Reflection suppression by polarization in backscatter LDA measurements near walls and in two-phase...

Reflection suppression by polarization in backscatter LDAmeasurements near walls and in two-phase flows

J. Gardavskj and Rolf Kleine

A method is proposed and verified by experiments to suppress disturbing reflections in laser Doppler ane-mometer (LDA) measurements. Disturbing reflections occur in a variety of practical applications of LDA,such as backscatter measurements near walls or in two-phase fluid or particle velocity measurements. Themethod uses circularly polarized light that changes the handedness when reflected and is quenched by a suit-ably oriented polarization analyzer. Scattered light, however, is partially transmitted by the analyzer togive undisturbed seeding particle signals. Verification experiments have been performed to demonstratethe method; systems for practical applications are proposed.

1. Introduction

The potential of optical techniques to measure flowvelocities has been demonstrated by the large numberof measurements in single- and two-phase flows re-ported during the last decade. Optical techniques allowlocal and time-resolved velocity measurements in alltransparent fluids and have the special advantage overconventional techniques in that no probes are used thatmight disturb the flow under study or that may be de-stroyed or decalibrated. Laser Doppler anemometry(LDA) is among the most advanced optical techniquesto measure flow velocities and is being used in hostileenvironments as well as in systems where conventionalprobe methods fail. Detailed data on complex turbu-lent flows have been obtained, and attempts are beingmade to apply the method in very restricted measuringconditions such as in close proximity to walls, where hotwires would suffer serious heat loss. Preferably LDAmeasurements are made in forwardscattering, becausegood Doppler signals can be obtained with low laserpowers. However, provision must be made for opticalaccess. When the windows are covered with particlesfrom the flow and measurements are made close to thewindow, scattered light from the wall will superpose onthe light scattered by the particles moving through themeasuring volume. Since the wall scattering is not

J. Gardavsky is with Charles University of Prague, Institute ofPhysics, Ke Karlovu 5, 12116 Prague 2, Czechoslovakia, and RolfKleine is with Universitat Karlsruhe, Sonderforschungsbereich 80,postfach 6380, 7500 Karlsruhe, Federal Republic of Germany.

Received 29 December 1981.0003-6935/81/234110-14$00.50/0.© 1981 Optical Society of America.

modulated it increases the dc and the shot noise,thereby reducing the SNR of the Doppler signal.

In many practical situations limited access or otherrestrictions require backscattering measurements. Inthese cases higher laser power does not always com-pensate for the lower scattering powers, which are 2-3orders of magnitude lower than in forwardscattering,and signal quality is therefore comparatively low. Acomplication arises when backscattering measurementsare made near a reflecting or transparent surface, in-terface, etc. Such situations are common in internalcombustion engines or gas turbines when the piston topsurface or a turbine blade approach and cross themeasuring volume. At some critical distance the noiseproduced by the wall reflectance would exceed the acpart of the Doppler signal and very low SNR would re-sult. At an even smaller critical wall distance the dcpart of the signal would force the photomultiplier tosaturation and no signals would be obtained closer tothe wall. These problems are normally found at walldistances of a few millimeters.

To some degree it is possible to suppress wall reflec-tance by means of spatial filtrations This paper de-scribes a new method making use of the scattering andreflection properties of circularly polarized laser light;this allows much higher suppression efficiencies thanspatial filtration. Circularly polarized light is com-monly used in optics to suppress glare from lenses,windows, filters, etc. We are not aware of any reportedattempts to use circularly polarized light for reflectancesuppression in LDA. The method has great potentialnot only for near-wall measurements but for two-phaseflow measurements, where spatial fringes formed by thetwo reflected laser beams are used to measure the ve-locity of big particles in the flow. Normally, when a

4110 APPLIED OPTICS / Vol. 20, No. 23 / 1 December 1981

photomultiplier detects both signals, they have to beseparated electronically to give independent informa-tion on the velocities of both phases. Presently am-plitude discrimination and bandpass filtering are usedto separate such signals. The method described hereallows optical suppression of reflections before detec-tion, enhancing the separation even when the size dif-ference between the particles is small and allowing op-timal use of the photodetectors.

Section II of this paper is an introduction to the in-fluence of wall reflection on the SNR of the single-particle LDA signal. Section III discusses reflectanceon flat metal surfaces and the change of state of polar-ization for linearly and circularly polarized light re-flected at metallic or dielectric surfaces. Also, themethod of polarizational suppression of reflection isexplained, and the suppression ratios and ellipsometricparameters are shown for various surfaces. SectionIII.D discusses the state of polarization of scattered lightfrom seeding particles and shows scattering distribu-tions from particles recorded with and without reflec-tance suppression. In Sec. IV backscatter measure-ments near surfaces of varying reflectance with andwithout reflectance suppression show the influence ofthe method on the SNR of the Doppler signals. SectionV describes an optimal LDA system for near-wallbackscatter measurements in single-phase flows; in Sec.VI the reflectance suppression is visualized photo-graphically in a bubble flow, and optical systems fortwo-phase flow applications are shown.

II. Signal to Noise Ratio of Near-Wall BackscatterLDA Signals

A typical system for backscatter LDA measurementsis shown in Fig. 1. The optical axis is normal to the wall.Two parallel laser beams are crossed and focused in themeasuring volume by the transmission lens. In mostflow situations where backscatter systems must be usedsingle-particle conditions will prevail. The scatteredlight is received through the transmission lens and fo-cused onto the front window of a photomultiplier (PM),which produces Doppler signals whose frequency isproportional to the velocity component parallel to thewall.

The front window of the PM serves as a field stop andmust be matched to the size of the image of the mea-suring volume diameter 2o to receive signals fromparticles not in the measuring volume. However,within a certain angle-of-view , defined as tan = o/F,

the pinhole transmits light originating near the opticalaxis outside the measuring volume. Thus, when themeasuring volume approaches the wall and the pointswhere the beams strike the wall shift closer to the axis,reflected and scattered laser light from the wall will bepartly transmitted through the pinhole. The minimumdistance from the wall, where the points on the wall startto enter the angle-of-view A, can be calculated usinggeometric optics. The minimum distance Ymin is geo-metrically defined by

(1 + ymir/F)uo = ymin tan.0 - az. (1)

With the formulas derived for the propagation of aGaussian beam,2 which is focused at y = 0 [see Fig. 1(b)],the radius of the beam at the distance z is

aZ sz F+z

(costp 1 11 2\F+z F ByI 1

So o F 64 (coso 1 11211+-B I -+-i_L A2 F F B5with A = 7r- 2/X and B = 1 + A2/62 .

