Experimental investigation of air flow around blunt aerosol samplers

J. Aerosol Sci. Vol. 28, No. 2, pp. 289-305, 1997 Copyright Q 1997 Elsevier Science Ltd

Printed in Great Britain. All rights reserved 0021.8502/97 $17.04 + 0.00

Pergamon

SOOZl-8502(96)00432-6

EXPERIMENTAL INVESTIGATION OF AIR FLOW AROUND BLUNT AEROSOL SAMPLERS

I-Ping Chung and Derek Dunn-Rankin

Department of Mechanical and Aerospace Engineering, University of California, Irvine, CA 92697, U.S.A.

(First receioed 29 February 1996; and accepted injnalform 11 July 1996)

Abstract-To evaluate better the aerosol collection behavior of blunt samplers, the flow field around a two-dimensional cylindrical sampler and a spherical sampler is investigated experimentally. The velocity along the symmetric axis of the cylindrical sampler is measured with LDV. The measurements are in very good agreement with viscous numerical calculations. The stagnation distance and the location of the separation point on the spherical sampler is studied by flow visualization. Smoke is employed to illuminate the fluid flow around the sampler. When the sampler faces the wind and the ratio of suction velocity to free-stream velocity Q/U, > 1, the experimental measurements are in excellent agreement with theoretical inviscid calculations. For U,/U, < 1, the theoretical predictions are higher than the measurements. When the sampler is oriented at an angle to the oncoming flow, the stagnation distance on a symmetric plane is consistent with the distance predicted by a theoretical inviscid calculation. Unfortunately, on an asymmetric plane, the experimental method breaks down and accurate results cannot be obtained. An appropriate experimental technique for this asymmetric study remains to be developed. The separation point on a spherical sampler at different orientations and different velocity ratios is also experimentally examined. With suction, the farther the sampler rotates, the stronger the influence of the suction on the location of the separation point. After rotating to angles greater than 90”, the separation point reaches an asymptotic value. This asymptotic value depends on the velocity ratio. At different velocity ratios, higher suction rates push the separation point further back, and the location of the separation point and its asymptotic value moves backward with an increase of the velocity ratio. When the sampler opening locates past the separation point, the flow near the opening entrance becomes turbulent, and laminar or inviscid flow models are inadequate to calculate aspiration efficiency. Copyright % 1997 Elsevier Science Ltd

NOMENCLATURE

D blunt sampler diameter, cm S linear distance between two stagnation points on a spherical sampler when it faces the wind, cm

St length of short axis of the area enclosed by the zero tangential velocity on a spherical sampler surface when it oriented at an angle to the wind, cm

Sz length of long axis of the area enclosed by the zero tangential velocity on a spherical sampler surface when it oriented at an angle to the wind, cm

St Stokes number, dimensionless (= prd*U,,/lSnD), where pP is the particle density, d is the particle diameter, and p is air viscosity)

u velocity component along the X-axis U,, free-stream velocity or wind velocity in the wind tunnel, m s-i Us sampler suction velocity, m s- i X direction along the symmetric axis Y direction perpendicular to the X-axis in a two-dimensional flow

Greek letters 6 sampler opening size which represents the width in a cylindrical sampler and the circular diameter in

a spherical sampler, respectively, cm 8 angle of a sampler oriented to the wind, degree

b, dimensionless parameter (= SzUs/D2U,)

INTRODUCTION

Blunt body sampling studies help determine the relationship between sampled aerosol and the aerosol inhaled into a human subject. The issues involve complex fluid and particle mechanics near the blunt object. For example, the bluntness of the sampler distorts the sampling air flow. The distortion affects the particle trajectories into the samplers because the air carries the particles. To understand blunt body sampling, therefore, it is important to

289

290 I-Ping Chung and D. Dunn-Rankin

understand first the fluid flow around the blunt samplers. In the past, many researchers examined the blunt sampler flow field using analytical and numerical calculations (for example, Ingham, 1981; Dunnett and Ingham, 1986, 1988; Ingham and Hildyard, 1991; Chung and Dunn-Rankin, 1992). To complement these theoretical activities, however, there are few experimental flow field investigations of blunt body sampling.

