Date post: | 09-Dec-2023 |
Category: |
Documents |
Upload: | independent |
View: | 0 times |
Download: | 0 times |
Global Rainbow Thermometry for Mean Temperatureand Size Measurement of Spray Droplets
Jeronimus P.A.J. van Beeck*, Laurent Zimmer**, Michel L. Riethmuller*
(Received: 23 April 2001; resubmitted: 13 July 2001)
Abstract
Global rainbow thermometry is a new technique for mea-
suring the average size and temperature of spray droplets.
For data inversion a global rainbow pattern is employed,
which is formed by constructive interference of laser light
scattered by an ensemble of spherical droplets. The non-
spherical droplets and liquid ligaments provide a uniform
background and hence do not influence the interference
pattern from which average size and temperature are
derived. This is a large improvement with respect to stan-
dard rainbow thermometry, investigated since 1988, which
is strongly influenced by particle shape. Moreover, the
technique is applicable to smaller droplets than the standard
technique because the global pattern is not spoiled by a
ripple structure. Data inversion schemes based on inflection
points, minima and maxima are discussed with respect to
spray dispersion and droplet flux. The temperature deriva-
tion from inflection points appears to be independent of
spray dispersion. Preliminary measurements in a heated
water spray are reported. The mean diameter obtained from
the rainbow pattern is smaller than the arithmetic mean
diameter measured by phase-Doppler anemometry. The
accuracy of the temperature measurement by global rain-
bow thermometry is shown to be a few degrees Celsius.
1 Introduction
Rainbow thermometry has been investigated since 1988
[1–3]. Up to now, the rainbow technique has measured the
temperature and size of individual droplets. Even though
the technique has been applied in the field of spray com-
bustion by Sankar et al. [4], confidence could not be
established because of major problems, related to the tem-
perature gradient inside the droplet [5–7], droplet non-
sphericity [8] and a ripple structure that strongly perturbs
the rainbow interference pattern, from which one deduces
the droplet parameters [5]. The last two problems are solved
by global rainbow thermometry. Especially the fact that
there is a solution to the non-sphericity problem is an
important innovation, because until now a droplet non-
sphericity of 1% could lead to an error in the temperature
measurement of � 40 �C. In recent years, several algo-
rithms for the detection of spherical droplets were proposed
but more of them was sufficiently general [4, 9].
2 Principle of Global Rainbow Thermometry
Global rainbow thermometry (GRT) primarily aims at
eliminating the non-sphericity effect that has existed in the
rainbow technique since its introduction in 1988. The basic
principles of GRT were published by the present authors in
1999 [10].
Figure 1 shows a photograph of the set-up. A CW argon-ion
laser beam illuminates a water spray. The typical laser
power for this spray is 100 mW and the beam is expanded to
a thickness of 15 mm. A transparent piece of paper is
installed in the focal plane of the large-diameter receiving
lens. All the droplets crossing the laser beam will contribute
to the angular scattered light distribution, visible on the
transparent screen. This results in the so-called global
rainbow pattern that is recorded from the other side of the
transparent screen by a digital video camera. The use of a
transparent screen makes the alignment of the video camera
a minor issue. To calibrate the magnification factor of the
camera, one records a millimeter paper that is attached to
* Dr. ir. J. P. A. J. van Beeck, Prof. M. L. Riethmuller, von Karman
Institute for Fluid Dynamics, Chaussee de Waterloo 72, B-1640
Rhode-Saint-Genese (Belgium).
E-mail: [email protected]
** Dr. L. Zimmer, National Aerospace Laboratory, CFD Technology
Center, 7-44-1 Jindaiji-Higashi, Chofu, Tokyo 182-8522 (Japan).
# WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2001 0934-0866/01/0412/0196 $17.50þ:50=0
196 Part. Part. Syst. Charact. 18 (2001) 196–204
the screen. Subsequently, the relationship between pixel
number and the scattering angle is found via the focal
length of the receiving lens.
