Fig. 5e) had a greater number of spherical particles than the
froth (Fig. 5d), with the latter containing several irregular shaped
components (reflecting its high LOI). The presence of agglomer-
ates and fine fly ash particles was also noted in all of these
materials.
For the LHC, less material was collected in U1 than for Trial 1,
reflecting the additional processes prior to introduction, with sim-
ilar quantities in U2 to U5. Again, the LOI increased slightly in U1
compared to the feed (tails) and reduced progressively from 7.9%
to 6.0% for U2 to U5. The material was also coarser in U1 and U2
than that entering the LHC, with fineness ranging from 11.7% to
0% for U3 to U5. Of the <5 lm particles, about 8.0% were present
in each of U1–U4 and about a third in U5.
This trial was repeated with Fly Ash 1 and the results compared
with Fly Ash 3 in Fig. 6. These indicate that following column sep-
aration the quantities of material, their LOIs and fineness were
similar for both (i.e. underflow and overflow). Slightly more mate-
rial was present in the froth following mechanical flotation of Fly
Ash 1 than Fly Ash 3. The LOI of the tails was also slightly higher
for Fly Ash 1, thus agreeing with the release analysis, and it was
marginally coarser. Given there was slightly less material in the
tails of Fly Ash 1, the quantities between U1 and U5 were corre-
spondingly less for this in the LHC. Similarly, its properties on entry
(increased LOI/coarseness) were also generally reflected in the
material at the various outlet points. Fewer cenospheres were
noted with Fly Ash 1 during separation than Fly Ash 3, which is
likely to reflect their dispersion during pond-storage.
7.3. Trial 3: column separation, 2  mechanical flotation and LHC
To achieve greater carbon separation a further Trial (Trial 3) was
carried out where material was recycled during mechanical flota-
tion (4c) to give the equivalent of four cells for the process, see
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441 439

80 80
(a) Particles <5 µm portion (b) Particles 5-10 µm portion
70 70
60 60
Quantity of Particles in Size Range, % of ‘Tails’

Quantity of Particles in Size Range, % of ‘Tails’

50 50
40 40
30 30
20 20
10 10
0 0
(c) Particles 10-20 µm portion (d) Particles >20 µm portion
70 70
60 60
50 Trial 1: Table 2 50
40 Trial 2: Table 3
40
Trial 3: Table 4
30 30
20 20
10 10
0 0
U1 U2 U3 U4 U5 U1 U2 U3 U4 U5

LHC Outlet Points

Fig. 8. Comparison of particle size range quantities from the LHC during the three trials.

Fig. 4. The results from this trial for Fly Ash 3 are given in Table 4. (reflecting less fly ash), but small differences between the other
The initial processes (column separation (3) and 1 cycle of mechan- separated materials.
ical flotation (4b)) gave similar results to those of Trial 2 (Table 3). The results from a repeat of the trial with Fly Ash 2 are shown
The flotation tails (Tails-Cycle-2) had an LOI of 4.2% compared to with Fly Ash 3 in Fig. 7. These indicate that a greater quantity of
8.0% (from 1 cycle) in Trial 2, with both of similar fineness. How- material was removed in the underflow for Fly Ash 2. While both
ever, while in Trial 2 approximately 56% of the material was intro- materials had a similar LOI in the tails, Fly Ash 2 had a lower ini-
duced to the LHC, this was 45% in Trial 3, with a corresponding tial LOI (indicating less effective carbon removal). The fineness of
reduction in the quantity of particles <5 lm. the tails was greater for Fly Ash 3 than Fly Ash 2 and more mate-
The results from the LHC were similar to those of the earlier tri- rial was introduced to the LHC for the former following mechan-
als, with an increase in LOI and coarseness of U1 material com- ical flotation. The LOI at the outlet points of the LHC was similar
pared to the tails and reductions thereafter to U5 generally between materials. However, Fly Ash 2 was significantly coarser
noted. Material in the LHC during Trial 3 had lower LOI (reflecting at each, reflecting its initial characteristics, although separated
differences on entry) and similar fineness to that in Trial 2. The material for both all passed the 45 lm sieve by U5. As for Fly
proportions of <5 lm material at each outlet point of the system Ash 3, cenospheres were noticeable with Fly Ash 2 during
were similar between the two trials, except U5, where less was ob- separation.
tained (corresponding to the reduced quantity in the feed). Tests
were also carried out for bulk oxide composition during the trial
8. Practical implications and end-use potential of recovered
and gave general agreement with Trial 1, indicating noticeable
material
changes in the main oxides for particle groups with higher LOIs

