DETERMINATION OF PARTICLE SIZE DISTRIBUTION-APPARATUS AND TECHNIQUES FOR FLOUR MILL OUST
KENNETH T. WHITBY. BME INSTRUCTOR, MECHANICAL ENGINEERING DEPARTMENT
•
UNIVERSITY OF MINNESOTA J. L. MORRILL, President
INSTITUTE OF TECHNOLOGY A. F. SPILHAUS, Dean
ENGINEERING EXPERIMENT STATION F. B. ROWLEY, Director
VOL. Llll NO.2 JANUARY 5, 1950 Entered at the post office in Minneapolis as semi-monthly second
class matter, Minnea_polis, Minnesota. Accepted for mailing at special rote of postage provided for in Section 1103, Act of October 3, 1917, authorized July 12, 1918.
ACKNOWLEDGMENT
. The author wishes to acknowledge the financial suppor of the Millers 1 National Federation and their guidance throug their members on the Curriculum Advisory Committee of the Insti tute of Technology, University of 1-linnesota.
The author would also like to express his appreciatio to Professor John M. MacKenzie of the Hechanic&l Engineering De· partment for the help and encoura.gement given in the performanc and publication of this investigation and to the Engineering Ex· periment Station, University of Minnesota, for cooperation i: furnishing facilities for this investigation.
1.4:90253
TABLE OF CONTENTS
. . • • 0 • 0 • 0 • 0 • • • 0 • • • 0 Introduction •
The Problem . . . . . . . • • • 0
Methods Studied by Other Investigators . . . . • 0
Methods Investigated • • • • • • • • Inclined Manometer Sedimentation Apparatus • • • Andreason Sedi~mentation Method • • • • • • Microscopic Determination of Particle Size Distribution
Apparatus • • • • • • •. • • • Preparation of Microscope Slides • • • • • • . . Measurement and Counting Procedure • . . . . Expression of Results • • • . . . Results • •
Determination of Particle Size Distribution by Direct Weight Sedimentation Apparatus • • • • •
Application of Stokes 1 Law • • • • .• Apparatus- • • • • • • • • • • • • Procedure • • • • . . • • . . Results . . . . Other Observations • . . . . . . . . .
Conclusions . . . . . . . . . . . . . Bibliography • . . . . . . .
Page
1
1
2
3 4 4 5 7 8 9
12 13
13 13 19 19 29 31 33
34
ILLUSTRATIONS
Figure
1. 2. 3. 4. 5.
Particle Size by Sedimentation--Andreason--Rye Dust Microphotograph of a Family Pastry Flour • Mi·crophotograph of Poor Roll Suction • • • • • • • • • Microphotograph of Rye Flour Dust. • • • • • • • • • • Microprojector Used for Determination of Particle Size
Page
5 6 6 7
Distribution •••••••••.•••••••••••• 8 6. Effect of Type of Mount--Microscopic Determination of
Particle Size of Regular Patent Flour ••••••••• 10 7. Path of Observation on Slide when Measuring Particles
with Microscope and Method ·of Measuring Particles on Microscope Screen. • • • • • • • • • • • • • • • • • • 10
8. Data Sheet for Microscopic Analysis •••••.... - ll 9• Microscopic Determination of Particle Size--Multistage
Count--Rye Flour Dust • • • • • • • • • • • • • • • • • 14 10. Microscopic Determination of Particle Size--Multistage
Count--Family Pastry Flonr • • • • • • • • • • • • • • 14 ll. Microscopic Determination of Particle Size--Multistage
Count--Poor Roll Suction • • • • • • • • • • • • • • • 16 12. Analytical Balance as Used f~ Determination of Particle
Size Distribution by Sedimentation • • • • • • • • . • • 17 13. Direct Weight Sedimentation Apparatus ••••••••• 18 14. Complete Sedimentation Apparatus for Particle-· Size
Measurement by Direct Weight Method. • • • • • ·• ·• • • 20 15. Direct Weight Sedimentation Apparatus with Flask Removed
Showing Cylinder • . . • • • • • • • • • • • • • • • • 20 16. Miscellaneous Data Sheet for Sedimentation Analysis. • 21 17. Particle Size by Sedimentation--Direct Weight Method-- ·
Rye Flour Dust--Effect of Concentration. • • 22 18. Data Sheet for Sedimentation Analysis ••••••.•• 24 19. Weight V s. Time Plot • • • • • • • • • • • • • • • • • 26 20. Particle Size by Sedimentation--Direct Weight Method--
Rye Flour Dust • • · • • • • • • • ._ • • • • • • • • • • 27 21. Particle Size by Sedimentation-,...;IJirect Weight Method--
Family Pastry Flour. • • • • • • • • • • • • • • • • • 27 22. Particle Size by Sedimentation--Direct Weight Method--
