COMPARATIVE STUDY OF THE COOLING OF A

HOT TEMPERATURE SURFACE USING SPRAYS

AND LIQUID JET
Alexandre Labergue, Thomas Aiguier, Michel Gradeck, Fabrice Lemoine

To cite this version:

Alexandre Labergue, Thomas Aiguier, Michel Gradeck, Fabrice Lemoine. COMPARATIVE STUDY
OF THE COOLING OF A HOT TEMPERATURE SURFACE USING SPRAYS AND LIQUID JET.
15th International Heat Transfer Conference (IHTC-15), Aug 2014, Kyoto, Japan. pp.2415-2429,
�10.1615/IHTC15.evp.009287�. �hal-01579529�

HAL Id: hal-01579529

https://hal.archives-ouvertes.fr/hal-01579529
Submitted on 1 Sep 2017

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 COMPARATIVE STUDY OF THE COOLING OF A HOT TEMPERATURE
SURFACE USING SPRAYS AND LIQUID JET

A. Labergue1’2, T. Aiguier1,2, M. Gradeck1,2, F. Lemoine1,2,

1: Université de Lorraine, LEMTA, UMR 7563, Vandoeuvre-lès-Nancy, France
2: CNRS, LEMTA, UMR 7563, Vandoeuvre-lès-Nancy, France

ABSTRACT

An experimental test-rig was designed to study the cooling of hot surfaces under semi-industrials conditions.
Two cooling methods were compared: liquid jet and full cone sprays. The wall is a 175 mm diameter nickel disk
heated by electromagnetic induction up to 800 °C. The heat removed during the cooling phase was investigated
with the use of infrared thermography while the droplet properties, in the case of sprays, were characterized
simultaneously with a Phase Doppler system. In the case of spray cooling, the Phase Doppler technique was
mainly applied to study the statiscal properties of the droplets in the vicinity of the heated surface. The influence
of the liquid mass flux distribution as well as the droplets properties on the cooling are studied by using five
different spray nozzles. The liquid mass flux distribution is also measured with the help of patternator. The main
results show that even if higher heat flux can be removed with the liquid jet the efficiency cooling is poor
because of its non uniformity. When spray cooling is considered, the results are in qualitative agreement with
previous studies performed in similar conditions.

KEY WORDS: Spray cooling, spray heat transfer, infrared thermography, inverse conduction problem, Phase
Doppler Analysis.

1. INTRODUCTION

The use of water for cooling of hot surfaces in industry is unavoidable when high heat flux dissipation is
required. Nuclear safety issues or the thermal processing of alloys in steel industry are good examples.
Cooling processes include pool boiling, impingement of liquid jets or sprays. Pool boiling and liquid jets
provide high dissipation rates but they have generally failed to ensure uniform and controlled cooling.
In steel industry, cooling by means of impinging water jets is still widely used. However, the spatial uniformity
of heat flux removed from the surface cannot be achieved using impinging water jets and thus the homogeneity
of the metallurgical steel phase cannot meet the quality standards. The use of sprays seems attractive for several
reasons: a better spatial uniformity of the cooling can be achieved associated lower water consumption for the
same efficiency in the obtained heat transfer, thus leading to energy saving. The experimental studies devoted
to water spray cooling cover a wide range of cooling conditions [1-7]. Several kinds of sprays are employed,
such as mono-sized sprays [1;2] or polydisperse sprays [3-7]. Moreover, several categories of nozzles are
available to generate various sprays with different geometries and liquid densities: full cone, hollow cone,
flat fan or air-mist nozzles. However, only few authors have performed comparisons between these different
systems [3]. Another key issue is the scale under which the experiments are conducted, either laboratory
scale or industrial scale in term of liquid flowrate. In steel processing industry, the surface temperatures are
generally above the Leidenfrost point and high liquid mass flux are required, which involved large-sized
droplets. For instance, Al-Ahmadi [3] and Yao [7] estimated the heat flux removed by sprays under
conditions that are close to industry operations (droplet diameters up to 25 mm and liquid mass flux up to
30 kg/m²/s respectively). However, the hydrodynamic characteristics of the droplet interaction with the
heated surface were not investigated.
Very often, thermocouples, combined to an inverse conduction method, and Phase Doppler system (PDA)
are combined to characterize respectively the heat flux removed from the droplet surface and the droplet

*Corresponding Author: [email protected]

1
properties (velocity and diameter distributions) [8-9]. More recently, infrared thermography was also applied
to estimate the heat flux extracted from a heated wall by water cooling [10-11]. But these latter concern
mainly liquid jets, single droplets or monodisperse droplet streams. Jia and Qiu [8] achieved to detect both
incoming and outcoming droplets in the case of a low liquid mass flux value (1.20 kg/m²/s in maximum).
Puschmann et al. [12] have used a 2D-PDA system and infrared measures to characterize simultaneously the
droplet properties and the heat flux removed by sprays having a maximum liquid mass flux of about 5
kg/m²/s. Nevertheless, the hydrodynamic characteristics of the impinging droplets were not studied.
Therefore, the present work aims to compare the cooling efficiency of several full cone sprays and of a
continuous liquid jet. An experimental set-up was designed in order to operate close industrials conditions: in the
case of sprays, both heat flux and characteristics of the droplets interacting with the droplet surface were
characterized simultaneously. In the case of a liquid jet, only the cooling efficiency was determined. The surface
to be cooled is a circular disk made of nickel, which is previously heated above the Leidenfrost temperature. A
2D-PDA system was used in the front face of the plate and an infrared camera is used to measure the
temperature field on the opposite surface. An inverse conduction model is used to estimate the removed heat
flux.