For small intersection angles cosp - 1, a

(2)

Ind when

az F+z Z Y- , = 1 + - 1 +-,ICO F F F

and with Eq (1), it holds that2F

Ymin F 1-tan --- -1ao cost

(3)

(4)

For F/uo >> tan2 o-o

Yrin- an (5)tang

With 1 = 2uo/sino as the length of the measuring volumeand sin0 tang0

Ymin" 1, (6)

the critical distance from the wall is equal to the lengthof the measuring volume, an intuitively reasonable re-sult.

Light hitting the photocathode will be transformedinto a photocurrent. For a single particle crossing themeasuring volume at the center, and in the absence ofa wall, the Doppler signal is

I (t) Iex [- -) [ + cos2r tu Ijj

where Io is the maximum photocurrent:

I = (4PL/iry2)(C00/h .) -.7Q

2d

x2sjnp - 1 5

glass 2d opaquewindow ,i , - wall

angle ofview 2p

Fig. 1. Modular LDA system for backscatter measurements.

(7)

(8)

which depends on the intensity at the center 4PL/bIr r,the Mie scattering cross-section C8,, and on the quan-tum efficiency 7Q. Due to the quantum nature of lightthe electron emission process is random, and the signalshows inherent noise that can be calculated with theformula for shot noise:

k 2eIAf.

The SNR of the Doppler signal is then defined as

SNR =I.-= -S 2ef

(9)

(10)

1 December 1981 / Vol. 20, No. 23 / APPLIED OPTICS 4111

Z

0

0.

©C% 'lo-

L)

U,W

E

as

0Cas

0CD0time

Fig. 2. Doppler signal of a single particle at far (a) distance from thewall and (b) small distance and (c) starting saturation at still smaller

distance.

In practice, additional noise from dark current, internalamplification, and from preamplification will sum to ih.Figure 2(A) shows a Doppler signal corresponding to Eq.(7) for a single particle far from the wall. At smallerdistances from the wall distance [Fig. 2(B)] a dc currentby reflectance from the wall adds to the Doppler cur-rent:

I(t) = I, + Is., (11)

and, since noise is proportional to the current, reducesthe SNR to

SNR, = SNR, k1 + 1 (12)

At high reflectance the SNR is approximated by SNRrr

Another important influence on the reflectance is thereduced detectability due to saturation of the last stagesof the phototube [see Fig. 2(C)]. Since the tube is acurrent amplifier fed from the divider chain current,which also sustains the voltages across the dynodes, thelatter will partially collapse at high photocurrents, i.e.,at high currents the Doppler signal sitting on the dc willbe amplified less than at low currents. To some extentthe linearity can be improved by using high chain cur-rent.

It is not shown here but is evident that the SNR de-pends also on the modulation depth of the signal, i.e.,particle size. The measurements in Sec. V are thereforevalid for the given particle size distribution. Earlymeasurements 3 have shown that LDA counters canmeasure with good accuracy at SNR > 8 dB corre-sponding to SNR > 4. More detailed studies 4 withsimulated Doppler signals have led to similar results,showing that better than 1% accuracy for the rms fre-quency and 0.01% for the mean frequency were obtainedat SNR > 4, as shown in Fig. 3.

The scattered light intensity reaching the PM fromthe piercing points of the two laser beams on a windowhas been estimated1 with simplified geometric opticsfor forwardscatter [see Fig. 4(a)]. The only contribu-

10-2

10-4

o0 10-2

- .N Test measurements with

artificial signals*a; ( Nin 3 2 )

I Y:- - -I1 -0-1 I \ 10 SNR

Mean frequency measurementswith artificial signals(Nmin= 32)

10- k

lo-4\

I -I. .. - I D-l I I lO' 102 SNR

Fig. 3. Counting errors for an LDA period-timing system dependenton the SNR.3

glass wall11 J 1 dust deposited

.Z< El /on indows

T Zggg nle or iew

1111| |1 single particle

(a)

/-~pngliew roughwlall

~~partid;4Ib)

Fig. 4. (a) Backscatter measurements near a window, at left. (b)Backscatter measurements near an opaque wall, at right.

tion from the wall was assumed to be scattered lightfrom particles sticking to the wall. With the scatteringcross section of the wall taken as 100 times that of asingle seeding particle, the minimum distance Ymin wascalculated for SNR > 4. For p < 50 the minimum dis-tance was below 0.8 mm and increased strongly up toseveral millimeters for 0 <-4°. Thus, for measurementsunder small angles, i.e., when the beams have to passthrough a small window, 1 mm from the window seemsto be the limit for practical measurements.

The situation is similar but more serious for back-scatter when measurements are made at an opposingopaque wall [Fig. 4(b)]. Light hitting the wall will bepartly scattered but also reflected depending on thesurface properties; reflected light is much more directedand therefore more intense than scattered light.

Ill. Spatial Distribution, Intensity, and State ofPolarization of Reflected Light

A. Spatial Reflectance Distributions

In most practical cases the system walls are metallic.Reflection at a surface can be either diffuse, when re-flection occurs with equal intensity in all directions ofthe hemisphere, or specular, i.e., mirrorlike, with astrong intensity maximum at the reflection angle : =20. Between these cases there are superpositions ofboth cases, which when specular reflection is the dom-inating part is called regular reflectance.


I . . . .. . . _

K

I V

-4 .11.1Avoo�,2)

0 20 40 60 80 100 120 140 160 0 20 40 60 80 :00 t20 140 160p0

0 20 40 60 80 100 120 140 160p

Fig. 5. Intensity distributions of reflected light from a wall dependent on the reflection angle : vs the incident light for varying angles ofincidence 0 (see Ref. 7). From left to right: or = 125,um, 400 m, and 1250 m: X = 0.55 Am, sandblast brass.

The spatial distribution of reflectance strongly de-pends on the surface roughness or. The type of surfacefinish determines whether the reflectance is specularor diffuse and if it is unidirectional or bidirectional.Measurements of spatial distributions are reported inRefs. 5-7. A typical feature of reflectance from metallicsurfaces is the transition from specular distributions atlow roughness to more diffuse patterns at high rough-ness. This is shown in Fig. 5 for sandblast brass 6 withvarying roughness. Regular reflectance patterns arecentered near the reflection angle / = 2 and 0 = 1800.For small angles of incidence 4) and small roughness thespatial distribution is almost symmetrical around thereflection angle. Only at high roughness and greaterangles of incidence is more light reflected in the back-ward ( < 20) than in the forward direction, the reasonbeing that reflection is taking place primarily on theroughness slopes oriented toward the incident beam.Since at high angles of incidence the roughness elementsshade each other and mechanical roughness appearssmall, the quantity correlating reflectance intensitiesand roughness is the optical roughness or cos4. 7 Thegeneral trend is a decrease of angular reflectance in-tensity with increasing roughness and transition to morediffuse reflectance. For practical LDA measurements

tj41/4 I

(D p

Fig. 6. Quenching of right-handed circularly polarized light afterreflection on a flat surface (normal incidence at left, grazing incidenceat right) by analyzers Ae, A consisting of a /4-retardation plate,and a linear polarizer with the indicated operations. e, I denotethe transmitted light due to scattering, passing the analyzers afterblocking the left- and right-handed polarized light components,

respectively.