Poor Roll Suction. • • • • • • • • • • • • • • • 28 ' 23. Microscope V s. Sedimentation--Average of Runs--Rye Flour
Dust . . . . . . . • . . . . . . . . . . . . . . . . . 29 24. Averages of Sedimentation and Microscopic Determination
of Particle Size--Family Pastry Flour • • • • • • • • • 30 25. Averages of Sedimentation and Microscopic Determinations
of Particle Size--Poor Roll Suction. • • • • • • • • • 30 26. Andreason Vs. Direct Weight Sedimentation--Rye Dus~ •• 32 27. Correlation between Stokes' Diameter and Microscopic
Diameter • • • • • • • • • • • • • • • • • • • • • • 32
INTRODUCTION
1 The problem of determining the particle size distribution of various materials has been thoroughly covered in many aspects by previous investigators. Therefore, as it exists today, the ~problem is chiefly one of adapting existing methods to the measurement of a particular substance over a specified range by developing apparatus and modifying the technique to give the desired accuracy at a minimum investment of equipment and time. With the above factors in mind the Milling Engineer~ng Division of the Department of Mechanical Engineering set out during the summer of 1948 to de-
1Velop methods for measurement in the subsieve range as a preliminary step in the study of mill dusts. An extensive library search Wd.S made, first to obtain a complete bibliography on the general subject of particle size measurement and second to search out information applicable to the problem at hand. Although this study was made with reference to a particular problem, it is felt that the bibliography and the outline of the subject is of sufficient
1 value to be published. Therefore, the bibliography and outline along with a subject classification are to be published in a separate circular.
THE PROBLEM
Since the smaller particles of mill dust are most difficult to collect, the study is devoted ·primarily to the'development of methods that would be most suitable for the subsieve particle size range (1 to 37 microns). However, it has been recognized that there is still a ·need for a rapid method that would successfully cover the particle size range of flour; that is from 5 to 150
1microns. Therefore, some work has been done to extend the range of methods that were found suitable for the subsieve range.
Mill dust and other mill products have cert~in characteristics which require careful consideration when selecting methods. Mill products have a low specific gravity, usually less than 1.5. They are easily reduced by'attrition when preparing microscope slides or making a ~ieve analysis. Care must be taken to select suspending mediums th~t do not affect the particles either by changing their size or density. Particle shape varies between specimens, sometimes over the range of sizes, an~ particles tend to aggregate. Iviill products have a wide range of particle size; ~flours, for example, may have particles ranging in size from 2 to 150 microns. All of these characteristics tend to impose certain limits on the application of the various methods.
2 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
The techniques and apparatus developed for any one class of material must compromise accuracy, speed, range and cost. For research work a high degree of accuracy is required, whereas a method used for production control may be able to sacrifice some accuracy to save time. Some methods do not have flexibility with regard to the range that may be covered. Other methods are suitable only for a ce_rtain range. For example, sieving is limited to the measurement of particles above 37 microns in size (12).
METHODS STUDIED BY OTHER INVESTIGATORS
Work (13) has de.scribed microscopic methods rather completely. He has outlined methods of measurement, computation and presentation of data. Though microscopic measurement involves many' problems and is tedious, it is fundamental because it permits direct observation and measurement of the particles.
H1.ldebrand, Ferrari, Borchardt and Ankar ( 6) have applied microscopic measurement of particle size distribution to mill products. Their method of preparing a representative, well dispersed sample on the microscope slide has overcome a major difficulty.