The research has shown that fly ash can be recovered from

long-term storage and by wet processing, separated into a range
Table 5 of particle groups, with properties suitable for various end-uses.
Classification of material for end-use applications.
For the fly ashes tested, carbon reduction was achieved, but
Application/designation Properties was influenced by the flotation method adopted. It has been sug-
Fineness, 45 lm sieve retention, LOI, % gested that column flotation is effective in the removal of fine
% by mass by mass minerals [15]. Furthermore, fly ash is prone to particle agglomer-
Concrete ation during wet storage, which may affect carbon particle distri-
Fine fly ash 0 6.0 bution and influence separation. It is also possible that with
Category S/category Ba 12.0 7.0 column flotation there was insufficient carbon to form a stable
Category N/category Ba 40.0 7.0
froth, thereby reducing the process’s effectiveness. The improved
Fly ash bound mixtureb 60.0 10.0
Groutc 60.0 14.0
results with mechanical flotation seem likely to relate to the
Coarse/high LOI >60.0 >14.0 method’s ability to remove coarser particles. The vigorous action
Carbon-rich residue Potential re-use as fuel of this process may also have been beneficial in breaking down
a
Fineness and LOI categories to BS EN 450-1, [7].
agglomerates.
b
To BS EN 14227-3 [8]. The characteristics of fly ash from the LHC were influenced by
c
To BS 3892-3 [9]. the properties of the feed material as indicated in Fig. 8, which
440 M.J. McCarthy et al. / Fuel 111 (2013) 430–441

Fine fly ash Fine fly ash Fine fly ash

Cat. S/Cat. B Cat. S/Cat. B
Carbon- Cat. S/Cat. B
rich Carbon-
residue Carbon-
rich
rich
residue
Fly ash residue
Fly ash Cat. N
Coarse / bound
Coarse / bound
High LOI mixture Fly ash
High LOI mixture Coarse / bound
High LOI mixture
Grout

Trial 1 Trial 2 Trial 3

APPLICATION/ QUANTITY, %
DESIGNATIONa Trial 1 Trial 2 Trial 3

Concrete
Fine fly ash 7.3 6.2 4.1
Category S/Category B 8.0 9.2 16.8
Category N/Category B - - 10.5
Fly ash bound mixture 27.8 22.2 14.0
Grout - 17.8 -
Coarse / High LOI 43.3 20.3 24.9
Carbon-rich residue 13.6 23.4 29.8
a See Table 5 for classification

Fig. 9. Material quantities for potential end-use applications from the three trials (Fly Ash 3).