A typical rainbow interference pattern is shown in Figure
2a. It is recorded when the water spray works at stable
ambient temperature. Horizontal profiles at different
instants of time are plotted in Figure 3. Note the stability of
the so-called Airy fringes, from which all droplet char-
acteristics are derived. Although, most probably, numerous
non-spherical droplets cross the laser beam, the Airy fringes
do not move. This implies that the pattern is formed by
constructive interference of the spherical droplets, because
for each of them the rainbow position is identical.
Destructive interference occurs for the non-spherical drop-
lets and liquid ligaments, because their rainbow patterns are
randomly oriented and thus yield a uniform background.
Consequently, there is no need for complex selection cri-
teria to look for rainbow patterns from spherical droplets.
The selection of spherical droplets is done automatically,
just like the rainbow in the sky, which is a static phenom-
enon, even though numerous raindrops are not spherical.
Apart from the fact that the global rainbow pattern (Figure
2a) is formed by spherical droplets only, it is interesting
that the visibility of this pattern is reduced compared with
that formed by a single droplet (Figure 2b). A ripple
structure such as that in Figure 2b cannot be observed at
all in the global rainbow pattern in Figure 2a. This ripple
structure contains no temperature information and there-
fore can only deteriorate the accuracy of the temperature
measurement; for a single fuel droplet smaller than 30mm,
the temperature uncertainty exceeds � 6 �C using standard
rainbow thermometry [5]. Therefore, the fact that the
ripple structure does not appear in the global rainbow
pattern is very favorable for an accurate detection of the
Airy maxima y1 and y2, from which droplet size and
temperature are derived (Figure 3). The disappearance of
the ripple structure has been discussed by Roth et al.
[11, 12], who studied the mean scattering diagram of a
polydispersed burning droplet stream. A droplet size varia-
tion of less than 1 mm made the ripple structure vanish
completely. However, the authors made no comments on
the natural selection of spherical droplets by means of this
mean interference pattern.
Fig. 1: Photograph of the set-up for global rainbow thermometry showingthe water spray (1), receiving lens (2), transparent screen (3) and videocamera (4). The laser beam was added during post-processing of thephotograph and does not represent the real thickness. a)
b)Fig. 2: (a) A typical global rainbow pattern in a water spray recorded by avideo camera. Only the so-called Airy fringes are visible. The horizontalaxis is proportional to the scattering angle. (b): A typical rainbow patterncoming from a single droplet in a water spray. Note the high-frequencyripple structure superimposed on the low-frequency Airy fringes.
Part. Part. Syst. Charact. 18 (2001) 196–204 197
For the patterns in Figure 3, the mean diameter is deter-
mined to be 49� 3 mm. This is calculated from the fringe
spacing y27 y1 as if the rainbow were formed by a single
droplet. Phase-Doppler anemometry measures an arithmetic
mean diameter of 60mm and a Sauter mean diameter of
130mm. From this, the question arises of what mean dia-
meter is deduced from the global rainbow pattern. This is
answered below.
3 Numerical Simulations of the GlobalRainbow Pattern
To invert the global rainbow pattern to mean droplet size
and temperature, data-inversion algorithms have to be
found, which will be based on simulations of the global
rainbow pattern. For these simulations, we assume a log-
normal droplet-size distribution at uniform temperature.
The density function f (d ) is given by
f ðdÞ ¼1
sdffiffiffiffiffiffi2p
p � e12
lnðd=dÞs
� �2
: ð1Þ
The integral from zero to infinity over f (d ) yields unity. d is
a parameter representing the mean diameter of the spray and
s is the dispersion around it. Because there is a finite number
of droplets, Ntot, the distribution has to be discretized. With
di Ddi=25d5di þ Ddi=2 the diameter range containing
one droplet with diameter di, it follows that
f ðdiÞDdi ¼ 1=Ntot: ð2Þ
Subsequently, the arithmetic mean diameter D10 results in
the following simple expression:
D10 ¼XNtot
i¼1
di f ðdiÞDdi ¼1
Ntot
XNtot
i¼1
di: ð3Þ
A similar simple expression is found for the commonly
used Sauter mean diameter D32:
D32 ¼XNtot
i¼1
d3i =
XNtot
i¼1
d2i : ð4Þ
The influence of Ntot, s and d on the global rainbow pattern
is investigated. Figure 4a and b depict droplet-size dis-
tributions according to Eq. (1) for s¼ 0.2 and 0.5,
respectively. Note that for s¼ 0.5 the distribution is less
symmetric and ranges from 35 to 280mm.