2. SPRAYS FACILITIES

An experimental test rig, sketched in Fig. 1, was designed in order to work under conditions as
closed as possible as industrial conditions.

Fig. 1 Side view (i.e. X-Z plane) of the test rig. 1 – Pump1; 2 – Water tank; 3 – Pump2; 4 – Manometer; 5 –
Flowmeter; 6 – 3D traverse system; 7 – insulation part (quartz); 8 – nickel disk; 9 – retractable inductor; 10
– mirror; 11 – infrared camera; 12 – IR camera protective cover; 13 – laser probe; 14 – PDA receiver; 15 –
main tank for water collecting.
As the spreading of the impact occurs on a large area, a 340.30 m3 tank (15) was made where
water remains permanently. So that, the cooling water from a tank supplies the nozzle by means of
a closed looped circuit. A first pump (1) transfers the water from the main tank (15) up to a second
tank (2). Then, a centrifugal regulated pump (3) ensures the circulation of the liquid at the desired
operating parameters (i.e. flow rate and injection pressure controlled respectively with a manometer
(4) and a flowmeter (5)). The cooling of a hot surface with water is basically a transient
phenomenon. Here, an experiment is divided into two steps: a heating phase of the surface to be
cooled (the spray is not flowing) and a cooling phase that consists in applying the spray. However,

2
the time to establish the spray flow is not negligible compared to the cooling time. This time could
be reduced by maintaining the centrifugal pump switched at the same operating point. Therefore, a
second nozzle is used so that during the heating phase, a second spray flows directly towards the
main tank. Two valves (Valve A and B) allow switching from a spray to the other one.

Table 1 Description of the five sprays and of liquid jet used with their experimental operating conditions.

Nozzle-disk Injection
Flow Injection
Spray angle distance, Z- temperature
 Sprays rate Pressure
(°) direction Tl (°C)
(l/mn) (bar)
(mm)

Nozzle 1 45 6.7 7 250 19

Nozzle 2 45 5.5 7 250 19

Nozzle 3 45 6.7 3.7 250 19

Nozzle 4 60 6.7 7 160 19

Nozzle 5 60 6.7 3.7 160 19

Nozzle-disk
Jet diameter Flow rate
 Liquid Jet distance, Z-
(mm) (l/mn)
direction (mm)
9 6.7 250 19

The disk (8) is made out of nickel (Radius R = 87.5 mm and thickness e = 5 mm) because the properties of
this metal are very close to those of steel. The disk is positioned horizontally below the nozzle A so that the
disk center is aligned with the nozzle axis. The disk is placed on a ring made of an insulating material (7).
The disk and both nozzles with their respective holders are fixed together on a 3D-traverse system (6). The
origin of the X-Y-Z axis is taken at the exit nozzle A. The Z-axis direction corresponds to the spray motion.
Above, the upper surface of the disk (designated by front face), measurements of the droplet size and
velocity are performed with a phase Doppler system (PDA) (Items (14) and (13) in Fig. 1). The temperature
of the opposite wall (designated by rear face) is measured by an infrared camera (11) through a 45° mirror
(10). Both the mirror and the camera are protected against expelled water with the insulating part (7) and a
protective cover (12) respectively. Heating of the disk is realized by an electromagnetic induction system
(Power Supply 12 kW; Five-Celes). A retractable inductor (9) is placed just above the front face of the plate
during the heating phase and is removed before the cooling with the sprays.
One of the objectives of the present work is to test several droplet size and velocity distributions as well as
different liquid mass flux distributions. Therefore, five Lechler nozzles (numbered from 1 to 5) leading to
full cone sprays are used. The differences between these sprays concern their flow rate for a similar injection
pressure and their cone angle value. Nozzles 1,2 and 3 present a cone angle of 45°, while nozzles 4 and 5
have a cone angle of 60°. Thus, to ensure that the spray covers all the surface area, the elevation Z (distance
between the exit nozzle and the front face of the disk) is adapted according to the disk radius and the value of
the spray angle. Table 1 summarizes the value of each of the parameter for the five nozzles. Fig. 3 presents
typical droplet size distributions measured with the PDA system for the five nozzles in the section center
(X=Y=0) and at the Z elevation from the nozzles according values given by Table 1. All distributions peak at
the same diameter value (d  40 µm). Nozzle 2 and nozzle 5 have the smallest and largest droplets
respectively.
When the liquid jet is used, a 9 mm-diameter tube is used with similar operating conditions as nozzle 1, i.e.
at Z = 250 mm above the disk center and with a flow rate of 6.7 l/mn (Table 1).

