Desalination 288 (2012) 126–134

Contents lists available at SciVerse ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Fouling and fouling mitigation on heated metal surfaces

S.N. Kazi a,⁎, G.G. Duffy b, X.D. Chen c
a
Department of Mechanical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
Department of Chemical and Materials Engineering, School of Engineering, The University of Auckland, Private Bag 92019, Auckland, New Zealand
c
Biotechnology and Food Engineering, Department of Chemical Engineering, Monash University, Victoria 3800, Australia

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

Article history: An apparatus was built to study heat transfer fouling on different heat exchanger pipe surfaces by visually
Received 2 September 2011 observing the progressive fouling deposition under the same solution conditions. Test pipes were centrally
Received in revised form 21 December 2011 located in a cylindrical tank with a concentric vertical agitator to give constant and uniform flow conditions
Accepted 21 December 2011
near the pipe surface. Pipes with either smooth or roughened surfaces provided quantitative data on the pro-
Available online 28 January 2012
gressive build-up and the composition of the deposits. The calcium sulphate deposition on four different
Keywords:
metal surfaces (copper, aluminium, brass, and stainless steel) was investigated. The results show that fouling
Heated surfaces increases with time but at a decreasing rate over set intervals. The deposition also increases with the increas-
Calcium sulphate ing thermal conductivity of the metal, or the total surface energy. Chemical reaction fouling along with par-
Fibre ticulate and crystallisation fouling occurred on reactive surfaces when corrosive chemicals were used, and
Fouling mitigation this was compared with crystallisation-only fouling on non-reactive surfaces. Bleached Kraft softwood pulp
fibres at various concentrations were added to the fouling solution to study their affects on fouling on the hy-
draulically smooth pipes. Fouling was retarded with only a low fibre concentration and reduced further as
fibre concentration was increased.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction 98 °C. A change of 10 to 20% in induction time was observed while

experimenting with a copper plate at a Reynolds number of 23,000, ini-
Mineral scale formation on heat exchanger surfaces rapidly increases tial Cb = 3.0 kg/m3 and bulk temperature Tb of 40 °C.
the thermal resistance and reduces operating service life. Panchal and Al Mutari [7] noted the increase of fouling deposition rate of CaSO4
Epstein [1,2] categorised 5 types fouling (biochemical interaction, chem- with the increase of concentration of the fouling solution. Amjad [8]
ical reaction, particulate, crystallization, and corrosion fouling), and 5 studied gypsum deposition on various metal surfaces and found
deposit steps (generation, transportation, attachment, removal, and that the order of the rate of deposition on different metals as; cop-
aging). Panchal [1] also mentioned that crystallisation fouling is formed per > brass > stainless steel. It is reported that molecules or atoms at
by the deposition of salts on the surface by attraction. Hasson [3] the depositing surface interact with those of the solution and the
reported that asymptotic deposit behaviour observed for mixed salts types of forces or interactions depend on the chemistry of both the
may be different from solutions of only single salts. Tlili et al. [4] using surface and the fouling liquid [9]. Generally the lower the surface
a gold-layered nickel foil found that calcium sulphate concentration free energy of the depositing surface, the weaker the adhesion of de-
has a stronger effect than temperature in the precipitation process. posit on it [10]. Low-energy surfaces are more resistant to build-up of
Demopoulos [5] reported that supersaturation is the important control- fouling and easier to clean due to weaker binding at the substratum–
ling parameter of aqueous precipitation. Mwaba et al. [6] experimentally liquid interface [11]. Fouling occurs when the total interactive force is
investigated calcium sulphate deposition on a flat plate and observed negative [12]. Zhao and Muller-Steinhagen [12] described surface free
that nucleation started on the downstream side and then propagated energy (surface tension) as a direct measure of intermolecular forces
upstream. They also observed that the rates of nucleate-front propaga- that affect the attraction of the bulk liquid to the surface layer. Gill
tion and scale layer growth increased with the surface temperature and Nancollas [13] studied crystallisation of calcium sulphate on dif-
and decreased with the flow velocity. They have defined super satura- ferent metal surfaces using a constant heat flux method. They ob-
tion as the ratio of the bulk concentration Cb to the saturation concentra- served more deposition with metals of higher thermal conductivity.
tion Cs. They also stated that crystalline calcium sulphate forms on the They reiterated that the metals with the higher thermal conductivity
surface as Gypsum (CaSO4·2H2O) in the temperature range of 40 °C to yielded higher initial steady-state temperatures in the order of
Cu > Al > Brass > Stainless steel 304.
⁎ Corresponding author. Tel.: + 60 3 7967 4582; fax: + 60 3 7967 5317.
Surface roughness leads to a much larger effective contact area
E-mail addresses: salimnewaz@um.edu.my (S.N. Kazi), gg.duffy@auckland.ac.nz than the projected surface area and hence deposit adhesion is stron-
(G.G. Duffy), dong.chen@monash.edu (X.D. Chen). ger [14]. Surface roughness may also enhance the initial deposition

