Energy xxx (2015) 1e11

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Cooling tower fog harvesting in power plants e A pilot study

Ritwick Ghosh a, 1, Tapan K. Ray b, 2, Ranjan Ganguly c, *
a
Boiler Maintenance Department, NTPC Limited, Farakka 742236, India
b
NTPC School of Business, PMI, Noida 201301, India
c
Department of Power Engineering, Jadavpur University, Kolkata 700098, India

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

Article history: Fresh water shortage is a major global problem of this century. Estimates have shown that a large part of
Received 30 April 2015 the human population will not have access to clean drinking water in a couple of decades from now.
Received in revised form Collection of fog can be a useful solution to this concern. Fog, a large source of potable fresh water, has
15 June 2015
potentials to substitute traditional sources. Attempts have been made over the last few decades to
Accepted 16 June 2015
Available online xxx
capture fog from nature by installing large fog water collectors along coastal mountains and highlands.
However, fog harvesting from artificial fog generators were not envisaged in these studies. In this pilot
study, we have explored the possibilities of fog capture from CT (cooling tower) plume in a thermal
Keywords:
Fog harvesting
power plant; CT plume accounts for one of the major sources of industrial water losses. Our study shows
Cooling tower that a recovery of about 40 percent water from the drift loss e amounting to a saving of nearly 10.5 m3 of
Drift loss capture water per hour from a 500 MW unit e could be achieved using the proposed fog harvesting strategy.
Metal wire woven mesh Unlike the natural fog harvesting schemes where the fog laden flow is predominantly horizontal, fog
Collection efficiency flow stream in a cooling tower rises against the gravity. Three parameters are found to influence the
collection efficiency predominantly: the shade coefficient of the mesh, effective dripping length of water
droplets along the fog net, and angle of inclination of the mesh with respect to the vertically rising fog
stream. The observed collection efficiency is more than twice as compared to those of other globally
operational fog collectors. Results offer the design bases for full-scale fog harvesting systems that can be
deployed in power plant cooling towers and a wide range of other artificial fog generators.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction 2005e2015 as the International Decade for Action “Water for Life”
[3]. Research for new alternative sources of adequate fresh water
Fresh water is essential not only for household living, irrigation, supply has gained impetus in the last two decades [4]. Fog is found
feedstock but also for several industries and is vital for sustaining to be one of the potential alternate sources of good quality fresh
urbanization. With increasing contamination of fresh water the water with a very low-cost collection system. As a result, many fog
concern for global water crisis is rising. Currently, 90% of energy collection initiatives have been attempted in more than 17 coun-
production relies on intensive and non-reusable water models, tries including the coastal deserts of Peru, Sultanate of Oman,
which are not sustainable [1]. To recognize this concern for global Namib desert, Yemen, and in the hills of Nepal [5,6]. Reviews show
water crisis, UNESCO has recently launched the International Year that the projects till date have mostly been installed in regions of
for Water Cooperation 2013 [2] and the UN declared the period low precipitation and droughts [7]. These projects face a common
limitation of seasonal fogs or limited number of fog days (param-
eters that are dependent on the climatic and topographic location
of the project site).
* Corresponding author : Department of Power Engineering, Jadavpur University,
Standard fog collector consists of a large screen held by support
Kolkata 700098, India. Tel.: þ91 33 2335 5813; fax: þ91 33 2335 7254.
E-mail addresses: ritnitg@gmail.com, ritwickghosh@ntpc.co.in (R. Ghosh), structures like poles or frames, perpendicular to the direction of the
rayt3@asme.org (T.K. Ray), ranjan@pe.jusl.ac.in (R. Ganguly). wind driven fog (Fig. 1(a)). The screen acts as an obstacle to the fog
1
Mailing address: Boiler Maintenance Department, Service Building, NTPC carrying wind stream when a fraction of the droplets present in this
Limited, Farakka, P.O. Nabarun, Murshidabad 742236, West Bengal, India. Tel.: þ91 wind stream hits the mesh fibers and gets deposited for collection.
94340 77763 (mobile), þ91 3512 226236; fax: þ91 3512 226124.
2
Mailing address: NTPC School of Business, PMI, Noida 201301, India. Tel.: þ91
A typical fog consists of water droplets ranging from 1 to 30 mm [8].
9650 99 2718 (mobile), þ91 120 2416849. As the fog-laden air passes through the mesh, the smaller droplets

http://dx.doi.org/10.1016/j.energy.2015.06.050
0360-5442/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
2 R. Ghosh et al. / Energy xxx (2015) 1e11

Cooling Tower
Plume

Fog laden Mesh Mesh Cooling

wind Tower Cell

Fog + Humid Air

Pole
Ground Cooling Tower

Fig. 1. Schematics of (a) standard atmospheric fog collector, (b) cooling tower fog collector.