3. EXPERIMENTAL DIAGNOSTICS

3.1 2D Phase Doppler Measurements

3
Droplet size and velocity were measured using a phase Doppler system (PDA). The technique was used to
characterize the properties of the five free sprays in absence of the target. For that purpose, the PDA
measurements volume is located at the same Z elevation value as those of the front face of the disk (i.e.
Z = 250 mm for nozzle 1,2 and 3 and Z = 140 mm for nozzles 4 and 5). The spray is moved in the X-Y
plane by using the 3D-traverse system in order to cover the whole section of each spray. In a second step,
PDA measurements were performed to characterize impingement of the droplets on the heated surface.
The phase Doppler system is a 2D commercial classic PDA manufactured by Dantec-Dynamics. An argon
ion laser, operating in multiline mode, is used to generate two wavelengths (v = 514.5 nm and v = 488 nm)
in order to obtain the two overlapped laser illuminated volumes. The illuminated volume was formed by
using a LDA transmitter probe (Dantec-dynamics Fiber-Flow probe) which is aligned along the X-direction
(label (13) in Fig. 1). The system includes a classic reception optical device and a P80 signal processor. The
receiver (label (14) in Fig. 1) is operating in the first refraction mode with an off-axis angle of 60°. The front
lens of the transmitter probe is 1200 mm while the focal length of the receiver is 510 mm, leading to a
maximum detectable droplet size of 327 µm.
In order to characterized the droplets interacting with the heated plate (incoming and outcoming droplets),
the PDA probe volume has to be relatively close to the front face. But the presence of a large droplets cloud
between the probe volume and the PDA receiver can strongly disturb the signal. Then, the PDA receivers as
well as the beams system were inclined of an angle an angle of 45° (Fig. 2). This rotation was corrected in
the calculation of the two velocity components so that to obtained them the X-Y-Z reference. Therefore, the
components along the Z-axis (w) and along the Y-axis (v) were measured. With this configuration, droplets
impinging the disk have positive velocities and droplets with negative velocities are considered as outcoming
droplets. Table 2 gives the characteristics of the PDA optical configuration.

Table 2 Characteristics and optical configurations for the 2D PDA measurements.

Green wavelength laser , bv (nm) 514.5

Blue wavelength laser, bb (nm) 488
Beam radius (green and blue) , r (mm) 1.35
Beam spacing, sb (mm) 40
Emission focal length, fe (mm) 1200
Reception focal length, fr (mm) 500
PDA magnification, r 3.8
Scattering angle,  (°) 60
Micrometer (mm) 1.5
Maximum detectable droplet diameter, dpmax (µm) 327
Illuminated volume size for v = 514.5 nm based on 1/e² Illuminated volume size for v = 488 nm
(mm) based on 1/e² (mm)
In X direction 36.80 In X direction 34.9
In Z and Y direction 0.58 In Z and Y direction 0.55

4
 0.25
X=Y=0
0.20 Nozzle 1; Z=250 mm
Nozzle 2; Z=250 mm
Nozzle 3; Z=250 mm
0.15
Nozzle 4; Z=160 mm

P(d)
Nozzle 5; Z=160 mm
0.10

0.05

0.00
0 50 100 150 200 250 300
d (m)

Fig. 2 Front view (i.e. Y-Z plane) of both laser and Fig. 3 Typical droplet size distributions for the five
receiver probes arrangement. studied nozzle measured without disk.

3.2 Heat flux measurements

Infrared thermography  The infrared camera (Cedip Jade III) is built around a focal plane array photonics
detectors (InSb) working in the spectral range [3-5 µm] and it is equipped with a narrow [3.97-4.01 µm] spectral
interval filter to fit with the expected experimental temperature range. It allows the measurements of up to
320×240 pixels/frame. The sampling rate depends of the phenomenon duration observed; here, the cooling phase
has a typical duration of about 20s to 30s and the desired number of frames to store per run is about 2000. Thus,
the frequency for all experiments conducted in this study is adjusted at about 60Hz. The calibration of the
detector itself was performed with a large blackbody. Then, to accurately measure the field of temperature of the
nickel disk's surface, one face of the disk (rear face) has been previously sandblasted to obtain a diffuse surface.
Then, the surface has been oxidized (with O2 of air) at high temperature. The layer of obtained oxide is so thin
and so stable that it has no effect on heat transfer and guaranties a stable radiation property. The value of total
measured emissivity is r = 0.64. A last measurement using a thermocouple is required for calibrate the whole
measurement chain (i.e. to account for the mirror and environment); for such calibration, a thermocouple is
inserted at the disk centerline in the rear face in order to have a reference temperature: the disk is firstly heated
up to 800°C then the infrared signal is measured simultaneously with the thermocouple signal while the disk
cooled down naturally (without additional cooling provided by spray or water jet).

Inverse heat conduction prolem  Knowing the temperature field at the rear face, an inverse heat conduction
model is used to estimate both the heat flux due to the spray or jet cooling and to determine the temperature
on the front face. The 3D real internal conduction problem is reduced to a 2D one by angular averaging. So
that only an estimate of the averaged value of the heat flux q”spray(r,t) is available. Then, the 2D internal
transient conduction reduced problem within the nickel circular disk is solved assuming some hypothesis:
constant nickel thermophysical properties, no conduction losses due to the disk holder and no losses at the
edge of the disk. A full description of the model is available in papers [10; 13; 15]. The reduced problem to
be solved is thus described by the following equation associated with the initial and boundary conditions:
 2 T 1  T  2 T 1 T
   (1.a)
r r r z a t
T  Tinit (r ) at t = 0 (1.b)
T
k  q spray
"
(r , t ) (1.c)
z z 0
T
k  hr (T ) (T  T )   r (T 4  T4 ) (1.d)
z z e

5
 T
k  0 , in r = R (1.e)
r r R

Laplace (time) and Hankel (space) transforms are performed so that equation (1a) can be solved :
~
Tn2
 ( n2  ) Tn  0 with u n  n   
p ~ 3
z a 
8 (n  )
4
un
with  n  and u n solutions of J1 (u )  0 . Finally, the quadrupole method [15] yields a linear
R
relationship between the rear face temperature (z = e) and the cooling heat flux (z = 0) :
~ ~
T  Z ( p) q~ n n n
(2)
where:
1 1
Z ( p) 
h EQ
INF cosh(e p / a )   ( p / a ) sinh(e p / a )) p
EQ
hINF is an equivalent heat transfer coefficient accounting for convection and radiation losses; p is the
Laplace parameter. Laplace inversion of equation (2) is a convolution product and can be expressed as :
t
~ ' ~
Tn (t )   e  a  n (t t ) Z n (t  t ' ) q~n (t ' ) dt '
2
(3)
0
Finally, the real temperature in time-space domain is obtained through the following relation :
2 
~ J 0 ( n r )
T (r , e, t )   T (e, t ) J (4)
0 ( n R )
n
R2 n 0
2