0011-9164/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2011.12.022
 S.N. Kazi et al. / Desalination 288 (2012) 126–134 127

rate by providing sites for crystal nucleation [15]. Yoon and Lund [16]
reported that no surface roughness effect was found in the Ra (the ar-
ithmetic mean of the departure of the measured roughness profile
from the mean line value) value range of 0.08 to 0.60 μm. McGuire
[17] and McGuire and Swartzel [18] in contrast showed that surface
roughness was not a factor for fouling with their surfaces having Ra
values of 0.04, 0.41, 1.93 and 2.31 μm for polished SS, rough SS,
PTFE coating and alumino-silicate coating respectively. Herz et al.
[19] also confirmed that calcium sulphate deposition was strongly
affected by the degree of surface roughness and the fouling layer
on a rough surface was more tenacious than on a smooth surface.
They also stated that the higher roughness causes shorter induction
periods.
Zhao and Muller-Steinhagen [12] treated metal heat exchanger
surfaces with selected elements by ion implantation, sputtering, elec-
trolytic deposition (or a combination of these). The fouling resistance
was reduced by up to 78% when CaSO4 was used.
Junghahn [20] reported that the free-energy change associated Fig. 1. Schematic diagram of the test apparatus.

with crystal nuclei formation was much less on a rough surface than
on a smooth surface. Other authors [21,22] however, reported higher 37 L/min with a maximum head of 4.5 m [130 W, 2,400 rpm motor].
deposition rates on rougher surfaces. Gunn [23] investigated 304- Full details are given in the PhD thesis by Kazi [25].
stainless steel tubes with different degrees of roughness correspond- The other flow loop is a simple 6 mm diameter copper coil for
ing to the bonded particles at 80, 120, 180 and 320 Tyler mesh. He once-through mains cooling water for keeping the bulk temperature
maintained the steady-state temperatures of the tube surface and at the required value. The temperature drop over the test section
the bulk liquid at 97 °C and 95 °C respectively and observed that the was approximately 2.5 °C.
rate of nucleation increased with increasing surface roughness. The fouling solution is contained in the agitated Perspex (Lucite)
Some anti-foulants have been successful in preventing fouling but cylindrical tank, 202 mm diameter and 807 mm high. The tank capac-
some have caused undesired product contamination or adverse envi- ity at the top mark is 17.5 L. It is graduated from a top mark on the
ronmental effects. There are some limited methods for reducing foul- tank; designated A0, (594 mm from the bottom). Other level mark-
ing by changing plant operational parameters using techniques such ings from the top A0 are: B0, C0, D0, E0 and F0, which are 90, 162,
as electromagnetic, electrostatic and acoustic fields, ultraviolet light, 234, 342 and 414 mm respectively below the top A0 level. The
radiation or catalytic treatments. The results have been mixed and marks were used to locate the fouling sections after different periods
the outcomes inconclusive. of time as the tank level was sequentially dropped to those marks as
The present work is in line with environmental-friendly procedures the experiments progressed.
by adding solid, biodegradable fibre to a fouling liquid to mitigate foul- The stirrer of the tank is an inverted U-shaped PVC impeller with
ing of metal surfaces. Calcium sulphate dihydrate (CaSO4∙2H2O) scale two wings (dimensions 660 mm long × 16 mm wide × 4 mm thick).
formation on different heat exchanger surfaces from aqueous solutions The speed of the stirrer is controlled with a variable speed drive.
have been studied in a reproducible experimental set-up. The effect of The system is isolated from the atmosphere. The bulk fluid tempera-
bleached Kraft softwood fibre [24] on CaSO4∙2H2O scale formation ture was controlled to within ± 1 °C.
from metastable supersaturated solutions was also been investigated
in single-pipe SS heat exchangers at different flow rates, salt concentra-
tions and temperature differences. Wood pulp fibres at various concen- 2.2. Test specimens
trations were discovered to mitigate fouling and reduce the deposition
rate, showing that the life of a heat exchanger could be prolonged. The experimental test pipes are 1200 mm in length each with a
The study of fouling on different metal surfaces in a shear field ap- wall thickness of 1.2 mm and an outer diameter of 12.7 mm. These in-
proaches the reality of heat exchangers used in industry. The present clude four metal pipes (stainless steel SS 316, brass, copper, and alu-
work reports the (visual) history of fouling deposition of calcium sul- minium) of the same dimensions. Test pipes made from these
phate with time under the same conditions of heat flux with different materials were connected directly and vertically to the outlet pipe
types of pipes and surface finishes. It compares crystallisation fouling from the pump, through a seal at the centre of the tank base, and
on stainless steel and aluminium surfaces as well as crystallisation then to the return pipe at the top above the tank. The test length of
and corrosion fouling on copper and brass surfaces. Interactive effects each pipe was the actual submerged length between the inside base
of particulate and corrosion fouling are also recorded along with the of the tank and the solution surface, which was lowered sequentially
additional effects of fibre in the fouling solution. during each experimental run.
The smooth test specimens (Table 1) were used in the as-received
condition but before installing in the test rig they were cleaned by
2. Experimental rubbing with a water-soaked cloth and flushing with hot water to re-
move any deposits of grease, oil etc.
2.1. The apparatus

Table 1
The schematic diagram of the experimental apparatus is presented
Hydraulically-smooth and roughened test pipes.
in Fig. 1. It consists of two flow loops and a stirred tank containing the
fouling solution and test pipe. One flow loop comprises a centrifugal Smooth specimen μm Ra Roughened specimen μm Ra
pump, rotameter, a replaceable pipe and test section, and a thermo- SS 316 0.245 SS 316 0.88
statically controlled water bath to maintain the heating water at a Brass 0.47 Brass 2.30
Copper 0.25 Copper 2.80
more or less constant temperature (55 ± 1.5 °C). The hot- water recir-
Aluminium 0.22 Aluminium 2.90
culation centrifugal pump (Little Giant Pump Co., USA) can operate at
128 S.N. Kazi et al. / Desalination 288 (2012) 126–134

