See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273024525

Performance of Coanda-Effect Screens in a Cold Climate

Article  in  Journal of Cold Regions Engineering · December 2014

DOI: 10.1061/(ASCE)CR.1943-5495.0000073

CITATIONS READS
2 1,911

3 authors, including:

Hanne Nøvik Leif Lia

Norwegian University of Science and Technology Norwegian University of Science and Technology
3 PUBLICATIONS   12 CITATIONS    43 PUBLICATIONS   162 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Upgrading Hydropower Plants to Pumped Storage Plants View project

Embankment dam safety under extreme loading conditions View project

All content following this page was uploaded by Hanne Nøvik on 01 August 2016.

The user has requested enhancement of the downloaded file.

 Performace of Coanda-Effect Screens in a Cold Climate
Hanne Nøvika , Leif Liaa , Henrik Opakerb
a
Department of Hydraulic and Environmental Engineering
Norwegian University of Science and Technology, S.P Andersens veg 5, 7491 Trondheim, Norway
[email protected], [email protected]
b
Current adress: Norconsult AS
Postboks 626, 1303 Sandvika, Norway
[email protected]

April 24, 2014

Abstract
This study assesses the performance of a Coanda-effect intake screen for a small hydropower plant in winter operation. The
unique self-cleaning screen is making use of the Coanda-effect for withdrawal of water and consists of panels with 2 mm wide
horizontal wedge wires with 1 mm spacing; underneath the screen is a canal conveying water to the power plant intake. A broad
understanding of the winter performance of the Coanda-effect screen in a cold climate was obtained from experiments in a
frost laboratory and monitoring program consisting of hydrological measurements, an online web camera and field observations
of an intake structure in Norway. Different freezing processes were observed, and critical hydrological conditions for the ice
blockage of the Coanda-Effect screen were evaluated. The Coanda-effect screen performed well under all normal conditions
without snow or ice, and was entirely self-cleaning. Two main types of ice formation on the screen were observed; Type I where
frazil ice particles stick to the screen wedge wires, but with remaining openings underneath the ice, and Type II where solid ice
form between the wedge wires at very low temperatures. There were some periods during the winter of 2011/2012 in which
there was a reduced intake capacity or complete blockage from ice. All of the complete ice blockage events were related to
supercooled water just upstream from the screen, thus indicating frazil ice, in addition to a temperature on the intake screen and
in the air below 0o C. The screen reopened without any operational measures after all events of ice blockages, and the intake
performed well even during periods with solid ice cover over the screen, whereby water entered the intake through the screen
wedge wires underneath the ice cover. In spite of some periods with reduced intake capacity, the intake screen is considered to
have performed reasonably well in a cold climate.
Keywords: Intake Screen, Frazil Ice, Hydropower

Introduction stalled on a supercritical slope and tilted by a few degrees in

the downstream direction to obtain a shearing offset to the
All intake structures for hydropower plants in rivers have next downstream wedge wire. Because of the Coanda ef-
a trash rack that prevents potential damage to the turbines fect, which is the tendency for a fluid jet to be attracted by
from objects transported along the river. The trash racks are nearby surfaces, the flow remains attached to the top sur-
easily clogged and cleaning of the intake structures account face of the wedge wires and the tilting of the wedge wires
for an important amount of the operation and maintenance enhances water to be efficiently directed through the screen.
costs of a number of small hydropower plants. The self- The screen capacity is specified by manufacturers to be at
cleaning Coanda-effect intake screen provides for reliable least 0.09 − 0.14 m3 /s/m, dependent on the screen height
operation of a power plant with a minimum of maintenance and the wedge wire spacing. More-or-less vertical boulder
(Strong and Ott 1988), and the screen has been success- bars in the flow direction are often included to protect the
fully applied at several small hydropower plants worldwide fine screen wedge wires from damage from larger objects
(Wahl 2003), with a typical arrangement as shown in Fig- transported by the river, with a typical spacing of 100 mm.
ure 1. The horizontal steel wedge wires on the screen are The hydraulic performance of the Coanda-effect screens
typically 2 mm wide with 0.5−1.0 mm spacing, and are in-

