Вы находитесь на странице: 1из 5

EFFECT OF PROBE DIAMETER ON ISOKINETIC

SAMPLING ERRORS
Hideto YOSHIDA, Kenzou YAMASHITA,
Hiroaki MASUDAand Koichi IINOYA
Department of Chemical Engineering,
Kyoto University, Kyoto 606

To examine the effect of probe diameter on isokinetic sampling errors, experimental studies
were conducted using methylene blue-uranine aerosol generated by a spinning disk and lycopodium
spores in the small and large inertia regions, respectively. From the results, it is found that
the standard deviation of concentration ratio, defined as the ratio of concentration measured by
a probe of any diameter to that by a one-cm diameter probe, is larger as the probe diameter de-
creases. It is also found in the small inertia region that the meanvalues of concentration ratio
measured by smaller diameter probes are different from one another and deviate significantly
from unity in some cases.

Introduction the main flow concentration, a large mixing chamber


was set after the aerosol generator. A sonic jet
Isokinetic sampling should be conducted when ionizer was also used. Particles deposited on the
measuring the concentration of gas-solid flow in filter and inside wall of the probe were removed by
a duct. However, there are no theoretical or ex- washing with a known volume of distilled water,
perimental studies of the effect of probe diameter
on sampling error in isokinetic sampling. Watson3}
pointed out that probe diameter should be large
enough, because the inlet wall thickness cannot be
neglected and the anisokinetic sampling errors be-
comelarge for a small-diameter probe.
In this work, experimental studies were conducted
by use of a 3-jum aerosol generated by a spinning disk
and of a 24-jum aerosol in the small and large inertial
regions respectively.
Effects of probe diameter, main flow concentration
and main flow velocity on isokinetic sampling errors
were examined.
Effects of velocity distribution at the sampling Fig. 1 Schematic diagram of experimental ap-
probe inlet are also discussed. paratus No. 1 (lycopodium spores)
1. Experimental Apparatus and Method
Twokinds of experimental apparatus were used.
Figure 1 shows that employed for the large inertial
region. Lycopodium spores were fed from a powder
feeder through a mixer-type disperser to the test
section. The amount deposited on a probe wall and
a filter was weighed gravimetrically. A sonic jet
ionizer was used to neutralize aerosol charges. Fig-
ure 2 shows the apparatus used for the small inertial
region. In this case methylene blue-uranine aerosols
generated by a spinning disk were used. To stabilize Fig. 2 Schematic diagram of experimental ap-
Received on July 28, 1977. Correspondence concerning this article should paratus No. 2 (methylene blue-uranine aerosol
be addressed to H. Yoshida. produced by spinning disk generator)
48 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
Table 1 Preliminary experimental results
(uo=7 [m/s], Dp=24 |>m], n=6)
Probe Mean Standard
diameter concentrati on deviati on
d [cm] c [g/m3] [g/m3]
0.194 0.248 0.121
0.308 0.237 0.125
0.367 0.247 0.133
0.409 0.213 0.118
0.446 0.236 0. 142
0.504 0.256 0. 109
0.590 0.266 0.130
0.715 0.349 0.0295
0.753 0.300 0.0626
0. 804 0.258 0.0474
0.893 0.268 0.0535
1.001 0.291 0.039
1.384 0.283 0.036
1.570 0.290 0.029

Table 2 Probe dimensions* Fig. 3 Test section


d [cm] a [cm] b [cm]
0.133 4.4 0.0082
0.194 4.5 0.0127
0.308 4.4 0.0057
0.367 4.2 0.0057
0.409 4.3 0.0083
0.446 4.2 0.008
0.504 4.2 0.010
0.590 4.1 0.0067
0.715 4.2 0.010
0.753 4.3 0.0083
0.804 4.2 0.010
0.893 4.1 0.010
1.001 4.2 0.007
* SeealsoFig.4
Fig. 4 Sampling probe