may easily flow through the openings of the mesh, thus escaping recorded a fog water collection of 11.8 Lm2 per fog day, Chile a
capture. Therefore, the screen is able to collect only a fraction of the highest of 7.8 Lm2 per fog day, South Africa 7 Lm2 per fog day and
droplets present in the fog laden wind [9]. The deposited droplets Oman (only for a period of two months annually) has a maximum
coalesce and flow down the mesh fibers by gravity to the collector. collection of 30 Lm2 per fog day. Despite moderate to good col-
This fraction of droplets hitting the mesh is also seen to be indi- lections, the commonly used fiber meshes have serious sustain-
rectly dependent on the amount of openness of the mesh, as well as ability issues. Strong winds either tear the mesh fiber elements or
the droplet distribution in fog. Two types of screens are commonly cause collapsing of supporting structures [12]; the damage of mesh
used [6], the SFC (standard fog collector) of 1 m2 area and LFC (large elements may be prevented by replacing polymer fibers with
fog collector) of 40e50 m2. When a site is identified for fog har- corrosion resistant commercially available metal wire fibers. These
vesting, preliminarily the SFC is used to find the measure of fog metal wire mesh may act as a better element for fog water capture
water collection, while the LFC is installed at established sites with than the commonly used polyethylene fibers, particularly when the
enough space for installation. metal surface is tuned with appropriate wetting characteristics [5].
The screen materials commonly used are generally of locally Attempts are being made for improving the fog collection technol-
available type, like double layered Raschel mesh (made of poly- ogy through modification of the physical and chemical characteris-
ethylene fibers) or single-layered mosquito mesh [6]. Double layer tics of fog collecting surfaces of wires/fibers by taking cues from the
of Raschel mesh with 35% openness that covers approximately 60% natural fog harvesting capability of a few desert animals and plants.
surface area of the collector [10] is used most predominantly The adaptive bio-mimicry seen in the darkling Namib Desert beetle
worldwide for fog collectors. However, there is a large variability in [13], a species of desert grass (Stipagrostis sabulicola) [14] and the
water collection and airflow correlation data reported in the liter- sticky spider webs [15] have contributed significantly to design the
ature. Some studies reports [11] improvement of collection over fog capture, transport and collection devices.
Raschel mesh with other mesh types. For example, indoor green- Although systematic efforts in the area of atmospheric fog
house aluminet shade net was found to achieve the same collection capture have started over a couple of decades ago, there has not
efficiency as the aforementioned Raschel mesh at lower mesh been any formal report on extending the philosophy to industrial
density. In the present case stainless steel wire mesh with 20% and fog capture. This paper proposes to explore potential alternative fog
33% shade coefficients are used. harvesting resources that are also economically feasible and
Mesh type and pattern directly influence the fog harvesting renewable but is not dependent on the climatic or topological
behavior on the woven wire mesh. Park et al. [5], showed that the factors for installation. The present work aims at characterizing
fog collection efficiency of a mesh depended on two dimensionless water harvesting potential of fog nets installed in the cells of
variables: (i) the ratio of the radius of the fog droplets ðrfog Þ to the cooling tower (see Fig. 1(b)) and evaluating its potential as a source
radius of the mesh wire ðRÞ i.e., R* ¼ ðrfog =RÞ and (ii) a geometric for sustainable fresh water supply. Power plant cooling towers are
groups variable, termed the shade coefficient (SC), which denotes known to produce dense fog plumes that arise due to the evapo-
the fraction of mesh projected area occupied by the mesh fiber rative cooling of the CW (circulating water). The cooling tower fog
material. It is intuitive that a large value of SC would imply a greater is looked upon as a technical hazard for the outdoor electrical
physical interception of oncoming stream by the wires/fibers. Thus a installation (e.g., in the plant switchyard) and has recently attracted
larger fraction of the oncoming fog would be intercepted by the concerns from environmental perspective too [16]. The CW
mesh fiber, facilitating collection. However, this does not necessarily required by a typical 500 MW unit of a power plant [17] is
mean an overall increase in fog interception by the mesh, since a approximately 54,000e60,000 m3 h1. Evaporative cooling of CW
higher mesh SC also causes a “shielding” effect by the mesh. An in a cooling tower incurs water losses in the form of vapor, drift (un-
increased SC results in large flow impedance offered by the mesh, evaporated tiny water droplets carried along with the vapor plume)
causing a greater diversion of wind stream past the mesh structure and blow down. The amount of makeup water to compensate these
and a drop in overall incidence of fog droplets on mesh. The opti- losses is about 900 m3 h1, which includes approximately
mum value of SC would depend on the structure (e.g., wire mesh or 27 m3 h1 of drift loss, for a single 500 MW unit. This huge amount
filament) and layout (e.g., triangular or orthogonal weaving) of of fresh water needs to be replenished from the local rivers, adding
mesh, as well as the geometry and orientation of fog collector. to the scarcity of the fresh water available for consumption. In
The overall collection rate of a fog collector depends on the water recent past, many global and national organizations like Electric
content in the fog, average wind velocity and the fog collection ef- Power Research Institute, Central Electricity Authority (New Delhi,
ficiency of the nets. Among the ongoing projects [7], Peru has India) have started promoting innovative techniques to minimize

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
 R. Ghosh et al. / Energy xxx (2015) 1e11 3