the system walls should be of polished metal, sincemetals have lower reflectance than nonconductingsurfaces and a dominating specular reflectance whenthe surface is smooth. Light should not be received inthe direction of the incident beams 3 2, since thereflectance is centered around the beams. At smallangles of incidence, i.e., intersection angles of the laserbeams ( < 10°), the angular width of the speculardistributions does not depend much on 4. Thus, fornormal backscatter systems the reflectance patterns atsmall angles of incidence can be taken to calculate thereflectance contribution to LDA signals. Optical sys-tems such as that shown in Fig. 1, which uses only thecenter of the transmission lens between the beams, willbe able to stop specular reflection to a large extent butnot completely.

B. Change of State of Polarization of CircularlyPolarized Light at Reflection

In the preceding section only the intensity distribu-tion of reflected light from walls was discussed. How-ever, when linearly polarized light is reflected, it be-comes in general elliptically polarized and changes itspolarization parameters depending on the angle of in-cidence and the refractive index of the surface.8 Themethod of reflectance suppression proposed here anddescribed in Fig. 6 uses circularly polarized light: First,vertically polarized laser light is converted into circu-larly polarized light with a quarterwave plate. Thegeneral state of polarization after reflection is recon-verted to linearly polarized light by a suitably orientedX/4-plate and is then quenched by a linear polarizer.

Figure 6 shows two typical situations of interest inlaser Doppler anemometry: the case of normal inci-dence ( - 0) at left and the case of grazing incidence(4 90°) at right, with the respective orientations ofthe /4-plate and the linear polarizer quenching thereflected light.

The change of polarization of light at reflection ex-pressed in the Jones calculus takes the form9

exp(JA) tan+, 4' [E. E'

I , 11 EY]= IEYJI(13)


55

50

I1

00

SY-bd t~h. O ppression ratio R /R

Oplexiglexs ~~0,00026 bl.k . st, didi Jj=0

x br.. ~ h 00010

+ anodized aluminium 0,CC1I_ ~~~~~~~~~~~0

0 di_ .t 0inn *002

0 di.lec1 mirro 10,0027 O

o0

00

0 0 0 0 0

a a a

+ aA + a

A +

I

10 20 30

00 O

T ?a 4

0

00

0

40 * 4X 4-

40

Fig. 7. Ellipsometric angle 4 = tan-'(RIRj) 1 12 and A = 3rp - 'rs

as functions of the angle of incidence (p (degrees) (a) above: for re-flection at an air-glass interface; X = 0.5461 ,um, mglass = 1.50 (b)below: for reflection at an air-gold interface; X = 0.5461 jim, mgold

= 0.35 - j2.45.

where A = 3rp -5rs and A = arctan(RP/R) 1 /2 are the

ellipsometric angles of the reflecting surface (see Fig.7), E.,,Ey are the electric vector components of the in-cident light, and the primed values are those of the re-flected light. 3

rp and rs are the phase angles and Rrand R, the vector components of the reflectance withpolarization parallel and vertical to the plane of inci-dence.

Incident circularly polarized light will change thehandedness at normal incidence;

±j] [Fj] at q5 =O.,

* (deg)

Fig. 8. Dependence of the angular difference boy = Ypol - 'Y/4 be-tween the settings of the linear polarizer and the X/4-retardation plate

on the angle of incidence for various surfaces.

independent of the kind of reflecting surface. Withincreasing 0 the reflected light will become more ellip-tically polarized, with the same handedness as for nor-mal incidence. For dielectrics at the Brewster angle 4B

(Fig. 8) and for metals near the principal angle 4p theincident circularly polarized light will be reflected aslinearly polarized light;

I -] I ° a] t P 0B

+j][1 t -

At greater angles the reflected light again becomes el-liptically polarized with handedness identical to thatof the incident light. At grazing incidence the state ofpolarization at reflection is

[I]. [L i at 0 = 900.

The ellipsometric angles A(4) and &(O) will be slightlychanged when the reflecting surface is rough or hasdeposits. 9 The value of the principal angle will increaseand the Brewster angle will decrease with increasingsurface roughness.10 Furthermore, the rough surfacewill scatter light waves, whose polarization state differsfrom that of incident light, and generate a partiallydepolarized diffuse reflectance that cannot be com-pletely suppressed by the circular polarization sup-pression as described in Fig. 6. To verify the suppres-sion efficiency of the method experiments were madewith metallic and dielectric surfaces as described in Sec.III.C.


'j,,A

35

I

I

-

x

x

C. Measurements of the Polarization Suppression ofReflected Light

The polarization reflection suppression shown in Fig.6 is analogous to the null-ellipsometer operation,9 wherethe angles Yx/4 of the X/4-plate and Ypol of the linearpolarizer vs the electric vector of the incident light (YE= 450) will be set to a position transmitting a minimumof the reflected light. The settings (Yx/4,ypol) determinethe ellipsometric angles A,4 of the reflecting surface.It is convenient to introduce the angular difference:

6 = YIpol - /4 = /2 arcsin 2 cosA 1 (14)1 tan4' + cotanj

for applications of the LDA method. The dependenceof 6y on the reflecting material, i.e., tanip, and on theangle of incidence 0 follows from Eq. (14) and Fig. 7. Atnormal incidence, where tank = 1,

& = 45° at = 0 (15)

independent of the type of reflecting material. Withincreasing ) the angular difference 6-y increases for di-electrics and decreases for metals. It will reach theextreme values

6-y = 1800 at( = OB for dielectrics, and (16)

boy = 0 at 0 (pp for metals. (17)

At higher values of 0 the angular difference &y decreasesfor dielectrics and increases for metals. At the grazingincidence

5-y = 135° atp = 90 (18)

independent of the reflecting material.The experimental setup for the measurement of b3y

was of the type shown in Fig. 6(a). A vertically polar-ized He-Ne 5-mW laser beam, expanded to 8-mm diamand transformed into circular polarization, was incidentat an angle 4) on the surface studied. The reflected lightpassed through a X/4-plate and a linear polarizer. Thereceiving PM, set at 20, collected the reflected light overa solid angle of 3°. The angles of the X/4-plate and thelinear polarizer were set to minimize the reflected in-tensity Ir,min(,ypolyX/4) at the suppression mode ofoperation and to maximize the reflected intensityIrmax (Ypo + 900, YX/4) at the transmission mode of op-eration of the X/4-plate and the linear polarizer. Thedegree of polarizational suppression RSupp was measuredat = 50:

Rsupp = Psupp Immn(YPo1,YX/4) (19)

PX Imax('Ypoi + 90,7X/4)

where PX is the reflectance of the surface measured at(typoi + 900,y\/4). Measured dependences of 6' on theangle of incidence 0 for selected dielectric and metallicsurfaces are shown in Fig. 8. At low angles of incidence,corresponding to the angular apertures in practicalbackscatter LDA measurements, the angular differenceboy is close to 450 in all cases and does not depend ap-preciably on . The measurement results in Fig. 8follow the trend implied by Eqs. (14)-(16) with in-creasing angle of incidence. The results obtained showthat the reflectance was suppressed by 3-4 orders of

too 90 80 M= 33IIO 70 X=19 69910

120 608\@t@~t88/((S

130 60

140 40

I SOa'((, 30

160~' ' 20

170 I i I i I I

0 4 3 2 1 0 -1 -2 -2- l 2 3 4

INTENSITY LOGARITHMIIC SCALE)

Fig. 9. Typical scattering diagram showing the polarization sepa-rated backscattering Ie (full curve), polarization-separated for-wardscattering Is (dotted curve), and scattering of linearly polarized

light III (dashed curve).

magnitude, with increasing suppression efficiencytoward specular reflectance, i.e., Plexiglas.

D. Circularly Polarized Light Scattered by SeedingParticles

Mie scattering of circularly polarized light by di-electric spheres was calculated and discussed in Ref. 11.In terms of the Mie scattering amplitudes S1 and S2, thescattering light intensities transmitted by the X/4-re-tardation plate and the linear polarizer have the form

I9 = Isc(YpOl,'Y4)- (S' -SOI',

IS = I(7pOI + 9 0,YX/4) I (S1 +l S2) 12 ,

(20)

(21)

where Ie and I(d are the parts of the scattered light in-tensity with conserved and reversed handedness, re-spectively. At exact backscattering and exact for-wardscattering the intensities Ie and I49 each go to zero.Of special importance are the intensities and angularpositions of the first backscattering and forwardscat-tering lobes, which will be detected by the LDA opticsoperating at backscattering and forwardscattering, re-spectively.

Figure 9 shows scattering diagrams calculated fromthe Mie theory. 1 For particles that lie in the me-dium-size range for LDA applications between 1 and 2/im, the first lobes IE in the exact backscatter have in-tensities comparable with the exact backscattering lobeof the linearly polarized light III used in conventionalLDA. Therefore, in the reflection suppression modethe backscattering intensities will be comparable withthose of conventional linear polarization. For for-wardscattering the suppression is less efficient since thefirst lobes of I are typically more than 2 orders ofmagnitude less intense than the exact forwardscatteringlobe of Il.

Backscattering of circularly polarized light was ex-perimentally investigated with the setup shown in Fig.10. An Ar-ion laser beam (1 W, X = 0.5145 im, circu-larly polarized light by a X/4-plate and expanded to


f = 250

'f = 300

/

_~- pinhole

c > imaging lense

rL | 0 /14 retardation plate

linear polarizer

L L' polarod aid filmpack

Fig. 10. Experimental setup for the visualization of the backscat-tering characteristics of the particles.

about 8-mm diam) entered a (f = 250-mm) lens parallelto the optical axis. At the waist of the focused beam,which had a 25-Am diam, a small air jet from a 1-mmdiam nozzle containing monodisperse seeding particleswas directed normal to the optical axis of the lens andformed a pencillike scattering volume.

The scattered light was collected over a total solidangle of 17°, passed through another X/4-plate and alinear polarizer, and imaged on Polaroid-Land film.Typical scattering patterns, with one dark and one lightlobe centered around the rectangular mirror shadow atleft, are shown in Fig. 11. Since the light is nearlyparallel after passing the imaging lens, the scatteringpattern is a plane cross section of the spatial scatteredlight distribution at the lens and allows the measure-ment of the angular position of the first lobe. Thecenter of the picture therefore corresponds to a scat-tering angle of 5.70. The dark area centered around themirror at the reflection suppression mode (Ie) corre-sponds to the low intensity of the scattered light com-ponent with conserved handedness around the exactbackscattering (compare with the backscattering in-

~PI/4

IYPo I -4/ =4 5 0I (PoI 9O; 4) = I

Fig. 11. Scattering patterns from -Am silicon-oil droplets at (above) the backscattering RS-mode and (below) backscattering RT-mode.


U1 .. /st)

05

lt srface

I 1/

_.v _F_

1 - \I' \

11

I--/��- N

I�

�- ,,

11i \packing of3mm straw tubesas flow straighteners

Fig. 12. Model wind tunnel with test surface at ba

Fig. 13. Experimental setup for measuring the backscattercharacteristics near a reflecting surface.

tensity Ie in Fig. 9). The annular light area claround the mirror at the reflection transmissio(Ie) corresponds to the dominating first backscelobe with reversed handedness centered arouexact backscattering.

The dark region between the exact backscattEand the first backscattering lobe Is is at 0 = passing from medium-sized (1-gum) silicon-oil pto large (5.7-gm) latex particles the dark/light boin the reflection quenching backscattering Ie shifan angular position at 5 to 9.3°, which can be undas an increase of the reflectionlike component iiscattering for larger particles and is in accordantheory."

IV. Backscatter LDA Measurements Near a Wwith Polarization Reflectance Suppression

The improvement of the LDA signal to noise rEtested by LDA measurements in exact backscatla reflecting wall. The flow system was a smamodel wind tunnel with laminar flow (Re = 400)plates of various material could be inserted fluthe wall at the back. The tunnel had a cross se(10 X 40 mm2 . Access to the flow was by a 10-mhole in the cover (see Fig. 12). Seeding part:

1-gm silicon-oil were added to the air, which had a meanvelocity of 0.35 m/sec.

The LDA system with a 1-W Ar-ion laser is shown inFig. 13. Parallel laser beams of = 0.514 gim and50-mm separation were focused and intersected by the(f = 250-mm) lens. The intersection angle was tan =0.1 and the fringe distance Ax = 2.5 gm. Backscatteredlight was collected by the same lens and focused on the500-gm pinhole of a PM by a (f = 300-mm) lens. Sincethe measuring volume had a 65-g, diam, the pinholewas about six times larger than the image of the mea-suring volume, and the optics was used with its fullaperture of 170 thus simulating worst-case condi-tions.