Wichser (12) has developed a sieving technique forthe size analysis of flour, but the method is limited to particle size ranges above 37 microns.
Methods involving the rate of sedimentation of the particles in a fluid have been developed by many investigators. A direct weighing sedimentation technique developed by Oden (10) makes use of a balance pan submerged in the particulate suspension. The time~eight data determined with this apparatus is thenanalyzed mathematically or graphically and Stokes' law applied to obtain the particle size distribution. Oden and other early investigators found that some of the particles passed around the pan,thus giving erratic results. Hildebrand and associates (6) solved this problem by using a pan larger in diameter than the settling column and placing it directly below the column. Although satisfactory accuracy was obtained, their apparatus was complex and rather difficult to· clean.
While experimenting with the Oden type apparatus, Jacxi>sen and Sullivan (7) developed a modification of the direct weighing method which solved many of the problems encountered by previous investigators. The essential improvement that·they made was to extend the edges of the weighing pan upward to within a millimeter of the surface of the suspension, thereby preventing the particles from passing around the pan. With relatively simple equipment they obtained good results on high density inorganic materials. The
METHODS STUDIED BI OTHER INVESTIGATORS 3
further development of this apparatus for measurement of the particle size distribution of mill products is the main subject of this paper.
Turbidimetric determinations have also been applied to measurement of particle size. Kent-Jones (8) has described the
~ use of a turbidimetric method for determining the size distribution of flours. Essentially this apparatus measures the projected area
. of the particles in suspension at a given level by measuring the intensity of a transmitted beam of light. By applying Stoke's 1 law it is possible to obtain the approximate particle size distribution from this data. The accuracy of this method is limited, however, because light transmission is a function of mass and of specific surface. Further research and calibration may extend the application of this method and it may eventually become an important method for particle size measurements.
Wichser. (12) and others ha've also used the Andreason sedimentation apparatus for particle size measurement of mill products.
I •
Accurate results can be obtained if temperature control and very low concentrations of solid material are used.
Continental millers have.also used the Werner sedimentation apparatus (ll) in which depth of the material deposited in the bottom of a sedimentation column is directly measured on a scale at given time intervals. Though the method is simple, accuracy is poor due to the l~rge concentrations neces~ary to get a measurable height of material settled in the calibrated tube.
Experiments by Stairmand (11) and Wichser (12) have shown that elutriators, although useable for separating the sample into fractions, have some serious disadvantages. Stairmand's work with
l the Roller air analyzer, for instance, has shown that fractions separated are not .sharply defined, that the upper limit for sizing is 80 microns, and that the time necessary for a complete ana-lysis may be long.
This brief survey of methods e.mployed by other investigators has been presented to show the need for further research correlating the results obtained by the various methods, and more specifically the need for a rapid simple method of measuring the size distribution of flour mill products ov~r a range of 2 to 200 microns.
METHODS INVESTIGATED
After a study of the methods used by previous investigators, a microscopic and three sedimentation methods were selected
4 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST·
for further investigation. Microscopic measurement was selected primarily because it give·s basic data on shape. The Andreason, Dotts, and Oden sedimentation methods were selected for simplicity of apparatus. The Oden direct weighing sedimentation method offered the grea.test possibility for development of a rapid automatic or semi-automatic device.
Inclined Manometer Sedimentation Apparatus
Dotts (4) has devised a simple apparatus for determination of particle size distribution by observation of the pressure ~hanges in a sedimentation column.
Satisfactory results have been reported for relatively high density materials. The sensitivity of the apparatus decreased rapidly, however, with a reduction in the density of the dust or its concentration in the suspending liquid. The low density of flour dust made concentration of at least 5 per cent by volume necessary in order to obtain measurable pressure changes. As will be shown later, about 2 per cent by volume is the highest concentration that can be used without serious interference between falling particles. Also the time required for the measurement of particles under 10 microns was excessive, being in the neighborhood of four hours. No data obtained with this apparatus has been included because it is believed that the results are too erratic.
