You are on page 1of 8



Kurt Lainer and Hans M. Tensi

Technical University of Munich


It is well known that optical techniques are

widely used in the measurement of fluid flow and
heat transfer. A laser shadow technique for the
investigation of vapor film thickness S during
immersion cooling of cylindrical probes is intro
duced. Normally only qualitative information's
can be obtained from the conventional shadow
graph technique. Therefore it is necessary to
make some efforts to get quantitative results.
The main reason for measuring the vapor film
thickness is to get more information's about the
heat transfer and cooling rate during the film
boiling stage. It is demonstrated that the well
known shadowgraph method can be combined
with modem data processing units to give quan
titative results about the film thickness S of verti
cally submerged probes during cooling in aque
ous polymer solutions.
During quenching in liquid media the three sta
ges of heat removal occur:
the film boiling or vapor blanket stage,
the nucleate boiling stage, and
the convective heat transfer.

During film boiling stage the surface tempera

ture is such high that the liquid quenchant is
vaporized and a stable film of vapor is formed
around the part. This vapor film has an insu
lation effect so that the cooling rate during film
boiling is relatively low. Below the so called Lei
denfrost-temperature the vapor film collapses
and the heat flux arises by partial film boiling
and nucleate boiling. When the surface tempera
ture drops below the boiling temperature of the
liquid, the surface is permanently wetted by the
fluid. Now the cooling rate is low and mainly
determined by the rate of convection and by the
viscosity of the liquid quenchant.
Theoretical investigations of the heat transfer
during stable film boiling are based on the theo
ry of the boundary layer. The solution of the
differential equations requires simplIfications
and assumptions for the boundary conditions
which leads to highly different results for the
vapor film thickness.
The deflection of a thin laser beam by a liquid
phase object is a measure of the density and tem
perature of the fluid. This deflection is proportio
nal to the temperature gradient in the thermal
boundary layer ahead of the vapor film. Thus it
is possible to make an optical determination of

the temperature in the boundary layer where;

temperature measurements by conventional
techniques, such as thermocouples, are out of
question. The deflection of a laser beam which
touch the vapor/liquid interface is therefore of
particular interest for determination the vapor
film thickness. A particularly attractive feature
of the shadowgraph technique is that they are
nonintrusive, i.e. they do not influence the i
thermal boundary layer in a noticeable manner. ,
This technique is of course only applicable if :
density variations exists in the fluid because i
otherwise no shadowgraphs can be obtained.
dowgraph system the linear displacement of the
perturbed light beam Is measured. The shadow

graph image can be understood by referring to

Fig. 1 in which a parallel light beam enters a

nonuniform test section. To Simplify the deriva

tion, variations of refractive index are assumed

to exist only in the y direction. At the exit of the

: test section the beam is not usually parallel

having been deflected by an angle y which is a

, function ofy.

dimensional system where there is no variation

i of index of refraction in the x direction. The

standard shadowgraph method is used for quan

titative density measurements only if the mea
sured contrast is accurately enough.

A simple modification of the shadowgraph,!

called the "Schmidt"-schlieren system [1,2,3,4]
has been used in heat transfer studies to obtain
surface heat fluxes. A deflection that is proportional to the index of refraction gradient at the
solid-liquid interface is measured. From this
shadowgraph picture the wall temperature
, gradient, and thus the surface heat flux, can be
obtained [1,4,5]. The light path in Fig. 1 enters
the test region parallel to the heated test surface.
There is no variation of properties in the x
direction and therefore only deflections in the y
x plane are considered. The path of the light
beam for small angular deflections is described

The integration of equation (1) gives the value of

the slope of a light beam at the exit of the test

=! On LTS


Integration of equation (2) gives the value of Yl at

this position for a ray that entered at yo.
1 On L2
Y1 - Yo = - - --I!.
n Oy 2



---to- +-+--""'~


With these integration's we have assumed that

! (:)

is constant along the light beam even

though the y position changes slightly. This is

; valid only if

Fig. 1 - Schematic of the basic shadowgraph system


The shadowgraph, like the schlieren and inter

ferometer methods, is best suited in a two-



is small [20]. If no other varia-

tion in index of refraction is present, after

leaving the test section the ray travels in a
straight line to the screen. The beam height at
the screen is given by:



1 On L2

Yse -Y.

o 8y

+ 1:.... On LTS (Lsc _ LTS)

o. 8y
and if Lrs
Yse - Yo





~. L TS Lse


l1l:I -





In most thermal boundary layers the maximum

I value of the temperature gradient, and thus the !

! maximum of the index of refraction gradient, is
at the fluid-solid interface. With heat flow from
the solid to the fluid, the light will normally be i
bent away from the surface, and the ray that
. passes just adjacent to the surface will undergo
maximum bending. If the screen is placed far ;
enough away from the test section, this ray will
also have the maximum deflection Ysc,mu:' Since

the value of



usually small or .
even zero near the surface, a significant light .
bundle adjacent to the surface will be deflected,
this maximum amount, producing a bright con- .
tour on the screen whose position can be mea- :
sured fairly accurately.