and the amountwas measured spectrophotometrically. To reduce anisokinetic sampling errors, flow rate
In both cases, the particle concentration c determined was checked by a flow meter and a wet gas meter.
included the amount deposited on the inside walls Small deviations of measured concentration caused
of the probe and the sampling line. Test sections of by anisokinetic sampling were corrected by use of
both apparatuses were made of transparent polyvinyl the results obtained previously4\ Experiments were
chloride in order to observe probe direction parallel carried out using thirteen brass probes of diameter
to wind direction and deposition of aerosol on the ranging from 0.1cm to 1.0cm. Figure 4 shows
inside wall of the duct. To reduce errors caused by the probes used, and the size of each part is shown in
misalignment of sampling probe with respect to flow,
the probe centeriine was set to coincide with a pre- Table2.
To reduce interference of the two probes, they
determined line of the test section. The errors as- were held more than three centimeters apart and
sociated with this point are discussed by Watson3) electrically earthed. Experiments were carried out
and Raynor2). A preliminary experiment shown in eight times for each probe, and concentration ratio
Table 1 revealed that the concentration measured c/c0 and standard deviation a were calculated. As
by a probe of diameter greater than 1 cm was nearly an example, the reliability of mean value and standard
constant. Therefore, the concentration measured deviation for the data obtained by smaller diameter
by the probe of 1 cm in diameter was taken as the probe were checked and are shown in Appendex 1.
basic main flow concentration. With two probes Sampling time selected was from 2 to 15 minutes.
(one reference probe and one test probe) set side The distance from the probe inlet to the sampling
by side in the test section, as shown in Fig. 3, experi- filter was 15 centimeters and sampling tube diameter
ments were carried out under the condition of iso- was 1 centimeter. Details of experimental conditions
kinetic sampling. Experimental concentration ratio are shown in Table 3.
was obtained from the ratio of concentration measured
2. Experimental Results and Discussion
by the test probe to that measured by the reference
probe. Figure 5 shows the experimentally derived relation
VOL ll NO. 1 J978 49
Table 3 Expiermental conditions between concentration ratio and probe diameter when
Particle Dp|>m] ag pp[g/cm3] lycopodium spores were used. Thoughthe standard
Lycopodium spores 24.0* 1.09 1.05 deviations for smaller diameter probes were large, the
Methylene blue-uranine 3. 47 1. 07 1. 42 concentration ratio, (c/c0), was nearly equal to unity.
Gas velocity in duct u0=5-13 [m/s] Figure 6 shows the experimental results on the con-
Duct diameter D =S [cm] centration ratio for methylene blue-uranine aerosol.
Upstream straight duct length B=2 [m] In this case, the mean values of the concentration
Flow Reynolds number in duct Red=(2.67-6.93) x 104 [-]
ratio deviated significantly from unity when the probe
* note: particle diameter measured by microscope is 32 ^m, diameter was smaller. The standard deviations of
but Stokes diameter measured by sedimentation is 24 /urn.
one-cm reference probe, <7i (c0), are also presented in
Figs. 5 and 6. In Figs. 5 and 6, the ninety-five percent
confidence interval of the mean value is also presented
for each probe. Tables 4 and 5 show the experi-
mental data for the case of lycopodium spores and
methylene blue-uranine aerosol, respectively. The
following empirical equations are obtained by use of
the method of least squares, because the standard
deviation should be inversely proportional to the
square root of sampling amount W.
(7=9.19X l0-3W~0-5d-°'n; for lycopodium spores (1)
Fig. 5 Experimental relation between concentra-
tion ratio and probe diameter (lycopodium spores) a=1.25 x \0~4W~0'5d~0-28; for methylene blue-uranine
(2)
where W=(tu/4) d2cuot x 10"4 [g]
A comparison of experimental data with the results
calculated by Eqs. (1) and (2) is shown in Fig. 7.
From Eqs. (1) and (2), it is found that the standard
deviation of concentration ratio is larger as the probe
diameter decreases. Therefore, the concentration
measured by isokinetic sampling is not always ac-
curate when the inlet diameter of probe is small.
The following discussion concerns the effects of thick-
ness of inlet wall tip and velocity distribution at
the probe inlet on isokinetic sampling errors.
If one assumes that half of the particles that are
impacted on the inlet wall tip are sampled into the
Fig. 6 Experimental relation between concentra-
tion ratio and probe diameter (methylene blue- probe, sampled concentration is at most 13 %higher
uranine aerosol) than the main flow concentration for the minimum
probe diameter15, but this amount is too small to ex-
plain the deviation of the mean value. (See also
Appendix 2.)
The effect of velocity distribution at the probe
inlet on isokinetic sampling errors is analyzed as
follows. Because of fluid viscosity, a velocity distri-
bution offluid is formed at the probe inlet. Owing to
this distribution, there may be distortion of fluid
streamlines even if sampling is carried out under iso-
kinetic conditions (u=u0). If distortion exists, then
sampling error might occur under isokinetic sampling.
Figure 8 shows velocity distributions calculated with
the Navier-Stokes equation for isokinetic sampling.
It is found that the velocity profile at the probe inlet
is nearly uniform and that the boundary layer deve-
Fig. 7 Comparison of experimental and cal- lops rapidly in the low Reynolds number region
culated standard deviation both inside and outside the probe. From the results
50 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
Table 4 Experimental results (in case of lycopodiumspores)
Main flow Concentration Probe diameter Concentration Sampling time Amountof Standard
Run* velocity
«o [m/s]
c0 [g/m3] d [cm] c [g/m3] t [sec] sample
W [g]
deviation
a [-]
T ~5 0^99 (K4O9 l7o4 180 12.3 x l0"3 0J28
2 5 1.06 0.133 1.05 1032 7.5 0.193
3 5 1.13 0.194 1.04 642 9.9 0.094
4 5 1.03 0.308 0.923 382 13.1 0.074
5 5 1.14 0.367 1.166 240 14.8 0.079
6 5 0.98 0.446 0.962 180 13.5 0.090
7 5 0.99 0.504 0.985 180 17.6 0.080
8 5 0.81 0.590 0.76 180 18.7 0.035