water consumption in power plants [17,18]. The UN has also entirely SC ¼ 20% (designated as Type 1) and SC ¼ 33% (Type 2). A third type of
dedicated the “World Water Development Report 2014” high- mesh design (Type 3) was chosen where a mesh of SC ¼ 33% was
lighting the interlocking risks and uncertainties between water and fitted to a triangular metal frame with mesh area 0.0525 m2. Details
energy production in all parts of the globe. of the two meshes are shown in Fig. 2. The lower edges of the fog net
Inspired by this scope of fog harvesting, the present study is were fitted with thin plastic channels (headers) to guide the captured
carried out to find viable solutions to minimize the drift loss water to the measuring cylinder (not shown in the image).
component of the total water loss from an operating cooling tower
of power plant. To the best of the authors' knowledge, this type of 2.2. Experimental procedure
industrial fog harvesting has not yet been reported in the literature.
Experiments are conducted to demonstrate the feasibility of water The entire experiment has been done on-site in a fully operating
harvesting from the cooling tower fog by mounting woven metal induced-draft wet cooling tower. All fog collection experiments
meshes at the exit plane of the cooling tower cell. The fog droplet were performed on top of the cooling tower cell at a RH (relative
capture efficiency is analyzed using different mesh configurations humidity) of 100% and a temperature of 28 ± 0.5  C. Each sample
that are produced by (i) varying the spacing between adjacent fog net was positioned 2 m inside (measured from the wall of the
wires in the woven mesh i.e. the shade coefficient of the mesh, and cell) and 30 cm below the exit plane of the cell (to eliminate any
(ii) varying the geometry of the fog net frame. The fog collection effect due to the presence of CT (cooling tower) wall and flow of
characteristics of the mesh are also reported for different orienta- external wind). The positioning of the fog collector is schematically
tions of fog nets. Our fog collection data have been compared with shown in Fig. 3. The fog velocity was measured (using a handheld
the existing fog collection projects operating around the world to digital anemometer, Make: Nunes Instruments, Precision of
evaluate the acceptability of this new resource. Results of the study 0.1 ms1) to be constant at ~ 4.8 ms1.
have been rationalized to evaluate the potential of this untapped The three types of fog nets as shown in Fig. 2 were held at
fog capture source. inclination angles (a) of 15, 25, 35, 45, 55, 65 and 75 with the
vertical (see Fig 3 inset). Water collection tests for each of the cases
2. Material and methods were performed over 10 min durations. All runs at a particular
orientation (of the fog net) were repeated for at least 5 times.
2.1. The meshes
3. Results and discussions
Single-layer woven wire mesh, made of corrosion-resistant
stainless steel (grade SS 304) wires of 0.3 mm diameter and pro- 3.1. Origin of collected water
cured from the local market, was used for the experiment. The mesh
was fitted to a rhombus metal frame, creating a fog net of surface area The sampling location at the cooling tower has stream of white
0.0814 m2. Meshes of two different shade coefficients were chosen, plumes or fog. The plumes contain both water vapor and water

Fig. 2. Rhombus fog net with (a) 20% SC mesh, and (b) 33% SC mesh. (c) Triangular fog net with 33% SC mesh. Inset figures show the wire mesh (scale bars denote 10 mm).

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
4 R. Ghosh et al. / Energy xxx (2015) 1e11

Fig. 3. Schematic diagram of the experimental setup. Inset shows the angular position of fog net with the vertical.

droplets (fog). The water evaporated in the cooling process is pure inclination a. As the fog-laden flow passes through the nets (along
in nature, while the droplets leading to drift loss carry a part of the with the possible diversion due to the flow impedance as discussed
mineral impurities from those present in the circulating water. in Section 1), water droplets are observed to appear on the net (see
Chemical analysis of the collected water was carried out for the residual droplets on the nets in Fig. 2). The intercepted water
verifications of hardness and impurities. The calcium carbonate droplets drain along the inclined mesh to the header below the
hardness was found to be 210 ppm, with a neutral pH and con- mesh frame, and finally gets collected in the measuring cylinder.
ductivity in the range of 710 mS/cm. This figure is close to the salt For meshes aligned at large angles with the vertical, droplets larger
concentration in the circulating water (the circulating water in than the capillary length for water [19] (~2.7 mL) are found to detach
power plants contains dissolved chemicals because of different from the wire mesh due to gravity before they could reach the
types of dosing done to arrest bio-fouling in the CW line). The collection header at the bottom edge of the mesh. As these droplets
analysis confirms that the collected water in the present study has are shed back into the flow they do not count in the collection data,
its source in the drift droplets only, and not in the evaporated water leading to a decline in collection. This effect of gravity-induced
(vapor phase). So, it is intuitive that the collection efficiency is droplet shedding is discussed later in this section.
evaluated by normalizing the collected water by the drift compo-
nent of water mass flowing out of the cooling tower. Therefore, the
collected water can either be recycled without treatment to sup-
plement the cooling tower make-up, or used as potable water after
suitable treatment.

3.2. Fog collection

Fog collection experiments have been carried out with meshes

of different SC, size and orientations to understand the dependence
of fog capture on these parameters. Fig. 4 shows the experimental
results of fog collection experiments for the three different type of
mesh configurations, viz., Type 1 (Rhombus, SC ¼ 20%), Type 2
(Rhombus, SC ¼ 33%) and Type 3 (Triangle, SC ¼ 33%) for different
angles of inclination (a) of the fog nets with respect to the flowing
fog stream (vertically rising). The data point error bars, signifying
the standard deviation of the results, are mostly within 4% varia-
tion, implying consistency in the water collection data. The plots in
Fig. 4 clearly indicate a dependence of fog collection data on the
shade coefficient, the frame dimension (Type 3 net has ~ 65% Fig. 4. Water collection for the three type of meshes at different orientations of the fog
collection area compared to the other two types), and the angle of nets. Error bars denote the standard deviation in collection data.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
 R. Ghosh et al. / Energy xxx (2015) 1e11 5

