Measurement and Estimation of the Particle

Size Distribution by the Buoyancy Weighing-Bar

Method and the Rosin-Rammler Equation

著者 TAMBUN Rondang, FURUKAWA Katsuhiko, HIRAYAMA

Masayuki, SHIMAZU Masamitsu, YAMANAKA Shinya,
OHIRA Yuichi
journal or JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
publication title
volume 49
number 2
page range 229-233
year 2016-02-20
URL http://hdl.handle.net/10258/00009944
doi: info:doi/10.1252/jcej.14we129
Measurement and Estimation of the Particle Size
Distribution by the Buoyancy Weighing–Bar Method and
the Rosin–Rammler Equation

Rondang TAMBUN1,2, Katsuhiko FURUKAWA3, Masayuki HIRAYAMA3,

Masamitsu SHIMADZU4, Shin-ya YAMANAKA4 and Yuichi OHIRA4

1Department of Chemical Engineering, University of Sumatera Utara, Jl. Almamater Kampus

USU Medan 20155, Indonesia
2Division of Chemical and Materials Engineering, Graduate School of Engineering,

Muroran Institute of Technology, Muroran, Hokkaido 050-8585, Japan

3Taihoku Civil Engineering and Construction Co., Furano, Hokkaido 076-0031, Japan
4Division of Sustainable and Environmental Engineering, Graduate School of Engineering,

Muroran Institute of Technology, Muroran, Hokkaido 050-8585, Japan

Keywords: Particle Engineering, Particle Size Distribution, Buoyancy, Rosin–Rammler Equation, Fine
Particle

Measuring the size distribution of fine particles (< 5 μm) in a few hours is difficult when using
the sedimentation method due to a decrease in sedimentation rate. Herein, we discussed the validity
of using a combination of the buoyancy weighing–bar method and the Rosin–Rammler equation to
estimate the particle size distribution. When the cumulative mass oversize exceeds about 0.15, the
Rosin–Rammler equation can be used to estimate the particle size distributions of suspended solids.

To control the sedimentation process of suspended solids

Introduction
from a water treatment plant, the particle size distributions of
To remove suspended solids by the sedimentation method, particles less than 5 μm must be determined. Generally, log–normal
information about the particle size distribution is important. or Rosin–Rammler distribution is often used for expression of
Typically, the particle size distribution is measured by the particle size distribution (Miwa, 1981; Hidaka and Kamiya, 2014).
Andreasen pipette method (Andreasen, 1928), a hydrometer Although the log–normal distribution has some advantages in terms
(Bouyoucos, 1927), the centrifugal sedimentation method of mathematical analysis, it is confined within relatively narrow
(Donoghue and Bostock, 1955), microscopy (Stober and Fink, particle size distribution (Miwa, 1981). Thus, we now demonstrate
1968), laser diffraction/scattering method (Minoshima et al., 2005), the buoyancy weighing–bar method and the Rosin–Rammler
or the Coulter counter method (Coulter, 1953). Although the equation (Allen, 1990) could be applied to estimation of particle
equipment is economical, the Andreasen pipette, microscopy, and size distribution containing fine particles. In this paper, we discuss
the hydrometer methods are difficult to perform and time the estimation validity of the particle size distribution using the
consuming. In contrast, the other methods quickly provide highly Rosin–Rammler equation.
accurate results, but require expensive equipment. Hence, a reliable
and affordable method to measure the particle size distribution in a 1. Theory
water treatment plant does not exist.
Here, we briefly describe the theory of the buoyancy
Our research aims to develop a method to measure the
weighing–bar method because it is described in detail elsewhere
particle size distribution using the buoyancy weighing–bar method.
(Obata et al., 2009; Furukawa et al., 2010; Ohira et al., 2010).
We have reported that the particle size distribution of settling
Equation (1) shows the buoyant mass balance for a weighing bar in
particles can be measured using the buoyancy weighing–bar method
a suspension.
(Obata et al., 2009; Furukawa et al., 2010; Ohira et al., 2010). In
xmax
this method, a change in density due to settling particles is
measured by weighing buoyancy against a weighing bar hung in the
W0 − W = (W0 − W∞ )
∫xi f ( x)dx