9 7 0.97 0.893 0.996 105 45.8 0.032


10 7 0.97 0.804 0.98 120 42. 1 0.027
ll 7 0.92 0.715 0.961 120 32.3 0.030
12 7 0.92 0.590 0.92 135 23.6 0.046
13 7 0.95 0.446 0.84 219 20.2 0.0526
14 7 0.95 0.409 0.96 240 21.2 0.066
15 7 1.02 0.367 0.97 325 23.2 0.064
16 7 1.00 0.308 0.90 400 18.9 0.123
17 7 1.00 0.194 0.93 600 ll.5 0.098
18 7 1.02 0.133 0.88 712 6.07 0.212
19 7 0.29 0.893 0.27 173 20.2 0.102
20 7 0.29 0.804 0.26 260 23.8 0.068
21 7 0.32 0.753 0.30 315 29.4 0.132
22 7 0.34 0.715 0.35 270 26.5 0.121
23 7 0.28 0.590 0.27 435 22.1 0.117
24 7 0.28 0.504 0.26 576 20.6 0.159
25 7 0.26 0.446 0.24 620 16.0 0.168
26 7 0.24 0.409 0.21 885 17.3 0.167
27 7 0.27 0.367 0.25 1125 20.6 0.202
28 7 0.25 0.308 0.24 1200 14.8 0.228
29 7 0.24 0.194 0.25 2175 ll.1 0.152
30 7 0.96 0.504 0.89 180 22.4 0.103
31 7 0.94 0.753 0.93 120 34.8 0.017

Table 5 Experimental results (in case of methylene blue-uranine aerosol)


Main flow Concentration Probe diameter Concentration Sampling time Amountof Standard
Run* «o
velocity
[m/s]
cQ
[g/m3]
d [cm] c[g/m3] t [sec] sample
W [g]
deviation
a [-]
1 5 0.46X10"3 0.893 O.37xlO~3 200 23.2xl0~6 0.173
2 5 0.62 0.804 0.53 220 29.6 0.208
3 5 0.59 0.753 0.56 220 27.4 0.284
4 5 0.47 0.715 0.45 247 22.4 0.265
5 5 0.56 0.590 0.59 270 21.7 0.200
6 5 0.37 0.504 0.44 330 14.3 0.242
7 5 0.38 0.446 0.41 336 10.7 0.215
8 5 0.46 0.367 0.52 444 12. 1 0.324
9 5 0.45 0.308 0.78 456 13.3 0.541
10 5 0.43 0.194 1.02 720 10.9 0,710
H 5 0.33 0.409 0.46 345 10.5 0.687

12 10 0.25 0.804 0.27 225 30.5 0. 142


13 10 0.28 0.715 0.28 240 26.9 0.133
14 10 0.28 0.590 0.24 345 22.3 0.126
15 10 0.28 0.504 0.26 468 24.4 0.224
16 10 0.23 0.446 0.27 516 21.8 0.307
17 10 0.21 0.409 0.21 600 16.5 0.164
18 10 0.19 0.367 0.19 696 14.1 0.235
19 10 0.16 0.308 0.17 864 10.7 0.281
20 10 0.16 0.194 0.33 900 8.8 0.395
21 10 0.30 0.753 0.29 225 29.3 0.072

22 13 0.23 0.715 0.25 240 30.7 0.073


23 13 0.22 0.590 0.24 336 28.3 0.173
24 13 0.16 0.504 0.17 456 20.4 0.158
25 13 0. 14 0.446 0.16 480 15.6 0.247
26 13 0.13 0.409 0.13 576 12.7 0.215
27 13 0.13 0.367 0.13 600 10.9 0.155
28 13 0.12 0.308 0.ll 900 9.15 0.143
29 13 0.12 0.194 0.21 900 7.19 0.287

* Experiments were carried out 8 times for each probe.