The inverse problem consists in estimating the surface heat cooling flux. In the real problem estimation,
equation (4) is truncated at a finite harmonic number, nmax . Equation (3) can be written as follows:
k
~ ~
Tn (t k )   e  a  n ( k l ) t Z n (k  l ) q~ln (tl ) t
2
(5)
l 1
where t is the time step of the camera or a multiple and t k  k t , tl  l t . Equation (5) can be written as
a linear matrix relationship:
~
 Tn (t1 )   S11n 0 0 0   q~1n 
     
~  .  ~  . . 0 0   .  (6)
θn   S nk q
 .  n
 . . . 0   . 
~   n n  ~ n 
Tn (t kmax )  S kmax . . S kmax kmax  qkmax 

We can estimate the heat flux harmonics q~n (t k ) without regularization by an ordinary least square method:
~ˆ
q~ˆn  (Snt Sn ) 1 Snt  n (7)
The return to time-space domain yields:
2 J (  r ) ~ˆ
nm ax
(8)
q" (r , t )   q (t )
0 n l

R J ( R)
l k 2 2 n k
n 0
0 n

3.3 Measurements of the liquid mass flux distributions

A patternator, composed of a row of twelve tubes connected to individual sampling vessels, is specially designed
in order to measure the distribution of the liquid mass flux for each nozzle. The diameter of each individual tube
is 7.5 mm, leading to a collection surface of 44.5 mm². The distance between two consecutives tubes is twice the
diameter tube value. The patternator is placed so that the top of the tubes are located at the same Z elevation of
the front face similarly to the PDA probe volume in the case of free sprays. The use of the 3D-traverse system
allows moving the spray in order to scan the spray section in the X-Y plane. The liquid mass flux is then

6
measured during a given time, on the order of 180 s per measurement. Typically, the liquid mass flux
distribution is described with a mesh having 11*11 measurement points for each nozzle.

4. ACQUISITION PROCEDURE AND DATA REDUCTION

As mentioned in section 2., each experiment has a heating phase and a cooling one. For the heating phase,
the inductor is placed as close as possible to the surface in order to ensure a quick heating. The heating phase
is monitored by the IR camera and is stopped when the average rear surface temperature is about 800°C.
Then, the cooling of the disk can begin. For this second phase, both PDA system and IR camera are
synchronized. The trigger output of the PDA processor is used in order to trigger the IR camera as
acquisition and the spray operation (valves A and B). The recording time tr depends on the cooling time,
reached when the average surface temperature is equal to the initial liquid temperature Tl = 19°C. As
different liquid flow rates are used, tr varies from 20 up to 30 s according the nozzle. During the cooling
phase, IR frames and PDA data are recorded on two separated computers.
The analysis of the results aims to correlate spatially the heat flux removed from the wall with the liquid
mass flux and droplet properties distribution. But, the absence of a perfect symmetry in the liquid mass flux
distribution (see Fig. 4) leads to an asymmetric and inhomogeneous cooling. Therefore, a data reduction is
necessary for the further analysis. Basically, it consists in calculating an average value for each of the
measured quantities, within an element of crown centered on the target (radius r, thickness dr). Thereby, all
the averaged quantities depend only on one spatial variable r (varying from r = 0 up to r = R) and is invariant
according the azimuth position on the surface.

5. CHARACTERISTICS OF THE FREE SPRAYS

Fig. 4a gives an example of the liquid mass flux distribution obtained with the nozzle 1 in the spray section
located at Z = 250 mm. It depicts clearly the non-uniformity of the liquid mass flux distribution. The
evolution of the averaged liquid mass flux along a radius for all the considered nozzles is showed in Fig. 4b.
Whatever the sprays, the liquid mass flux varies from about 2.5 up 8 kg/m²/s. Close to the center, the nozzle
5 and the nozzle 2 have respectively the highest and the lowest liquid mass flux value. Towards the edge of
the disk, all curves tend to merge and seem to increase slowly with r. Obviously, the liquid mass flux of the
jet is relevant, but, the comparison with the case of nozzle 1 should be performed on the whole surface area
of the disk. This will be discussed and presented in the section 7.

Nozzle 1
8 Nozzle 2
Nozzle 3
7 Nozzle 4
Nozzle 5
6
5
4
3
2
0 10 20 30 40 50 60 70 80
r (mm)
(a) (b)
Fig. 4 Typical liquid mass flux distribution measured in a cross-section obtained with nozzle 1 (a) and along a
radius for the five nozzles after data reduction (b).

7
Fig. 5 presents the statistical properties of the droplets (derived from all the droplet size classes), obtained with
the 2D-PDA system: mean values of the two velocity components, Sauter Mean Diameter (SMD) and mean
Weber number based on SMD. The evolution of the w velocity component (a), the v velocity component (b), the
SMD d32 (c) and the mean Weber number We (d) are plotted along a radius. The Weber number is base on the
SMD and the vertical velocity. It appears that the nozzles 2 and 3 have the lowest values of velocities and SMD
and consequently of the Weber number. As observed in figure5(c), the dependence of the Weber number as a
function of the radius remains low. As expected, nozzles 4 and 5 have the highest horizontal velocity (v
component) due their higher cone angles.