Preprint submitted to Journal of Cold Regions Engineering 1

 Background
Suspended ice particles in turbulent water are called frazil
ice, and tend to cling to obstructions in rivers (Ashton
1986). Useful overviews of this phenomenon are given by
Andersson (“Frazil ice at water intakes”) and Daly (1994).
Frazil ice is often defined as being suspended disc-like ice
particles in the water column with a typical diameter of
0.1 − 2.0 mm. Frazil ice occurs in two distinctively dif-
ferent states: active and passive (Ashton 1986). If the wa-
ter surrounding the suspended ice particles is supercooled,
a few hundredths of a degree being sufficient, the frazil
ice is defined as being active (ibid.). Active frazil ice is
highly adhesive, whereas the passive state is slushy and less
Figure 1: Features and typical arrangement of a Coanda- sticky. Active ice particles will initially accumulate on sur-
effect screen structure (Wahl and Einhellig 2000). With faces with temperatures below freezing; as a result, frazil
permission fraom ASCE. ice layers will then grow in all directions, which can create
bridges between adjacent objects that may lead to block-
ages (Daly 1991). Frazil discs agglomerate into frazil flocs
and frazil slush floating partly above the water surface at
lower stream velocities, or in a slush ice form in streams
was thoroughly investigated by Wahl (2001), whereas the with very high turbulence intensity (Osterkamp and Gosink
flow over the screen was further modeled by Wahl and Ein- 1983). Latent heat is released as frazil crystals grow, which
hellig (2000). The self cleaning effect has been tested by reduces water column supercooling. An ice cover insulates
laboratory experiments (Huber 2005) and the advantageous a water body from extensive heat loss to the atmosphere,
properties for the passage of fish documented by Blevins so the temperature of supercooled water will gradually in-
et al. (2005). Investigations of a pilot project conducted by crease to the freezing point under an ice cover. According
ENTEC (1998) in Switzerland at an elevation 1850 meters to Daly (1994), Morse and Richard (2009), and Richard and
above sea level (masl) revealed that the screen performed Morse (2008) there is a correlation between ice cover and
well without any loss of water, despite a snow and ice cover the amount of frazil ice ending up on an intake screen.
on the screen throughout most of the winter. The winter per- Snow may agglomerate into a slush ice that floats on the sur-
formance of the Coanda-effect screen exposed to frazil ice, face if the water temperature is 0 ◦ C or colder (Osterkamp
which is suspended ice particles in turbulent water (Ashton and Gosink 1983). An increase in water discharge may feed
1986), is less documented. snow from the river border into the water, thus having the
The first intake structure with a Coanda-effect screen in same effect on the river as precipitation.
Norway was commissioned in May 2011 at the 3.5 MW Trash racks at water intake structures are exposed to frazil
Dyrkorn Small Hydropower Plant (SHPP), which is owned ice accumulation and may become completely blocked
by Tafjord Kraftproduksjon AS. The Coanda-effect screen (Daly 1991), with frazil ice blockages often taking place
at Dyrkorn Hydro SHPP performed very well under all nor- on cold winter nights and at inaccessible locations under
mal conditions without snow and ice and was completely water. Some case studies of frazil ice blockages of water
self cleaning, which is in accordance with previous stud- intakes have been reported, including one by Daly and Et-
ies (Huber 2005; Strong and Ott 1988; Wahl 2001). The tema (2006) who studied frazil ice formation on intakes in
Dyrkorn River, above the intake, is steep, with an average the Great Lakes, emphasizing the role of surface ice covers
slope of 1:10, has several waterfalls and riffles that make in preventing frazil ice formation. The recorded air tem-
the river prone to supercooling and frazil ice formation, perature at the initiation of the frazil ice formation in their
and the intake pond is small and shallow. Concerns about study ranged between −10 ◦ C to −21 ◦ C. Multiple frazil
frazil ice and frequent freeze and melt processes called for ice blockages at a water intake in a tidal reach of the St.
enhanced knowledge about the winter performance of the Lawrence River has been studied by Morse and Richard
Coanda-effect screen. The objective of this project is to (2009) and Richard and Morse (2008), who recorded su-
assess the different ice formation processes on a Coanda- percooled water and no ice cover for most of the frazil
effect screen and to find critical hydrologic parameters re- ice events. There was rarely much precipitation before the
lated to ice blockage during winter conditions in a cold cli- frazil ice events, while the critical air temperature for ice
mate. blockage was dependent on the time of the day, surface ice

Preprint submitted to Journal of Cold Regions Engineering 2

conditions and the tide. The significant risk for ice block-
ages was for air temperatures in the range between 2.3 ◦ C
to −29.3 ◦ C, dependent on the conditions.
Frazil ice can be created in the laboratory by seeding su-
percooled water with ice particles (Carstens 1966; Ettema
et al. 1984), and the rate and quantity of frazil ice growth
increases with an increasing turbulence level and with a
decreasing temperature of the supercooled water accord-
ing to Ettema et al. (1984). Furthermore, higher turbulence
yields more rapid frazil ice growth, smaller ice particles and
a shorter time with supercooled water (Clark and Doering
2006).