- - 2 is

8y w = Lse LTS

The thermal gradient on the wall surface is:


Oy W


Yse. max

= (dn)





displacement on the screen the zero position for'

y must be known. This must usually determined.!
when no disturbance is in the test section so the!
light beams passing the wall are not deflected.' If '.
a slit is placed in the light beam before the test
section, so that primarily the solid-surface inter- .
face is illuminated, a somewhat sharper image of
. the maximum deflection can be obtained,
although care must be taken to avoid defraction
If a thermal boundary layer is present, all light

within the boundary layer will be deflected at

least to some extent If the distance to the screen
is large, a shadow will appear which is represen
tative of the thickness of the thermal boundary
layer. Since the boundary layer does not have a
finite thickness, this shadow height at the screen
is somewhat a function of the screen position
even for large values of Lsc.
MODIFIED CASE: The shadowgraph method
was adjusted in the case of film thickness mea
surement during immersion cooling,. Not the
maximum light deflection in y-direction Ysc,mu:
was used to determine the vapor film thickness
(s. Fig. 2 and 3). The difference between the
maximum deflected and a reference light beam
in z-direction was used to determine the vapor
film thickness {so Fig. 3). Then the vapor film
thickness l) can be determined by a simple geo
metric relationship.
undef1ected beam
-- deflected beam

The heat flux at the wall in the y direction is


and the dimensionles Nusselt number can be

calculated directly:

Nu = _ Yse.max

Tw -T..,

Lse L TS


" Cooling mediun

boundary layer


It should be noted that for measurement of the

Fig. 2 - Light deflection in the thermal boundary layer of a

cylindrical probe during immersion cooling in avaporable
fluids (top view)

The difference between the maximum deflected

and a reference laser beam in z-direction is used
with a. to determine the vapor film thickness a
according to Fig. 3 in quite an easy manner.
tan a.

The used light source is a 15 mW Helium-Neon

laser with a wavelength of 633 nm. The original
laser beam is reduced by a circular aperture to
1.2 mm in diameter. With a combination of two
cylindrical lenses the beam is spread up to a
band of parallel light beams (40 x 1.2 mm). With
a slit aperture the laser band was reduced to 0.8
mm. The band of parallel laser light was used to
illuminate only a half of the probe during a
stable vapor film exists. The deflection of light
caused by density gradients within the thermal
boundary layer outside the vapor-liquid interface
is projected onto a screen and recorded with a
high resolving CCD-camera. With modem data
processing units it is possible to take about 50
pictures/so The pictures were evaluated with the
use of digital image processing methods after the
experiment was fInished.


= ..------~ -



o = light direction
measurement light band '
(with an angle Q)
reference light band
(with an angle of 90"
to the specimen's axis)
deflected light beam

cooling medium


thickness a is determined on a vertically immer
sed cylindrical probe. The dimension of the spe

boundary layer
vapor-liquid interface
vapor film

Fig. 3 - Light deflection in the thermal boundary layer and

princip of measurement in vertically submerged probes

cimen is 25 mm dia x 100 mm. The experiments

has been carried out with aqueous polymer solu
tions in different concentrations.

EXPERIMENTAL SETUP: The optical arrange

ment of the laser shadow technique is shown in
Fig. 4.
optical bench
mounting system

deflection of a laser beam caused

by density gradients in the ther
mal boundary layer nearby the
vapour-liquid interface
laser beam

(15 mW, 633 om)

filter for contrast


Fig. 4 - Experimental setup of the laser shadow technique for determination the vapor film thickness during immersion cooling (top

1,05 s

34,12 s


10,15 s

34,58 s

13,62 s

35,43 s

19,37 s


37,08 s

28,95 s

33,79 s

39,90 s

43,76 s

.A_ 34. 12 duri ng q uenching a cylindrical CrNi-aUd $pecimen

Fig. S - Wetting process with an Nexplosil18 N.collGpse oftM vapor fl!m aJK>r
(0 2S x 100 mm) in aqueous polymer solutwn (11 = 0 m/s, T B = 30 C)

The wetting process of a cylindrical CrNi-steel

specimen in 5% aqueous polymer solution is
shown in Fig. 5. After immersion of the specimen
the high surface temperature causes a stable va
por film. The stability of the interface vapor
liquid increases with increasing cooling time. A
short time before the vapor film collapsed in
nearly an "explosive" manner the interface
vapor-liquid is extremely stable as can be seen
in Fig. 5. The time interval of wetting .Mw = (tc
tJ is infInitely short and a foam of polymer sur
rounds the specimen. After a short time liquid is
brought into contact with the hot surface which
evaporates immediately and vapor bubbles leave
the surface. Evaporation and strong convection
cause a high heat transfer from the part to the
fluid. When the surface temperature drops below