suspension. Then, the particle size distribution is calculated using xi

∫xmin
v( x)t
+ (W0 − W∞ ) f ( x ) dx , (1)
the length of the weighing bar and the change over time in the h
apparent mass of the weighing bar (Obata et al., 2009). This method Where, W is the buoyant mass, h is the submerged length of the
is very economical as the measuring apparatus consist of an weighing bar, v(x) is the settling velocity of the particle, and f(x) is
analytical balance, a personal computer, and a weighing bar. the mass frequency with particle size x. The apparent mass of
Moreover, the measurement results are almost identical to the weighing bar G in the suspension is given by Eq. (2).
sedimentation balance method (Odén, 1925). However, particles
less than about 5 μm cannot be measured directly. G = VB ρ B − W = VB ( ρ B − ρS ) , (2)
1
Where, ρS is the suspension density, ρB is the weighing bar density size: 1.9–2.4 μm, Calcium carbonate, heavy) were used as
in the suspension, and VB is the weighing bar volume. authorization particles (JIS Z8901, 2006). Sodium
Differentiating Eqs. (1) and (2) with respect to time t provides Eq. hexametaphosphate (NaHMP, Kanto Chemical Co., Inc.) with a
(3). concentration of 10 kg ･ m–3 and a NaHMP solution with a
 dG  concentration of 1.0 kg･ m–3 were used as the dispersant and
G = GR +  t . (3)
 dt  dispersion liquid, respectively. The particle densities measured by
The value of GR is calculated from the tangent line based on Eq. (3). the pycnometer (JIS Z8807, 1976) were 2.61×103 kg･m–3 (class 3),
The cumulative mass oversize is given by Eq. (4). 2.14×103 kg･ m–3 (class 10), 3.00×103 kg･m–3 (class 11), and
xmax GR − G0 2.80×103 kg ･ m–3 (class17). The density and viscosity of the
R=
∫
x
f ( x)dx =
G∞ − G0
. (4)
NaHMP solution measured by the pycnometer (JIS Z8807, 1976)
and Ubbelohde viscometer (JIS K8803–01, 2011) were 1.00×103
Particle size x is given by the following equation using the Stokes
formula. kg･m–3 and 0.89 mPa･s, respectively.
About 2 L of seepage water at the construction site near the
18µ L v( x) 18µ L h
x= = , (5) Sorachi River in Furano, Hokkaido, Japan was obtained as a natural
g (ρP − ρL ) g ( ρ P − ρ L )t suspension. The suspended solid concentration, ignition loss and
Where, g is the gravitational acceleration and μL is the viscosity of solid density were analyzed according to JIS K0102 (2008) and JIS
the dispersion liquid. The particle size distribution of the suspended A1202 (2009). The concentration of suspended solids in the seepage
particles is calculated using the particle size x at each time t, and water was 7.8 kg･m–3, while the density and ignition loss of the
subsequently plotting the cumulative mass oversize R. suspended solids were 2.71×103 kg･m–3 and 0.041, respectively.
The Rosin–Rammler equation describes the relationship The concentration of NaHMP added as a dispersant, was 0.78 kg･
between particle size x and cumulative mass oversize R (Allen, m–3.
1990). The cumulative mass oversize R can be expressed as a To prepare a suspension, 1 L of the dispersion liquid and the
function of particle size x as Eq. (6). particles to be tested were mixed in a graduated cylinder. A
  n  weighing bar was hung from the analytical balance via a hanging
x
R = exp −    , (6) wire, which did not extend due to the weight of the weighing bar.
  xe  
  After thoroughly stirring the suspension, the weighing bar was set
Where, xe is the absolute size constant and n is the distribution in the suspension, and this was recorded as 0 s. Data, which
constant. To determine these parameters, Eq. (6) can be rewritten as consisted of time and the corresponding apparent mass of the