Research Methodology
Figure 2: Model of the Coanda-effect screen in the frost
The freezing processes on the Coanda-effect screen are laboratory. Water enters through the screen in a 15-20mm
studied in a frost laboratory, by field visits, and a contin- long wet zone just downstream from the acceleration plate.
uously filming web camera at Dyrkorn hydropower plant.
The critical hydrological conditions for ice blockage and
reduced capacity of the Coanda-effect screen are evaluated produced slush mixture was comparable enough to passive
based on measurements from the winter of 2011/2012. We frazil slush.
have defined a reduced intake capacity for the Coanda- Before adding the snow slush mixture to the tank, the in-
effect intake screen to be a combination of a reduction in take model with a continuously running pump was set in
power production, with > 10%, and an increased mea- the frost laboratory maintained to a preselected Ta in the
sured water discharge in the river, with > 10%, during the range between −3 ◦ C to −14 ◦ C. The temperature on the
same period. In order to find causal links between hydro- screen, Ts , and the water temperature, Tw , were measured
logic conditions and the performance of the Coanda-effect with an ASYC II MX 54 thermometer, with an accuracy of
screen, measurements of temperatures of air, water tem- ±0.5 ◦ C, hence supercooled water was not detected. The
perature and screen temperatures were logged simultane- wet zone where water enters through the screen was ap-
ously with observations of the ice coverage of the intake proximately 15 mm high. While the pump was running, the
pond, recorded precipitation, water discharge in the river Tw was close to 0 ◦ C, whereas the Ts , was closer to the
and power production. Ta in the frost laboratory. For each test, 0.0045 m3 of the
slush mixture was added to the upper tank of the test rig
and stirred with a hand blender, and the suspended ice then
Laboratory Test approached the screen along with the water. The freezing
The freezing processes on the Coanda-effect screen was as- processes on the Coanda-effect screen were observed and
sessed in a model intake located in a frost laboratory at the formation of the ice accumulation on the screen was
NTNU. The physical model consisted of a 300-mm-wide described. The entire screen was inspected by visual obser-
and 400-mm-high full scale Coanda-effect screen with a vations and eventual ice particles between the screen wedge
1 mm wedge wire spacing supplied by Dulas Ltd, as well wires were assessed.
as a circulating water supply system, see Figure 2. A pump
with capacity of 0.002 m3 /s circulated water from the lower Field Monitoring Program
tank of 0.04 m3 to the upper 0.008 m3 tank, and the water
flow through the wedge wires in a 15 − 20 mm long wet The Dyrkorn SHPP is located next to the Storfjord along the
zone just downstream from the acceleration plate. western coast of Norway, and utilizes a net head of 220 m,
Instead of making frazil ice directly from supercooled wa- with the intake structure situated at elevation 225 masl.
ter, a slush mixture of snow and water with ice particle sizes The installed Coanda-effect screen from Dulas Ltd is 1.2 m
in the range from 1 − 10 mm was prepared. The ice par- high, with wedge wires openings of 1 mm and a guaranteed
ticles in the snow-water slush mixture were not as adhe- capacity of 0.14 m3 /s/m. 15 modules with a width of 1 m
sive as active frazil ice discs. But as the ice particles in have been installed, providing a total theoretical capacity of
the slush mixture stuck to the Coanda-effect screen during 2.1 m3 /s, where as the design discharge at Dyrkorn SHPP
the experiments, it was assumed that the behavior of the is 1.5 m3 /s. The locations of all the temperature measure-

Preprint submitted to Journal of Cold Regions Engineering 3

 night.
Precipitation was recorded at the metering station at Sjøholt
and the report was retrieved from the Norwegian Meteoro-
logical Institute (eKlima 2012) with one measurement ev-
ery day. The metering station at Sjøholt is located 31 km
from the Dyrkorn power plant at an elevation 11 masl, but
is considered to produce an estimation of the precipitation
at Dyrkorn. The water discharge in the Dyrkorn River was
logged by water level measurement at a sharp-crested weir
type gaging station situated 300 m downstream from the in-
take. A Druck pressure sensor (GE 2012) was installed in a
vertical pipe at a frost free depth, whereas the weir profile
was prone to ice. Weekly inspection and downloading of
data assured for regularly supervision of the gaging station.
The hourly electricity production in MWh was logged, and
Figure 3: Location of temperature measurement devices the water discharge through the turbine calculated, based on
and video camera at the Coanda-effect screen intake struc- a production discharge curve.
ture at Dyrkorn SHPP.