The roughened test pipes (Table 1) were prepared by sand blast- Table 2
ing the surfaces uniformly along the entire length and washing Conditions of the experimental runs.

them before installing in the test rig. Surface roughness values were Parameters Range of values
determined with a TAYLOR HOBSON Surfronic 3 instrument. These
U (outside pipe surface) 0.1 m/s
roughness results are reported as μm Ra. The instrument was calibrat- Tbulk 40 ± 1 °C
ed using a smooth glass surface. The longitudinal surface roughness ΔT 15 ± 1 °C
variation of a typical brass roughened test pipe was within the limit CaSO4 concentration, bulk (Cb) 3.6 g/L
Cs at 55 °C 2.16 g/L
of variation of 12%, indicating the uniformity of the roughened sur-
Average fibre length 2.53 mm
faces made in the workshop. Fibre coarseness (mass/unit length) 0.246 mg/m
RFN (Relative fibre number) 80
2.3. Data acquisition Cfibre 0.005 to 0.1%
pH of solution 7.5

A Picolog software program was employed to record the mean

tank temperature and the inlet and outlet temperatures of the hot
water passing through the vertical test pipe. Both inlet and outlet bottom of the tank. Heat transfer data were recorded during each
temperatures were measured with K-type thermocouples, while a step. Similarly the heat transfer to a 0.01% fibre suspension in water
T-type thermocouple was used to obtain the bulk temperature. The only was measured to compare with the results for water using the
program was set to record temperatures every 10 min and set to ter- same sequence.
minate each experiment after 4000 min of data logging.

2.7. Experiments with supersaturated calcium sulphate and fibre

2.4. Experimental procedure
suspension
The Perspex tank was filled with distilled water and the hot water
The experimental conditions are summarised in Table 2. The tank
flow from the water bath started. The water bath controller was set
level was reduced sequentially so that the test pipes were exposed to
and the flow rate through the test pipe was set at a predetermined
the solution for various times. After 25 min of running, the level of
value. The upward cooling water flow was adjusted to obtain a tem-
tank was reduced from A0 to 1/2B0. After 100 min the level was low-
perature difference (ΔT) of 15 ± 1 °C between the bulk liquid temper-
ered from 1/2B0 to B0 and then to C0 after 400 min. From A0 to C0 the
ature in the tank and the mean surface temperature of the test pipe.
nucleation and initial deposition could be investigated. From C0 and
The average pipe surface temperature was estimated by averaging
sequentially via D0, E0, and F0 to the bottom of the tank, the time in-
the inlet and outlet temperatures of the hot water flowing through
terval was 1000 min, so that the history of deposition could be ob-
the tube. The mixer speed was set to a constant 45 rpm so that the
served visually and recorded.
tangential liquid velocity at the external tube surface was approxi-
On completion of an experimental run the test pipe was removed,
mately 0.1 m/s; roughly equivalent to previous pipe flow experimen-
preserved in a container, and air-dried slowly for 7 days. The deposit
tal conditions. Heat transfer coefficients were calculated for the
patterns, formed deposits, and foreign products were first examined
various experiments carried out at different tank levels. The calculat-
visually. The tube surfaces were marked at the prescribed intervals
ed heat transfer coefficient for water calibration provided a basis for
and photographed. The deposits were removed by scraping the sam-
comparison with values obtained from other experimental condi-
ples weighed. The small amount of remaining deposit at each marked
tions, including those with fibre suspensions.
length was dissolved in water and the concentration was determined
by titrating with EDTA (Ethylene-diamine-tetra-acetic acid). The dis-
2.5. Preparation of calcium sulphate solution
solved amounts were added to the scraped amount to get the total
deposition at each marked interval. The amount in a specified area
Stoichiometric proportions of calcium nitrate tetrahydrate
in each interval was measured and presented in a graphical form.
Ca(NO3)2∙4H2O and sodium sulphate Na2SO4 were dissolved sepa-
The deposits on the surface were analysed using X-ray diffraction
rately in distilled water to prepare the aqueous solution of calcium
(XRD) to determine the composition of the various deposits. Differ-
sulphate:
ent crystals have different ‘d’ spacings. If X-rays of known wave-
CaðNO3 Þ2 ∙4H2 O þ Na2 SO4 →CaSO4 ∙2H2 O þ 2NaNO3 þ 2H2 O ð1Þ length λ and measuring angle θ are used (where θ is half of the
angle between the diffracted beam and the transmitted beam), the
Distilled water was heated to 40 °C and then calcium nitrate and spacing ‘d’ of the various planes in a crystal can be determined.
sodium sulphate solutions were gradually added sequentially while Therefore by measuring the ‘d’ spacings, different compounds of de-
continuously stirring. The desired concentration of the solution was posit could be identified. The XRD patterns are presented by plotting
maintained throughout the experimental runs by sampling from the counts (intensity) as a function of 2θ to identify the composition of
sample line at 6 h intervals. Samples were analysed by titrating with deposits [27].
EDTA [26] and the tank solution concentration was adjusted accord- Bleached Kraft soft wood fibres were used to investigate the ef-
ingly, either by adding more liquid chemicals or by adding more fect of fibres on fouling mitigation. Fibres were soaked in water for
water. When applicable, the fibre suspension concentration was 24 h, blended for 20 s to disperse them in water, and then added
maintained by addition or solution dilution. to the fouling solution. Full details are given in the PhD thesis by
Kazi [25].
2.6. Experiments with water only and with a 0.01% fibre suspension