Andreason Sedimentation Method
The Andreason sedimentation apparatus (2) consists of a glass cylinder about 25 om high and 5 em in diameter with a special withdra~al pipette attached to a ground glass stopper. The pipette is fj.tted with a three-way stopcock so that samples of the suspenJion of dust in liquid can be withdrawn periodically into an evaporating dish. The residue after evaporation is weighed and, by application of Stokes' law, the particle size distribution is determined.
Though the apparatus is simple, care is required to get reproducible result.s. It was found that at a concentration of 5 gms per 100 ml the curve was shifted so that the distribution appeared 25 per cent finer than at a concentration of 2 gms per 100 ml. Stairmand (11) has shown that concentrations as low as 0.2 gm p~r 100 ml are necessary to completely eliminate interference between falling particles. A further discussion of the effects of concentration will be made in connection with the direct weighing method. Temp~yre control of the suspension to within 0.2 F was found necessary for good results. The time necessar,y
MICROSCOPIC DETERMINATION OF PARTICLE SIZE DISTRIBUTION 5
•oo v-~ p -
~~
..L /
v I
90
80
I
ld 70 (/)
~60 l: .... 0::50
.LLI I I(
z ~40
.... ~30 0 0:: ~20
10
0 0
I o -RUN I
' a - RUN 2
v g.-~
10 20 30 40 50 60 PARTICLE SIZE - MICRONS
FIG. I
PARTICLE SIZE BY SEDIMENTATION - ANDREASON RYE DUST
70
for comple~e evaporation was also found to be a further disadvantage. However, if a good analytical balance and an oven are available, the Andreason method may be useful where low cost is of primary importance.
Figure 1 presents the particle size distribution curve of ~ rye flour dust determined by the Apdreason apparatus at a concentration of 2 gm per 100 ml.
Microscopic Determination of Particle Size Distribution
IIJ!easurement by microscope can be consid~red fundamental ~because it permits visual comparison of size and direct observation of particle shape. The latter is particularly iinport&nt when measuring mill products because particles in different size ranges may have different shapes. Figures 2 and 3 show clearly that the large particles, 75 to. 150 microns in size, are much more angular in
6 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
·•·· ...•. ·•··.·.· ..... ··.•.·•··· ...............•.... • .. · ......... ····.·· .. ··.·•.· .. •· •... ··.• .. · .. •·.··.······· ft.;··~,~,·~
Figure 2. Microphotograph of a family pastry flour 335X.
Figure 3. Microphotograph of noor roll suction 335X.
• •
..... v
,
APPARATUS 7
shape than the round starch granules, 15 to 40 microns in size. As will be shown later, this variation in shape"with size· affects the size distribution as determined by microscope or sedimentation. Figure 4, a microphotograph ·of a rye dust, sho~s the predominance of starch granules in this relatively fine sample.
·It might be noted at this point that the particles in the microphotograph are much closer together than they would be on a slide to be used for counting. The more dense dispersion was ne-cessary to get a satisfactory photograph. ·
Figure 4. Microphotograph of rye flour dust 480X.
Apparatus . -d
When measuring particles with a microscope, microprojection is almost a necessity. Measurement with a filar eyepiece or eyepiece graticules is extremely hard on the eyes and very slow. ¥dcrophotography is also ruled out by the large number of fields that it is necessary to count.
Figure 5 is a photograph of tpe microprojector setup used for most of this work. An ordinary microscope equipped with a substage condenser and a mechanical stage is used. The photograph shows a ten cent mirror mount above the occular to reflect the :iJnage
8 APPARATUS AND TECHNI UES FOR FLOUR MILL DUS
Figure 5. Microprojector used for determination of particle
size distribution.
onto the screen. Though fairly satisfactory, this was later replaced by a reflecting prism when it became available. The screen consists of a sheet of trdnsparent coordinate paper, ruled in em and mm, placed between two sheets of window glass in an ordinary picture frame. It was found that ruling of the screen into 4 em squares further facilitated the counting procedure
A 6 volt, 108 watt Genco ribbon filament lamp gives sufficient light t-o enlarge the image about 5001 diameters.. Though an arc lamp wou..Ld give a much bright- 1
er image ana permit magnification of 1000 diameters, it is believed that tungsten filament lamps are less expensive and less troublesome to operate for long periods.