Results
ments and the web camera are shown in Figure 3. The tem-
peratures in the air, Ta , and water, Tw , were logged using a Two main types of ice formation on the Coanda-effect in-
Vemco Minilog-II-T (Vemco 2011). The sampling interval take screen were observed, and are described below. Criti-
was specified at 30 minutes, and the outdoor temperature, cal conditions for ice-blockage and reopening of the screen
Ta , was measured at the gate housing wall. The logger for was then assessed.
the air temperature inside the intake chamber, Tac , was lo-
cated underneath the Coanda-effect screen, attached to the Freezing Process of the Coanda-effect Screen
concrete wall.
The water temperature, Tw , in the river was measured and Field and laboratory observations indicate that there are
logged upstream from the screen with a high accuracy in or- two main types of ice clogging of the Coanda-effect intake
der to identify periods of supercooling. A temperature sen- screen. Type I clogging occurs when soft and wet ice slush
sor, SeaBird 39, with an accuracy of ±0.002 ◦ C (SeaBird- clog by adhesion to the intake screen, Figure 4a). The ice
Electronics 2011), was located on the upstream side of particles sticks to the top surface of the screen and do not
the weir, 200 mm below the highest regulated water level enter between the screen wedge wires; hence the screen is
(HRWL), and the sampling interval was set to 15 min. more or less open underneath the soft ice cover. Type II
The temperature on the screen wedge wires, Ts , was mea- clogging is when the screen is so cold that solid ice forms
sured with a thermistor sensor attached to the downstream between the wedged wires, Figure 4b). Type II clogging
side of the screen wires and logged with a Comark EV was not observed in field during the winter of 2011/2012.
N2012 data logger located inside the gate housing. The The ice clogging was of Type I for all of the experiments
accuracy of the sensor was ±0.5 ◦ C (Comark 2011), and in the frost laboratory with air temperature, Ta , in the range
foam insulation was placed between the thermistor and the of −2.9 ◦ C to −6.8 ◦ C and screen temperature, Ts , in the
air. The measurements of Ts were taken on the upper part range of −0.4 ◦ C to −2.4 ◦ C. The ice particles started to
of the screen, and were most likely influenced by the water, build up on the wedge wires, just downstream from the
while the temperature on the lower part of the screen is ex- 15 − 20 mm long wet zone. Some ice was pushed further
pected to be lower than in the wet zone. Unfortunately, the down on the screen as the accumulated ice grew higher, Fig-
sensor cable was torn off by ice the 22nd of December 2011 ure 4a). At the end of the experiment, the ice slush accumu-
during an ice blockage event. lated up to a thickness of between 40−50 mm, with less ice
A web camera, type AXIS 214 PTZ, was mounted on the further down on the screen. The capacity of the screen was
gate house for a continuous observation of the Coanda not noticeably reduced. By inspection of the rear side of the
screen and the intake pond. There was no continuously screen, only a few ice particles were observed between the
recording, but a manual screen shot of the intake pond was screen wedge wires. Most of the ice particles were laying
taken daily for an estimation of the intake pond ice cover- on the top surface of the screen. After the test, the accumu-
age. Because of a lack of light, no pictures were taken at lated ice was removed and water was poured over the screen

Preprint submitted to Journal of Cold Regions Engineering 4

Figure 4: Two main types of ice clogging of the Coanda-effect intake screen, here from the laboratory tests. a) Type I: Soft
ice is accumulated on the top surface of the screen wedged wires, just downstream from the wet zone where water enters
through the screen. b) Type II: Formation of solid ice between the screen wedged wires before any ice particles were added
to the approaching water.

Figure 5: Active frazil ice clogging of Type I at the Coanda-effect screen in field. The spacing between the screen wedge
wires are 1 mm. a) The frazil ice particles stick to the top surface of the screen wires. b) There are no frazil ice particles
between the screen wires underneath the removed ice.

Preprint submitted to Journal of Cold Regions Engineering 5

 up to 10 − 12 mm, some with sharp edges and others with
smoother edges. Visual observations of the screen wegde
wires underneath the accumulated frazil particles were done
for sample locations. No ice particles were found in be-
tween the screen wires, Figure 5b). Hence a Type I clog-
ging was observed. The ice particles clung to the surface
of the screen, with thickness of the accumulated frazil ice
slush up to 0.2 m, thus leading to a reduced flow capacity
over the screen. It has been observed from the web cam-
era that frazil ice starts to accumulate at the Coanda-effect
intake screen in field, from the lower part of the rack and
upwards , indicating Type II of clogging.
Other types of freezup of the Coanda-effect intake screen in
field were observed as solid and compact ice on the boulder
bars, Figure 6. Ice covers start to accumulate above the hor-
Figure 6: Down-looking view on the Coanda-effect intake izontal wedged wires, initiated at the boulder bars, without
screen with solid ice growth on the boulder bars. The flow noticeably affecting the flow capacity of the screen. It has
capacity of the screen is not noticeably affected. also been observed that the entire screen becomes simul-
taneously clogged. A simultaneous clogging of the entire
screen was often the case when it was snowing directly on
in order to check the capacity underneath the ice compared the screen in combination with supercooled water. Snow
to a warm, clean screen. The capacity was not reduced. may be deposited on the lower part of the screen, while the
Type II ice clogging of the screen was observed during upper part was still open.
the frost laboratory experiments with Ta in the range of
−13.8 ◦ C to −14.0 ◦ C and Ts in the range of −5.9 ◦ C to Winter Performance 2011/2012
−7.1 ◦ C. For Type II clogging, water froze to solid ice be-
tween the wedge wires, and the entire screen was covered Figure 7 gives an overview of the hydrological conditions
with 1 − 3 mm of ice, with the exception of the wet zone, and the power production at the Dyrkorn SHPP during win-
before any of the prepared slush mixture was added, Fig- ter of 2011/2012. The connection between the hydrologi-
ure 4b). The solid ice could be formed by condensed water cal parameters and the flow capacity of the Coanda-effect
or from water spurt, since the solid ice formed downstream screen can be assessed from the data set, and will be dis-
from the wet zone. During these coldest tests, Ta in the cussed below.
range of −13.8 ◦ C to −14.0 ◦ C, active frazil ice particles The air temperature in the intake chamber, Tac , was closely
were observed in the upper test tank before any of the pre- related to Tw , Figure 7a). The air underneath the Coanda-
pared slush mixture was added. The observation of active screen is separated from the air in the inner intake cham-
frazil ice particles in the upper tank, suggests that Type II ber, as the transition between the collecting canal and the
clogging also could be caused by small active frazil ice par- inner chamber is fully submerged. The Tac differed signifi-
ticles entering in between the wedge wires. When the slush cantly from Tw only during a period around the turn of the
mixture was added, the ice slush slid over the already ex- month from January to February, when there was a draw-
isting thin, solid ice layer; hence the ice slush was build- down of the water level in the intake pond because of low
ing up from the lower end of the screen, and then forming water discharge and longer periods with maintenance work
more compact ice. The wet zone remained open, helping and shutdown of the power plant. The collected data show
to provide for maintaining the capacity of the screen. The that the temperature on the screen wedge wires, Ts , fol-
zone that was free of ice was reduced over time. Frazil lows Tw more closely than the outdoor Ta . The measure-
ice clogged the pump after a while in low temperatures, ments also indicate that for the periods with Ta << Tw , the
thereby reducing the possibilities for long test runs. As ex- screen wedge wires became colder than 0 ◦ C. On the walls
pected, the laboratory experiments of ice formation on the inside the intake structure, a small amount of border ice has
screen revealed that the type of freezing process is depen- been observed during the coldest periods, though it has not
dent on the air temperature. We had opportunities to study caused any operational problems.
the Coanda-effect screen also in field when active frazil ice For long periods during the winter, especially from early
started to cling to the wedge wires, Figure 5a). In field, the January to the end of February, the discharges were equal
ice particles observed on the screen had disc-like form with to or lower than the authoritys requirements of minimum
a thickness of approximately 2 − 4 mm and a diameter of released environmental flow of 0.050 m3 /s, which means