Vincent and Mark (1982) and Vincent et al. (1982) are the first group to study experimentally the flow field around blunt samplers. They used an electrolytic tank analog to map the flow field near the entrance of two-dimensional blunt samplers. The samplers they studied include a flat-nosed blunt sampler, a 45” inclined leading edge blunt sampler, and a two-dimensional cylindrical sampler. The electrolytic tank analogue provides a two- dimensional potential flow field, which may be different from the viscous velocity field, as discussed by Chung and Dunn-Rankin (1992). Chung et al. (1994) illuminated smoke tracks and particle trajectories to investigate particle aspiration efficiency in a two-dimensional cylindrical sampler. However, their experiments concentrated on the illumination of the limiting streamlines and the limiting particle trajectories. The limiting streamlines and the limiting particle trajectories are the fluid streamlines and particle trajectories which separ- ate the fluid and particles entering the sampler from those which are not sampled.

In order to contribute some experimental insight into studies of the fluid flow controlling the aspiration performance of blunt samplers, this paper presents two experiments that characterize the flow field and its dynamics near a blunt sampler. The first experiment uses laser Doppler velocimetry (LDV) measurements to determine the velocity along a symmetric axis of a two-dimensional cylindrical sampler. As stated above, aerosol aspiration efficiency is based on the limiting streamlines and the limiting particle trajectories. While the central line velocity may not directly relate to the calculation of aerosol aspiration efficiency, the limiting particle trajectories are strongly influenced by the fluid velocity field around the sampler. The validation of the central line velocity will reflect the accuracy of velocity field calculations and hence the accuracy of aspiration efficiency calculations. For a flow around a two-dimensional cylindrical sampler, the central line velocity determined from an electrolytic tank potential analogue (Vincent and Mark, 1982) differs from that determined by a viscous numerical simulation (Chung and Dunn-Rankin, 1992). Velocity measurements can be used to resolve this discrepancy and validate the viscous simulation of such Rows.

The second experimental section presents visualizations of the fluid flow field around a spherical sampler by using illuminated smoke. The flow field studies include measurements of the distance between stagnation points in front of a spherical sampler and the determination of the separation point location when a spherical sampler is rotated at an angle to the oncoming flow. Instead of solving for the complicated fluid velocity field and tracing particle trajectories, Vincent (1987) developed a physical model and semi-empirical formulas to calculate aspiration efficiency for blunt samplers. Dunnett and Ingham (1988) extended this model to predict the aspiration efficiency of spherical blunt sampler oriented at an angle to the oncoming flow. The basic feature of the physical model is to employ the stagnation point characteristics, the point on the sampler body where the airflow divides between that which is sampled and that which passes by the sampler, in empirical formulas leading to a calculation of aspiration efficiency. The distance between stagnation points in front of a blunt sampler, S, is a key parameter in these semi-empirical formulas. This paper presents experimental measurements of S on a spherical blunt sampler in order to validate the theoretical predictions coming from inviscid theory.

In viscous laminar flow, the fluid separates from a sphere at about 80” measured from the forward stagnation points, due to an adverse pressure gradient caused by friction between the fluid and the body surface, as illustrated by White (1986). The fluid flow characteristics before the separation point and after the separation point are different, and this difference will produce different aerosol sampling performance in these two regions. After the separation point, the semi-empirical formulas developed by Vincent (1987) or Dunnett and Ingham (1988), or any other Auid flow model based on inviscid theory is unlikely to apply. Moreover, the location of the separation point varies with suction rate and suction location,

Investigation of air flow around blunt aerosol samplers 291

which means the location of the separation point on a spherical sampler is different from that on a solid sphere. Suction influencing laminar boundary layer development has been extensively studied in the context of aerodynamics control (for example, Iglisch, 1944; Thwaites, 1946, 1948, 1949; Watson, 1947; Kozlov, 1970; Fasel, 1994; and many others). However, there is no theoretical analysis or experimental investigation of a spherical sampler with localized suction oriented at an angle to the oncoming flow. This paper experimentally determines the location of the separation point on a spherical sampler at different orientations to the oncoming flow and at different suction rates. The results presented here will help determine when models, laminar or turbulent, are appropriate for calculating aspiration efficiency of a spherical sampler at an angle to the oncoming flow.

TWO-DIMENSIONAL CYLINDRICAL SAMPLER

Experimental apparatus

The experimental apparatus is schematically displayed in Fig. 1. The apparatus contains four parts: the wind tunnel, the sampler, the particle seeder, and the optical components. The wind tunnel dimensions are marked in Fig. 1. It is an open-circuit design with a square test section of 20 cm x 20 cm. The detailed description of the wind tunnel can be found in a previous paper by Chung et al. (1994).