shows the distribution of particles in different size ranges during
the three trials. For Trial 1, where limited initial processing was
used, greater quantities of material of all particle sizes were ob-
tained in U1, with reductions generally obtained from U2 to U5
thereafter (except at <5 lm for U5). The presence of large quanti-
ties of fine particles noted in U1 was unexpected and may relate
to their agglomeration as mentioned above and their distribution
in the sample. The results of Trials 2 and 3 both gave similar behav-
iour, with particles <20 lm generally increasing from U1 to U5 (ex-
cept at 10–20 lm for U5), with the reverse occurring for those
greater than this size. The results, therefore, highlight the effect
of material characteristics on entry to the LHC and role of initial
processing on the separation achieved.
Given the influence of the recovered material and processing
techniques on separation, the selection of particular systems and
optimisation to meet specific end property requirements should
be possible. To illustrate end-use potential of material following
the three trials, the separated particle groups of Fly Ash 3 are con-
sidered with respect to the classification given in Table 5, and the
yields from this are shown in Fig. 9. As indicated, the quantity of
carbon-rich residue increased between Trials 1, 2 and 3, reflecting
the targeted separation of this. Trial 1 gave more fly ash bound mix-
ture and high LOI/coarse material than the other trials, while Trial 3
gave greater quantities for use as a cement component in concrete.
Trial 2, which also gave material suitable for grout, had similar
quantities for each of the potential end-uses considered. Clearly
the above is based on fineness and LOI, and the materials would
need to meet other property requirements in relation to these.
Preliminary research [18] examining the performance of the
processed materials, indicates general suitability of these in var-
ious applications. However, with the material’s use in concrete,
there appeared to be little benefit to strength with finer fractions,
unlike with dry (recently produced) fly ash [25] (although re-
duced water requirements were obtained with these). This may
be due to the tendency for finer material, generally recognised
as contributing most to fly ash cementing properties [21,26], to
be retained at different processing points, and the effect of wet
storage conditions on reactivity. End-use applications of the pro-
cessed materials, represents an area being further investigated by
the Authors.
9. Conclusions
1. Fly ash in UK pond and stockpile storage areas has a wide range
of physico-chemical characteristics, with most being of high LOI
(>10%) and relatively coarse (>40% retained on a 45 lm sieve).
The presence of rough particle surfaces and agglomerates was
also noted, which seem to relate to the wet storage conditions
and surface reactions. For this heterogeneous type of fly ash
to be used holistically, then carbon and particle size separation
would be necessary to concentrate particle groups.
2. Initial tests indicate that the use of coarse screening and lamella
hydraulic classifier (LHC) only, did not give carbon reductions
(measured as LOI) to levels suitable for fly ash use as a cement
component in concrete. The results suggest that carbon parti-
cles behaved similarly to fine fly ash during separation and thus
were difficult to differentiate and hence additional methods (i.e.
froth flotation) were necessary to obtain low LOI/high fineness
particle groups.
3. The main trials used various combinations of wet-processing
methods, i.e. pre-screening, column (hydraulic) separation,
froth flotation and LHC. These indicate that initial processing
 M.J. McCarthy et al. / Fuel 111 (2013) 430–441
and the froth flotation method used were important factors
influencing the properties of the particle groups obtained. The
data also show that there was a balance in achieving low LOI
and high fineness fly ash during separation and their yields.
The effects noted during the trials were generally consistent
between fly ash from pond and stockpile storage areas.
4. Scanning electron microscopy and LASER diffraction indicate
that particles of <5 lm, generally regarded as the most reactive,
were difficult to concentrate and were present throughout the
separated particle groups. Minor changes in chemical composi-
tion were generally found for those groups mainly comprising
fly ash. Greater differences were apparent in high LOI particle
groups. Noticeable variations between non-crystalline compo-
nents of the coarsest and finest fractions from the LHC (U1
and U5) were obtained.
5. For the three trials carried out, it was possible to separate mate-
rials to give characteristics for a range of end-uses, e.g. cement
component in concrete, grout and road bases. Further work is
needed to examine their influences in various applications.
The results suggest that there is potential to target specific
material properties, using appropriate combinations of separa-