An experiment was conducted with water alone to study the heat 3. Results and discussion
transfer coefficient of a SS 316 (smooth) test pipe surface for calibra-
tion of the test rig. The test was performed at ΔT 15 °C, U = 0.1 m/s The experimental procedures are validated in Sections 3.1–3.5.
(estimated) and other parameters as described in Table 2. The liquid The data were also reproduced with good agreement between two
level in the tank was reduced step-wise and sequentially at a constant similar runs. The results show that the current set-up has yielded
interval of 60 min from A0 via 1/2B0, B0, C0, D0, E0, F0, and to the similar results to previously-used pipe and annular-flow systems.
 S.N. Kazi et al. / Desalination 288 (2012) 126–134 129

5 200
Heat Trans. Coeff. kW/m2K
Water
CONCENTRATION EFFECT
0.01% fibre
4 150 Δ CaSO4 3.0 g/l

Deposition g/m2
Ο CaSO4 3.6 g/l
3
100

2
50

1
0 1 2 3 4 5 6
0
Time min x 102 0 1000 2000 3000 4000 5000

Fig. 2. Variation of heat transfer coefficient for water only, and a 0.01 mass % fibre in water
Time min
suspension. Bulk temperature 40 °C, ΔT 15 °C, 45 rpm stirrer (0.1 m/s equivalent).
Fig. 4. Fouling deposition as a function of time for two different concentrations of
CaSO4 solution. ΔT 15 °C and bulk temperature of 40 °C.

3.1. Water and fibre–water suspension experiments using SS surfaces

The calculated values of the heat transfer coefficients for water as It is observed that deposition increases with the increasing solu-
well as a 0.01 mass% suspension of fibre are plotted as a function of tion concentration as well as the time of exposure of the surface to
time in Fig. 2 for direct comparison. the solution. This is caused by the increased amount of crystallisation
It can be seen that wood pulp fibres in water reduce the heat in the boundary layer and increased total attractive force to the sur-
transfer coefficient. Similar results for fibre suspensions flowing in face. Again similar results were obtained by Kazi et al. [33], Middis
pipes have been previously reported [28–32]. [28] and Bansal [34].
Kazi et al. [31] studied heat transfer in the drag reducing regime
of wood pulp fibre pipe flow and reported that both drag and heat
transfer reduction increases with fibre flexibility as well as fibre 3.4. Fouling in CaSO4 solution in the presence of fibre (fibre concentration
concentration. effect)

3.2. Data reproducibility The effect of varying the fibre concentration in the CaSO4 solution
(3.6 g/L) was investigated for up to 4000 min exposure using the SS
Fig. 3 represents deposition of CaSO4 on the SS smooth surface as a 316 pipe. The results are presented in Fig. 5.
function of time for two different fouling runs under the same exper- Fibre concentration was varied from 0 to 0.1% in three steps. The
imental conditions (ΔT 15 °C, bulk temperature 40 °C, and CaSO4 con- results show similar trends of deposition. The slopes of the curves de-
centration 3.6 g/L). It is observed that data reproducibility is good and crease with increasing fibre concentration. By adding 0.005%, 0.01%
the data remain within the confidence level of 95% with an average and 0.1% fibre separately to the supersaturated CaSO4 solution, the
rms (root mean square) error less than 5%. The small variation is fouling depositions were reduced 24.3%, 78.4% and 99.7% respectively
due to the incursion of foreign particles in the solution during the at the specified final time of observation of 4000 min.
course of experiment as reported by Gill and Nancollas [13]. The results are in concert with those obtained in pipe flow studies
previously [30,33]. Fibres reduce the rate of deposition by ‘shielding’
3.3. Fouling in CaSO4 solution—solution concentration effect (without the surface from approaching crystals formed in the boundary layer,
fibre) by sustained micro-mixing, and by the continuous physical collision
with the heated surface. Fibres affect the turbulent eddies which re-
The deposition of calcium sulphate from two different supersatu- duce the rate of mass transfer of the fouling elements to the heated
rated solution concentrations on the stainless steel SS 316 (smooth) wall [33].
surface as a function of time was studied and the results are pre-
sented in Fig. 4.

200 200

Run 1 FIBRE EFFECT

Run 2
150 150 NO FIBRE
Deposition g/m2
Deposition g/m2

0.005% FIBRE

100 0.01% FIBRE

100
0.1% FIBRE

50
50

0
0 1000 2000 3000 4000 5000 0
Time min 0 1000 2000 3000 4000 5000
Time min
Fig. 3. Fouling deposition on stainless steel SS smooth tubes as a function of time for
two different fouling runs at same experimental conditions. Experiments were per- Fig. 5. Fouling deposition as a function of time for different fibre concentrations on a SS
formed at ΔT 15 °C, bulk temperature 40 °C, and CaSO4 concentration 3.6 g/L. 316 (smooth) surface, ΔT 15 °C, solution concentration 3.6 g/L, bulk temperature 40 °C.
130 S.N. Kazi et al. / Desalination 288 (2012) 126–134

400 400
Cu (0.01%) Brass (0.01%) SS (0.01%)
Copper
Cu (0%) Brass (0%) SS (0%)
Aluminium