As shown in Fig. 2a, the two-dimensional cylindrical sampler in this experiment has a diameter, D, of 5.7 cm with a 0.5 cm wide of slit opening, 6. The height of the sampler is the same as that of the wind tunnel to prevent flows around the end of the cylinder from affecting the bulk flow near the sampler. More detailed information about the two-dimensional cylindrical sampler has been discussed in the paper of Chung et al. (1994).

Particles are the source of Doppler-shifted scattered light used for measuring the flow velocity. To obtain quality data with a LDV system, proper seeding of particles is necessary. A cyclone-type particle seeder illustrated in Fig. 3 is designed for this purpose. The dry compressed air tangentially enters the seeder and swirls the small particles up to the outlet of the seeder. The particles are nominally 1 ,nm aluminum oxide coated with flow agent. The flow agent prevents particles from agglomerating. Detailed statistical information about particle size and the physical property of the flow agent supplied by the manufacturer

- -

exhaust air

Fig. I. Schematic of experimental apparatus and the dimensions of the wind tunnel. In order to display LDV system schematically, the LDV system has been oriented 90”.

292 I-Ping Chug and D. Dunn-Rankin

k-D = 5.7 cm-4

-

T 15cm

!

6= 0.5 cm

‘cm

Fig. 2. Dimensions of (a) the cylindrical sampler and (b) the spherical sampler.

particle out let

compressed air inlet

............... ................. .................... ....... ..> ...................... ............................ ........................ :;. ... . . ......

.:‘-.::‘:‘:: . .................. . ....... : .................... .................. ..:..::.: ................

.......... . . ...*..... ..... . :.: ..: ..:. ....... .- : ...... ... ....... ...................... . . ... . ... ...... .: ..t ..~ .: ..... .... ... . . . ..: . . ... . ....................... ..:..::. :: .............. ... ::*: g.* *.: .. .. .. ... ..... ....... ........ .... :...:;.....{‘.-- ..... .. y......;. ~~..:~...;...~...~...~...~...~~......~...~ .............. .... ........... : ..:..: .. : :.~.~...~..~...~:.~. .:. ..: .......... .......... ............................ ;. ..... . ............... .: .......... : , ..... .:.::.::. ............... ..z .... -.; ..........

........... ..‘..... ............. .

................................. ................ :..*z ...... . ....... .: ..-. ...... .. ..... .... . ..- ........... . : . . . : :::. .. .: .................... ............ ..~.... . :.. : ........................... ..................... ..a... ...... .. ........................................................................................................ .: ..: .................. .:..:..:.::::..:..: ....... ............................. I 1

Fig. 3. A schematic of the particle seeder.


(Micro Abrasive Co., Westfield, MA) are listed in Table 1. To ensure stable feeds and individual particles, we always dry the particles overnight in an oven before the experiment. In addition, a small vibrator is attached to the seeder for further powder bed agitation and seed flow stabilization. Although treated carefully, it is not possible to avoid agglomeration and coagulation completely, but when the percentage of agglomerated particle is low, there is a very small potential for these larger particles to pass through the sample volume. Because the Stokes number is small (St < 0.002 for pm particle), even particle agglomerates do not have a large velocity difference from the fluid. The LDV measurement is the mean value of more than 1000 velocity realizations which significantly reduces the error caused by any few agglomerated particles.

An optical configuration for the LDV system is schematically shown in Fig. 4. The principle of the LDV system is well recognized. Two equal intensity coherent laser beams cross to generate a fringe pattern sample volume. When particles pass through the sample volume, the modulated scattering signals can be correlated to velocity. In the system, we put all optical components on a 3 m long rail for easy traverse, and for flexibility we used a 600 pm core diameter optical fiber to transmit signals to the LDV signal processor.

In this experiment, the sampler suction rate is fixed at 5.5 ms-‘, while the wind speed is adjusted at values of 1.5, 4.5, and 8.3 ms-‘. The wind speed is measured by a pitot tube. Consequently, the velocity ratios of suction velocity to wind speed (U,/U,) are 3.6, 1.2, and 0.7, where U, is the suction velocity and U, is the wind speed. The values are chosen to show three different cases with velocity ratios of Us/U, > 1, UJU, - 1, and us/u, < 1.

Table 1. Statistic information of aluminum oxide particles and physical information of flow agent

(supplied by Micro Abrasive Co.)