Since it was desired to have the apparat11s portable and flexible, the various components are merely set on a laboratory table as shown. Calibration of the image on the screen is accomplished by calibrating the filar eyepiece with a stage micrometer and then using the filar to calibrate the image on the screen. This is much more convenient than using the stage micrometer directly.
Preparation of Microscope Slides
The preparation of microscope sl~des to be used for the determination of particle size distribution poses a number of problems. Since the actual weight of the sample on the slidemay be only a fraction of a milligram, the problem of obtaining a representative sample is a formidable one.
The necessity of a good dispersion in which the large and small particles are discrete and evenly distributed over the slide has been shown by many investigators. However, many of the
, MEASUREMENT AND COUNTING PROCEDuRE 9
dispersion methods developed by previous investigator~ are not applicable to ~11 products because of the wide range of pa~icle sizes that may be encountered. Hil'"debrand, Ferrari, Borchardt and Anker ( 6) obtained satisfactory dispersions of flour by using a fine pointed glass rod to disperse·· a milligram or so of flour on a slide, with dilute collodion. This method is satisfactory for
'fairly large particles above 20 microns, but it was found that samples of dust with a mean size under 20 microns tended to aggregate a·s the material was stirred. Also, it was found that this method had one other disadvantage. The size of the sample placed on the slide was so small that difficulty was experienced in obtaining a representative dispersion.
To overcome some of these difficulties, another dispersion technique has been developed. The dispersion medium is prepared.by dissolving half a ml of cellulose acetate (airplane dope) in 200 ml of amyl acetate. Then,· approximately 50 mg of the dust is added to 10 ml of this solution contained in a dropping bottle. After shaking vigorously, a drop is quickly placed in the middle-of a clean slide and allowed to evaporate without being disturbed. This technique permits a larger and more representative sample to be used and gives a better dispersion. However, as the drop spreads on the slide, it tends to carry some of the finer particles to the outside giving a non-unifor.m dispersion along a radius from the center of the drop. At first it was believed that this might cause a serious error in the count. Figure 6 shows, however, that counts on slides prepared by each method are not significantly different. For this test a very carefully prepared sample of a patent flour was used to obtain a representative sample for each slide.
Measurement and Counting Procedure
Work (13) has described a good counting technique for uniformly dispersed particles. Basically the procedure outlined is as follows:
Measurement of particles is started on one edge of the dispersion. As soon as one field is measured, the slide is moved to bring the adjacent field into view. This process is continued until the disp·ersion has been crossed. If one trd.verse gives an insufficient number of particles, the slide is then moved just enough to bring a new field into view, as shown in Figure Gdnd another row of fields counted. This procedure is continued until about 600 particles have been counted.
Figure 8 shows a sample data sheet. As can be seen from the first count column, a large numb~r of small particles must be
l_
10
100
90
80
70
~ 60 (/)
z 50 ~ ::1: .... 40 a: I.LI
~ 30 IL
oe 20
10
0
0 ~ ~ ~
ll
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
.. v~ ~
/9 STIRRED MOUNT 11/
v%' " ~y
AI
,f/ ~ DROPPER BOTTLE MOUNT v A
~ ~
~ ~ ~
loo"&.A
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 SIZE- MICRONS
FIG. 6
EFFECT OF TYPE OF MOUNT MICROSCOPIC DETERMINATION OF PAR TICLE SIZE OF REG. PAT. FLOUR
/-r---. ...... r~
~ ~ l -- 7
r-- r-v r- t l,
381" --I~
PATH OF OBSERVATION ON SUDE MET HOD OF MEASURING PARTICLES
WHEN MEASURING PARTICLES ON MICROSCOPE SCREEN
WITH MICROSCOPE
FIG. 7
! MICROSCOPIC ANALYSIS 11
I
Sample
Date
Remarks
Mount
Siz~ Range
0-10
10-20
20-30
30-40
40-50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
150-160
160-170 170-180
TOTAL
MICROSCOPIC ANALYSIS
D-4 Run ---=1 ___ From Poor Roll Suction
May 10, 1949 Observer K.T.W.