Eq. (7). submerged weighing bar, were recorded on a personal computer.
 x  Then, the particle size distribution was determined and estimated
1
ln ln  = n ln  = n ln x − n ln xe . (7) based on the aforementioned theory.
R  xe 
Graphing lnln(1/R) versus ln x can determine the distribution 3. Results and Discussion
constant n from the slope, while the slope and intercept can
determine the absolute size constant xe. The particle size distribution 3.1. Combination validity of the Rosin–Rammler equation
can be calculated using Eq. (6), the distribution constant, and the The particle size distributions of JIS Test Powders 1, class 3
absolute size constant. (Silica sand) measured by the buoyancy weighing–bar method
within 2 h are reported in a previous paper (Obata et al., 2009). The
2. Materials and Methods minimum particle size that can be measured in the NaHMP solution
is about 5 μm. To estimate particle sizes less than 5 μm, the absolute
Schematic diagrams of the experimental apparatus and the size constant xe and distribution constant n in the Rosin–Rammler
weighing bar are reported elsewhere (Obata et al., 2009; Furukawa equation are calculated using previous data (Obata et al., 2009).
et al., 2010; Ohira et al., 2010). The experimental apparatus Figure 1 shows the Rosin−Rammler plot. Because the slope
consisted of an analytical balance, personal computer, graduated and intercept are 0.883 and −2.24, the absolute size constant xe and
cylinder, weighing bar, etc. The weighing bar was an aluminum distribution constant n are 12.6 μm and 0.883, respectively. Figure
slit–cylinder (O.D.: 40 mm, I.D.: 30 mm, cylinder length: 210 mm, 2 shows the particle size distributions of JIS Test Powders 1, class 3
slit depth: 5.0 mm, slit length: 200 mm, submerged length: 200 (Silica sand). The open circles and dashed line denote previous data
mm), and according to JIS Z8807 (1976), its density was 2.70×103 (Obata et al., 2009) and the particle size distribution calculated by
kg･m–3. The graduated cylinder (diameter: 65 mm, Sanplatec Co., the Rosin–Rammler equation, respectively. The dashed line
Ltd.) was 1 L. An analytical balance (GR–300, A&D Co., Ltd., estimates that the cumulative mass oversize R at 1 μm is about 0.9.
minimum readout mass 0.1 mg) had a below–balance–weighing The filled circles in Figure 2 show the standard data for the particle
hook for hanging measurement, and was equipped with an size by the sedimentation balance method from Japan Industrial
RS–232C interface, which was used to connect the balance with a Standard (JIS), while the squares denote the particle size
personal computer. Data were collected in 3 s intervals. To avoid distribution measured within 24 h. The particle size distribution
external effects such as airflow and temperature changes, the calculated by the Rosin–Rammler equation is close to the standard
experimental apparatus was placed in an insulated vessel with a data. Additionally, the particle size distribution measured within 24
heater connected to a thermostat to maintain a constant temperature h agrees well with that calculated by the Rosin–Rammler equation.
of 298 K. Additionally, a different principle, the laser Hence, we conclude that the Rosin–Rammler equation and data
diffraction/scattering (MT3000EX, Nikkiso Co., Ltd.), was used to measured within 2 h by the buoyancy weighing–bar method can
measure the particle size distribution. estimate the particle size distribution.
JIS Test Powders 1, class 3 (median size: 6.6–8.6 μm, Silica
sand), class 10 (median size: 4.8–5.7 μm, Flyash), class 11 (median 3.2. Estimation accuracy
size: 1.6–2.3 μm, KANTO (Japanese) loam), and class 17 (median The Rosin–Rammler equation and two data points can