Mean diameter Median diameter Mode diameter Standard deviation Aluminum oxide density Flow agent density

1.04 pm 0.98 pm 1.02 pm 1.68 pm 3.5 gcme3 2.9 gcmm3

He-NC laser

I J

l-3 oscilloscope

1 1

Fig. 4. The optical configuration of a LDV system.


RESULTS AND DISCUSSION

With our optical configuration of the LDV system, the calculated measuring sample volume is about d, = 0.17 mm and 1, = 2.85 mm, where d, and I, are the lengths of the two axes of the ellipsoidal measuring region. The measuring volume is defined by the intensity distribution of the interference fringe which has a surface with l/e2 of the maximum amplitude. In our system, d, is the direction along the symmetric axis and I, is in the direction perpendicular to the axis. In the experiment, the minimum step for measurements along the axis is 0.5 mm. Therefore, the spatial resolution along the axis is adequate. In the direction perpendicular to the axis, the length of the measuring volume is about 3/5 of the suction opening, which produces some spatial averaging. This spacing can be reduced by increasing the angle between the two incident beams. In our system, however, all the components are mounted on a moving rail with limited space to improve the configuration. Fortunately, the numerical calculations show that the variation of the axial velocity component (or u-velocity) in the perpendicular direction (Y-direction) is small. Therefore, the spatial resolution in the perpendicular direction is also acceptable.

Figures 5-7 display experimental measurements of the velocity along the symmetric axis of a two-dimensional cylindrical sampler facing the wind at three different velocity ratios. In the figures, the value of the velocity and the position have been normalized by the free-stream velocity, UO, and the cylinder diameter, D, respectively. In the numerical

2.0

1.8- us/u0 = 0.7 delta = 0.088D

1.6- - wind tunnel boundary, Re = 26000 -- free stream boundary, Re = 26000 0 experimental data

1.4-

0.8

0.6

ooI i6.0 16.4 16.8 17.2 17.6 18.0 18.4 18.8 19.2 19.6 21

Fig. 5. The comparison of velocity along the symmetric axis of a cylindrical sampler between experimental measurements and viscous numerical calculations at US/ I/, = 0.7, S/D = 0.088, and

Re = 26,000. The Reynolds number is based on the wind velocity and the sampler diameter.

.O


2.0

1.8- us/u0 = 1.2 delta = 0.0-38D

1.6- - wind tunnel boundary, Re = 14000

-- free stream boundary, Re = 14000

0 experimental data 1.4-

1.2-

0.8

0.6

-f. i6.3 16.4 16.8 17.2 17.6 18.0 18.4 18.8 19.2 19.6 2

X/‘D

Fig. 6. The comparison of velocity along the symmetric axis of a cylindrical sampler between experimental measurements and vicous numerical calculations at U,/ U, = 1.2, S/D = 0.088, and

Re = 14,000.

.O

simulation, the coordinate system is set as the center of the sampler at position of X = 20.50; hence the suction opening is at X = 200 where X is the direction along the symmetric axis. Numerical calculations also show that the central velocity is influenced by the suction flow only about 0.5D away from the opening. The velocity measurement, therefore, starts 40 upstream of the opening.

The points in the figures represent the experimental measurements, and the lines report the numerical calculations. Every experimental measurement represents the mean value of approximately 1000 data points with typical standard deviation of 2%. Here, we show two types of calculations. The dashed line is the calculation of the sampler using free-stream boundaries and the solid line is that with wind tunnel boundaries. The free-stream boundaries are defined as zero-velocity gradient at 200 away from the sampler and the wind tunnel boundaries are no-slip at the walls 30 away from the sampler. The computational domain for the free-stream boundaries and details of numerical calculation procedures can be found in a previous paper by Chung and Dunn-Rankin (1992).

The figures show that the experimental data are close to the wind tunnel boundary calculation. The constraint of the tunnel wall causing more dramatic velocity field vari- ations has been analytically discussed by Allen and Vincenti (1994) by using image doublets theory, which represents a cylinder in a tunnel with an infinite array of image doublets. In our experiment, the ratio of the cylinder diameter to the width of the test section of the wind


2-o- 1.8- Us/U0 = 3.6

delta = 0.088D

1.6- - wind tunnel boundary, Re = 4900

-- free stream boundary, Re = 4900

q experimental data 1.4-

1.2- I

$J 1.0 3

0.8

0.6

0.4

::i. i6.0 16.4 16.8 17.2 17.6 18.0 18.4 18.8 19.2 19.6 i

X/D

0

Fig. 7. The comparison of velocity along the symmetric axis of a cylindrical sampler between experimental measurements and viscous numerical calculations at Us/U0 = 3.6, S/D = 0.088, and

Re = 4900.

tunnel (i.e. blockage factor) is 0.285. With this blockage factor, the central velocity is 10% higher than that in the free-stream boundary case based on our numerical analysis.