Preprint submitted to Journal of Cold Regions Engineering 6

Figure 7: Overview of the hydrologic parameters at Dyrkorn SHPP during the winter of 2011/2012. Subplots: a) Outdoor air
temperature, Ta , air temperature inside the intake chamber, Tac , water temperatures in the chamber, Twc and temperature on
the Coanda-effect screen, Ts b) Water temperature upstream from the weir, Tw c) Ice coverage d) Daily precipitation e)Water
discharge in the Dyrkorn river f) Hydro power production at Dyrkorn SHPP

Preprint submitted to Journal of Cold Regions Engineering 7

that the power plant have to be shut down regardless the Morse 2008; Morse and Richard 2009). Frazil ice often ap-
Coanda-effect screen flow capacity. proaches the screen is floes or as frazil slush. When the in-
There were several stops in the production during this first take pond was partially covered by ice at the initiation of the
winter of operation, Figure 7f). Some of the stops are due event, there was a tendency to accumulate more frazil ice on
to the river discharge being lower than the minimum re- the parts of the screen without ice cover just upstream. Pre-
lease requirement, and some shutdowns related to damage cipitation was recorded at Sjøholt the day the blockage oc-
on the electricity grid and other maintenance work. But cured, most likely in the form of snow at Dyrkorn, in six out
some stops in production revealed a reduced flow capacity of eight reported stoppages. But there were several events
of the Coanda-effect screen, and will be discussed below. during the winter with Ta < 0 ◦ C and precipitation with-
out any reduction of the flow capacity of the Coanda-effect
screen. It is hence deduced that snow alone is not a critical
Ice Blockages in field parameter for blockage of the intake screen. One example
During its first year of operation, eight events of complete of a rapid frazil ice blockage event was on the 7th of De-
ice blockages at the Coanda-effect screen caused a shut- cember, when it was observed that frazil ice started to block
down of Dyrkorn SHPP, Table 1, with the duration of the the Coanda-effect screen one hour after the first measure-
stops being between 3.0 − 37.5 h. In addition to the com- ment of supercooled water, and 20 minutes later the power
plete blockages there were four events with reduced intake plant had to be shut down. The entire screen was clogged
screen capacity. The length of the power production stop with soft frazil ice and the ice then froze to a solid ice cover.
and start was logged with the exact time, whereas the tem- After 39 hours of a complete blockage, and two hours af-
perature measurements were taken every 15 minute; con- ter the last measurement of supercooled water, the water
sequently, the highest accuracy of the reported durations of was flowing under the ice cover and the power production
ice blockages was ±0.25 h. started again.
All of the reported events with a complete ice blockage of Another example of an ice blockage event, with a more
the Coanda-effect screen were associated with supercooled gradual reduction of the flow capacity of the Coanda-effect
water before the ice clogging started, as shown in the sum- screen, was on the 5th of January. It was snowing that day,
mary of the hydrological conditions during the ice block- and a period of 25 hours with supercooled water started.
ages in Table 1. Frazil ice is active when the surrounding There was an ice cover on the intake pond, and the power
water is supercooled (Ashton 1986; Daly 1991; Stickler and production stopped 11 hours after the first measured super-
Alfredsen 2009). The active frazil ice particles sticks to the cooled water, although the power production was reduced
top surface of the screen wedge wires and causes an ice after only two hours with supercooled water. A water re-
blockage of Type I. The frazil ice blockage of the screen lease higher than the environmental flow was measured for
wedge wires started shortly, 0.3 − 3.3 h, after supercooled 20 hours, which means an unintended loss of water, and