Deposition g/m2
300
300
Deposition g/m2

Brass

SS 316
200
200

100

100
0
0 1000 2000 3000 4000 5000
Time min
0
0 1000 2000 3000 4000 5000
Fig. 8. Fouling deposition as a function of time on Cu, brass, and SS 316 smooth surfaces
Time min
exposed to solution with-and-without fibre in suspension. ΔT 15 °C, bulk temperature
40 °C, 0.01% fibre, solution conc. 3.6 g/L, 0.1 m/s.
Fig. 6. Deposition of CaSO4∙ 2H2O as a function of time and metallurgy of heat exchang-
er pipe. ΔT 15 °C, bulk temperature 40 °C, solution conc. 3.6 g/L and 0.1 m/s external
surface velocity.
to the free energy change associated with crystal nuclei formation
which is much less on a rough surface than on a smooth surface.
3.5. Fouling with CaSO4 on various smooth metal surfaces (without fibre) Present observations also qualify the inference of Junghahn that
edges and sharp points are centres of high surface energy causing
The influence of the nature of the metal (copper, aluminium, preferential nucleation sites. Rough surfaces produce greater deposi-
brass, and stainless steel) on the deposition of calcium sulphate tion than the smooth surfaces due to protected zones in the cavities
scale was studied in a series of experiments, and the results are pre- or pits where flow velocities are very low and nucleation rates are
sented in Fig. 6 as a function of time. The deposition on different higher [14].
metal substrates was consistent with order of their thermal conduc- Once crystallisation is commenced, the higher adhesive forces be-
tivity values, i. e. copper > aluminium > brass > stainless steel. It was tween the fouling layer and the surface influence the removal rate
also observed that the fouling on metal surfaces tends towards an as- and lead to higher asymptotic fouling values for the rough surfaces.
ymptotic limit at 4000 min, which is consistent with previous pipe
flow investigations [24]. Forster and Bohnet [35] found that the depo-
3.7. Fouling with CaSO4 on smooth surfaces in the presence of fibre
sition on different metal substrates was consistent with the values of
the polar component of the surface free energy, i.e. copper > alumin-
Three different smooth test specimens (copper, brass, SS 316)
ium > SS > brass. Amjad [7] also obtained similar results for calcium
were investigated in an aqueous CaSO4 solution of concentration
sulphate deposition on copper, brass and stainless steel surfaces at a
3.6 g/L with 0.01% fibre in suspension. Fig. 8 shows the deposition of
higher solution concentration of 4 g/L.
CaSO4 as a function of time. The deposition increases with increasing
thermal conductivity of the material i.e. Cu shows the highest deposi-
3.6. Fouling with CaSO4 on roughened metal surfaces (without fibre)
tion followed by brass, and then SS 316.
Fibres in turbulent motion away from the boundary layer affect
Data for the deposition of CaSO4 on smooth and roughened metal
the boundary layer which reduces the rate of mass transfer of the fou-
surfaces are presented in Fig. 7. Clearly it is observed that fouling de-
lant to the heated tube surface [24]. It is observed that the addition of
position increases with increasing roughness. The order of the rank-
even a very small amount of fibre in the aqueous solution of CaSO4
ing of deposition on the roughened surfaces is the same as the
consistently reduces the foulant deposition on different metal sur-
order of deposition recorded on smooth surfaces of those metals,
faces at any specified time interval.
and the trends in the fouling data are also asymptotic. The results
are in agreement with the observation of Junghahn [20] with respect
3.8. Effect of fouling liquid on surface topography

400
Table 3 summarises the effect of corrosion on different smooth
Cu SB Cu S
metal surfaces. No corrosion effects on SS and Al surfaces were ob-
Al SB Al S
served. Brass and Cu surfaces however experienced corrosion and
300 Brass SB Brass S
that generated a change at the surfaces which caused both crystallisa-
Deposition g/m2

SS SB SS S
tion and corrosion fouling. Although there were no measurable
roughness changes in Al and SS smooth surfaces, the results could
200 be compared with total-fouling deposition on smooth surfaces. The
achieved roughness changes all lie within the tolerance limit [16,17]

100
Table 3
Surface roughness change of smooth test specimens.

0 Time (min) SS 316 Aluminium Brass Copper

0 1000 2000 3000 4000 5000
0 000 0.25 0.22 0.48 0.25
Time min 1 000 0.25 0.22 0.64 0.46
2 000 0.25 0.22 0.98 0.95
Fig. 7. Fouling deposition as a function of time for Cu, Al, brass and SS test pipes both 3 000 0.25 0.22 1.38 2.38
smooth (S) and sand-blasted roughened (SB). ΔT 15 °C, bulk temperature 40 °C, solu- 4 000 0.25 0.22 1.38 2.38
tion conc. 3.6 g/L and 0.1 m/s surface velocity.
 S.N. Kazi et al. / Desalination 288 (2012) 126–134 131