The experimental measurements are in very good agreement with the viscous numerical calculations.

SPHERICAL SAMPLER

Experimental apparatus

The experimental apparatus is similar to that described in the cylindrical sampler experiment as shown in Fig. 1 except that the cylindrical sampler is replaced with a spherical sampler, and the LDV system is replaced with a laser sheet optical arrangement and a still camera. A smoke generator is used instead of a particle seeder.

A spherical sampler with a diameter, D, of 6.35 cm and an orifice opening, S, of 0.635 cm shown in Fig. 2b is used in the flow visualization experiments. The sampler is mounted on a post so that its opening can be set any direction relative to the flow direction. A protractor is glued to the bottom of the wind tunnel for an indication of the sampler orientation. A flexible tube, connected to the back of the sampler, sucks air through the orifice opening to a vaccum pump. Before the vacuum pump, a flowmeter is installed to indicate the suction rate.

Investigation of air flow around blunt aerosol samplers

(b)

(4

Fig. 8. Smoke visualization of the flows around a spherical sampler at orientation of (a) 0, (b) 30. (c) 45, (d) 60, and (e) 90’. The velocity ratio (US/U,) is 1.5 and d/D = 0.1. The orifice opening is

marked by an arrow.


(a)

Fig. 13. Smoke visualization of the flow when the opening of a spherical sampler locates (a) before the separation point (100”) and (b) after the separation point (120”). The velocity ratio (U,/ U,,) is 2.5

and 6/D = 0.1. The orifice opening is indicated by an arrow.


To identify the fluid streamlines around the sampler, we burn incense and draw smoke into the wind tunnel. The incense smoke size has been measured in the range 0.3-0.5 pm using a light scattering method (Chung and Dunn-Rankin, 1996). This size is small enough to follow fluid flow very closely.

A 5 W argon-ion laser provides a high-intensity light source to illuminate the smoke. The laser beam passes through a cylindrical lens and a spherical lens to generate a laser sheet. The optical arrangement has been discussed further in the paper of Chung et al. (1994).

In this experiment, the wind velocity is fixed at 2.85 m s-i but the sampler suction rate is varied so that the velocity ratios (UJU,) are at values of 0.5, 1.0, 1.5, 2.0, and 2.5.

RESULTS AND DISCUSSION

Figure 8 is a series of photographs showing the typical flow field of a spherical sampler at five different angles to the oncoming flow: 0 (facing the wind), 30,45,60, and 90”. The smoke delineates the streaklines around the sampler. Note that the actual flow is a three-dimensional flow, the streaklines shown here are the intercross lines of the fluid flow and the central symmetric plane. When the sampler faces the wind, the flow field becomes an axisymmetric, allowing a two-dimensional representation of a three-dimensional flow. Figure 9 is a sketch of this case. The linear distance between the two stagnation points, S, is the parameter we measured. The stagnation points are identified by zero tangential velocity along the sampler surface.

When the spherical sampler faces the wind, Ingham (1981) assumed a potential flow and approximately analyzed the distance between stagnation points. He applied linear characteristics of the potential flow superimposing a uniform flow around a sphere and a point sink on a sphere front surface to obtain an approximate solution for the stagnation distance. Based on this analysis, the equation of S is

; 2: 0.69341$"~, (1)

where &J = 2i2U,/D2Uo. The symbols 6 and D are defined in Fig. 9, and they are the sampler opening size and the sampler diameter, respectively. Ingham also found that in equation (1) the multiplication constant is 0.798 for a two-dimensional cylindrical sampler and 0.9226

separation point

Fig. 9. The schematic of the flow stagnation points and the separation point on a spherical sampler when facing the wind.