the boiling temperature of the quenchant, the

specimens surface is pennanently wetted by the
fluid. In this stage the cooling rate is low and
mainly determined by the rate of forced convec
The light deflection recorded during this immer
sion cooling process is shown in Fig. 6. The left
picture of the first row shows an undeflected
light band projected onto the screen. The next
picture was recorded a short time before the
specimen achieved his final quenching position.
The next four pictures shows the light deflection
caused by the thennal boundary layer outside
the vapor-liquid interface during film boiling
stage. The picture recorded 33.79 s after immer
sion beginning shows a very stable interface
vapor-liquid shortly before the process of

three pictures can be used to determine the tem

perature distribution of the laminar thennal
boundary layer as well as the surface tempera
ture of the specimen during convective heat
transfer. The maximum light deflection is a mea
sure for the surface heat flux density and heat
transfer coefficient during convective cooling

wetting occurs. A light deflection caused by the

polymer foam shows the exposure after 48.16 s.
All pictures in the last row shows a light deflec
tion caused by the thennal boundary layer
during convective heat transfer. The surface
temperature of the cylinder 120 s after immer
sion beginning reaches almost bath temperature
so the light deflection is negligible. The last

0.55 s

1.25 s

23.15 s

58.70 s

33.88 s

48.16 s

79.90 s

119.66 s

Fig : ~ - U:ght deflection .projeC:ed onto a screen during immersion roofing for calculation the vapor film thickness with
a d19J.tal Image processing wut

aqueous polymer solution

eEl 0.8



+ - - - t - - - - i - - - i - - - - t - - - - t - - - t - - - - i - - - - j - - - - ; TB 3O-C

v == Om/I


El 0.4 -FII;i,.....,....,













cooling time t [8]

Fig. 7 - VaporjiI.m thicknesB duri"8 imrn.ersion c:ooU"8 in S% aqueous solution d4termiraed b)1 the shadow rn.etluJd (TB - 30 "C,
11 = 0 mls)

According to equation (10) the recorded light de

flections were used to determine the thickness l)
of the vapor film during immersion cooling. The
vapor film thickness over the cooling time t is
shown in Fig. 7. The measurement position was
at half height of the specimen (z = 50 mm). The
vapor film thickness was determined to 0.35 mm
at the beginning of cooling process and increases
during immersion to a value of 0.40 mm shortly
before the film collapsed.

The vapor film thickness l) of a vertically immer

sed cylindrical specimen were experimentally
detennined by the use of an optical method.
Vapor film thicknesses of 0.35 to 0.40 mm were
measured and calculated during immersion
cooling of a cylindrical specimen into an aqueous
polymer solution. It has been shown that the
shadowgraph technique is applicable for
measuring the thickness of a vapor layer during
stable film boiling. In particular it has .been
demonstrated that the vapor film thickness l) can
be obtained by a comparatively simple
experimental setup which uses the light deflec
tion within a thermal boundary layer. The ex
perimentally detennined measurements can be

compared with numerically calculated vapor

film sicknesses from theoretical models based on
the theory of the boundary layer. Further investi
gations will be done to determine the parameters
which affect the stable film boiling stage and the
wetting behaviour as well as the heat transfer
during immersion cooling.

This paper is a partial prepublication of doctoral

thesis by K. Lainer (1996) approved by the Fa
culty of Mechanical Engineering, Technical Uni
versity of Munich.
The authors are grateful to the Deutsche For
schungsgemeinschaft (DFG) Bonn for the finan
cial support of this study (contract number Te


length of the heated test section [mm]

distance between screen and heated
test section [mm]
temperature [K,OC]
wall temperature [OC]
bulk temperature rC]
index of refraction [-]
wetting time [s]



, tr
. Yt




i SC

time interval of wetting [s]

time when wetting starts [s]
time when wetting is finished [s]
beam deflection in y-direction
onto the screen [mm]

maximum beam deflection in y-

direction onto the screen [mm]

distance of the light beam entering

the test section [mm]

distance of the light beam leaving

the test section [mm]

vapor film thickness [mm]

thermal conductivity [WImK]

light deflection after leaving the

test section [0C]

Nusselt number [-]

characteristic length [mm]
cartesian coordinates
test section

[1] Goldstein R.J.: Optical techniques fortem

perature measurements. In: Measurements

in Heat Transfer. Hemisphere, Washington,

DC (1976).

[2] Schmidt E.: Schlierenaufnahmen des Tem

peraturfeldes in der Nahe winneabgeben

der Korper. Forschung. Ing. Wiss., Bd. 3,

Seite 181, 1932

[3] Boelter L.M.K., Cherry V.H.: Measurement

of Heat Transfer by Free Convection from

Cylindrical Bodies by the Schlieren Method.

ASHVE Trans., Vol. 44, p. 499, 1938.

[4] Dong Z.F., Ebadian M.A.: A modified for

mula for calculating the heat transfer coeffi

cient by the shadowgraph technique. Int. J.

Heat Transfer, Vol. 35, No 7, pp 1833-1836,


[5] HaufW., Grigull U., Mayinger F.: Optische

MeJ3verfahren in der Winne- und Stoffiiber

tragung. Springer Verlag Berlin-Heidelberg

New York, 1991


. c



1 ':-;