----------------------------------------------Stirred - Dope
_1 ___ 2 ___ 4_ 5 1 1 st 2 1nd Corr. -%Count Count Count Size
443 x3. 26=1445 72 •.
113 368 18.3
27 88 4.38
10 33 1.643
9
6
2
1
1
12
10
9
7 6
21 1.045
16 .796
.547
.398
.348
.149
Magnification Med. __ ...;;._ _ _ _.;...6- 7 _8 9 _1Q_ Cube of % Wt. % Finer Me'an mean size (5 )x(6) Size Than Size
500 360 1.79 1.79 7.94
4500 823 4.10 5.89.16.5
17500 766. 3.82 9.71 26.0
45500 748 3.73 13.44·35.7
94500
170500
279500
427500
620500
814500
1
2
3 2
1
1
1
11
8
7
3
3 .149 1160000
988 4.92 18.36 45.5
1357 6.76 25.12 55.4
1530 7.62 32.74 65.35
1702 8.49 41.23 75.3
2160 10.76 51.99 85.3
1213 ~.05 58 •. 04 95.2
1730 8.62 66.66 105.2
2265 11.27 77.93 115.1
1
23 = (No. above 40)
3 .149 1520000
1 .0497 1950000
1 .0497 2460000
3100000
3720000
1 .0497 4500000
77 = (No. above 40)
5700000
<~j = 3.26)
2009
970 4.83 82.76 125.1
1220 6.08 88.84 135
145
155
2240 11.16 100.00 165
175
20072
Figure 8. Data sheet for microscopic analysis.
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
counted to measure sufficient particles in the larger size ranges. Experience has indicated that at least 40 particles in the upper two-thirds of the size range of the sample must be counted for satisfactory accuracy. To increase the accuracy and reduce the labor involved, a multiple count method suggested by Work (13) is used. The procedure is to count particles in all size ranges on the first count until perhaps a total of 600 particles has been · measured. Then, ·a second count is made,counting only particles above a certain range; in this case, 40 microns. Then, these two counts are added together, the sum being placed in column 4. Next the ratio of the sum of particles above 40 microns in column 4 is multiplied by the number of particles in each size range below 40 microns in column 1, to give the rest of the corrected count in column kw Greater accuracy could be obtained by making further subdivisions and making more counts, but two or three stages have been found to be sufficient for most sampLes.
The method of measurement of the particles on the screen is important. Since only two dimensions are visible, it must be assumed that the third dimension_is the average of the other two. Actually it is known that this is not true because the particles will tend to arrange themselves with their least dimension perpendicular to the slide. It will be shown later that this error will be relatively constant for a given material and is, therefore,ofinterest only when comparing distributions determined by different methods. Of'the several diff~rent ways it is possible to measure the particle •. Faires (5) gives measurement along horizontal lines dividing the particle into equal parts as a practical method, F~gure 7. It is rap~d and give~ results fairly close to the average dimension.
Expression of Results
Usually it is desirable to express the particle size distribution in cumulative percentages by weight finer than size, Figure 9. The problem is to translate the percentage of total count by size range to _per. cent finer than a given. size by weight •. The first step is to determine a particle size representative of the average weight of the range~· Since the mass of a particle is proportional to the cube of its· size, the cube mean size is selected as representative of the average weight. It will be noted from column 10, of the data sh•et, Figure 8, that the. cube mean size is not the average size in the range. Thus, the particle diameter corresponding to the average weight of a 10 and a 20 micron particle ·is 16.5 microns and not 15. In practice this cube mean diameter is calculated as follows:
~~----
RESULTS 13
Cube of 10 - 1000
Cube of 20 == 8000
Sum - 9000
Average = 4500
Cube mean size - '<f4500 = 16.5
The second step in the translation of results,from per cent of total count in each size range to per cent finer than a given size by weight, is multiplication of the per cent by siz~ range (column 5) by the cube of the cube mean size (column 6).
Column 8 gives the percentage by weight in each size range, column 9 gives the cumulative percentages finer than the size in column 10.