2
 Fig. 1 Rosin–Rammler plot (JIS Test Powders 1, Fig. 3 Particle size distributions of JIS Test
class 3(Silica sand)) Powders 1, class 10 (Flyash) and class 11
(KANTO (Japanese) loam)

Fig. 2 Particle size distributions of JIS Test Fig. 4 Particle size distributions of JIS Test
Powders 1, class 3 (Silica sand) Powders 1, class 17(Calcium carbonate,
heavy)

theoretically be used to calculate the particle size distribution. Powders 1, class 17 (Calcium carbonate, heavy), the calculation
However, more data are necessary due to measurement errors. The results from the Rosin–Rammler equation and the JIS standard
cumulative mass oversize for JIS Test Powders 1, class 3 (Silica value. The value of the cumulative mass oversize is up to about 0.10
sand) is up to about 0.6 within 2 h. In this section, the lower limit of within 2 h. In this case, the calculated results and the JIS standard
the cumulative mass oversize measured by the buoyancy values do not agree. Because the particle size distributions can be
weighing–bar method and the estimation accuracy are estimated by the Rosin–Rammler equation when the value of
experimentally investigated. cumulative mass oversize exceeds 0.15, the particle size distribution
Figure 3 shows the particle size distributions of JIS Test within 6 h is measured. The triangles and dashed line in Figure 4
Powders 1, class 10 (Flyash) and class 11 (KANTO (Japanese) depict the particle size distribution and the calculated result using
loam) with a measurement time of 2 h as well as the JIS standard the Rosin–Rammler equation, respectively. The value of the
values and the calculated values using the Rosin–Rammler equation. cumulative mass oversize is up to about 0.30 within 6 h. The
The values of the cumulative mass oversize are up to about 0.25 calculated results and the JIS standard values agree well. These
(class 10) and about 0.15 (class 11), but the particle size results demonstrate that the particle size distribution can be
distributions less than about 5 μm remain unknown within 2 h. estimated with the Rosin–Rammler equation when the cumulative
Using these data and the Rosin–Rammler equation, the absolute mass oversize R exceeds 0.15.
size constant xe and distribution constant n are calculated. The
calculated line agrees well with the JIS standard values. 3.3 Application for natural particles
Figure 4 shows the particle size distributions of JIS Test Figure 5 shows the particle size distributions of the
3
 (2) The Rosin–Rammler equation cannot estimate the particle size
distribution of JIS Test Powders 1, class 17 because the
cumulative mass oversize is up to about 0.1 within 2 h.
However, when the cumulative mass oversize is 0.15 or more,
the Rosin–Rammler equation can be used to estimate the
particle size distribution.
(3) The particle size distribution of suspended solids in seepage
water can be estimated using data from the buoyancy
weighing–bar method.

Acknowledgements

The authors wish to thank Messrs. K. Yamamoto and Y. Inaba.

Nomenclature

C = particle concentration [kg･m– 3 ]

f(x) = mass frequency of the particle size x [ - ]
g = gravitational acceleration [m･s – 2 ]
G = apparent mass of the weighing bar [kg]
GR = intercept of Eq.(3) [kg]
Fig. 5 Particle size distributions of suspended h = submerged length of the weighing bar [ m ]
solids in seepage water at the n = distribution constant [ - ]
construction site near the Sorachi River in R = cumulative mass oversize [ - ]
Furano, Hokkaido, Japan t = time [s ]
v(x) = settling velocity of particle size x [m･s – 1 ]
VB = submerged volume of the weighing bar [m3 ]
W = buoyant mass of the submerged weighing bar [kg]
suspended solids in seepage water at the construction site measured x = particle size [ m ]
over 2 h. Particle sizes from 5 to 71 μm can be measured by the xe = absolute size constant [ m ]
xi = particle size of ith [ m ]
buoyancy weighing–bar method. The particle size distribution
μL = viscosity of the dispersion liquid [Pa･s]
(dashed line) can be estimated with the Rosin–Rammler equation
ρL = density of the dispersion liquid [kg･m– 3 ]
because the cumulative mass oversize R is up to about 0.70. ρB = weighing bar density [kg･m– 3 ]
Absolute size constant xe and distribution constant n are 10.9 μm ρP = particle density [kg･m– 3 ]
and 1.02, respectively. The particle size distribution measured by ρS = suspension density [kg･m– 3 ]
the laser diffraction/scattering method (solid line) and that estimated
Subscripts
by the Rosin–Rammler equation give similar results, confirming
max = maximum
that the buoyancy weighing–bar method can be used to estimate the min = minimum
particle size distribution of suspended solids in seepage water. With 0 = initial t = 0
respect to the distribution constant n in the Rosin–Rammler ∞ = infinity t = ∞