Table 4 aqueous calcium sulphate with fibre suspensions of 0%, 0.005%,

Surface roughness change of rough test specimens. 0.01% and 0.1% concentration respectively. There is absolutely no sig-
Time (min) SS 316 Aluminium Brass Copper nificant deposition on the stainless steel tube surface with 0.1% fibre
in fouling liquid. It is also observed that fibre consistently mitigates
0000 0.88 2.90 2.30 2.80
1000 0.88 3.20 2.51 3.11 fouling on metal surfaces (Refer to Fig. 5).
2000 0.88 3.80 2.72 3.48 Fig. 10(d) and (e) represents the deposition on brass and copper
3000 0.88 3.80 2.95 4.07 (smooth) surfaces after 4000 min of exposure to 3.6 g/L aqueous
4000 0.88 3.82 2.96 4.08
CaSO4 with fibre suspension of 0.01%. It is observed that the order
of total deposition on SS 316 Fig. 10(b), brass Fig. 10(d), and copper
of surface alterations up to a comparable time of exposure to fouling Fig. 10(e), all for 0.01% fibre present, remains the same as the case
liquid. where no fibres are present.
Similarly, changes in the surface roughness of roughened metal The deposition on SS 316 surface is CaSO4·2H2O, but on brass sur-
surfaces are also due to corrosion as presented in Table 4. The rough- face it is accompanied with a green layer of CuSO4·2H2O formed due
ened surfaces of selected materials Cu, Al, brass, and SS 316 follow to the reaction of Cu component of brass with CaSO4 solution. In the
similar trends as discussed earlier for the case of smooth surfaces. case of the copper surface, CaSO4·2H2O formed on the surface is
also accompanied with the green stain of CuSO4·2H2O. In addition
3.9. Visualisation of the surface and crystal morphology to the green stain, reddish and black stains of CuO and Cu2O are
also obtained on the copper surface due to reaction of Cu with
The photographs in Fig. 9 show the white deposits formed on CaSO4 solution.
stainless steel SS 316 and aluminium surfaces, indicating only crystal- XRD spectra of the deposits of copper and brass are presented in
lization fouling. The deposits formed on the brass surface however Fig. 11 after 4000 min exposure. The deposition samples include
have green stains along with white heavy encrustations of calcium those from smooth and sand-blasted roughened copper pipes with
sulphate, showing some surface chemical attack and hence some no fibre in the fouling solution (see Fig. 7), as well as deposits from
chemical reaction fouling. the smooth copper and brass pipes with 0.01% fibre in suspension
The green crystals are CuSO4 formed by reaction between the Cu (see Fig. 8). The results show that the deposits on the hydraulically
component in brass and CaSO4. Similarly the deposition on the copper smooth copper pipe with-and-without fibre in solution are comprised
pipe surface also has a green layer along with the white deposition of hydrated calcium and copper sulphates and cuprous oxide Cu2O.
showing combined effect of crystallization and chemical reaction Visually there are black patches on the surface.
fouling. Thus in industry, it could be inferred that the cooling water The deposit from the roughened, sand-blasted copper surface is a
affect on corrosion-prone heat exchanger surfaces are the combined composite of CaSO4·2H2O and CuSO4·2H2O. Visual inspection of the
total effect of fouling. surface shows occasional reddish stains along with black patches
In these experiments fibre is used to mitigate fouling deposition but the XRD could not identify exactly if it were cupric oxide CuO,
on metal surfaces. So when fibre is used, the depositions on SS and al- as only a minute amount was present.
uminium surfaces were of similar nature. XRD analysis of the composite deposition on brass smooth surface
Fibre is mixed with deposits and forms a composite structure. On after the fouling run with 0.01% fibre in suspension (see Fig. 8) shows
the brass surface, CuSO4 and CaSO4 formed composites with fibre, but hydrates of CaSO4 and CuSO4. Clearly the nature of the composite de-
on the copper surface, reddish and black stains of CuO and Cu2O also posits varies with different metal surfaces but the fouling resistance is
formed in the composites. However the order of total deposition on the cumulative effect of deposition on the surface.
different metal surfaces does not change due to chemical reactions. The surface chemical reaction also affected the wall thickness and
Figs. 9(a) and 10(a) to (c) represent the crystalline deposition on surface roughness, and hence the fouling rate was also changed. After
the SS 316 (smooth) surface after 4000 min of exposure to 3.6 g/L experiments of 4000 min duration, the roughness of the initially

Fig. 9. (a–d). Deposition on smooth SS 316 (a) aluminium, (b) brass, (c) and copper (d) surfaces. ΔT 15 °C, bulk temperature 40 °C, solution conc. 3.6 g/L, 4000 min.
132 S.N. Kazi et al. / Desalination 288 (2012) 126–134

Fig. 10. (a–c). Crystal growth on SS 316 (smooth) surface exposed to solution of CaSO4, conc. 3.6 g/L, ΔT 15 °C, bulk temperature 40 °C, 4000 min exposure and 0.005%, 0.01% and
0.1% fibre respectively. (d–e). Deposition on brass and copper (smooth) surface exposed to solution of CaSO4, conc. 3.6 g/L, ΔT 15 °C, bulk temperature 40 °C, 0.01% fibre and
4000 min exposure.

smooth brass and copper surfaces were both increased from 0.48 to
1.38 μm Ra, and from 0.25 to 2.38 μm Ra respectively, but the
changes were not very high or far beyond acceptable limits
(Table 3) [16–18,6]. There were no measurable roughness changes
for the aluminium and SS surfaces.
The crystal size of deposits on different metal surfaces was studied
under the microscope. Fig. 12(a–d) represents the comparison of
crystal size by microscopic view of the crystals formed on different
metal (smooth) surfaces (SS 316, brass, aluminium, and copper)
after 1000 min exposure to a solution of CaSO4 having a concentra-
tion of 3.6 g/L. It is found that crystals formed on copper, aluminium,
brass, and SS surfaces increase in average magnitude in the same
order of thermal conductivity: copper > aluminium > brass > SS 316

900

750 1 Cu(S)
2 Cu(R)
600
3 Cu(0.01)
Counts

450 4 Brass(0.01)