To calculate the scattering diagram of the global rainbow
pattern, the scattered light intensity of each single droplet is
summed, thereby neglecting optical interference between
different droplets and assuming perfect sphericity. The
rainbow pattern for a single droplet is represented by the
normalized Airy function Ai(x,d); here, x¼ y7 yrg, where
yrg is the scattering angle corresponding to the geometrical
rainbow angle, which depends only on the droplet’s
refractive index, and hence temperature [13]. Using the
Airy function implies that only the Airy fringes and no
ripple structure is taken into account. The mean scattered-
light intensity distribution Rnbw(x) then becomes
RnbwðxÞ /XNtot
i¼1
Aiðx; diÞ2� d
7=3i : ð5Þ
The normalized Airy function accounts for the angular
variation of the scattered-light intensity, whereas the factor
di7/3 takes into account the dependence of this intensity on
the droplet diameter within the primary rainbow region
[14]. Remember that the effect of droplet temperature and
laser-light wavelength l on the rainbow pattern lies in the
geometrical rainbow angle yrg (and therefore x). l is set to
514.5 nm for all the simulations in this paper.
Figure 5 shows three global rainbow patterns formed by an
ensemble of water droplets at 40, 60 and 80 �C. A tem-
perature difference of 40 �C leads to an angular displace-
ment of the pattern of 1.2�. From this, an appreciation of the
accuracy of temperature measurement can be discerned.
Even though the simulation is performed for a specific
spray, one can understand that the application of the rain-
bow technique lies in fields where the droplet temperature
varies at least over a few decades of degrees Celsius.
Figure 6 shows normalized global rainbow patterns for
s¼ 0.2 and three different mean diameters d (Eq. (1)). The
total number of droplets Ntot is fixed at 100, which appears
to be sufficient to reach a converged solution. Note the
perfect similarity between the curves. Because for smaller
droplet diameters the rainbow pattern broadens, angular
displacements are more difficult to detect. Hence the
accuracy of the temperature measurement degrades for
smaller droplet diameters.
Fig. 3: Profiles of experimental global rainbow pattern at four differentinstants in time. Note the stability of the pattern. The scattering angle isnverse proportional to the pixel number.
198 Part. Part. Syst. Charact. 18 (2001) 196–204
Figure 7 depicts global rainbow simulations for constant d,
but with different dispersions s. In contrast to Figures 5 and
6, the signals are not at all similar. When s increases, the
principal rainbow maximum tends to move towards the
geometrical rainbow angle yrg at x¼ 0. Moreover, the width
of the principal maximum decreases, which makes its
detection more accurate. However, the visibility of the other
Airy fringes diminishes for higher spray dispersion, which
can also be observed in the experimental signal in Figure 3.
Eq. (5) will be the model used for the creation of the
data-inversion schemes (Section 4). However, one must
realize that the agreement between the model and the
experimental rainbow pattern is not perfect. This could be
due to the following facts:
� The distance between a droplet and the receiving lens is
not constant. From Figure 1, one understands that this
distance depends on the position of the droplet in the
laser beam. This effect could be eliminated by adding a
spatial filter to the receiving optics.
� The contribution of nearly spherical droplets is not
taken into account. It is likely that these droplets fill the
‘‘valleys’’ between the Airy maxima, thus reducing the
fringe visibility.
4 Data Inversion Algorithms For GlobalRainbow Thermometry
Based on the simulations of the global rainbow pattern
(Section 3), one can evaluate what kind of mean diameter
and temperature are obtained from the global rainbow. The
aim is to look for data-inversion algorithms that are inde-
pendent of the spray-dispersion paremeter s (Eq. (1)).
Three different schemes are studied. They vary in the
information taken from the signal: this can be the Airy
maxima y1 and y2, the minimum ymin between both max-
ima, or the two inflection points yinf 1 and yinf 2 around y1
(see Figure 6).