After splitting, the laser beams were transformed intocircular polarization by two X/4-retardation plates.

ck. The backscattered light was passed through the secondX/4-plate, and-the linear polarizer was placed in frontof the PM (see Fig. 13). By rotating both the polarizerand the X/4-plate the transmitted part of the reflectedlight was minimized at (Ypol,'YX/4), the reflection sup-pression (RS) mode, or transmitted at (pol + 90',yx/4),

oscilloscope the reflection transmission (RT) mode. Two surfacematerials were tested; one sample was from 2-mmPlexiglas with black color on the back, the other was astainless steel sheet. The signal was observed on anoscilloscope, and the SNR was measured with a HP8552B spectrum analyzer. The analyzer was used in the

Z [ log mode, with 10 dB/div of the vertical scale, as shownspectrum in Fig. 14. Good Doppler signals were stored for com-rznalyzer parison with the analyzer results, and the modulation

ed signal depth was evaluated. Although the optical system hadno optical frequency shifting and the measured spectramerged with the internal zero peak at small distancesfrom the wall, the SNR could still be measured with

--- I reasonable accuracy.2.1 LVI VU

n modeLttering.nd the

,ring Ie9°. ByarticlesundaryIts fromerstoodin back-ce with

all

itio was;er near11-scale, where3h withction ofrm openiides of

0.05 MHz 50psec

- 10dB

0.1 MHz 20psec

y=10mm

y= 5mm

y=Omm

, I . .

0.1 MHz 50p sec

Fig. 14. Typical backscattering LDA signals (at right) and the cor-responding frequency spectra (at left) for the polarization reflection

suppression (RS) mode.


I= }--r-- -

FrIEIEiI

II

II

r --- or __ ffi~~~~~~~~~~~~~~~~~~~~~

I r - - - - 4r ' 'fIl

tree flw

-30

-25

-20

_-5

0 2 4 6 8 10y (m m-

Fig. 15. SNR of backscattering signals plotted in dependence of thewall distance for reflection suppression and transmission mode and

for Plexiglas and stainless steel.

200~~~~~~~

100

o plexiglas .RS-mode

+ " RT- mode

A stainless steel RS- mode

RT- mode

0 1 2 3 4 5 6 7 5 9 102' - ~Y( )

Fig. 16. Direct current-component of backscattering signals in de-pendence of the wall distance for reflection suppression and reflection

transmission modes and for Plexiglas and steel.

Table 1. Comparison of Measured and Theoretical Increase of SNR

Plexiglas Stainless steelx = 10mm x = 5mm x = 9mm

Is (mV) 10 6 (10)IrRS(mV) 3.75 15 55IrRT(mV) 17.5 80 240SNR(dB) -3.01 -6.12 -5.85

(theoretical)SNR(dB) -2.3 -5.6 -6.5

(measured)

A 1-W laser was used to simulate a real case.Therefore, to prevent saturation a low PM supply-voltage and a divider chain with high current (4 mA)were used. Figure 14 shows a superposition of severalDoppler signals and the corresponding spectra in theRS-mode near the Plexiglas surface at varying distancesfrom the wall, where measurements down to zero dis-tance from the wall are possible. In the RT-mode nosignals could be obtained at x < 4 mm.

Figure 15 shows the SNR evaluated from the spectra,which varied from -30 dB with no wall to -14 dB at thewall for the RS-mode. With the RT-mode -14 dB wereobtained at x = 4 mm. In the RS-mode measurementswere still possible when the measuring volume had ap-parently penetrated the wall since the reflected beamsformed a new measuring volume at x > 0. This isshown in Figs. 15 and 16 by one point measured at x =-0.5 mm, with a SNR reduced by only 13 dB comparedwith x = +0.5 mm.

Figure 16 shows the dc signal on which the high-fre-quency Doppler signal was superposed. Both forPlexiglas and stainless steel the RT-mode suppressesthe dc by a factor of 5. Rewriting Eq. (12),

ASNR = 20 log SNR = 10 log bI I )XSNRRT VIs+ IRs'5J

(22)

the increase in SNR can be calculated and comparedwith the measured results. With I, estimated from thebest Doppler signals, such as Fig. 14, this was done forx = 5 and 10 mm, and results are shown in Table I.

The theoretical results are slightly higher than themeasured values indicating the presence of other noisesources. For stainless steel a higher dc was measuredbecause the more diffuse reflection and more scatteredlight was collected by the optics than for Plexiglas. Thesudden drop in dc near the wall seems to be due to sat-uration of the tube when the specular part of the re-flectance is starting to pass through the pinhole.However, the SNR is not significantly influenced bythis.

V. Practical LDA Systems for Single-Phase Flows

As shown in Secs. II-IV the scattered circularlypolarized light in the RS-mode Ie has almost the sameintensity in the dominating backscattering lobe as thelinearly polarized light III commonly used. However,the dominating lobe is centered annularly around theincident beams, while for linearly polarized light it iscentered directly on the incident beams.


o peliglasRS-mode

* IT --ode

A stoinkss steetS-mode

aX

, I

I

.118 (a)

INTENSIlY LOG SCALE

5 1 . 1 0 - -2. I I 0 . . 5; 4 3 2 ; 0 J -2 1 0 1 2 3 4 5

t q11.14

t~~~~~~~~~~~~~~~~~~ 19=92

INTENSITY LOG SCALE)

1 0 -1 -2

Fig. 17. Mie-scattering diagrams for one beam of linearly (III) and,circularly polarized light (Io) incident on a particle with m = 2.42 and

increasing Mie parameter q.

Therefore, the position of the first lobe is of impor-tance for the strength of the scattering signal and wascalculated by Mie theory. 0 For particles with refrac-tive index m = 1.33-2.0 and X = 0.5145 ,4m the firstbackscattering lobe I has an angular position at -30'for dp = 1 um, at 15° for dp = 2m, and at 8 for dp =3 Aum vs the exact backscatter direction. When light iscollected only from the region Al < 2, the availableaperture range between the beams is between 6 and 120for normal LDA optics. Therefore, to collect the firstlobes of scattering distributions Ie of the two beams 0should correspond to the angular position of the firstlobe. This is an impracticable demand for manybackscattering applications.

Figure 17 shows Mie scattering patterns lIe and III independence on the particle diameter for one beam only.Here it is evident that the lIe Doppler signal would be-come weaker both for smaller and larger particles sincethe first lobes have a greater position, and that particlesin the range of q 10 are the best suited for reflectancesuppression measurements, which corresponds to-1.8-,um diam at X = 0.5 ,um. Figure 17 shows also that

, (b)

2*=13 1e (d)

Fig. 18. Superimposed Mie-scattering diagrams for two beams in-cident on a particle with q = 11.14 and m = 2.42 for (a) linearlypolarized light III and (b)-(c) circularly polarized light at increasing

intersection angle 20.

signals of very small and very large particles, which ei-ther give weak signals or do not follow the flow exactlyenough, can be rejected optically by polarizationalsuppression when the angle is optimized for me-dium-size particles.