Nozzle 1 Nozzle 1
20 Nozzle 2 4 Nozzle 2
Nozzle 3 Nozzle 3
18 Nozzle 4 Nozzle 4
Nozzle 5 3 Nozzle 5
16
w(m/s)

v (m/s)
14

12 1

10
0
8
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
r (mm) r (mm)

(a) (b)

200
600
180
500
160 400
d32 (m)

We

Nozzle 1 300
140
Nozzle 2
Nozzle 3 200 Nozzle 1 Nozzle 3
120 Nozzle 4 Nozzle 2 Nozzle 4
100 Nozzle 5
Nozzle 5
100 0
0 10 20 30 40 50 60 70 80 90 0 20 40 60 80 100
r (mm) r (mm)
(b) (c)
Fig. 5 Droplet statistical properties along a radius obtained from the PDA measurements after data reduction for
the five nozzles: mean vertical velocities (a), mean horizontal velocity (b), SMD (c) and mean Weber number
based on SMD (d).

6. STUDY OF THE COOLING

6.1 Influence of the wall on the incident droplets properties

To analyze the cooling characteristics of a heated surface, the incoming droplet properties are often measured in
absence of the surface. However, the presence of the surface is likely to influence the trajectories of the droplets.
This section aims to verify this assumption. The PDA results about the incoming droplets in the case of an
impinging spray are compared with those obtained in the case of a free spray (without the disk). Fig. 6a depicts
the incoming droplets size distribution in impinging conditions and in absence of the disc for nozzle 1. Both are

8
measured at Z = 240 mm, i.e. 10 mm above the disk surface. Clearly, a significant difference can be observed:
the pdf of the droplet size measured in the case of the impinging spray exhibits a shift towards larger droplets
sizes compared to the case of the free spray. This can be potentially attributed to the deviation of the flowfield by
the disk, resulting in a deviation of the small droplets (d< 50 µm) which follow the deviated flow due to their
small Stokes number. Fig. 6b, describes the evolution of the horizontal velocity (v component) as a function of
the droplet diameter. It tends to confirm this explanation: the horizontal velocity of the smallest incoming
droplets increases widely in the case the impinging spray. It corresponds to the deviation of the smallest droplets
towards the disk edges. Therefore, the effect of the surface to be cooled on a spray flow can modify significantly
the droplet distribution of size and velocity in the impingement region.

0.25 5
Nozzle 1
Nozzle 1 spray with impact
0.20 4
Free spray
spray with impact
Free spray
0.15 3
P(d)

v(m/s)
0.10 2

0.05 1

0.00 0
0 50 100
150 200 250 300 0 50 100 150 200 250
d (m) d (  m)
(a) (b)
Fig. 6 Comparison of the incoming droplets parameters measured in the case of a free spray and in presence of
the disk for nozzle 1: pdf of the size distributions (a) and distribution of the horizontal velocity as a function of
the diameter (b).

6.2 Investigation of characteristics of the impinging droplets

This section is devoted to the study of the interaction of the droplets onto the heated wall. The following results
are derived from PDA measurements recorded during the cooling phase, at a distance of 10 mm above the front
face of the wall and at X=Y=0. . Fig. 7 describes typical size distributions of incoming and outcoming droplets
for the nozzle 1, incoming droplets having a positive vertical velocity and outcoming droplets having a negative
one. Even if outcoming detected droplets seems to be generally smaller than the outcoming one, both
distribution are peaking at the same value (d  40µm).

0.16 1.00
0.14
0.12 0.75
Nozzle 1
0.10
Incoming droplets
do/di

0.08 Oucoming droplets 0.50

P(d)

0.06 d32,o/d32,i
0.04 0.25 d10,o/d10,i
0.02
0.00 0.00
0 50 100 150 200 250 300 Nozzle 1 Nozzle 2 Nozzle 3 Nozzle 4 Nozzle 5
d(µm)

Fig. 7 Typical incoming and outcoming droplets size Fig. 8 Ratio of the outcoming Sauter mean diameter to
distribution the incoming Sauter diameter and ratio of the
outcoming mean diameter to the incoming mean
diameter, for the five nozzles.

9
The ratio of the outcoming Sauter mean diameter D32,o to the incoming Sauter diameter D32,i for the five
sprays is presented in Fig. 8 as well as the ratio for the mean diameter (D10,o/ D10,i). It appears that these
ratios are relatively close to one, especially for the mean diameter. Nevertheless, the outcoming droplets that
the PDA device detected can be incoming droplet that rebounded and also incoming droplet that splashed.
The Weber distributions of incoming droplets for the five sprays are depicted in Fig. 10. It shows that all
distributions peak for a Weber number of about 100. Therefore, we can assume that the droplet splashing is
surely the main phenomenon. Thereby, results of Fig. 7 and 8 can be related to the disintegrations by
splashing of the biggest droplets in smaller one.

0.45
0.40
0.35
Nozzle 1
0.30
Nozzle 2
P(We)

0.25 Nozzle 3
0.20 Nozzle 4
0.15 Nozzle 5

0.10
0.05
0.00
0 200 400 600 800 1000 1200 1400
We
Fig. 9 Incoming droplet Weber distribution for the
five studied nozzle

6.3 Study of the heat transfers rate

For numerical reasons, the IR images have not been post-processed up to r = R. Thus, all the following
results are presented for a maximum radius value of 70 mm.