The first algorithm that is discussed is the one most often
applied when processing a rainbow coming from a single
droplet. The droplet diameter, DAiry , is derived from the
distance between the first two Airy maxima, y1 and y2,
using the Airy theory for the rainbow [3–5, 8, 13]:
DAiry ¼ 1016:2lðy2 y1Þ3=2; ð6Þ
which is valid for a refractive index m¼ 4/3. Once DAiry is
known, the temperature is calculated from the geometrical
rainbow angle, deduced from y1 and DAiry [3, 8, 13]:
yrg ¼ y1 46:18ðl=DAiryÞ2=3; ð7Þ
where the second term on the right-hand side is valid for
m¼ 4/3. The relationship between yrg and temperature is
shown in Figure 5.
To evaluate the first algorithm in a polydisperse spray, Eqs.
(6) and (7) were applied to Eq. (5), which is the model for
the global rainbow pattern. Figure 8a shows the resulting
DAiry as a function of s for d¼ 100 mm and different
amounts of droplets. Up to s¼ 0.3, DAiry equals the Sauter
mean diameter D32. For higher s, the Airy diameter
depends on the total number of droplets forming the global
rainbow. This number is unknown in the experiments,
hence Eq. (6) becomes useless for s > 0.3. Figure 8b depicts
the results of Eq. (7) for different d with Ntot¼ 200. The
deviation of yrg from its value at s¼ 0 reaches 0.4� for
s¼ 0.3 and d¼ 20mm, which corresponds to an uncertainty
in the temperature measurement of about � 16 �C for water.
Consequently, the algorithm employed for a single droplet
(standard rainbow thermometry) cannot be applied for a
polydisperse spray to calculate the mean droplet size and
temperature, especially when the mean droplet diameter is
smaller than 100 mm. The problem with this algorithm is
Fig. 4: Droplet size distribution for (a) s¼ 0.2 and d¼ 100 mm and (b) s¼ 0.5 and d¼ 100 mm.
Part. Part. Syst. Charact. 18 (2001) 196–204 199
related to the fact that the secondary Airy maximum y2
disappears for large s (Figure 7).
The second algorithm employs the maximum of the prin-
cipal Airy fringe y1 and the first minimum ymin (Figure 6).
From these parameters one can deduce DAiry and yrg using
the Airy theory for the rainbow:
DAiry ¼ 462:6lðymin y1Þ3=2; ð8Þ
yrg ¼ y1 46:18ðl=DAiryÞ2=3: ð9Þ
Figure 9a and b depict DAiry and Dyrg, respectively. There is
no improvement in the temperature measurement (Figure
9b) with respect to the algorithm based on y1 and y2.
However, concerning the droplet-size measurement (Figure
9a), DAiry follows D32 up to s¼ 0.25. For 0.25< s< 0.4,
DAiry is closer to the arithmetic mean diameter D10. For
even larger s, DAiry depends weakly on Ntot. The same
behavior is found for other d, because the rainbow signals
are similar (Figure 6).
The most interesting results are obtained for the data-
inversion algorithm based on the inflection points around
y1, i.e. yinf 1 and yinf 2 (Figure 6). The relationship between
these points and DAiry and yrg is
DAiry ¼ 531:6lðyinf 2 yinf 1Þ3=2; ð10Þ
yrg ¼ yinf 1 13:91ðl=DAiryÞ2=3: ð11Þ
Roth et al. [15] noted that yinf 1 lies very close to the geo-
metrical rainbow angle yrg, thus being a good indicator for
the droplet temperature. Figure 10a shows that DAiry is
independent of Ntot up to s¼ 0.5, but it exceeds the Sauter
mean diameter D32 considerably. This is related to the
decrease in the width of the principal rainbow maximum
when s increases (Figure 7). But look at Figure 10b! Dyrg is
never larger than 0.025�, which means that this algorithm
ensures an accuracy in the temperature measurement of less
than � 1�C.