The aperture of practical anemometers should bechosen for a certain particle size and intersection angle20. For LDA systems of a given aperture the intensityof the collected scattered light, which is contributing tothe modulation of the Doppler signal, is defined as


q = 5. 81

fl 2/I e2sin(2q sin) dUff Iel + Ie2 2q sin4)(23)

where the first term depends on the ratio of the re-spective angular scattering intensities and the secondon the Mie parameter and fringe spacing. For a givendiameter and wavelength, S has a maximum where Iel= Ie2. The scattering diagrams shown in Fig. 18, whichwere calculated for q = 11.1 and m = 2.42, correspond-ing to a particle of 1.8 gm at X = 0.5145,gm, show dashedregions where small contributions to the signal modu-lation are obtained. Thus, with small intersection an-gles the detection angles v3 giving the best modulationare more concentrated near the optical axis, and smallapertures should generally be chosen. Only for shortmeasuring distances and large intersection angles canthe aperture angle be larger than the intersection angle20 [Fig. 18(d)].

Polarization-separated reflectance suppression canbe easily incorporated into standard LDA systems. Anadvantage of the modular systems developed in ourlaboratories is that they can be simply expanded by twoadditional modules as shown in Fig. 19. The firstmodule is introduced in the beam transmission part andwill hold two X/4-retarders fixed in the same place. Thesecond module is introduced in front of the photomul-tiplier and contains a second X/4-retarder and a linearpolarizer. Both optical elements should be indepen-dently rotatable at high precision since there are smallchanges of the elliptical parameter , and corre-spondingly b-y, depending on the material of the re-flecting surface. The linear polarizer should be a prismpolarizer (Glan Thompson) with high suppression(10-5) and transmission (90%). Additional improve-ment of the reflection suppression will be achieved bya spatial filter. Film retarders and film polarizers areinconvenient due to their less definite polarizationproperties and low transmittance, resulting in poorseparation of reflection and scattering.

VI. Application of Polarization-SeparatedReflectance Suppression in Two-Phase FlowMeasurements

A. General ProblemDifferential two-beam anemometers can be used to

measure the velocity of a fluid and the velocity of largeparticles carried by the fluid as a second phase.13 Thelarge particles may be bubbles in a liquid or rigidtransparent or reflecting particles; they may be muchlarger than the measuring volume. Both beams, whenreflected on the particle by a reflecting sphere, formspatial fringes in the regions of superposition since thelight originates from different reflection points on theparticle [Fig. 20(a)]. When the particle becomessmaller the reflection points are closer together, and thefringe spacing increases. When the particle movesacross the laser beams near the intersection point Tt willmake the spatial fringes move through the space aroundthe particle. Similar movements occur for reflectionon a bubble, where total reflection occurs at near tan-gential incidence [Fig. 20(b)], or on a transparent

PR BS BR Q Pi4l I I I I

83 LM

/4 Glon Thonpson ,paCiO titterotatable

X/4 rotatable

Fig. 19. Modular backscatter LDA systems with extension modulesfor polarization-separated reflectance suppression: PR = polariza-tion rotation; BS = beam splitter; BR = double-bragg cell module;

PM = photomultiplier; and LM = lens module.

(a I

(b)

reflecting sphere C

I P~~~~~c

.,p

Fig. 20. Reflections on spherical particles: (a) spatial fringes inbackscattering; (b) total reflection on a bubble; and (c) shadow' regionand source and sink of the fringes in forward direction for an opaque

sphere.

sphere, though in this case reflection will occur. For areflecting sphere the modulation frequency measuredby a PM oriented in near forwardscatter [see Fig. 20(c)]is

fD = X (uI cos3 u sin3), (24)

which for 0 - 0 and correspondingly 3 - 0 reduces tothe normal formula for the Doppler frequency.Therefore, small intersection angles should be used formeasurements of the velocity of big particles. LDAsystems for two-phase flow measurements and experi-mental results have been presented13; these demon-strate that velocities of both phases can be measuredwith LDA systems having only one PM. One mainproblem of the measurement, however, is to determineif a signal is generated by a small or large particle. Sincereflection is generally stronger than scattered light,


A

Fig. 21. Amplitude-separated Doppler signals generated by a re-flecting sphere (B) swinging across an air jet with Mie particles (A).

strong Doppler signals are attributed to large particlesand small signals to small particles; amplitude dis-crimination is used to separate the signals electronically.Figure 21 shows signals generated by a reflecting sphereand seeding particles, respectively, at the output of anamplitude discriminator. The high signals on channelB correspond to reflected light, while the small signalson channel A, measured during the absence of largeparticles, correspond to the seeding particles. Addi-tional discrimination in the frequency range by filterbanks can be used when a considerable frequency dif-ference exists between the signals of both phases.However, electronic separation of phase informationbecomes doubtful when the size of the second-phaseparticles approaches that of the seeding particles andno satisfactory amplitude or frequency differencesexist.

B. Optical Signal Separation

Polarization suppression of reflected light, i.e., oflarge particle signals, can be used to separate the signalsoptically. This was verified by a simple test with acircularly polarized laser beam traversing a squarechannel filled with high-viscosity and oil containing ahigh concentration of small bubbles (1-2-mm diam).At right angles a Polaroid-Land camera recorded thebubbles through a /4-retardation plate and a linearpolarizer. Figure 22 shows the trace of the beamthrough the liquid in the RT-mode and with reflectancesuppression in the RS-mode. There is obviously strongsuppression of direct reflection from bubbles.

Figure 23(a) shows a backscatter system with re-flectance suppression. Each phase is measured withone photomultiplier. PM1 receives the IE componentof scattered light but no reflected light, while PM2 re-ceives the reflected light. In this way the normally weaksignal of the first phase is free from strong reflected lightcontributions, and PM1 can be operated at high am-

Fig. 22. Influence of reflectance from bubbles in the RT-mode(above) and the RS-mode (below). A laser beam with circular po-larization is traveling from right to left through a liquid containing

small bubbles.

Fig. 23. Optical systems for two-phase flow signal separation bypolarization-separating reflectance suppression: (a) above: forbackscatter measurements, (b) below: for forwardscatter

measurements.


plification. In Fig. 23(b) the second-phase signal is stillsuperimposed by scattered light Ia. Since the ratio ofIe/I is much larger in the forward direction, the sup-pression of Iq, in the output signal of PM2 is more effi-cient in the forwardscatter orientation of the receivingoptics than in backscatter.