water first occurred, for most of the events. hence a reduced intake capacity of the intake screen . Three
The air temperature was observed to be in the range of hours after the last recorded supercooling, the production
−5.4 ◦ C to −1.7 ◦ C at the time of complete blockage. Dur- was back to normal, without any surplus of released water.
ing some freezing events Ta increased, whereas during The frazil ice and snow on the screen grew to a protective
other events, Ta decreased. Ta,min is included in Table 1. solid ice cover, and on the 8th of January the intake per-
The temperatures in the chamber, both Tac and Twc , were formed without any problems, in spite of a Ta = −8.5 ◦ C.
close to 0 ◦ C, but never below. For the two first events, the At that point, the intake pond was 100 % covered with ice.
temperature at the screen, Ts was < 0 ◦ C at the initiation
of the clogging and below 0 ◦ C approximately 1.5 h before Cold Air Temperature and Supercooling
the clogging started. The measurement device for Ts broke
on the 22th of December. During the winter of 2011/2012, 79 % of the 950h with
There was less than 50 % ice cover of the intake pond before Ta < 0 ◦ C and 66 % of the 322 h with Tw < 0 ◦ C, had
the clogging started for seven out of the eight ice blockage a more or less open Coanda-effect intake screen. A to-
events. For the event with 95 % ice cover, there were 11 h of tal of 51 supercooling events were recorded, and the water
supercooled water before the stop, compared to 0.3 − 3.3 h flow to the power plant was reduced or stopped for 61 %
for the events with less ice cover. For the event with 95 % of this time period. For all the periods with either Ta or
ice cover, the power plant started up while there still was Tw < 0 ◦ C, the lowest Ta is plotted against the lowest Tw,
supercooled water. Ice cover of the intake pond seemed and each period is defined with respect to the performance
to have an impact on the performance of the Coanda-effect of the Coanda-effect intake screen, i.e. complete blockage
screen, which is consistent with previous studies of frazil (8 events), reduced flow capacity (4 events) or normal flow
ice blockage of intake screens (Daly 1991; Richard and conditions (91 events), Figure 8. Events with Tw > 0.5 ◦ C

Preprint submitted to Journal of Cold Regions Engineering 8

Table 1: Overview of the hydrological parameters during the events with complete stop of production because of ice blockage
winter 2011/2012
Date of blocage 12/07 12/20 12/31 01/05 03/03 04/03 04/04 04/07
Duration of stop [h] 38.5 36.5 10.5 4.0 7.0 6.0 3.0 8.5
Time with Tw < 0 ◦ C before stop [h] 1.5 0.8 1.0 11.0 1.3 1.0 0.3 3.3
Totale time with Tw < 0 ◦ C [h] 37.5 11 11 25 6 3 0.3 10
Time from last Tw < 0 ◦ C before start [h] 2.5 10.5 -1.0 -10.5 3.5 4.5 2.5 1.0
Ta at stop [◦ C] -5.5 -3.9 -2.4 -2.9 -5.1 -4.0 -1.7 -5.4
Ta,min [◦ C] -9.7 -6.4 -6.0 -6.4 -5.9 -5.5 -1.7 -8.4
Tw,min [◦ C] -0.016 -0.015 -0.019 -0.012 -0.023 -0.019 -0.007 -0.010
Ts,min [◦ C] -0.2 -0.2 - - - - - -
Pond Ice coverage [%] 50 40 0 95 0 10 10 10
Daily percipitation at stop [mm] 7.1 0 17.2 11.7 0 37.1 8.2 0.8
Discharge in the river [m3 /s] 0.80 0.40 0.95 0.70 1.20 1.10 0.85 0.40
Type of ice blockage [-] Type I Type I Type I Type I Type I Type I Type I Type I

exceeds the axis of Figure 8. All periods with both Ta and

Tw > 0 ◦ C are excluded from the data set. All events
with a complete ice blockage of the Coanda-effect screen
or a reduced flow capacity comprise measurements of both
Ta < 0 ◦ C and Tw < 0 ◦ C, Figure 8. Even though all the
events with reduced flow capacity are related to Tw < 0 ◦ C
and hence frazil ice, there are periods with measurements of
Tw < 0 ◦ C, but without a blockage of the screen. Based on
the results shown in Figure 8, it is deduced that the tempera-
ture on the screen wedge wires, Ts , and hence also Ta , must
be below 0 ◦ C, in addition to supercooled water,Tw < 0 ◦ C,
to cause the risk of a reduction of the Coanda-effect screen
flow capacity caused by frazil ice blockage. It must be em-
phasized that there were no long periods with deep frost,
such as Ta < −10 ◦ C during the winter of 2011/2012.