Study of the heat flux removed from the wall  Fig. 10 depicts typical heat flux extracted by the nozzle 1 and
4 along a radius (horizontally) and as a function of the time (vertically). The colorbar is kept identical for
useful comparison). The main comments are the following:
- For nozzle 1 and up to t  7 s, the heat flux is almost spatially uniform. After, non homogeneous regions
appear along the radius. For 0 < r <30 mm at t  13 s, we can observe the apparition of the Leidenfrost
regime (the heat flux is minimum) before to reach the critical heat flux (CHF) at t  18 s. From t  20 s,
the heat flux decreasing (nucleate boiling regime) is again almost homogeneous throughout the radius.
- For nozzle 4, the CHF is reached quickly (t  3 s). Transition between wetting and non-wetting regime
(Leidenfrots point) cannot be clearly identified. Before the CHF point, the heat flux presents strong
spatial inhomogenities. Conversely, past the CHF point, the cooling rate is relatively homogeneous.
The boiling curves for the five sprays are presented in Fig. 11 and are plotted at two radius values:
r = 2.5 mm (a), and r = 70 mm (b). At r = 2.5 mm, the plots for nozzle 4 and 5 are clearly above the others.
Indeed, at this position, nozzle 4 and 5 present the highest liquid mass flux value (Fig. 4b). When r increases,
the boiling curves corresponding to the different sprays tend to superimpose; this can be related to the fact
that from r  35 mm, the liquid mass flux for the five nozzles is similar. Moreover, for nozzle 1, 2 and 3 it is
easy to observe that the Leidenfrost temperature tends to increase when r increases. This observation is also
directly correlated to the liquid mass flux which increases with r (Fig. 4b). This is in well agreement with the
observation of Al-Ahmadi and Yao [3].

10
 (a) (b)
Fig. 10 Evolution of the heat flux extracted to the wall by the sprays along a radius and as a function of the time
for nozzle 1 (a) and for nozzle 4 (b).

r = 2.5 mm
1.4x106 1.4x106 r = 70 mm Nozzle 1
1.2x106 1.2x106 Nozzle 2
Nozzle 3
1.0x106 1.0x106 Nozzle 4
q" (W.m-2)

Nozzle 5
q" (W.m-2)

8.0x105 8.0x105

6.0x105 Nozzle 1 6.0x105

Nozzle 2
4.0x105 Nozzle 3 4.0x105

2.0x105 Nozzle 4 2.0x105

Nozzle 5
0.0 0.0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Tw(°C) Tw(°C)
(a) (b)

Fig. 11 Boiling curves of the five nozzles plotted at X = 2.5 mm (a) and X = 70 mm (b).

Cooling efficiency  Usually, in order to quantify the ability for a spray to cool, the cooling efficiency  is
introduced:

(9)

It represents the amount of heat extracted by the spray divided by the total heat transfer which is required for
complete evaporation of the liquid, which includes the power required to heat the liquid up to its saturation
temperature and to heat the vapor up to the wall temperature. The evolution of the cooling efficiency along
the radius is presented in Fig. 12 for the five sprays at a given time during the cooling when the wall
temperature Tw is 560°C. In the central part of the plate, from r = 0 up to r  40 mm, it is observed that the
nozzle 2 has the highest cooling efficiency (  11%) whereas the other nozzle presents relatively similar
cooling efficiencies, on the order of 6 %. In the outer part of the plate (for 40 < r < 70 mm), the cooling
efficiency is the same regardless the nozzle used and decreases slowly with increasing the radial coordinate.
These observations can be correlated with the liquid mass flux (Fig. 4b). In the central part, the nozzle has
the lowest liquid mass flux and the highest cooling efficiency. In the second part, all nozzles have their liquid
masse flux that increases and their cooling efficiency that decreases.
Therefore, it should be conclude that an increase of the liquid mass flux is correlated to a decrease of the
cooling efficiency. In a previous study carried out by Yao and Cox [16], similar conclusions were found.
These authors introduced the spray Weber number based on the SMD and the liquid mass flux:

11
 (10)

The liquid mass flux is preferred because the direct effect of the droplet velocity on the cooling efficiency is
not clearly established. Their work highlighted that the cooling efficiency decreases with the spray Weber
number. However, it is expected that the wall temperature Tw as well as the liquid temperature Tl can also
influence the effectiveness. Finally, based on experiments performed on several sprays, they proposed a
correlation for the cooling efficiency, based on Wes, Tl and Tw :

(11)

This correlation is valid over a wide range of operating conditions and is well adapted for those of the
present study. Knowing the distribution of the liquid mass flux, the SMD (Fig. 4b and Fig. 5b respectively)
and the wall temperature, the quantity Wes Tsat/( Tw Tl) can be derived, the initial liquid temperature being
taken at 19°C (Table 1). Therefore, it is possible to compare the cooling efficiency obtained in the present
study (eqn. (9)) with those obtained by Yao and Cox [16] (eqn. 11). The cooling efficiency from both present
study and those of Yao and Cox [16] are plotted in Fig. 13 as a function of Wes Tsat/( Tw Tl) for the five
nozzles and for a constant wall temperature of 560 °C. As the wall temperature is fixed, the sole effect of the
liquid mass flux can be observed. As expected, the cooling efficiency decreases with the liquid mass flux and
is clearly above the values obtained by Yao and Cox [16]. The cooling efficiency corresponding to nozzle 2
is almost twice the values determined for the other nozzle.