It should be mentioned that equal size intervals were used in this work because unequal size intervals somewhat complicate the calculations.l
Results
Figures 6, 9, 10, and 11 show cumulative distribution curves obtained by microscopic measurement. Figure 6 was made· from a very carefully prepared sample and shows that results can be reproduced in successive examinations if a repr~sentative sample can be prepared. The other curves, however, show that random samples taken from a several hundred gram sample can give varying distributions·. Thus, it is necessary to take great care in selecting the sample,and also necessary to obtain the aver,age of several runs for dependable results.
Determination of Particle Size Distribution by Direct Weight Sedimentation Apparatus
Applic~tion of Stokes' Law
The direct weight sedimentation method originally worked , out by Oden (10) is similar to other sedimentation methods in that
lwork, L. T., ''Methods of Particle Size Determination," ~ ceedings of the American SocietY. for Testing Materials, 28:771. 1928.
14
100
90
80
"" N
cn1o z Cl %:60 ~
a: ~50 i&: ~40 z
"" ~30 ~
20
10 L r/'
~/ !-"''
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
-"'
L ~ ~ ~ ......... ;y
v 'r v' / L
LV _,..v /.
v v/ p
/ I/~ /
~/
v // v
~ v ....... ~ / L
~
........ --~ 6- SLIDE NO. I
-0- SUDE NO.2 V-A 1 1 _l 0
0 2 4 6 8 10 12 14 J6 18 20 22 24 26 28 30 32 34 36 38
"" N
100
90
80
c;; 70
z ~60 ~
! ffi 50 z iL ._40 z "" ~ 30
:f 20
10 I~
SIZE IN MICRONS
FIG. 9 MICROSCOPIC DETERMINATION OF PARTICLE SIZE
MULTISTAGE COUNT - RYE FLOUR DUST
........ ~ ~ ........G ~ ./ ............;
~ ~ /6 /
,../ v ~ ~
/ v ~ ~ /
v / v-v y
~ v v v
~ v V'4 v / A/
v v v
~~
v
.. ~ I 0o 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 SIZE - MICRONS
FIG. 10
MICROSCOPIC DETERMINATION OF PARTICLE SIZE
MULTISTAGE COUNT- FAMILY PASTRY FLOUR
APPLICATION OF ~TOKES' LAW 15
it uses Stokes' law to relate the velocity of the settling particles to their-.diameter. In symbol farm, Stokes r law is:
(1)
where Vm = terminal velocity of the particle falling freely in a fluid-em/sec •
. D = diameter of the particle em
v = kinematic viscosity of the fluid - Stokes'
fJ = density of the particle - gm/cc
~ = density of the fluid - gm/cc
K8 = a constant depending on the acceleration of gravi~y and the shape of the particle
~ = viscosity of fluid - pois~
Now: g = 980 cm/sec"2
V= Po
(2)
Also from experiment we know that for a sphere falling in a fluid Ks= 54J5~ If we now rearrange (1), solve for the diameter and introduce the constants, Stokes' law becomes:
8 d'= IBxlOeh ~ y . (.P - ~) g t 1{
(3)
where d = diameter o~ particle - microns
h = height of fall - em
t = time to fall a distance h - sec
SK = correlation factor
16
100
90
80
LU N7Q c;;
~60 3: .... a: 50 LU z ii:40 .... z ~30 a: Ill CL.
20
10
~
~ ~ ~
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
J'L ---v ~ v~ aL
I ? y v RUN I---... r--..y
~ v v /
Vo _L
_p/ ~ v
L
~ ~ ~-""""-
1'----RUN 2
/ ?7
~
s~ 0
o ro 20 30 40 50 eo 70 eo so 100 110 120 130 140 150 160 170 r8o SIZE - MICRONS
FIG. II
MICROSCOPIC DETERMINATION OF PARTICLE SIZE
MULTISTAGE COUNT- POOR ROLL SUCTION
Stokes' law, however, relates diameter and terminal velocity of a particle only for streamline flow. Allen (l) has given the following formula for'calculating the largest particle which will fall according to Stokes' law under given conditions.