Reopening of the Coanda-effect Screen

Frazil ice eventually led to a reduced intake capacity or
complete blockage of the Coanda-effect screen, but the
structure reopened without any intervention after all events
of freezing. When the air temperature is low enough, accu-
Figure 8: The performance of the Coanda-effect screen in- mulated soft ice on the wedge wires will freeze to a solid
take screen categorized according to the flow capacity. All ice cover over the screen, and after a while, the water will
events with either Ta < 0 ◦ C or Tw < 0 ◦ C, are plotted with start to flow under the ice. Since there are no ice particles
the minimum air temperature, Ta,min , against the minimum in between the wedge wires, the intake will reopen without
water temperature, Tw,min , during the specific period. any intervention. In Figure 9a the Coanda-effect screen is
completely clogged with frazil ice and the water is flowing
over the ice. In the picture taken two days later, Figure 9b, a
solid ice cover is protecting the screen, the water is flowing
under the ice and a full flow capacity is reestablished.
The cause of the reopening of the screen is not fully con-
firmed. We suggest that the reopening process is a combi-
nation of melting caused by the heat flux from the water to
the ice particles, mechanical erosion of ice particles by the

Preprint submitted to Journal of Cold Regions Engineering 9

Figure 9: Ice cover on the Coanda-effect screen. a) A complete frazil ice blockage of the screen (2011-12-07). b) Solid ice
cover on the screen (2011-12-09). The water is passing under the solid ice cover into the intake, and the full flow capacity is
reestablished.

water flow momentum and the shear forces and uplifting is when the screen is so cold that solid ice forms between
forces from hydraulic pressure on the ice particles. the wedged wires. Based on laboratory experiments, as
Based on the data received in field, the reopening of the long as Ta > −7 ◦ C and Ts > −2.5 ◦ C, the ice forma-
Coanda-effect screen, which is confirmed by the startup of tion will take place mainly above the wedge wires, a Type I
the SHPP, took place without any intervention for all the ice blockage. According to Andersson and Daly (1992), the
complete ice blockage events, and for most of the complete initialization of ice accumulation is at the upstream side of
ice blockage cases, 1 − 10 h after the last measurement of a screen bar, and the ice will extend upstream, increase in
supercooled water, Table 1. For some of the ice blockage width until the ice creates a bridge between two successive
events, supercooled water was measured up to 10 hours af- bars. Their finding correspond well with Type I ice clog-
ter the screen was reopened. It can therefore be deduced ging. The Coanda-effect screen was completely self clean-
that once the frazil ice production stops, after an event of ing and the screen reopened without any manual interven-
ice clogging of Type I, the screen reopens without any in- tion after all observed ice blockage events in field. Frazil ice
tervention since the spacing between the wedge wires are in combination with air and screen temperature below 0 ◦ C
free from ice. The intake may also start performing again is critical for ice to affect the intake screen flow capacity. If
if an ice cover protects the screen from more approaching frazil ice reaches screen wedge wires with a Ts < 0 ◦ C, the
suspended ice and the clogging ice particles are removed intake may become blocked, even though no ice particles
by erosion or melting underneath an ice cover. The ice may enter between the screen wires. As long as the frazil ice
also be washed away by an increased water discharge in the formation in the river stops or suspended ice is prevented
river exceeding the screen capacity. from reaching the screen, the Coanda-effect intake screen
will reopen and start performing in a cold climate.
In order to improve the performance of a Coanda-effect in-
Conclusions take screen in a cold climate, it is recommended to pro-
vide for an early stage solid ice cover over the screen,
This study reveals that the Coanda-effect intake screen per- by placing a fabric or something similar above the boul-
formed well in a cold climate, even during periods with a der bars, may be extended upstream from the weir. A
solid ice cover over the screen, water entered the intake solid ice cover may reduce the amount of frazil ice reach-
through the screen wedge wires underneath the ice, as re- ing the screen and hence a Type I clogging. An ice cover
ported by ENTEC (1998). But ice formation may temporar- will also protect the screen from condensation and a Type
ily reduce the intake flow capacity of the screen or lead to II ice blocking. It is also a theoretical possibility of in-
complete ice blockage. Two main types of ice clogging of stalling a heating system to the screen wires. Previous
the Coanda-effect intake screen has been observed. Type studies on intake screen heating systems suggest a required
I clogging occurs when soft and wet ice slush clog by ad- heat of 2 − 8 kW/m2 in order to retard the frazil ice ad-
hesion to the top surface of the screen and no ice particles hesion (Logan 1974). Heat is not sufficient to stop larger
enter between the screen wedge wires. Type II clogging