300

O.OE

0.04

o-o""

I-Ping Chug and D. Dunn-Rankin

- Ingham's Theoretical Prediction 0 Experimental Data

I 1 I I 1

0 0.005 0.010 0.015 0.020 0.025 0. Phi

30

Fig. 10. The comparison of stagnation distance, S, on a spherical sampler between experimental measurements and inviscid predictions by Ingham (1981). The sampler faces the

wind and S,iD = 0.1.

for a disk. The values he calculated for the cylinder and disk were in very good agreement with experimental results obtained by Vincent et al. (1982), but there was no experimental work for a sphere. Figure 10 compares our experimental results with Ingham’s prediction for the stagnation distance on a spherical sampler. The solid line is the theoretical prediction and the points are the experimental measurements with 5% error. The abscissa represents the parameter 4, which in our case also represents the velocity ratio (US/U,) because S2/D2 is fixed (= 0.01). The figure shows that experiment and theory are in excellent agreement when US/U, > 1 (two points on the right), but that there is a discrepancy when US/U, d 1 (two points on the left). When Us/U0 3 1, the limiting streamsurface into the opening is a convex surface, and the stagnation points locate outside the edge of the opening. When US/U0 < 1, the limiting streamsurface into the opening becomes a concave surface. The locations of stagnation points are unknown for this case. What we are really interested in, however, is the flow sucked into the opening. Therefore, the size of the orifice opening becomes the limiting value (i.e. the equivalent to S) for this case. Based on the comparison shown in Fig. 10, measured value of S are below Ingham’s predictions for Us/U0 < 1 and US/U,, = 1 cases.

When the sampler is oriented at an angle to the oncoming flow, the fluid motion becomes three-dimensional. In this case, Dunnett and Ingham (1988) predicted that the area


1.

0.

0.

n \ c-3 rn

0.

0.

0.

- Dunnett & Ingham’s Theoretical Prediction

0 Experimental Data

40 50 60 Angle

70 80 90

Fig. 11. The comparison of S,/D between experimental measurements and inviscid predictions by Dunnett and Ingham (1988) for a spherical sampler oriented at an angle to the wind, where

S/D = 0.1.

enclosing the flow sucked into the opening is egg-shaped. For the long axis of the egg, they found the distance to be

s2 8 - 2: sin - . D 0 2 (2)

When the sampler is at angle to the oncoming flow, there is always one forward stagnation point located at the position facing the wind and the other zero tangential velocity point is located at the edge of the opening as shown in the photographs of Fig. 8. The curvature of the limiting suction streakline varies with the suction velocity. The higher the suction velocity, the larger the curvature of the limiting suction streakline. At very high suction velocity, the limiting suction streakline may move the zero tangential velocity to outside the edge of the opening. For our cases, however, the highest suction velocity (UJ UO = 2.5) did not result in such large curvature, and the zero tangential velocity remains at the edge of the opening. Therefore, the values of SZ for different velocity ratios are the same. Figure 11 is a comparison of Sz/D vs 8 between the experimental measurements and the Dunnett and Ingham predictions expressed in equation (2). The experimental points in the figure show that every angle contains five measurements at five different velocity ratios. Because S2 is the same for all five velocity ratios, the data shown in the figure also represent the experimental error for five repeated measurements. The comparison confirms that the

If-

)-

I-

I-

I I-

I-

:A 1

I-Ping Chug and D. Dunn-Rankin

0

0

0

0

X Sampler at 0 degree to the wind + Sampler at 30 degree to the wind 0 Sampler at 45 degree to the wind 0 Sampler at 60 degree to the wind 0 Sampler at 90 degree to the wind

I I I 1 1 0.5 1.0 1.5 2.0 2.5

Velocity Ratio (Us/Uo)

Fig. 12. The variation of the separation point on a spherical sampler with the sampler orientation and the velocity ratio (Q/U,).

estimation by Dunnett and Ingham (1988) is a good one. At least for 4 < 0.025, the prediction is an excellent agreement with our experimental results. For the short axis of the egg, it is difficult to obtain accurate results using the laser sheet. When the laser sheet illuminates the cross plane of the short axis, the flow crosses the sheet, making it difficult to distinguish the enclosed area. An appropriate experimental technique for this experiment needs further effort.

The orientation of the suction orifice and the suction rate influence on the location of the separation point of flow over a spherical sampler are demonstrated in Fig. 12. The measurement error in the experiment is less than 0.5”. Our experiments show that the variation of the sampler suction rate has little effect on the separation location when the sampler faces the wind. The separation location is the same as that on a solid sphere, 80”, as described by White (1986). When the sampler opening rotates, however, the separation point moves backward. Note that the flow is three-dimensional, the effect of the suction on the separation location may be different at different planes; here we concentrate the discussion on the central symmetric plane. Without suction, the boundary layer development on the central symmetric plane skips the distance of the sampler opening because there is no friction surface there. In the figure, for example, when the sampler rotates to 60 with no suction, the separation angle moves to 82”. The 2” movement is approximately equal to the suction opening angle. With suction, the separation point moves backward further because the starting point of boundary layer development moves backward, starting at the edge of the suction opening.