Figure 23(b) shows a near-forward arrangement.Right-handed circularly polarized light at grazing in-cidence conserves the handedness and is again verticallypolarized after the X/4-retarder and received by PM2.Only the Ie component of the scattered light reachesPM1. The scattered light Is received by PM2 is severalorders of magnitude weaker than the scattered light IE,as seen in Fig. 9.

VIl. Conclusions

Ellipsometric methods have been used to characterizeand measure reflection and scattering properties ofreflecting surfaces, seeding particles, and bubbles. Thepossibility of separating reflection and scattering bymeans of polarization in LDA has been theoreticallyanalyzed in terms of the SNR and experimentally ver-ified.

Appendix A: Nomenclature

A 1r2/dp particle diameter,e charge of an electron,E. ,Ey electric vector components of the incident

light,E'X,E, electric vector components of the re-

flected light,F focal length of the transmission lens,f focal length in general,fD Doppler frequency,Af frequency bandwidth of the detector,hp energy of a photon,I length of the measuring volume,Io maximum intensity in the measuring

volume,Is ac amplitude of the PM signal due to

scattering,I, dc amplitude of the PM signal due to re-

flection,Ir,RS;Ir,RT amplitude of the reflectance signal for the

RS- and RT-modes,III intensity of the scattered linearly polar-

ized light,Ie;Ie intensity of the circularly polarized light

scattered with reversed and conservedhandedness,

m refractive index,q 7rdp/X Mie-factor,RP;R8 reflectance with polarization parallel (p)

and vertical (s) to the plane of inci-dence,

S1,S2 Mie scattering amplitudes normal andparallel to the plane formed by the prop-agation vectors of incident and scatteredlight,

St

U1

SNRSNRS ,SNRr

ASNRxy

Ax

Appendix B:

signal strength Doppler signal,time,velocity component in the y direction,parallel to the optical axis,velocity component in the x direction,normal to the optical axis,diameter of the laser beam at the centerof the measuring volume and at a distancez = y/cos along the beam,signal to noise ratio,signal to noise ratio for pure scatteringand scattering plus reflection,increase of SNR by the RS-mode,coordinates parallel and normal to thewall,fringe distance.

Greek Symbols

axfs absorbance of light at a surface,/3 (1) angle between incident and reflected

light in the plane formed by their propa-gation vectors, and (2) angle of intersec-tion of reflected light from second-phaseparticles,

'Ypol angle between polarizer transmissionvector and the polarization direction ofincident light,

'YX/4 angle between the X/4-retardation plateand the polarization direction of the in-cident light,

61' Ypol - YX/4 angular difference between thelinear polarizer, and the X/4-plate,

6 distance from the waist in the laser to thefocusing transmission lens,

6,p ,6rs phases of the reflected light with polar-ization normal and vertical to the plane ofincidence,

X wavelength,

71 modulation depth of the Doppler signal,?7Q quantum efficiency of the PM,4) angle of incidence vs the normal on the

surface,lp principle angle of incidence for metals,

where all light is reflected as linearlypolarized,

4B Brewster angle, equivalent to p for di-electrics,

a beam waist in the laser tube,0o beam waist in the measuring volume,0-, beam waist at distance z from the mea-

suring volume along the beam,a.,- mechanical roughness of a surface mea-

sured parallel to the surface,A = rp -brs phase change at reflection,4' azimuth or angle of rotation of the main

polarization, direction,p reflectance,PX reflectance for monochromatic light.


References1. H. Mischina, N. S. Vlachos, and J. H. Whitelaw, Appl. Opt. 18,

2480 (1979).2. F. Durst, A. Melling, and J. H. Whitelaw, Principles and Practice

of Laser Doppler Anemometry (Academic, New York, 1976).3. J. A. C. Humphrey, A. Melling, and J. H. Whitelaw, "Laser-

Doppler-Anemometry for the Verification of Turbulence Mod-els," in Proceedings, Conference on Engineering Uses of Co-herent Optics, Strathclyde University K. J. Habell, Ed. (Cam-bridge U.P., London, 1975).

4. W. Hbsel and W. Rodi, "Errors occurring in LDA-Measurementswith Counter Type Signal Processing," in Proceedings, 1975 LDASymposium, Copenhagen, P. Buchhave et al., Eds., (LDASymposium Copenhagen 1975, P.O. Box 70, DK-2740 Skovlunde,Denmark).

5. D. C. Look and T. J. Love, Investigation of the Effects of SurfaceRoughness Upon Reflectance, AIAA Paper 70-820 (1970).

6. D. C. Look, AIAA J. 12, 656 (1974).7. T. F. Smith and R. L. Suiter, "Bidirectional Reflectance Mea-

surements for One-Dimensional Randomly Rough Surfaces," atAIAA Fourteenth Thermophysics Conference 1979, Orlando, Fla.,AIAA Paper 70-1036.

8. F. A. Jenkins and H. E. White, Fundamentals of Optics, StudentEdition (McGraw-Hill Kogakusha, Tokyo, 1976).

9. R. M. A. Azzam and N. M. Bashara, Ellipsometry and PolarizedLight (North-Holland, Amsterdam, 1977).

10. I. Ohlidal, K. Navrdtil, and F. Lukeg, Physics N.Y. 16, 1(1974).

11. J. Gardavsky and J. Bok, Scattering of Circularly Polarized Lightin Laser-Doppler-Anemometry, Applied Optics 1981 (to bepublished).

12. F. Durst and M. Zar6, "Laser Doppler Measurements in Two-Phase Flows," in Proceedings, 1975 LDA Symposium Copen-hagen, P. Buchhave et al., Eds. (LDA Symposium Copenhagen1975, P.O. Box 70, DK-2740 Skovlunde, Denmark).

13. F. Durst, "Studies of Particle Motion by Laser Doppler Tech-niques," in Proceedings, 1978 Dynamic Flow Conference, Mar-seille and Baltimore, L. S. G. Kovashay et al., Eds. (DynamicFlow Conference, P.O. Box 121, DK-2740 Skovlunde, Den-mark).

Patter continued irom page 3.999106 Btu (37.8 billion J) of the available energy, for a collector-array efficiencyof 16%. During periods when the collector array was active, a total of 103.92x 106 Btu (109.64 billion J) was measured in the plane of the collector array.Therefore, the operational collector efficiency was 35%.

Definitions of performance factors, solar-energy-system performanceequations, and long-term average weather conditions are collected in appendixesto the report. Although the insolation was below the long-term average for theLincoln, Nebraska, site, the net fossil-energy savings for the 12-month reportperiod were measured at 11.31 X 106 Btu (11.93 billion J) or the equivalent of300 liters of oil. Unquantified system losses into the heated space from thestorage bin and ductwork increase the actual savings beyond this figure.