300 1

2
150
3

4
0
0 40 80 120 160
2 Theta

Fig. 11. Counts as a function of 2 Theta of XRD for crystal and fibre crystal composite
samples deposited on different (Cu and brass) metal surfaces. The experiments were Fig. 12. (a–d). Microscopic view of crystals formed on SS 316, brass, aluminium and
conducted at ΔT 15 °C, bulk temperature 40 °C, CaSO4 concentration of 3.6 g/L and copper smooth surfaces. 1000 min exposure to solution of CaSO4, solution conc.
fibre concentration of 0 to 0.01%. 3.6 g/L, ΔT 15 °C and bulk temperature 40 °C.
 S.N. Kazi et al. / Desalination 288 (2012) 126–134 133

4. Conclusions

It can be concluded that:

1. The extent of fouling on different metal surfaces increases with in-

creasing thermal conductivity of the metal (copper > alumi-
nium > brass > stainless steel).
2. Increasing surface roughness increases fouling.
3. Fibres in solution retard fouling on smooth surfaces. The induction
period is also increased as previously observed in separate pipe
flow, and annular-flow experimental systems.
4. The effect of fibres on fouling mitigation is consistent, and irre-
25 minutes 100 minutes spective of the tube material used.
5. Surface characteristics are altered by corrosive attack and this pro-
motes fouling and reduces the effectiveness of mitigation. Thus
when different metal surfaces are compared, a single category of
fouling could not be specified.
6. Movement of insoluble corrosion products and their detachment
and re deposition need to be studied to generate realistic models
for industrial use. The effect of soluble corrosion products on the
generation of fouling precursors also needs to be studied.

Nomenclature
400 minutes Cb Bulk concentration of CaSO4 in g/L
Cfibre Fibre concentration in percent
Cs Saturation concentration of CaSO4 at surface temperature in
g/L
Ra The arithmetic mean of the departure of the measured
roughness profile from the mean line value
RFN Relative number of fibres per unit mass
Tbulk Solution bulk temperature in °C
U Velocity in m/s
ΔT Temperature difference in °C

Acknowledgement

The authors gratefully acknowledge the International Conference

2,000 minutes 4,000 minutes
on Heat Exchanger Fouling and Cleaning VIII-2009 (Peer-reviewed)
Fig. 13. Growth of crystal size with time (minutes) on a smooth brass test piece. for accepting this paper for presentation in June 14–19, 2009,
Experiments performed at CaSO4 concentration 3.6 g/L, ΔT 15 °C and bulk temperature Schladming, Austria.
40 °C.

References
surfaces. Similar results were reported by Liu et al. [36] for CaCO3 de-
[1] C.B. Panchal, Review of fouling mechanisms, Proc. of Int. Conference on mitigation
position on heated surfaces. They stated that bigger crystals could be of heat exchanger fouling and its economic and environmental implications,
found on surfaces where rate of deposition is higher. Banff, Alberta, Canada, , 1999, pp. 8–15.
The crystal sizes of deposits on the brass surface have been sys- [2] N. Epstein, Thinking about heat transfer fouling: a 5 × 5 matrix, Heat Transfer Eng.
4 (1983) 43–56.
tematically studied. It is reported (see Fig. 6) that the total-fouling [3] D. Hasson, Progress in precipitation fouling research—a review, Proc. Of Engineering
mass per unit surface area increased with time in the specified cases Foundation Conference, Understanding Heat Exchanger Fouling and its Mitigation,
in the range of investigation. Lucca, Italy, , 1997.
[4] M.M. Tlili, P. Rousseau, M. Ben Amor, C. Gabrielli, An electrochemical method to
Figs. 13 and 12(c) represent the gradual development and growth study scaling by calcium sulphate of a heat transfer surface, Chem. Eng. Sci. 63
of crystal size with time on a smooth brass test piece. The largest (2008) 559–566.
sized crystals are obtained at 4000 min of exposure of the tube spec- [5] G.P. Demopoulos, Aqueous precipitation and crystallization for the production of
particulate solids with desired properties, Hydrometallurgy 96 (2009) 199–214.
imen to the fouling liquid. It is also observed that between 1000 to [6] M.G. Mwaba, C.C.M. Rindt, A.A. Van Steenhoven, M.A.G. Vorstman, Experimental
4000 min the deposit contains many crystals closely packed together, investigation of CaSO4 crystallization on a flat plate, Heat Transfer Eng. 27
and large crystals are surrounded by many smaller randomly-oriented (2006) 42–54.
[7] N.N. Al-Mutairi, Fouling studies and control in heat exchangers, M.S. thesis,
crystals. Similar results were obtained by Tianqing et al. [37] for the de-
Department of Chemical Engineering, College of Engineering, King Saud University,
position of CaCO3. It can be noted that the photos in Fig. 13 also indicate Saudi Arabia. (2007).
the crystal morphology. [8] Z. Amjad, Calcium sulphate dihydrate (gypsum) scale formation on heat exchanger
surfaces: the influence of scale inhibitors, J. Colloid Interface Sci. 123 (1988)
However, in this case the surface attack is not as marked as in the
523–536.
case of the deposit on a rough brass surface. Thus for short-term ex- [9] M. Fletcher, The physiological activity of bacteria attached to solid surfaces, Adv.
perimental purposes, the total deposition on brass and copper are Microb. Physiol. 32 (1991) 53–85.
the cumulative effects of crystallization and surface reaction fouling. [10] E. Lindner, A low surface free energy approach in the control of marine biofouling,
Biofouling 6 (1992) 193–205.
Extra care should be given in industry during long term operation if [11] R.E. Baier, Substrata influences on the adhesion of micro-organisms and their
reaction fouling occurs because tube corrosion will be ongoing. resultant new surface properties, in: G. Bitton, K.S. Marshall (Eds.), Adsorption
134 S.N. Kazi et al. / Desalination 288 (2012) 126–134