Table 2. The limiting separation location of a spherical sampler at various velocity ratio with

S/D = 0.1

Limiting separation location (deg)

0.5 105 1.0 110 1.5 - 112.5 2.0 - 115 2.5 - 118.5

The experiments also indicate that for the cases of opening orientation at 30 and 45”, the suction rate has little effect on the separation location until the suction rate increases to U,/ U0 = 2.5. When the sampler opening is at 60 and 90”, the displacement of the separation location increases with an increase of the suction rate.

From the experimental results, we can conclude that the larger the angle the sampler rotates, the stronger the influence on the separation location. Furthermore, the higher suction rate pushes the separation point further back. What is the limit separation point that can be achieved by suction? Equivalently, when does the sampler opening locate behind the separation point? The answer to these questions are important because the sampling behavior is different when the opening is behind the separation point compared to when it is ahead of the separation point. The flow field around a blunt sampler is a combination of the disturbance from the existence of the blunt sampler and that from the sampler suction. Usually, the former influence is much larger than the latter, but the suction can affect the boundary layer development. With the opening oriented at 90” and beyond, a point is reached where the suction force is overcome by the adverse pressure. The fluid then separates from the surface, leaving the suction opening behind the separation. When this happens, the sampler draws the aerosol from the turbulent region behind the blunt sampler instead of that from the upstream region.

Figure 13 gives one example of the photographs that show flow patterns near the sampler opening when the opening before and after the separation point. Before the separation point, the flow has clear laminar streaklines travelling into the opening. After the separation point, the flow becomes turbulent, and it is difficult to trace the fluid flow. We report the limiting locations of the separation point for different velocity ratios in Table 2. Beyond these angles, a different fluid sampling model should be used for the calculation of aspiration efficiency. Interestingly, the limiting value of the separation location is the same as that obtained for a sampler at an angle of 90” to the wind. This result indicates that the separation location reaches an asymptotic value that does not change as the sampler rotates past 90”.

CONCLUSIONS

This paper describes two experiments relevant to blunt-body aerosol sampling. First, the velocity along the symmetric axis of a two-dimensional cylindrical sampler facing the wind at three different velocity ratios (U,/ U,) is measured by a LDV system. The measurements can be used to validate the theoretical velocity calculations which reflect the accuracy of aerosol aspiration efficiency. The three velocity ratios are 0.7, 1.2, and 3.6. The measurements are compared with viscous numerical calculations. Two different boundary cases are calculated in the numerical prediction; one is boundary-free and the other one assumes wind tunnel boundaries. The experimental measurements are in very good agreement with the case using wind tunnel boundaries.

Next, a flow visualization method is used to investigate the flow field around a spherical sampler. The flow field includes measurements of the stagnation distance in front of the sampler and the variation of the separation points with the orientation of the sampler and


the velocity ratios (US/U,). The stagnation distance is a key parameter of semi-empirical formulas for aspiration efficiency calculations, and the relative position of the separation point to the suction opening determines which model, laminar or turbulent, is appropriate for the calculation of aerosol aspiration efficiency. Smoke is employed to delineate the fluid flows, and the velocity ratios are at values of 0.5, 1.0, 1.5, 2.0, and 2.5.

For the case of the spherical sampler facing the wind, the stagnation distance is compared to Ingham’s prediction (198 1). When US/U, > 1, the comparison between the experimental data and the inviscid prediction shows excellent agreement. For the cases of Us/U0 = 1 and Us/U, < 1, Ingham’s estimations are higher than the experimental measurements.

When the spherical sampler rotates at an angle to the oncoming flow, the area enclosed by the flow sucked into the sampler opening becomes egg-shaped. For the long axis of the egg, the experimental data are consistent with the inviscid prediction by Dunnett and Ingham (1988). For the short axis of the egg, our flow visualization method is limited by the asymmetric flow and it is difficult to obtain accurate results.