Finally, it is interesting to mention that the spray-dispersion
parameter s correlates with the ratio of DAiry from the
inflection points to that derived from y1 and ymin, as is seen
in Figure 11. The resulting curve can therefore be used to
Fig. 5: Simulations of global rainbow patterns for different mean watertemperatures and fixed dispersion s.
Fig. 6: Simulations of global rainbow patterns for different meandiameters at constant dispersion s. x¼ y7 yrg.
Fig. 7: Simulations of global rainbow patterns for different dispersions atconstant mean diameter d. x¼ y7 yrg.
200 Part. Part. Syst. Charact. 18 (2001) 196–204
determine the width of the droplet size distribution.
Unfortunately, for s> 0.4, the relationship depends also on
the total number of droplets Ntot.
5 Experimental Results
The above data inversion schemes were applied to the
experimental signals in Figure 3. DAiry , based on the first
Airy maximum and minimum (Eq. (8)), yields 48mm, but
from the inflection points (Eq. (10)) a mean droplet diameter
of 56mm is obtained. The precision in these diameters is
about �5 mm, hence the curve in Figure 11 yields s between
0.2 and 0.3. For this spray dispersion, the diameter 48mm
should be close to D32, but phase-Doppler anemometry
(PDA) gives 130mm. This large discrepancy could be
explained by the fact that the rainbow method is much more
sensitive than PDA to droplet shape (see Tropea [16]). PDA
measures a mean Sauter diameter of the spherical and
spheroidal droplets. On the other hand, the global rainbow
pattern is formed by the most spherical droplets only, hence
the smaller ones. This leads to a mean Sauter diameter for
the rainbow technique that is smaller to that of PDA.
All simulations in the previous section were made at con-
stant temperature and the whole picture could change when
the global rainbow is formed by droplets with different
temperatures. Nevertheless, it is interesting to see what the
technique gives so far. Temperature measurements are
presented in Figure 12, which shows the droplet tempera-
ture in the water spray of Figure 1 at a distance of 10 cm
from the spray nozzle. The initial nozzle temperature was
monitored by a thermocouple and was varied between 19
and 68 �C. The ambient temperature was 21 �C. The method
of the inflection points was used (Eq. (11)). Each rainbow
temperature was obtained after averaging 50 video images,
taken at different instants of time, which ensured con-
vergence of the mean temperature measurement with
respect to the number of droplets forming the rainbow.
Although the precision seems to be a few degrees Celsius,
the tendency is correct. The spray droplets cool.
6 Conclusions
Experimental and numerical results on global rainbow
thermometry have been reported. Low CW laser power is
needed to create an average, so-called global, rainbow
pattern. This pattern is formed by all the spherical droplets
that cross the laser beam during the integration time of the
video camera. The non-spherical particles are supposed to
create a uniform background. Because the selection of
spherical droplets happens naturally, the destructive influ-
ence of droplet shape on the accuracy of the temperature
measurement is overcome. Moreover, the pattern is smooth
and not spoiled by an additional ripple structure that would
deteriorate the temperature derivation from the global
rainbow pattern.
Simulations of the global rainbow pattern were performed
using the Airy function and a log-normal distribution for
the spray droplet diameters. All droplets are assumed to be
spherical, at the same temperature and at the same distance
from the detector. The simulations show that there exists a
similitude for global rainbow patterns related to different
mean droplet diameters. However, with respect to spray
dispersion, the similitude does not hold because signal
visibility varies significantly. An important result is that
the principal rainbow maximum, from which the mean
temperature is derived, becomes narrower for increasing
width of the droplet size distribution. This renders the
detection of the rainbow displacement, and therefore tem-
Fig. 8: (a) The Airy diameter, calculated from y27 y1, as a function of the spray dispersion s for various amounts of droplets Ntot (Eq. (6)). (b) Thedeviation of yrg, based on the first two Airy maxima y1 and y2, from its value at s¼ 0, as a function of s and different mean diameters d (Eq. (7)).
Part. Part. Syst. Charact. 18 (2001) 196–204 201
perature measurement, more accurate. From the simulations
it follows also that the larger the spray dispersion, the more
droplets are needed to reach a stable signal.