of Micro-organisms to Surface, Wiley-Interscience Publishers, New York, 1980, [25] M.S.N. Kazi, Heat transfer to fibre suspensions-studies in fibre characterisation
pp. 59–104. and fouling mitigation, Ph. D. thesis, Chemical and Materials Engineering, The
[12] Q. Zhao, H. Muller-Steinhagen, Influence of surface properties on heat exchanger University of Auckland, Auckland, New Zealand. (2001).
fouling, Mitigation and Heat Exchanger Fouling Conference, Banff, Canada, , 1999, [26] D.A. Skoog, D.M. West, Analytical Chemistry, an Introduction, Holt, Rinehart and
pp. 217–228. Winston, Inc., New York, 1974.
[13] J.S. Gill, G.H. Nancollas, Kinetics of growth of calcium sulphate crystals at heated [27] B.D. Cullity, Elements of X-ray Diffraction, Reading, Addison – Wesley, Massachusetts,
metal surfaces, J. Cryst. Growth 48 (1980) 34–40. USA, 1959.
[14] B.H. Rankin, W.L. Adamson, Scale formation as related to evaporator surface conditions, [28] J. Middis, Heat Transfer and Pressure Drop For Flowing Wood Pulp Fibre Suspensions,
Desalination 13 (1973) 63–87. Ph. D. thesis, Chemical and Materials Engineering, The University of Auckland,
[15] J. Taborek, T. Aoki, R.B. Ritter, J.W. Palen, J.G. Kundsen, Predictive methods for Auckland, New Zealand. (1994).
fouling behaviour, Chem. Eng. Prog. 68 (1972) 69–78. [29] J. Middis, H.M. Muller-Steinhagen, G.G. Duffy, Heat transfer and frictional pressure
[16] J. Yoon, D.B. Lund, Magnetic treatment of milk and surface treatment of plate heat drop characteristics of model nylon fibre suspensions, Appita J. 47 (1994) 154–158.
exchanger: effects on milk fouling, J. Food Sci. 59 (1994). [30] J. Middis, S.T. Paul, H.M. Muller-Steinhagen, G.G. Duffy, Reduction of heat transfer
[17] J. McGuire, The Influence of Solid Surface Energies on Macromolecular Adsorption fouling by the addition of wood pulp fibres, Heat Transfer Eng. 19 (1997) 1–9.
from Milk, Ph.D. thesis, North Carolina State University. (1987). [31] S.N. Kazi, G.G. Duffy, X.D. Chen, Heat transfer in the drag reducing regime of wood
[18] J. McGuire, K.R. Swartzel, Influence of solid surface energetics on macromolecular pulp fibre suspensions, Chem. Eng. J. 73 (1999) 247–253.
adsorption from milk, American Institute of Chemical Engineers, National Meeting, [32] G.G. Duffy, S.N. Kazi, X.D. Chen, A new way to monitor wood pulp fibre quality
1987, p. 31. using heat transfer coefficient, Appita Annual Conference, Melbourne, Australia,
[19] A. Herz, M.R. Malayeri, H. Muller- Steinhagen, Fouling of roughened stainless steel 2000.
surfaces during convective heat transfer to aqueous solutions, Energy Convers. Manage. [33] S.N. Kazi, G.G. Duffy, X.D. Chen, Heat exchanger fouling mitigation with flexible
49 (2008) 3381–3386. fibres, 26th Annual Chemeca Conference, Port Douglas, Australia, , 1998.
[20] L. Junghahn, Methoden Zum Herabsetzen oder Verhindern der Krustenbildung, [34] B. Bansal, Crystallisation Fouling in Plate Heat Exchangers, Ph. D. thesis, Department
Chem. Ing. Tech. 36 (1964) 60–67. of Chemical and Materials Engineering, The University of Auckland, Auckland, New
[21] T.R. Bott, J.S. Gudmundsson, Rippled silica deposits in heat exchanger tubes, 6th Zealand. (1994).
Int. Heat Transfer Conference, , 1978. [35] M. Forster, M. Bohnet, Modification of the characteristics of heat transfer surfaces as
[22] B.D. Crittenden, E.M.H. Khater, Fouling from vaporising kerosine, J. Heat Transfer a fouling mitigation strategy, Mitigation and Heat Exchanger Fouling Conference,
109 (1987) 583–589. Banff, Canada, , 1999, pp. 229–238.
[23] D.J. Gunn, Effect of surface roughness on the nucleation and growth of calcium [36] T. Liu, Y. Shi, X. Wang, Formation and remove process of calcium carbonate fouling,
sulphate on metal surfaces, J. Cryst. Growth 50 (1980) 533–537. Mitigation and Heat Exchanger Fouling Conference, Banff, Canada, , 1999.
[24] M.S.N. Kazi, G.G. Duffy, X.D. Chen, Fibre modified scaling in heat transfer fouling [37] L. Tianqing, S. Yuwen, X. Wang, Formation and remove processes of calcium carbonate
mitigation, Chem. Eng. Commun. 189 (2002) 742–758. fouling, Mitigation and Heat Exchanger Fouling Conference, Banff, Canada, 1999.