As for the separation point, the experiments show that when the sampler faces the wind, the suction rate has little effect on the location of the separation point. The separation point is located at the same 80” position as that of a uniform flow around a solid sphere. When the sampler is oriented at an angle to the wind, the separation point moves backward. With suction, the larger the angle the sampler rotates, the stronger the influence on the separation location. However, at orientations of 30 and 45”, suction rate has little effect on the separation location until the suction velocity increases to US/ U0 = 2.5. When the sampler is at 60 and 90”, the displacement of the separation point increases with an increase of the suction rate. The location of the separation point does not move when the sampler rotates past 90”, indicating that the separation location reaches its asymptotic value. This asymptotic value increases with an increase of the velocity ratio (Us/U,).

When the sampler is rotated beyond the asymptotic separation point, the laminar and inviscid flow theories are inadequate to calculate aspiration efficiency because the flow is clearly turbulent and dominated by different flow parameters.

Acknowledyement~This research was supported by the Center for Indoor Air Research. The authors thank Messrs. J. Yang and C. Choi for helping to set up experiment and conduct experiments.

REFERENCES

Allen, H. J. and Vincenti, W. G. (1944) Wall interference in a two-dimensional flow wind tunnel, with consideration of the effect of compressibility. Nat. Adv. Comm. Aero. Rep. 782.

Chung, I. P. and Dunn-Rankin, D. (1992) Numerical simulation of two-dimensional blunt body sampling in viscous flow. .I. Aerosol Sci. 23, 217-232.

Chung, I. P. and Dunn-Rankin, D. (1996) In situ scattering measurements of mainstream and sidestream cigarette smoke. Aerosol Sci. Technol. (in press).

Chung, I. P., Trinh, T. and Dunn-Rankin, D. (1994) Experimental investigation of a two-dimensional cylindrical sampler. J. Aerosol Sci. 25, 935-955.

Dunnett, S. J. and Ingham, D. B. (1986) A mathematical theory to two-dimensional blunt body sampling. J. Aerosol Sci. 17, 839-853.

Dunnett, S. J. and Ingham, D. B. (1988) An empirical model for the aspiration efficiencies of blunt aerosol samplers orientated at an angle to the oncoming flow. Aerosol Sci. Technol 8, 245-264.

Fasel, H. F. (1994) Numerical simulation of receptivity and transition in a boundary layer on a flat plate with suction hole. NASA Contractor report, NASA CR-195708.

Iglisch, R. (1944) Exact calculation-of laminar boundary layer in longitudinal flow over a flat plate with homogeneous suction. Schr. Dtsch. Akad. Luftfahrtforsch. SB, l-51 (trans. as NACA Tech. Mem. 1205).

Ingham, D. B. (1981) The entrance of airborne particles into a blunt sampling head. J. Aerosol Sci. 12, 541-549. Ingham, D. B. and Hildyard, M. L. (1991) The fluid flow into a blunt aerosol sampler oriented at an angle to the

oncoming flow. J. Aerosol Sci. 22, 235-252. Kozlov, L. F. (1970) Laminar boundary layer in the presence of suction. NASA Technical Translation, F-602. Thwaites, B. (1946) An exact solution of the boundary-layer equations under particular conditions of porous

surface suction, on the flow past a flat plate with uniform suction, investigation into the effect of continuous suction on laminar boundary-layer flow under adverse pressure gradient. Rep. Memor. Aero. Res. Coun. Lond. No. 2241, 2481, 2514.

Thwaites, B. (1948) The development of laminar boundary-layers under conditions of continuous suction, Part I: on a similar profiles. Rep. Memor. Aero. Res. Coun. Lond. No. 11830.

Thwaites, B. (1949) The development of laminar boundary-layers under conditions of continuous suction, Part II: approximate methods of solution. Rep. Memor. Aero. Res. Coun. Lond. No. 12699.


Vincent, J. H. (1987) Recent advances in aspiration theory for thin-walled and blunt aerosol sampling probes. J. Aerosol Sci. 18, 487-498.

Vincent, J. H., Hutson, D. and Mark, D. (1982) The nature of air flow near the inlets of blunt dust sampling probes. Amos. En&-. 16, 1243.

Vincent, J. H. and Mark, D. (1982) Applications of blunt sampler theory to the definition and measurement of inhalable dust. Ann. Occup. Hygiene 26, 3-19.

Watson, E. J. (1947) The asymptotic theory of boundary-layer flow with suction. Rep. Memor. Aero. Res. Coun. Lond. No. 2619.

White, F. M. (1986) Fluid Mechnnics, 2nd Edition, pp. 429-430, McGraw-Hill, New York.

Documents

Experimental investigation of air flow around blunt aerosol samplers