To analyze the dropletemesh interaction from the plot (Fig. 4), trend is more predominant in mesh with higher SC (Type 2 mesh).
the experimental results are classified under two groups. The first Therefore, the difference of collection between the two mesh types
one draws comparison between meshes with same capture surface widens at larger values of a.
area but different SC i.e., Type 1 and Type 2. Both the nets show The second group of results in Fig. 4 compares fog capture by
highest collections at a ¼ 15 , collection for Type 1 being higher mesh types having same SC (33%) but different collection surface
(26.3 mL) than that of Type 2 (20.7 mL). For Type 1 fog net, areas, i.e., Type 2 and Type 3. Although Type 3 has ~35% less mesh
collection efficiency drops to 16.3 mL as a is increased to 35 , in- area compared to the Type 2 (and the same SC), it shows a higher
creases slightly at higher angles (to 19.7 ml at a ¼ 55 ). Beyond collection rate than the later. Type 2 mesh offers a larger sliding
a ¼ 65 the collection drops monotonically (to 10.5 mL at a ¼ 75 ). length (distance between the top edge and the collector) for the
Type 2 fog net shows a decreasing trend, dropping down to 5 mL at water droplets as compared to Type 3 mesh does. It is hypothesized
a ¼ 65 . Therefore, collection with Type 1 mesh exceeds that from that the probability of bigger droplets getting detached due to
Type 2 mesh for all values of a. The difference in capture may be gravity-induced shedding will be higher in the case of meshes
explained in light of the shade coefficient of the fog net (the total having larger sliding length in the path of the of fog stream (the
mesh area being same in both the cases). As already mentioned in basis of the hypothesis has been summarized and substantiated in
Section 1, an increased SC has two conflicting influences on the Appendix A). The phenomenon of gravity-induced sliding and
collection. Higher SC implies smaller gaps between mesh fibers and shedding of larger droplets is described schematically in Fig. 6.
hence a larger fraction of oncoming fog droplets would be inter- Although the phenomenon of gravity-induced premature
cepted by the mesh fiber/wire. However, a fog collector with larger shedding of droplets from the mesh could be observed while per-
SC would also offer greater flow resistance (higher drag), diverting forming the experiment in the cooling tower, imaging of the phe-
the oncoming air stream. Thus, the quantity of incident fog may nomenon in situ was heavily impaired by the white plumes rising at
wane at higher SC values, offsetting the former effect. It has been the location. The schematic in Fig. 6(a) explains the cause of
observed in previous studies that for SC values exceeding a reduced collection on larger sliding surfaces i.e., Type 2 mesh with a
threshold value of ~0.55, the “shielding” of oncoming air by the fog sliding length of 340 mm (see Fig. 2). As seen in Fig. 6(a), a large
net increases to an extent that the overall collection by the fog net fraction of collected mass was shed out from the mesh by gravity
decreases with SC [9]. Besides this aerodynamic effect, increasing before reaching the collector. While Fig. 6(b) resembles Type 3
the wire mesh SC is also found to entail another practical impedi- mesh (case with a relatively smaller sliding distance of 240 mm),
ment to fog collection in the present work. At larger values of SC the where the droplets were collected without any noticeable shed-
chances of clogging of a mesh pore by the captured water droplets ding. This corroborates to the fact that collection for a smaller mesh
increases. Clogged pores of the mesh not only increase the effective frame (Type 3) is more.
drag (and thus buttress the “shielding” effect), but also render the To correlate our study on cooling tower fog collection with the
particular mesh region ineffective for fog collection. Fig. 5 shows a previous fog harvesting results in the literature, various collection
few instances of mesh clogging in our experiment. Poor visibility of efficiency terminologies have been taken into account. The effi-
the fog stream precluded the possibility of exact quantification of ciency of water collection in existing studies have been attributed
the mesh clogging in situ. Nevertheless, when the mesh was taken to three major parameters e the amount of water content in the fog
out of the stream after each set of experiment, the Type 2 mesh stream hitting the mesh, the amount of capture due to interaction
showed nearly 2.6 times higher clogging than the Type 1. Although of the fog laden wind and the mesh, and the amount of drainage
the SC values, for both Type 1 (20%) and Type 2 (33%) meshes at from the surface of the mesh to the collector. Following the analysis
orientations of a ranging between 25 to 75 are lower than the of Rivera [9] the total collection efficiency can be expressed as a
threshold SC (~0.55), Fig. 4 shows a decreased collection for higher product of (i) aerodynamic collection efficiency ðha Þ (ii) capture
SC. This clearly indicates that the clogging of mesh affected the efficiency ðhcapt Þ and (iii) drainage efficiency ðhdr Þ, such that
collection more than the aerodynamic factor.
At small values of a (i.e., the mesh is more inclined with the
hcoll ¼ ha ,hcapt ,hdr : (1)
horizontal) removal of clogged droplets is facilitated by the gravity-
aided drainage along the mesh fibers into the collector when the Aerodynamic efficiency ðha Þ denotes the fraction of droplets in
difference between the collections by Type 1 and Type 2 meshes is the unperturbed fog that may collide with the mesh. Capture effi-
the minimum. With increase in a, the gravity-aided drainage is ciency (hcapt , also termed as deposition efficiency [5]) denotes the
reduced, making the meshes more vulnerable to clogging. This fraction of fog droplets that actually impinge on the mesh fibers

Fig. 5. Instances of mesh clogging by collected water (shown by blue arrows) on Type I (SC ¼ 20%) and Type II (SC ¼ 33%) meshes. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
6 R. Ghosh et al. / Energy xxx (2015) 1e11

Fig. 6. Gravity-induced shedding of collected droplets from inclined surface having (a) a longer sliding length (b) a shorter sliding length.

and gets deposited (the remaining fraction of droplets flows and where, SCeff denotes the effective shade coefficient of the tilted
escapes through the openings of the mesh). Out of the deposited mesh, which differs from the SC value of an orthogonally oriented
mass of fog water, a fraction travels back to the flow stream after mesh (See Appendix B). Fig. 7 shows the variation of ha with the
colliding with the mesh fiber and re-entrains due to wind drag. mesh inclination angle a for two different SC, viz., 20% (Type 1) and
Another fraction of the captured water is lost as the deposited 33% (Types 2 and 3). The value of ha increases sharply as the angle a
liquid drip out from the mesh under the force of gravity (see Fig. 6). decreases, with the highest ha of 0.36 for SC ¼ 20% and 0.54 for
These two losses are accounted for through the drainage efficiency SC ¼ 33% occurring at a ¼ 15 . At low values of a, the drag coeffi-
ðhdr Þ parameter. cient offered by the mesh is low, thus the bulk flow is less diverted
An approximate theoretical model has been proposed by Rivera by the mesh (implying that a larger fraction of air passes through
[9] for predicting aerodynamic efficiency for fog nets oriented the mesh). At the same time the inclined mesh offers an enhanced
perpendicular to the fog flow. The aerodynamic efficiency of a fog effective shade coefficient. Both these factors attribute to an
net in such a configuration depends on the drag coefficient ðCD Þ and increased interception of the incident droplets by the mesh wire,
pressure loss coefficient ðC0 Þ as [9] resulting in the increased ha at low a. The aerodynamic efficiency
decreases monotonically for higher values of a as shown in Fig. 7,
SC displaying a nearly 1.5 times larger ha value for 33% SC mesh. As
ha ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : (2)
1þ ðC0 =CD Þ