Preprint submitted to Journal of Cold Regions Engineering 10

 amount of frazil ice. For Dyrkorn hydropower plant, the re- GE, Sensing (2012). 1730/1740 Series Druck Stainless
quired effect of a heating system would be approximately Steel Level Pressure Sensors. Web Page. URL: http :
15 m × 1.5 m × 5 kW/m2 ≈ 113 kW, which is technical / / www . ge - mcs . com / download / pressure -
feasible, but may be not profitable. level/920_093c_e.pdf.
Huber, Dorothe (2005). “BEDUIN Project ’Better design of
intakes’”. Norwegian University of Science and Technol-
Acknowledgments ogy. McS Thesis.
Logan, T.H. (1974). Prevention of ice clogging of water
The writers wish to thank Tafjord Kraftproduksjon AS for intakes by application of heat. Report. U.S. Bureau of
their welcoming reception and for providing access to the Reclamation.
intake chamber, to the online camera and for their technical Morse, B. and M. Richard (2009). “A field study of sus-
support on-site. We also want to thank Brødrene Dahl AS pended frazil ice particles”. In: Cold Regions Science and
for providing us with the model test rig. Technology 55.1, pp. 86–102.
Osterkamp, T. E. and J. P. Gosink (1983). “Frazil ice for-
mation and ice cover development in interior Alaska
References streams”. In: Cold Regions Science and Technology 8.1,
Andersson, Annika. “Frazil ice at water intakes”. pp. 43–56. ISSN: 0165-232X.
Luleåtekniska universitet. PhD thesis. Richard, M. and B. Morse (2008). “Multiple frazil ice
Andersson, Annika; and Steven; Daly (1992). Laboratory blockages at a water intake in the St. Lawrence River”.
Investigation of Trash Rack Freezup by Frazil Ice. CR- In: Cold Regions Science and Technology 53.2, pp. 131–
REL Report 92-16. U.S Army corps of Engineers. 149.
Ashton, G.D (1986). River and Lake Ice Engineering. Lit- SeaBirdElectronics (2011). Data sheet SeaBird39. Web
tleton, Colorado: Water Resources Publications. Page. URL: http : / / www . seabird . com /
Blevins, B.B et al. (2005). Roadmap for PIER Reserach products/spec_sheets/39data.htm.
on Fish Passage at California Hydropower Facilites. Re- Strong, J. J. and R. F. Ott (1988). “Intake screens for small
port. hydro plants”. In: HYDRO REV. 7.5 , Oct. 1988, pp. 66–
Carstens, Torkild (1966). “Experiments with supercooling 69.
and ice formation in flowing water”. In: Geofysiske pub- Vemco (2011). Data sheet MiniLog2t. Web Page. URL:
likasjoner [0072-1174] 26, pp. 1–17. http : / / www . vemco . com / pdf / minilog2t .
Clark, S. and J. Doering (2006). “Laboratory experiments pdf.
on frazil-size characteristics in a counterrotating flume”. Wahl, T. L. (2001). “Hydraulic performance of Coanda-
In: Journal of Hydraulic Engineering 132.1, pp. 94–101. effect screens”. In: Journal of Hydraulic Engineering
Comark (2011). Data Sheet Comark EV N2012. Web 127.6, pp. 480–488.
Page. URL: http : / / www . comarkltd . com / – (2003). Design Guidance for Coanda-Effect Screens. Re-
specification.tpl?product_id=183. port. U.S Department of the interior Bureau of Reclama-
Daly S. F, Ed. (1994). Report on Frazil Ice. IAHR Work- tion. Technical Service Center Water Resource Reserach
ing group on Thermal Regimes. Special Report 94-23. Laboratory.
US Army Corps of Engineer Cold Regions Research and Wahl, T. L. and R. F. Einhellig (2000). “Laboratory test-
Engineering Laboratory. ing and numerical modeling of Coanda-effect screens”.
Daly, S. F. (1991). Frazil Ice Blockage of Intake Trash In: Joint Conference on Water Resource Engineering and
Racks. Cold Regions Technical Digest No. 91-1. USA Water Resource Planning and Management. Vol. 104.
Cold Regions Research and Engineering Laboratory.
Daly, S. F. and R. Ettema (2006). “Frazil ice blockage of
water intakes in the Great Lakes”. In: Journal of Hy-
draulic Engineering 132.8, pp. 814–824.
eKlima (2012). Norwegian Meteorological Institute. Web
Page. URL: www.eklima.no.
ENTEC (1998). Schlussbericht Pilotprojekt COANDA Fein-
rechen fr kleine Wasserkraftanlagen. Report.
Ettema, R et al. (1984). “Laboratory experiments on frazil
ice growth in supercooled water”. In: Cold Regions Sci-
ence and Technology 10.1, pp. 43–58. ISSN: 0165-232X.

Preprint submitted to Journal of Cold Regions Engineering 11

View publication stats