Tw = 560 °C, Tl=19°C Nozzle 1

0.12 0.12 Tw = 560 °C, Tl=19°C Yao & Cox (eqn (11))
Nozzle 2
Nozzle 1
Nozzle 3
0.10 0.10 Nozzle 2
Nozzle 4 liq
uid Nozzle 3
Nozzle 5 ma Nozzle 4
0.08 0.08 ss
flu Nozzle 5
x
0.06 0.06

0.04 0.04

0.02 0.02

0.00 0.00
0 10 20 30 40 50 60 70 80 0.0 5.0x10-6 1.0x10-5 1.5x10-5
r (mm) WesTsat/(Tsub+Tsat)
Fig. 12 Profiles of cooling efficiency for the five Fig. 13 Comparison of cooling efficiency obtained in
nozzles along a radius when the wall temperature is the present study with the correlation of Yao&Cox,
560°C. when the wall temperature is 560°C.

Fig. 14 depicts the cooling efficiency obtained for the five nozzles at two different wall temperatures. As in
Fig. 13, the correlation of Yao & Cox is slightly below of the value found in the present experiments. It is
observed that the cooling efficiency decreases as the wall temperature increases; one potential explanation
could be in relation with increase of vapor film thickness due to higher wall temperature, on other one could
be related to homogeneous nucleate boiling inside the droplets leading to their disintegration.

12
 Yao & Cox (eqn(11))
0.12
Five Nozzles at Tw =560 °C
500
Five Nozzles at Tw =400 °C
0.10
400
0.08

Tleid(°C)
300
0.06
Yao & Cox (eqn (12))
0.04 200
Nozzle 1 Nozzle 3
Nozzle 2 Nozzle 4
0.02 100 Nozzle 5

0.00 0
0.0 5.0x10-6 1.0x10-5 1.5x10-5 2.0x10-5 0.0 5.0x10-5 1.0x10-4 1.5x10-4 2.0x10-4
WesTsat/(Tsub+Tsat) Wes

Fig. 14 Comparison of the experimental cooling Fig. 15 Evolution of the Leidenfrost temperature as a
efficiency obtained for 5 sprays and the correlation of function of the spray Weber number and comparison
Yao&Cox, for two wall temperature values. with the correlation of Yao & Cox.

The Leidenfrost temperature TLeid was also investigated. The leidenfrost temperature can be determined using
the boiling curves when the heat flux experiences a minimum value. This leidenfrost point is easier to detect
for the lower liquid mass flux. This observation is often mentioned in the literature and is related to the
absence of a stable boiling film. However, the Leidenfrost temperature tends generally to increase with the
liquid mass flux [3]. Fig. 15 presents the evolution of the leidenfrost temperature as a function of the spray
Weber number for the five nozzles. For each nozzle, the Leidenfrost temperature is obtained at several
positions along the radius. It appears that TLeid increases with Wes, i.e. with the liquid mass flux as mentioned
just above. Yao & Cox [16] suggested a correlation for TLeid:
(12)
This correlation is plotted in Fig. 15 and the best agreement is found with the experimental data for the
highest liquid mass flux.

7. COMPARISON BETWEEN THE LIQUID JET AND THE NOZZLE N°1

Fig. 16 depicts the heat flux extracted by the liquid jet along a radius and as a function of the time (the
colorbar is the same as in Fig. 10). Here, the impact area of the jet has an extension up to r = 4.5 mm. The
diagonal band shows very well the advancement of the reweting front with a velocity of about 3 mm/s.
As expected, the cooling is widely faster than nozzle 1 and 4 (Fig. 10). For instance at r = 0, the heat flux
falls down from about 14106 W/m² to 14106 W/m² in only 2 seconds. Moreover, the heat flux distribution
presents an important spatio-temporal inhomogeneity. Fig. 17 compares the boiling curve (black symbols)
and the cooling efficiency (blue symbols) for the liquid jet and for nozzle 1. In order to make the two cases
comparable, the heat flux of both liquid jet and nozzle 1 is integrated on the whole disk area and on the
cooling duration (i.e. up to a final mean surface temperature of Tl = 19°C). Therefore, q” and Tw in Fig. 17
are the heat flux and the wall temperature respectively averaged on the whole disk area (from r = 0 up to
r = 70 mm). Even if the averaged heat flux for the liquid jet presents has strong oscillations (due to its spatio-
temporal inhomogeneity), it is anyway greater than for the spray. On the other side, the spray has the greatest
cooling efficiency. Indeed, the liquid mass flux of the liquid jet (based on the whole disk area) is about 7.5
kg/m²/s, i.e. almost the twice of the spray. These first results demonstrate two main drawbacks of the cooling
using liquid jet: strong inhomogeneity and low efficiency.

13
 Heat flux
0.08
Nozzle 1
Liquid jet 0.07
4.50x105
0.06

q" (W.m-2)
0.05
3.00x105
0.04
0.03
1.50x105
0.02
Cooling efficiency 0.01
Nozzle 1 Liquid Jet
0.00
0 50 100 150 200 250 300 350 400
Tw(°C)

Fig. 16 Evolution of the heat flux extracted from the

Fig. 17 Comparison of the Boiling curve and cooling
wall by the liquid jet along a radius and as a function
efficiency for the liquid jet and for nozzle 1.
of the time