The pressure loss coefficient ðC0 Þ, which accounts for the pres-
sure drop across the mesh due to the flow restrictions offered by
the fiber assembly as the fog laden flow passes through the pores of
the mesh, is a direct function of SC. For wire mesh structure, this is
expressed as

C0 ¼ 1:3SC þ ½SC=ð1  SCÞ2 : (3)

The drag coefficient ðCD Þ, which accounts for the flow resistance
offered by the entire fog net assembly, remains almost same for all
the three types of meshes considered in the study ðCD x1:18Þ [20].
In conventional studies the fog net is set normal to the direction of
flow of fog-laden winds (Fig. 1(a)). But in our case the fog capturing
net is placed at different inclined orientations for different values of
a (with respect to vertical). Therefore, Eq. (2) is modified (see
Appendix B for details) as

SCeff
ha ¼ h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii ; (4)
1 þ ðC0 sin a=CD Þ Fig. 7. Aerodynamic efficiency of fog net at different orientations of mesh.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
 R. Ghosh et al. / Energy xxx (2015) 1e11 7

discussed in Appendix B, the highest pressure drop is offered by the

33% SC mesh at a ¼ 75 ; for the operating regime considered this
pressure drop is ~2.48 Pa, which is less than 1.4% of the driving
pressure (~177 Pa) of the cooling tower fan.
Fig. 7 provides an approximate theoretical range of ha , and does
not account for the effect of mesh clogging. Also, the individual
values for hcapt and hdr are not available in the present configura-
tion, since in-situ measurement of fog re-entrainment and prema-
ture drainage of droplets (Fig. 6), was not feasible. Therefore, the
deposition efficiency is calculated from Eq. (1) using the experi-
mentally observed value of collection efficiency and the theoreti-
cally calculated aerodynamic efficiency (Eq. (4)) as
 
hdep ¼ hcapt ,hdr ¼ hcoll =ha ; (5)

where the hcoll is calculated from the water collection data as

 mw . 〞
hcoll ¼ mdrift : (6)
t,A sin a
Here, the mass mw of the collected water over the duration of
time t is taken from the experiment (Fig. 4), the drift mass flux
m〞drift
for the cooling tower is estimated from the general design
data of the cooling towers as per Central Electricity Authority, New
Delhi [17]. A sin a denotes the projected area of the mesh on a plane
perpendicular to the fog flow direction (refer to inset of Fig. 3). The
rationale of computing the collection efficiency is summarized in
Appendix C.
Fig. 8(a) shows the variation of total collection efficiency as
function of a for the three types of mesh. In line with the water
collection data (Fig. 4), the collection efficiency also decreases with
the increase of a. The deposition efficiency should be dependent on
the relative openness of the mesh along the direction of fog-laden
wind. It also depends on the extent of coalescence of smaller
deposited droplets to form larger drops (as the droplets get drained
along the inclined mesh). Fig. 8(b) shows the variation of the
deposition efficiency at different inclinations of the mesh in our fog
capture experiment. As seen from this figure, the deposition effi-
ciency of the Type 1 (20% SC, rhombus mesh) fog net is the highest Fig. 8. (a) Total Collection Efficiency (hcoll) and (b) Deposition efficiency (hcap· hdr) of
while the Type 2 (33% SC, rhombus mesh) exhibits the least value. fog nets at different mesh orientation angles (a).
This may be attributed to the difference in the capture efficiencies
of the two mesh types, since the relative spacing between the fibers
differed, and so did the extent of “mesh clogging” (see Fig. 5). When
unnecessary loss of captured water owing to gravity-induced
the performance between Type 2 and Type 3 meshes are compared
shedding of droplets from the mesh. Due to this positional limita-
in Fig 8(b), a significant improvement is observed by switching to
tion in our study, tilted mesh installation techniques were adopted
triangular mesh. The effective shade coefficients of both the meshes
for efficient droplet capture. To compare the performance of our fog
are same at a given orientation, so the capture efficiency should
harvesting method, the water collection per unit mesh surface area
remain same for both types. This implies that the observed differ-
for the three types of fog nets in this study are compared with a few
ence in deposition efficiency is solely attributed to the different
existing fog harvesting projects [7]. Fig. 9 summarizes the cases of a
drainage efficiencies (due to gravity induced shedding effect). The
few salient fog harvesting studies around the world for comparing
plot gives an indirect measure of the percentage increase in
with our cooling tower fog collection data. The highest collection
drainage efficiency for the triangular mesh in the range of
data for (a) sites with perennial fog availability (Cape Varde [7]),
15 < a < 75 .
and (b) sites with a specific number of annual fog days (Oman [21])
are taken into account. Previously, Fessehaye et al. [7], in their re-
3.3. Comparison with traditional fog-harvesting technique view have presented the average rate of fog collection for all loca-
tions where the fog harvesting technique is in use along with the
The literature on fog harvesting have reported water collection average number of fog days per year. The natural sources for fog
in terms of two parameters (i) the total number of fog days (the harvesting have seasonal variations depending on the type of fog
number of days in a year when fog occurs at a place) and (ii) the formation. The recorded average highest rate of fog collection over
amount of overall collection recorded in terms of liter per square a period of 365 fog days (i.e. around the year), at Cape Verde, Serra
meter of fog net per hour. In the present study the fog stream is Malgagueta is ~12 Lm2 per day or 0.5 Lm2 h1 [7]. While
rising upward in a direction directly opposite to gravity (the considering the cases with seasonal average highest rate of fog
existing practices [8] have fog stream flowing parallel to the collection, it is 30 Lm2 per day or 1.25 Lm2 h1 in Oman with an
ground). Thus placing a fog net normal to the fog stream will cause average of 60 fog days [21].