VOL. ll NO. 1 1978 51
1.748 < (c/co)o < 2.378
Reliability of the standard deviation is calculated by
(n-\)o* 9 fo-IV2 (A2,
-9/ . <o"o <-vr+ ; vA-4)
xAp) xA\ -p)
Substituting /i=8, (7-0.395, /7-0.025 into Eq. (A-2)
0.239<(70<0.636

Appendix 2
Assuming that half of the particles that are impacted on
inlet wall tip are sampled into the probe, the following equa-
tion is obtained1} :
Fig. 8 Velocity profiles of fluid under isokinetic
sampling (u =uQ) cod i+R+2\Rj

of numerical calculation of particle trajectories, it is In the case of the minimumprobe diameter (d=0.133 cm)
found that the ordinate of critical trajectory equals b 0-0082
R-0.0665~°-1233L
0, Jr , _£__113
J' c, åå"
probe radius. The measured concentration is in-
dependent of the velocity distribution at the probe Nomenclature
inlet as long as sampling is carried out under iso-
kinetic conditions (u=u0). 5 = upstream straight duct length [m]
6 = thickness of inlet wall tip [cm]
As the above discussion indicates, experimental c = meanmeasured concentration [g/m3]
results cannot be explained as the effects of the thick- c0 = meanmeasured concentration for 1 cm
ness of inlet wall tip or velocity distribution at the reference probe diameter [g/m3]
probe inlet. cod = main flow concentration [g/m3]
c/co,(c/co)o = concentration ratio of sample and
As Eqs. (1) and (2) contain the terms d~0-31 and population [-]
d~0'28, the variation of standard deviation cannot be D = inner duct diameter [cm]
explained by the term Walone: it depends upon both Dp = particle diameter |>m]
probe diameter and sample amount. d{=2R) = inner probe diameter [cm]
n = number of trial [-]
Conclusion p = level of significance [-]
R = probe radius [cm]
Experimental study of isokinetic sampling errors Re{=duopIin) =flow Reynolds number in probe [-]
was carried out, and the following conclusions were Red(=Du0pli^) =^ow Reynolds number in duct [t-].
obtained. r(=r/i?),z(=z/7?) =dimensionless cylindrical co-ordinates [-]
t = sampling time [sec]
1) With respect to precision, it is found that stand- t(p) = ^-distribution function [-]
ard deviations of the concentration measured by
w0 = main flow velocity far ahead of probe
probes of less than 5 mmin diameter become more axis [m/s]
than 20 percent of the mean values for 3.5 /um meth- u, u ' = sampling velocity and average sampling
ylene blue-uranine aerosol and 10 percent for 24 jum vel oci ty [m/s]
lycopodium spores. The empirical equations, Eqs. W = amount of sample [g]
(1) and (2), are respectively obtained. li = fluid viscosity [g/cm - sec]
2) With respect to accuracy, it is found that mean v(=n-1) = degree offreedom [-]
values of the concentration measured by smaller Pp, P = particle and fluid density [g/cm3]
diameter probes also deviate more than 50 percent a, ffQ = non-dimensional standard deviation of
from the true values for a fine aerosol. Therefore, sample and population [-]
tfifco) = standard deviation of 1 cm reference
both precision and accuracy becomeworse in the case probe diameter [g/m3]
of smaller diameter probe. ag - geometric standard deviation of aerosol [-]
3) A sharp-edged wall of probe inlet is preferable Xv(p) = ^-distribution function [-]
for accurate sampling, and the fluid velocity distri-
bution at the probe inlet does not affect the isokinetic Literature Cited
1) Belyaev, S. P. and L. M. Levin: Aer. ScL, 3, (1972).
sampling error. 2) Raynor, G. S.: Amer. Ind. Hyg. Assoc. /., 31, 294 (1970).
3) Watson, H.H.: Amer. Ind. Hyg. Assoc. Quart., 15, 21
Appendix 1 (1954).
Reliability of the meanvalue is calculated by 4) Yoshida, H., T. Ohsugi, H. Masuda, S. Yuu and K.
_£_)-,,0,) " =<(JL\ <(JL)+ti,(j,> ° (A-l)
Iinoya: Kagaku Kogaku Ronbunshu, 2, 336 (1976).
c0I K*Wn \CoJo \coj ^Wn

Substituting
u=8, <j=0.395,
the/?=0.05,
data ofc/co=2.O63
Run 20 in Table 5 into Eq. (Arl) (Presented at the 42th Annual Meeting of The Soc. of Chem.
Engrs., Japan at Hiroshima, April 2, 1977.)

$2 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN

Вам также может понравиться