tion techniques, depending on local requirements.
Acknowledgements
The Authors would like to acknowledge, The Department for
Environment, Food and Rural Affairs (DEFRA) UK, and partners,
Ash Resources Ltd. (Pty), British Cement Association, British Precast
Concrete Federation, Castle Cement Ltd., European Coal Combus-
tion Products Association (ECOBA), Quarry Products Association,
ScotAsh Ltd., and United Kingdom Quality Ash Association (UK-
QAA) for their financial and technical contributions to the Project.
The input of Professor R K Dhir is much appreciated. Thanks are
also given to technical staff of the CAER, University of Kentucky,
in particular Mr. John Wiseman.
References
[1] European Coal Combustion Products Association. <http://www.ecoba.com/
ecobaccpprod.html> [accessed 02.13].
[2] McCarthy MJ, Dhir RK. Towards maximising the use of fly ash as a binder. Fuel
1999;78(2):121–32.
[3] Dhir RK, McCarthy MJ, Zheng L, Tella G. Feasibility of recovery and
beneficiation of stockpiled and landfilled PFA for use in construction. CTU
Report 3105; 2005, p. 155 [University of Dundee, Scotland, UK].
[4] Groppo JG, Robl TL. Ashes to energy: the Coleman station project. In:
International fly ash symposium, Lexington, Kentucky, USA; 2001. p. 205–215.
441
[5] McCarthy MJ, Tittle PAJ, Dhir RK. Lagoon PFA: feasibility for use as a binder in
concrete. Mater Struct 1998;31(10):699–706.
[6] McCarthy MJ, Tittle PAJ, Dhir RK. Characterisation of conditioned PFA for
use as a cement component in concrete. Mag Concr Res 1999;51(3):
191–206.
[7] British Standards Institution, BS EN 450–1. Fly ash for concrete: definitions,
specifications and conformity criteria. London, UK: BSI; 2012.
[8] British Standards Institution, BS EN 14227–3. Hydraulically bound mixtures–
specifications: fly ash bound mixtures. London, UK: BSI; 2004.
[9] British Standards Institution, BS 3892–3. Pulverized-fuel ash: specification for
pulverized fuel ash use in cementitious grouts. London, UK: BSI; 1997.
[10] Jones MR, McCarthy MJ, Newlands MD. Fly ash route to low embodied CO2 and
implications for concrete construction. Denver, Colorado, USA: World of Coal
Ash Conference; 2011.
[11] Jones, MR, Sear LKA, McCarthy MJ, Dhir RK. Changes in coal-fired power station
fly ash: recent experiences and use in concrete. AshTech 2006. Sear LKA,
editor., In: Proceedings of the International Conference organised by UKQAA,
Birmingham, UK; 2006.
[12] British Standards Institution. BS EN 451-2. Method of testing fly ash:
determination of fineness by wet sieving, BSI, London, UK; 1995.
[13] Sear LKA. The properties and use of coal fly ash: a valuable industrial by-
product. London, UK: Thomas Telford; 2001. p. 261.
[14] British Standards Institution. BS 8500-2. Concrete. Complementary British
standard to EN 206-1: specification of constituent materials and concrete, BSI,
London, UK; 2006.
[15] Wills BA. Mineral processing technology. 6th ed. Oxford: Butterworth-
Heinemann; 1997. p. 486.
[16] Robl TL, Groppo JG, Jackura A, Tapp K. Field testing of an advanced multi-
product coal by-product processing plant at Kentucky Utilities Ghent power
plant. AshTech 2006. Sear LKA, editor., In: Proceedings of the international
conference organised by UKQAA, Birmingham, UK; 2006.
[17] Robl TL, Groppo JG. Method for hydraulically separating carbon and classifying
coal combustion fly ash. US patent publication no US7963398 B2; 2011.
[18] Jones MR, McCarthy MJ, Zheng L, Robl TL, Groppo JG. Experiences of processing
fly ashes recovered from United Kingdom stockpiles and lagoons, their
characteristics and potential end uses. Lexington, Kentucky, USA: World of
Coal Ash Conference; 2009.
[19] Hwang JY, Sun X and Li Z. Unburned carbon from fly ash for mercury
adsorption. I. Separation and characterization of Unburned carbon. J Miner
Mater Charact Eng 2002;1(1):39–60. .
[20] Berry EE, Hemmings RT, Langley WS, Carette GG. Beneficiated fly ash :
hydration, microstructure and strength development in Portland-cement
systems. In: Third international conference: fly ash, silica fume, slag and
natural pozzolans in concrete, Trondheim, Norway, ACI-SP-114-11, Ed VM.
Malhotra; 1989. p. 241–273.
[21] Jones MR, McCarthy A, Booth APPG. Characteristics of the ultrafine component
of fly ash. Fuel 2006;85(16):2250–9.
[22] Ghosal S, Self SA. Particle size–density relation and cenosphere content of coal
fly ash. Fuel 1995;74(4):522–9.
[23] Watt JD, Thorne DJ. Composition and properties of pulverized-fuel ashes. I.
Composition of fly ashes from some British power stations and properties of
their component particles. J Appl Chem 1965;15(12):585–94.
[24] Vempati RK, Rao A, Hess TR, Cocke DL, Lauer HV. Fractionation and
characterization of Texas Lignite Class F fly ash by XRD, TFA and FTIR and
SFM. Cem Concr Res 1994;24(6):1153–64.
[25] Dhir RK, McCarthy MJ, Magee BJ. Impact of EN 450 fly ash on concrete
construction in the UK. Constr Build Mater 1998;12(1):59–74.
[26] Mehta PK. Influence of fly ash characteristics on the cube strength of Portland-
fly ash mixtures. Cem Concr Res 1985;15(4):669–74.