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
8 R. Ghosh et al. / Energy xxx (2015) 1e11

 For all angular orientations, water collection of the Type I mesh

(SC ¼ 20%) exceeded that of the other two types (SC ¼ 33%). The
higher shade coefficient mesh exhibited a tendency of high
mesh clogging, which affected the collection efficiency.
Although the aerodynamic efficiency of SC ¼ 33% mesh was
higher than that of SC ¼ 20% mesh, the product of capture and
drainage efficiencies of the SC ¼ 33% mesh was inferior enough
to offset the effect in overall collection efficiency.
 Although the Type 3 mesh had the same shade coefficient, and
only ~65% mesh area as compared to the Type 2, the former
exhibited a higher collection rate. The triangular shaped mesh
(Type 3) provided less drainage path length for the collected
water, leading to a smaller amount of gravity-induced prema-
ture shedding, and hence, more water collection.
 The additional pressure drop offered by even the 33% SC mesh
is less than 1.5% of the driving pressure of the cooling tower
Fig. 9. Comparison between different fog harvesting studies. fan.

The present water collection has been found almost twice as

high when compared to other successful fog harvesting projects
The water collection data reported in Fig. 9 for the three cases of
across the globe. Implementation of this pilot study in an operating
cooling tower fog are readings available at the mesh orientation of
power plant is expected to recover about 40 percent water from the
a ¼ 15 . This inclination showed the highest collection efficiency drift loss e amounting to a saving of nearly 10.5 m3 of water per
(refer to Fig. 8 (a)), owing to the highest aerodynamic efficiency
hour from a 500 MW unit. The collected water can either be
(Fig. 7) and better deposition efficiency (Fig. 8(b)). The Type 3 fog
recycled to supplement the cooling tower make-up, or used as
net shows (Fig. 9) the highest collection per unit mesh surface area
potable water after suitable treatment. In either case, the overall
(2.28 Lm2 h1) while the minimum collection was reported for
fresh-water consumption of the plant will be curbed, underscoring
Type 2 fog net (1.5 Lm2 h1). All cases of cooling tower fog
the importance of the scheme in the face of the globally increasing
recorded higher collections than the existing fog harvesters.
scarcity fresh-water.
Moreover, the cooling tower fog has an inherent advantage of
round the year availability. The rate of capture is also expected to be
constant throughout the year as the drift loss component of the Appendix A
thermal power plant cooling tower does not have any seasonal
variations [22]. Therefore our present study is established to be a Dependence of collection efficiency on the slant height “H”
potential source for fog harvesting from power plant cooling tower
and a wide range of other artificial fog sources. To understand the influence of gravity-induced shedding of the
collected fog droplets; experimentation was conducted using a
rectangular mesh of area 0.0624 m2 with aspect ratio of 2.5. The
4. Conclusion mesh has a shade coefficient of 33%. Two cases were considered
when the mesh was held at a ¼ 25 with vertical, and oriented such
A pilot study, highlighting the feasibility of cooling tower fog that: Case (a) shorter length of the mesh frame is horizontal
harvesting as a prospective avenue of fresh water conservation, (Fig. A.1(a)) and Case (b) the longer side held horizontal
has been conducted in a fully operating power plant. Water har- (Fig. A.1(b)). Thus, despite having same cross sectional area,
vesting potential of fog nets, mounted at the exit of the cooling orientation Case (a) offers 2.5 times longer path to the collected fog
tower cells, was characterized for different fog net designs and the water to shed under gravity.
influence of operating parameters on the collection efficiency Fog collection for 10 min yields 5.0 ml of water for Case (b) of
were identified. Three types of metal wire fog nets were studied: Fig. A.1, which is nearly double of that observed for Case (a) (2.6 ml).
Type 1 (Rhombus shaped, with shade coefficient of 20%), Type 2 Since the effective shade coefficients and the drag for the two cases
(Rhombus, with SC ¼ 33%) and Type 3 (Triangular, SC ¼ 33%). Their are identical, the aerodynamic efficiency and the capture efficiency
water collection data and capture efficiencies were compared at do not differ for the two cases. The only influencing factor that
different angles of inclination of the fog nets with respect to the differs in the two cases in Fig. A.1 is the dripping length for the
fog laden flow stream. collected water. The result provides an indirect evidence of
increased water loss due to gravity-induced shedding for fog nets
 Chemical properties (hardness, pH and electrical conductivity) offering longer dripping path of water droplets. This clearly shows
of the collected fog water matched closely with that of the that fog nets with dimensions having a greater width but a lesser
cooling tower circulating water quality, implying that the height are more suitable for droplet capture.
collected water has origin primarily in the cooling tower drift The same phenomenon has been attributed to the difference of
component, and not in the water vapor escaping with the collection between the Type 2 and Type 3 meshes as explained in
cooling tower plume. section 3.2 of the main paper. The maximum sliding length of the
 Water collection rate decreased with increase in the angle of mesh in the direction of fog flow stream for the Type 2 mesh is
inclination of the fog net with respect to the vertically rising fog ~340 mm while for Type 3 is ~240 mm (see Fig. 2 of the main pa-
flow for the three types of fog nets. At large angles of inclination per). Therefore, despite the fact that Type 3 mesh has only 65%
with the vertical, gravity-induced shading of water droplets is mesh area when compared to Type 2 mesh, the water collection is
increased, leading to a decline in collection. more when Type 3 mesh is used.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
 R. Ghosh et al. / Energy xxx (2015) 1e11 9

Fig. A.1. Variation of water collection with mesh aspect ratio. The water collection of the same rectangular mesh, held at a ¼ 25 with vertical, differ as the aspect ratio differs: (a)
height ¼ 2.5  width, and (b) width ¼ 2.5  height.