Three data-inversion algorithms were investigated. The
standard algorithm, based on the two important interference
maxima, fails for a polydisperse spray because the lower
maximum vanishes. The second algorithm derives the mean
droplet size and temperature from the first maximum and
first minimum. A mean diameter is obtained somewhere
between the arithmetic and Sauter mean diameter, but the
temperature measurement is not reliable. A correct tem-
perature is derived from the third algorithm, employing the
inflection points around the principal rainbow maximum.
For this algorithm, the accuracy is always less than � 1 �C.
The mean droplet diameter obtained from this method is
much larger than the Sauter diameter. That is why the ratio
of this diameter to that derived from the first maximum and
minimum can give an idea about the spray dispersion.
The resemblance between experimental rainbow patterns
and the simulations is not perfect as far as relative peak
intensities are concerned, and this should be studied further.
One reason could be the influence of nearly spherical drop-
lets, which is neglected in the simulations. Nevertheless,
rainbow temperatures have been measured with success in a
heated water spray. The precision is a few degrees Celsius.
The mean droplet diameter, measured by the rainbow
Fig. 9: (a) The Airy diameter, calculated from y17 ymin, as a function ofthe spray dispersion s for various numbers of droplets Ntot (Eq. (8)). (b)The deviation of yrg, based on y1 and ymin, from its value at s¼ 0, as afunction of s for different mean diameters d (Eq. (9)).
Fig. 10: (a) The Airy diameter, calculated from yinf 27 yinf 1, as a functionof the spray dispersion s for various number of droplets (Eq. (10)). (b)The deviation of yrg, based on yinf 2 and yinf 1, from its value at s¼ 0, as afunction of s for different mean diameters d (Eq. (11)).
202 Part. Part. Syst. Charact. 18 (2001) 196–204
method, appears to be considerably smaller than that
obtained by phase-Doppler anemometry. This is probably
because global rainbow thermometry only sees the smaller,
most spherical droplets.
7 Acknowledgments
The authors thank M. Santonastasi and J. Sanz for their
contributions to this paper.
8 Symbols and Abbreviations
Ai Airy function
d droplet diameter
Ddi droplet diameter range within discretized droplet
size distribution
D10 arithmetic mean droplet diameter
D32 Sauter mean droplet diameter
DAiry mean droplet diameter deduced from the rainbow
pattern
d mean droplet diameter used in f (d)
f (d ) density function for droplet size distribution
GRT global rainbow thermometry
y scattering angle
yrg geometrical rainbow angle
Dyrg deviation of yrg with respect to yrg at s¼ 0
y1 main maximum in the rainbow pattern
y2 secondary maximum in the rainbow pattern
ymin minimum between y1and y2
yinf 1 inflection point between yrg and y1
yinf 2 inflection point between y1 and ymin
PDA phase-Doppler anemometry
Rnbw mean scattered-light intensity as a function of x
x y7 yrg
Ntot, n total number of droplets
l wavelength of light
s dispersion parameter in f (d )
Subscripts
Airy Airy theory
i droplet-diameter class index
inf inflection point
min minimum
rg rainbow according to geometrical optics
tot total
9 References
[1] N. Roth, K. Anders, A. Frohn: Simultaneous Measurement ofTemperature and Size of Droplets in the Micrometer Range. Proc.7th International Congress on Optical Methods in Flow and ParticleDiagnostics ICALEO 88, L.I.A., Vol. 67, Sunnyvale, CA 1988,294–304.
[2] S. V. Sankar, K. H. Ibrahim, D. H. Buermann, M. J. Fidrich, W. D.Bachalo: An Integrated Phase Doppler/Rainbow RefractometerSystem for Simultaneous Measurement of Droplet Size, Velocity,and Refractive Index. Proc. PARTEC 3rd Int. Congress on OpticalParticle Sizing, Yohokama, Japan 1993, 275–284.
[3] J. P. A. J. van Beeck, M. L. Riethmuller: Simultaneous Determinationof Temperature and Size of Droplets from the Rainbow Using AiryTheory. Proc. 7th Int. Symp. on Applications of Laser Techniques toFluid Mechanics, Vol. 2, pp. 21.5.1–21.5.6, Lisbon, Portugal; alsoin R. J. Adrian et al. (eds): Developments in Laser Techniques andApplications to Fluid Mechanics, Springer, Berlin 1994,330–339.