8. CONCLUSIONS

An experimental set-up was developed in order to investigate water cooling of high-temperature

surfaces under semi-industrial conditions. The cooling by a liquid jet and several polydisperse full
cone sprays are compared. The wall, a 175 mm diameter nickel disk, is heated by an induction
device up to 800°C before to be quenched. The heat flux was estimated with infrared thermography
combined with an inverse conduction method synchronized a phase-Doppler system. In first, the
effect of the presence of the wall on the motion of the incoming droplets was highlighted. The
trajectories of the smallest droplets, which behave as passive tracers, are deviated by the plate,
which reduce their impingement probability on the heated target. The comparison between the
statistics of the incoming and outcoming droplets shows a slight decrease of the mean Sauter
diameter and is attributed to the disintegration of the largest droplets to smaller one.
When spray cooling is considered, the cooling efficiency decreases with the sprayed liquid mass
flux and the Leidenfrost temperature increases with the wall temperature, which is in qualitative
agreement with other studies of the literature. The comparison of the liquid jet with a spray show
that even if higher heat flux can be removed with the liquid jet, their use leads to strong spatio-
temporal inhomogeneities and low cooling efficiency.
Future works will be devoted to investigate the case of sprays having higher liquid mass flux. In
addition, high speed camera vizualistion will be performed in order to describe more thoroughly the
interactions of the spray with the heated plate and to identify the different hydrodynamic regime
(rebound, splashing, and film deposition).

ACKNOWLEDGMENT

This work has been supported by the French National Agency (ANR) in the frame of the research program
IDHEAS (ANR NT09 432160).

NOMENCLATURE

We We number (-) Cp specific capacity (J/Kg.K)

Re Reynolds number (-) d droplet diameter (m)

14
f focal length (m)  dynamic viscosity (Pa.s)
k thermal conductivity (W/m.K)  surface tension (kg/m²)
h hentalpy (J/kg)
liquid mass flux (kg/m2.s1) Subscripts
flow rate (l/mn1) b laser beam
q" heat flux (W/m2) d droplet
u,v,w droplet velocity component (m/s) l property of liquid phase
p liquid pressure (bar) Leid temperature at which minimum film
T temperature (K) boiling heat flux occurs
s spray
Greel symbols sat liquid saturation properties
α thermal diffusivity (m²/s) sub surface subcooling
 spray angle (°) v property of vapor phase
 emissivity (-) w conditions at front disk wall
 laser wavelength (m) r rear face
mass density (kg/m3)

REFERENCES

[1] Choi K.J., Yao S.C.,“Heat transfer mechanisms of horizontally impacting sprays,” Int. J. Heat Mass Transf., 30(2), pp. 1291-
1296, (1987). Journal Paper
[2] Deb S., Yao S.C., “Analysis on film boiling heat transfer of impacting sprays,” Int. J. Heat Mass Transf., 32(11), pp. 2099-
2112, (1987). Journal Paper
[3] Al-Ahmadi H.M., Yao, S.C., “Spray cooling of high temperature metals using high mass flux industrial nozzles,” Exp. Heat
Transf., 21, pp. 38-51, (2008). Journal Paper
[4] Cox T., Heat transfer experiments and general correlation for sprays of very large droplets, Ph.D. Thesis, Carnegie Mellon
University, (1998). Dissertation
[5] Gaugler R.E., Experimental investigation of spray cooling of high temperature surfaces, Ph.D. Thesis, Carnegie Mellon
University, (1966). Dissertation
[6] Hoogendoorn C.J., Valentine W.S. “Leidenfrost temperature and heat transfer coefficients for water sprays impinging o,n a hot
surface,” Proc. of 5th Int. Heat Trans. Conf., vol. 4, pp. 135–138, (1974). Conference Proceedings
[7] Cox T.L., Yao S.C., “Heat transfer of sprays of large water drops impinging on high temperature surfaces,” J. Heat Transf.,
121(2), pp. 446-450, (1999). Journal Paper
[8] Jia W., Qiu H.H. “Experimental investigation of droplet dynamics and heat transfer in spray cooling,” Exp. Thermal and Fluids
Sc., 27, pp. 829-838(2003). Journal Paper
[9] Chen R.H., Chow L.C., Navedo J.E., “Effects of sprays characteristics on critical heat flux in subcooled water sprays cooling,”
Int. J. Heat Mass Transf., 45, pp. 4033-4043, (2002). Journal Paper
[10] Gradeck M., Ouattara J.A., Rémy B, Maillet D., “Solution of an inverse problem in the hankel space – infrared thermography
applied to estimation of a transient cooling flux,” Exp. Thermal and Fluids Sc., 36, pp. 56-64(2012). Journal Paper
[11] Dunand P., Castanet G., Gradeck M., Lemoine F. and Maillet D., “Heat transfer of droplets impinging onto a wall above the
leidenfrost temperature,” Compte Rendus Mécanique., 341 (1-2), pp. 75-87 (2013). Journal Paper
[12] Puschmann F., Specht E. and Schmidt J., “Measurements of sprays cooling heat transfer using an infrared technique in
combination with the Phase Doppler technique and a patternator,” Int. J. Heat and Tech. ., 19, pp. 51-56, (2001). Journal
Paper
[13] Ouattara A., Gradeck M., Maillet D., Remy B., Lebouché M. “Solution of IHCP estimation of a nonuniform source using an
inverse heat conduction method based on an analytical solution,” Proc. of Eurotherm., (2008). Conference Proceedings
[14] Gradeck M., Ouattara A., Maillet D., Gardin P., Lebouché M. “Heat transfer associated to a hot surface quenched by a jet oil in
water emulsion,” Exp. Thermal and Fluid Sc., 35 (5), pp. 841-847 (2011). Journal Paper
[15] Maillet D., André S., Barsale J., Degiovanni A., Moyne C., Thermal Quadrupoles - Solving the heat equation through integral
transforms, Chisester: Wiley, (2000). Edited Book
[16] Yao S. C., Cox T.L. “A general heat transfer correlation for impacting water sprays on high-temperature surfaces,” Exp. Heat
Transf., 15, pp. 207-219, (2002). Journal Paper

15