Appendix B Therefore for the case of an inclined mesh, following Fig. B.1 (a),
the aerodynamic efficiency may be defined as
Evaluation of aerodynamic collection efficiency of a tilted mesh
A0
ha ¼ SC ; (B7)
Aerodynamic efficiency (ha ) denotes the fraction of droplets in ASina eff
the unperturbed fog that may collide with the mesh. The aero-
dynamic efficiency of a fog net oriented perpendicular to the di- where the term SCeff denotes the “effective” shade coefficient of the
rection of the fog-laden wind may be expressed as a function of the tilted mesh.
drag coefficient ðCD Þ and pressure loss coefficient ðC0 Þ as described Now, using equations B5 and B6 in the expression of Aero-
in Eq. (2). However, the expression would change if the angle of dynamic efficiency (B7), one gets
incident differs from 90 . Following the approach of Rivera [9] we
SCeff
apply the principle of superposition and strike a balance between ha ¼ h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii : (B8)
pressure drop across the fog net due to the large-scale drag on the 1 þ ðC0 Sina=CD Þ
net and the smaller length-scale pressure drop across the mesh
pores. Fig. B.1 shows the superposition diagram of the flow. The effective shade coefficient SCeff , representing the fractional
The velocity of unperturbed wind is V0 (which is 4.8 m/s in the solid area offered by the mesh fibers perpendicular to the flow inci-
present case). Wire meshes have shade coefficient, thus it has been dent to the mesh, differs from that evaluated by Rivera [9], since the
assumed that the wind stream passing through the openness has a mesh inclination to the direction of flow alters the effective opening
velocity V1 and wind stream offset due to the shaded region is (gap) between the wires in a mesh for the fog to pass through.
having a velocity V2. Using the principle of superposition, For an orthogonally woven wire mesh comprising of wires of
diameter 2R interspaced by a gap 2D, the shade coefficient is
Vo ¼ V1 þ V2 : (B1) evaluated as (Fig. B.2 (a)) [5]:
The pressure drop across the mesh structure can be expressed in

 
*
SC ¼ 2 1  1 2D* D ; (B9)
terms of the component of velocity V1 normal to the mesh and the
pressure drop coefficient, C0 . The pressure drop across the porous
where D* ¼ ð1 þ R=DÞ. Fig. B.2 (b) shows a non-orthogonally
mesh (Fig. B.1 (b)) may be expressed as,
(a < 90 ) aligned fog capture net. Since the “shade” area offered
1 by the cylindrical mesh fibers remain unchanged, while the effec-
Dp ¼ C rV 2 Sin2 a; (B2) tive opening (gap) perpendicular to the flow direction decreases as
2 0 1
a decreases, the effective value of shade coefficient increases.
On the other hand, the pressure difference across the inclined From the geometric disposition of mesh fibers and the void
fog net due to the large scale drag (offered by the solid obstruction spaces described in Fig. B.2 (a) and (b), it ensues that for a tilted
in Fig. B.1(c)) can be expressed in terms of the drag co-efficient as mesh the effective shade coefficient is
derived in Holmes et al. [23]:
h  . i.
1 SCeff ¼ 2 1  1 2D*eff D*eff : (B10)
Dp ¼ CD rV22 Sina (B3)
2 The variation of SCeff with respect to positioning of the mesh in
The two pressure drop terms are equated as per the basic hy- the range of 15 < a < 75 is plotted in Fig. B.2(c) for both Type I
pothesis of the superposition theory [9]. (SC ¼ 20% at a ¼ 90 ) and Type II, III (SC ¼ 33% at a ¼ 90 ) meshes. It
is evident from the figure that the effective shade coefficient in-
C0 V12 Sina ¼ CD V22 (B4) creases from its base value (i.e. SC at a ¼ 90 ) as the angle of
inclination decreases. At a ¼ 15 , the SCeff values for Type I and
The equation (B4) may now be rearranged as
Type II, III meshes are 46% and 75%, respectively. Reckoning
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi V0 ¼ 4.8 ms1, and using Eq. (B3) and (B5), the pressure drop across
VO Co Sina the meshes may be evaluated. For the range of a considered in the
¼ 1± (B5)
V1 CD study, the pressure drop varies between 0.28 Pa and 2.48 Pa for the
Type II and III meshes, and between 0.16 and 1.57 Pa for Type I
Using continuity equation, for constant flow density
mesh. Compared to the pressure developed by the forced draft type
A0 V0 ¼ AV1 sin a (B6) cooling towers, which is ~177 Pa for the present case, the additional
pressure burden offered by the mesh is less than 1.4%.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
10 R. Ghosh et al. / Energy xxx (2015) 1e11

Fig. B.1. Superposition scheme representing (a) the real flow, where the actual flow velocity (V0) is assumed to be the sum of (b) velocity V1 for the case of the mesh where all the
flow is forced to pass through, and (c) velocity V2 for the case where the screen behaves as a non-permeable plate. Flow velocities V1 and V2 cause identical pressure drops across the
inclined mesh/ plate.