equation (Eqs. (6) and (7)), a log–normal distribution is appropriate Literature Cited
for modeling the particle size distribution if n>3 (Miwa, 1981). Allen, T.; Particle Size Measurement, 4th ed., pp.163–164, Chapman and Hall,
While, the powder having a broad particle size distribution (e.g., London, UK (1990)
ground products and dust particles) does not accord with Andreasen, A. H. M.; “The Grinding of Materials. Theoretical and Experimental
log–normal distribution. As the n values of samples used in this Researches on Particle–Size Distribution Incident to the Disintegration
Process,” Kolloid Beihefte, 27, 349–358 (1928)
study are consequently less than 3, we should apply the
Bouyoucos, G. J.; “The Hydrometer as a New Method for the Mechanical Analysis
Rosin–Rammler equation to estimate its size distribution. At the of Soils,” Soil Science, 23, 343–354 (1927)
present stage, we do not find a clear criterion for the setting value of Coulter, W. H.; “Means for Counting Particles Suspended in a Fluid”, US patent No.
cumulative mass oversize. Thus, the physical meaning of the value, 2656508 (1953)
Donoghue, J. K. and W. Bostock; “Technique for Particle–Size Analysis by
a range in application of Rosin–Rammler equation, and their
Centrifugal Sedimentation,” Trans. Inst. Chem. Eng., 33, 72–75 (1955)
relation should be demonstrated to precisely estimate particle size Furukawa, K., Y. Ohira, E. Obata and Y. Yoshida; “Measurements of Mineral
distributions by using the proposed method. Further investigation is Particle Size Distributions by a Buoyancy Weighing Method,” Journal of
needed in order to apply the proposed method to a wide variety of MMIJ, 126, 577–582 (2010)
samples. In this paper, however, we proposed the Rosin–Rammler Hidaka, J. and H. Kamiya; Basic Powder Technology (Kiso Funtai Kougaku),
Nikkan Kogyo Shimbun, Tokyo, Japan (2014)
equation as an example to estimate the particle size distribution. We JIS Z8807; Measuring Methods for Specific Gravity of Solid, pp.1–12, Japanese
will report the application of this method to other samples on the Standard Association, Tokyo, Japan (1976)
future article. JIS Z8901; Test Powders and Test Particles, pp.1–23, Japanese Standard
Association, Tokyo, Japan (2006)
JIS K0102; Testing Methods for Industrial Wastewater, pp.1–366, Japanese
Conclusions
Standard Association, Tokyo, Japan (2008)
JIS A1202; Testing Methods for Density of Soil Particles, pp.1–8, Japanese
The particle size distribution is important information in the
Standard Association, Tokyo, Japan (2009)
sedimentation method. Evaluating the estimation validity of the JIS K8803–01; Methods for Viscosity Measurement of Liquid, pp.1–40, Japanese
particle size distributions using the Rosin–Rammler equation has Standard Association, Tokyo, Japan (2011)
lead to the following results: Minoshima, H., K. Matsushima and K. Shinohara; “Experimental Study on Size
(1) The particle size distributions of JIS Test Powders 1, class 3, Distribution of Granules Prepared by Spray Drying: The Case of a Dispersed
Slurry Containing Binder,” Kagaku Kogaku Ronbunshu, 31, 102–107 (2005)
class 10, and class 11 can be measured by the buoyancy Miwa, S.; General Powder Technology (Funtai Kougaku Tsuron), Nikkan Kogyo
weighing–bar method within 2 h, and estimated using the Shimbun, Tokyo, Japan (1981)
Rosin–Rammler equation. Obata, E., Y. Ohira and M. Ohta; “New Measurement of Particle Size Distribution

4
 by Buoyancy Weighing–bar Method,” Powder Technol., 196, 163–168 (2009) Weighing–bar Method: A Particle Size Distribution Measurement Using New
Odén, S.; “The Size Distribution of Particles in Soils and the Experimental Methods Settling Method,” J. Sediment. Soc. Japan, 69, 17–26 (2010)
of Obtaining Them,” Soil Science, 19, 1–35 (1925) Stober, W. and A. Fink; “Controlled Growth of Monodisperse Silica in the Micron
Ohira, Y., K. Furukawa, R. Tambun, M. Shimadzu and E. Obata; “Buoyancy Size Range,” J. Colloid Interface Sci., 26, 62–69 (1968)