This work was done by the Federal Systems Division of IBM Corp. forMarshall Space Flight Center. Further information may be found in DOE/NASA CR161495 N80-29851/NSPI, "Solar Energy System PerformanceEvaluation-Seasonal Report for SEECO Lincoln, Lincoln, Nebraska" [$71.A paper copy may be purchased from NTIS.

Solar-heated swimming schoolDuring the winter of 1977, with the low availability of natural gas, the Wil-

mington Swim School in Wilmington, Delaware, was judged to be a nonessentialuser and as such faced possible cutbacks or the elimination of its gas allocation.Solar energy now supplies much of the total annual building energy load at thisschool, thus alleviating the dependence on natural gas. A new report describesthe operation, installation, and performance of the solar-energy system.

The active solar-energy system is composed of 230 sq. m of liquid flat-platecollectors connected to a 13,600-liter concrete storage tank located below groundnear the building. An extension of the building completed in 1979 incorporatesa vertical-wall passive collection system that provides -25% of the heated-fresh-air for the office area. The active collector area is the maximum that couldfit on the building roofs.

A microcomputer-based control system selects the optimal application ofstored energy among the space heating, domestic-hot water, and pool-heatingalternatives. Selection is based on seasonal energy availability and the specificthermal requirements of each load. For example, if winter space heating re-quires water hotter than 491C, stored water below this temperature will be usedfor pool heating, which only requires temperatures higher than 290C; any heatstored below this temperature may be used for domestic water preheating.

This work was done by Cooperson Brack Associates for Marshall Space FlightCenter. Further information may be found in DOE/NASA CR-161538N80-31878/NSPI, "Solar Energy System Demonstration Project at WilmingtonSwim School, New Castle, Delaware-Final Report" [$8]. A paper copy maybe purchased from NTIS.

One-year assessment of a solar space/water heater for a dor-mitory building

System 4, a solar-energy system for space heating and domestic-hot-waterpreheating, was evaluated during 1 year of operation at a training-center dor-mitory in Mississippi. As stated in a new report, the system satisfied 32% ofthe building heating load. The system is designed to supply 48% of heatingneeds. The relatively low contribution of solar energy in the installed systemis explained by a combination of correctable and uncontrollable factors: Thesolar array, which is close to a dirt road, became coated with dust during dryspells and did not receive full insolation. Air leakage into the collector arrayand an open bypass valve also reduced efficiency. In addition, the winter wassomewhat cooler than usual.

System 4 is a prepackaged unit. Solar energy is absorbed in flat-plate col-lectors with air as the heat-transport medium. The collector area for the unitevaluated in the report is 24 m2. A blower circulates air from the collector arrayto a rock storage bed. Another blower circulates air to the building from eitherthe collector array or the rock storage bed. A heat exchanger at the bed absorbsheat for domestic hot-water use. Auxiliary heat is furnished by a 4-kW electricheater in each of two hot-water tanks and a 20-kW strip heater in an airduct.

The report describes the performance of the system and each of its subsys-tems (collector array, storage, hot water, and space heating). It presents op-erating energy requirements (the energy needed to transport solar energy tothe point of use) and energy savings. Appendixes contain definitions of per-formance factors and solar terms, a listing of equations used in performanceassessment, and a tabulation of long-term average weather conditions.

This work was done by the Federal Systems Division of IBM Corp. forMarshall Space Flight Center. Further information may be found in DOE/NASA CR-161509 [N80-30893/NSP], "Solar Energy System PerformanceEvaluation-Seasonal Report for IBM System 4 at Clinton, Mississippi" [$81.A paper copy may be purchased from NTIS.

Fire-station solar-energy system: screen-walled solar collec-tors

About half of the space heating and 75% of the heat for domestic-hot waterat a Kansas City fire station are provided by a solar-energy system, describedin a 157-page report. A historical narrative of the project is included, alongwith detailed drawings, charts, and product literature.

The fire station has two areas: an operations area occupied by the firemenand the apparatus bay occupied by the fire equipment. The operations areacovers 260 sq. m and has a winter design temperature of 211C; the apparatusbay covers 560 sq. m with a winter design temperature of 101C. The solarcollectors are an integral part of the apparatus bay. Two arrays of flat-plateair solar collectors are mounted on the roof, and one large array extends fromthe roof to the ground, forming the south wall. Ninety-six collectors are inte-grated into the wall.

East and west screen walls create an attractive profile for the fire stationwithout significantly affecting collector performance. A regional associationawarded a certificate of design excellence to the architect for the conceptionof these screens. A concrete-box storage subsystem, a domestic-hot-waterpreheat tank, blowers, pumps, heat exchangers, air ducting, controls, andplumbing complete the solar-energy system.

This work was done by the city f Kansas City, Missouri, for Marshall SpaceFlight Center. Further information may be found in DOE/NASA CR-161513[N80-30895/NSP, "Solar Heating and Domestic Hot Water System Installedat Kansas City Fire Station, Kansas City, Missouri-Final Report" [$111. Apaper copy may be purchased from NTIS.

Solar space-heating system at Yosemite: low insolation andequipment breakdown

A report on a solar-heating system in the Visitors Center at Yosemite NationalPark, Calif., assesses system performance from May 1979 through April 1980.The installation has 91 sq. i of liquid flat-plate solar collectors, water energystorage, heat exchangers, pumps, controls, and plumbing.

Design expectations that over half the annual heating demand would besupplied by solar energy were not met due to large building-heat loss, below-.average insolation, and maintenance downtime during the test period. Severallarge pine trees that add to the attractiveness of the Visitors Center also shadethe building and cause significant performance penalties. Nonetheless, fos-sil-energy savings amounted to 109 X 106 Btu (116 billion J); the potentialsavings with fewer equipment failures is higher.

The system has four modes of operation:Collector to storage via a heat exchanger, when the temperature difference

between the two is sufficient;Storage to space heating via a second heat exchanger upon demand, when

storage temperature exceeds 58°C;Mixed solar and conventional heating, at storage temperatures down to 320 C;

andConventional heating, using an oil-fired boiler to deliver hot water for building

space heating to a third heat exchanger.This work was done by the Federal Systems Division of IBM Corp. for

Marshall Space Flight Center. Further information may be found in DOE/NASA CR-161539 [N80-31883/NSP, "Solar Energy System PerformanceEvaluation-Seasonal Report for Colt Yosemite, Yosemite National Park,California" [$81. A paper copy may be purchased from NTIS.


Date post:	01-Oct-2016
Category:	Documents
Upload:	rolf
View:	214 times
Download:	0 times

Reflection suppression by polarization in backscatter LDA measurements near walls and in two-phase...

Documents