[4] S. V. Sankar, D. H. Buermann, W. D. Bachalo: An AdvancedRainbow Signal Processor for Improved Accuracy in DropletTemperature Measurements. Proc. 8th Int. Symp. on Applications ofLaser Techniques to Fluid Mechanics, Vol. 1, Lisbon, Portugal1996, 9.3.1–9.3.9.
[5] P. Massoli: Rainbow Refractometry Applied to Radially Inhomo-geneous Spheres: The Critical Case of Evaporating Droplets. Appl.Opt. 37 (1998) 3227–3234.
Fig. 11: The ratio of the droplet diameter derived from the inflectionpoints, to that derived from the first Airy maximum and minimum, plottedas a function of the non-dimensional width of the droplet sizedistribution, s.
Fig. 12: Rainbow temperature in a water spray at 10 cm from the injec-tion point as a function of the injection point temperature. Ambienttemperature, 21 �C.
Part. Part. Syst. Charact. 18 (2001) 196–204 203
[6] K. Anders, N. Roth, F. Frohn: Theoretical and Experimental Studiesof the Influence of Internal Temperature Gradients on RainbowRefractometry. Proc. PARTEC 4th Int. Congress on Optical ParticleSizing, Nuernberg, Germany 1995, 419–428.
[7] F. Corbin, A. Garo, G. Gouesbet, G. Grehan: Refractometried’arc-en-ciel: application au diagnostic des gouttes avec gradientd’indice, in Recueil des Actes du 5e Congres Francophone deVelocimetrie Laser, Rouen, France 1996, E1.1–E1.8. Universite etInstitut des Sciences Appliquees de Rouen, URA CNRS 230/CORIA.
[8] J. P. A. J. van Beeck, M. L. Riethmuller: Nonintrusive Measurementsof Temperature and Size of Single Falling Raindrops. Appl. Opt. 34(1995) 1633–1639.
[9] J. P. A. J. van Beeck, M. L. Riethmuller: Rainbow Pheno-mena Applied to the Measurement of Droplet Size and Velocityand to the Detection of Nonsphericity. Appl. Opt. 35 (1996) 2259–2266.
[10] J. P. A. J. van Beeck, D. Giannoulis, L. Zimmer, M. L. Riethmuller:Global Rainbow Thermometry for Droplet-Temperature Measure-ment. Opt. Lett. 24 (1999) 1696–1698.
[11] N. Roth, K. Anders, A. Frohn: Refractive-Index Measurements for theCorrectionofParticleSizingMethods.Appl.Opt.30(1991)4960–4965.
[12] N. Roth, K. Anders, A. Frohn: Simultaneous Determination ofRefractive Index and Droplet Size Using Mie Theory. Proc. 6th Int.Symp. on Applications of Laser Techniques to Fluid Mechanics,Vol. 1, Lisbon, Portugal 1992, 15.5.1–15.5.5.
[13] J. P. A. J. van Beeck: Rainbow Phenomena: Development of ALaser-Based, Non-intrusive Technique for Measuring Droplet Size,Temperature and Velocity. PhD Thesis, Eindhoven University ofTechnology, ISBN 90-386-0557-9, 1997.
[14] H. C. van de Hulst: Light Scattering by Small Particles. DoverPublications, New York 1981, 243–246.
[15] N. Roth, K. Anders, A. Frohn: Size Insensitive Rainbow Refracto-metry: Theoretical Aspects. Proc. 8th Int. Symp. on Applications ofLaser Techniques to Fluid Mechanics, Vol. 1, Lisbon, Portugal1996, 9.2.1–9.2.6.
[16] C. Tropea: Developments of Specialized Phase-Doppler Techniquesand Refractive Index Measurements. Lecture Series on OpticalDiagnostics of Particles and Droplets, von Karman Institute, Sint-Genesius-Rode, Belgium 1999.
204 Part. Part. Syst. Charact. 18 (2001) 196–204