Appendix C

Rationale of computing the collection efficiency

The LWC (liquid water content) of the fog laden wind per
square meter area of the cooling tower cell is very low (~360 ml).
Although attempts were made to perform onsite measurement of
LWC to find the collection efficiency (by isolating a given volume
of plume flow and separating the water content in it), the
methods were not successful due to the extremely low value of
LWC. The collection efficiency was thus calculated using the
standard cooling tower operational data for a 500 MW unit [17].
The quantity of fog (drift droplets) available for collection is given
in Table C.1.
With the present harvesting system, the highest collection
efficiency of about 40% was achieved at the cooling tower exit.
Even with this moderate collection rate, the overall water
collection becomes almost 5 times (see section 3.3, Fig. 9) of the
conventional perennial natural fog harvesting projects. Thus the
present study underscores the potential of the cooling tower fog
harvesting as a significant alternate fresh water conservation
scheme.

Table C.1
Cooling tower data and water collection efficiency.

Sl. Parameters Quantity

1. 500 MW CW water intake 54,000 m3 h1

2. Drift loss (0.05%) 27 m3 h1
3. Amount of loss per cell (16 cells) 1.69 m3 h1 ¼ 1690 L h1
4. Cross-sectional area of diffuser of 78.5 m2
a particular cell (radius ¼ 5 m)
5. LWC per unit area 21.5 Lm2 h1
Fig. B.2. Schematic diagrams showing the variation of SC with inclination (a) at 6. Highest collection (see Fig. 8) 8.56 Lm2 h1
a ¼ 90 and (b) at 15<a < 75 (c) Variation of the effective SC at the different mesh Water collection efficiency (as defined in Eq. (6)) 40%
orientations (following Eq. (B10)).
Bold signifies the main achievement and significance of this research work.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050
 R. Ghosh et al. / Energy xxx (2015) 1e11 11

References [13] Parker AR, Lawrence CR. Water capture by a desert beetle. Nature 2001;414:
33e4.
[14] Roth-Nebelsick A, Ebner M, Miranda T, Gottschalk V, Voigt D, Gorb S, et al.
[1] The United Nations world water development report, vol. 1. UN Water; 2014.
Leaf surface structures enable the endemic Namib desert grass Stipagrostis
Water and Energy. URL: http://www.unwater.org/publications/publications-
sabulicola to irrigate itself with fog water. J R Soc Interface 2012;9:
detail/en/c/218614/.
1965e74.
[2] Bokova I. UNESCO e CL 3990. April 10, 2012.
[15] Zheng Y, Bai H, Huang Z, Tian X, Nie F-Q, Zhao Y, et al. Directional water
[3] General assembly, UN documents e A/RES/58/217. December 23, 2003.
collection on wetted spider silk. Nature 2010;463:640e3.
[4] Schemenauer RS, Cereceda P. Fog-water collection in arid coastal locations.
[16] Veldhuizen H, Ledbetter J. Cooling tower fog: control and abatement. J Air
Ambio 1991;20:303e8.
Pollut Control Assoc 1971;21:21e4.
[5] Park KC, Chhatre SS, Srinivasan S, Cohen RE, McKinley GH. Optimal design of
[17] Report on minimisation of water requirement in coal based thermal power
permeable fiber network structures for fog harvesting. Langmuir 2013;29:
stations. New Delhi: Central Electricity Authority; January, 2012. URL: http://
13269e77.
www.cea.nic.in/reports/articles/thermal/min_of%20water_coal_power.pdf.
[6] Schemenauer RS, Cereceda P, Osses P. Fog water collection manual. FogQuest:
[18] Technology innovation water use and availability program. Electric Power
Sustain Water Solutions 2011:5e9.
Research Institute; March, 2013. URL: http://www.epri.com/Our-Work/
[7] Fessehaye M, Abdul-Wahab SA, Savage MJ, Kohler T, Gherezghiher T, Hurni H. Fog-
Documents/Nuclear/TI_Water_Use_Availability_Program_1025771.pdf.
water collection for community use. Renew Sustain Energy Rev 2014;29:52e62.
[19] de Gennes P-G, Brochard-Wyart F, Quere D. Capillarity and wetting Phe-
[8] Schemenauer RS, Joe PI. The collection efficiency of a massive fog collector.
nomena, drops, bubbles, pearls, waves. New York: Springer Verlag; 2004.
Atmos Res 1989;24:53e69.
p. 33.
[9] Rivera JD. Aerodynamic collection efficiency of fog water collectors. Atmos Res
[20] White FM. Fluid mechanics. 2nd ed. New York: McGraw-Hill; 1986. p. 420.
2011;102:335e42.
[21] Abdul-Wahab SA, Al-Damkhi AM, Al-Hinai H, Al-Najar KA, Al-Kalbani MS.
[10] Schemenauer RS, Cereceda P. A proposed standard fog collector for use in high
Total fog and rainwater collection in the Dhofar region of the Sultanate of
elevation regions. J Appl Meteorology 1994;33:1313e22.
Oman during the monsoon season. Water Int 2010;35:100e9.
[11] Shanyengana ES, Sanderson RD, Seely MK, Schemenauer RS. Testing green-
[22] Jagadeesh T, Reddy KS. Performance analysis of the natural draft cooling
house shade nets for the collection of fog for use as a water supply. J Water
tower in different seasons. IOSR-Journal Mech Civ Eng 2013;7:19e23.
Supply: Res Technology-AQUA 2003;52(3):237e41.
[23] Holmes R, Rivera JD, Jara E. Large fog collectors: new strategies for collection
[12] Rivera JD, Lopez-Garcia D. Mechanical characteristics of Raschel mesh and their
efficiency and structural response to wind pressure. Atmos Res 2014;151:
application to the design of large fog collectors. Atmos Res 2015;151:250e8.
236e49.

Please cite this article in press as: Ghosh R, et al., Cooling tower fog harvesting in power plants e A pilot study, Energy (2015), http://dx.doi.org/
10.1016/j.energy.2015.06.050