3 36 ,.l g Pa(P-Pa)
(4)
Though there are more complete equations that will give the velocity or a falling particle for all types of flow, these equo.tions are much too complicated for general application.2 For this reason, practico.lly all methods of determining parti~le size by sedimento.tion choose liquids of such density and specific gravity that the lindting diameter exceeds that of the largest po.rticle to be measured.
Table I gives the approximate density, viscosity and the diameter corresponding to the upper limit of Stokes' law, for several fluids used in this work. In general, the limit can be raised by increasing viscosity and density.
2Dalla Valle, J. M., Micromeritics--The Technology~ Fine Particles (2nd Edition), Pitman Publishing Company, 1948, p. 22.
APPLIC~TION OF STOKES' LAW
TABLE I
Limiting Stokes' Diameter and Physical Constants for Various Liquids Used
Benzene • Gasoline 1 Kerosene 1 part SAE - 10 oil to
1 part Kerosene
.8775
.720
.8054
.8464
*Assume a dust density of 1.44.
.00638
.0055
.01856
.0812
Limiting d - microns*
l.44 134 291
784
Figure 12. Analytical balance as used for determination of particle size distribution by sedimentation.
1
WEIGHT SCALE
MOVABLE CHAIN SUPPORT II l.l:d
BALANCE BEAM
DEWAR FLASK
SEDIMENTATION CYLINDER
.. ------~ .... .,.. ....... ...,... - ftr-flllllt.•aar-al-rA'9'1A&I ._,...ftA 0 A Til~
BALANCING PAN
f--1 (X).
~
~ t-3
~
~ ~ 0
~ ~~ til
~I ~I ~
~I ~
!APPARATUS 19
Andreason (2) found that Stokes' law could be applied to angular or cubical particles of th~ same weight as spherical particles. By calculating the particle size as the edge length of a cube of the same volume as a sphere of diameter d, his particle size conformed to the results of sieve analysis. This shape factor (SK) is found to be equal to 1.612.
All sedimentation results presented in this paper'are calculated using ~ = 1· .. 612; unless otherwise noted.
Apparatus
Briefly, the apparatus consists of a thin walled metal cylinder in the fonn of a cup 10.3 CM in height and 3. 5 em in diameter, suspended from a beam·balance so that the.upper lip is
~ not more than 2 mm below the surface of the suspension. The suspension may be contained in a Dewar vacuum flask or a constant temperature bath. An analyticai balance may be used, Figure 12. although the simple beam balance shown in Figures 13, 14, and 15 is much more convenient to operat~ since no small weights need be handled. Some of the early runs were made with a constant temperature bath that would maintain the temperature within 0.1 F. However, it was found that an ordinary Dewar vacuum flask will
~ maintain the temperature within 0.2 F under ordinary conditions.
If liquids that do not have known specific gravity and viscosity constants are used, it will be necessary to have a viscosimeter of the Ubeholdt or similar type. Density of the dust and of the liquid can be obtained in the usual manner with a 100 I
t ml pycnometer bottle.
Procedure·
Prep,aration of the susp,ension--The liquid selected for lthe suspending medium must meet certain requirements. Viscosity
and specific gravity should be such that Stokes'. law applies. The liquid must not affect the particles in any way. A non-vola~ile liquid is preferred because evaporation from the surface will ca.tise convection currents in the suspension.
Benzene was used for measuring distribution in size ranges below 50 microns because its viscosity and density remain constant from one batch to &nether. Kerosene was used successfully for particles up to 150 microns, and a half-and-half m:Lxture of kerosene and lubricating oil for suspensions containing particles larger than 150 microns.
Figure 14. Complete sedimentation appar&tus for pa~ticle size measure
ment by direct weight method.
~
Figure 15o Direct weight sedimentation apparatus with
flask removed showing cylinder.
SEDIMENTATION ANALYSIS 21
SEDIMENTATION ANALYSIS
Sample 4 Run __ 2 ________ From ________ _
Date May 6, 1949 Observer __ L_e.;;..s;g~g~---
CALCULATION OF DENSITIES Temp.
(1) Wt. of bottle - 8.4178 gms.
(2) Wt. of bottle and liquid - 16.2706 gms.
(3) Volume of bottle - 9.8000 cc
(4) Wt. of liquid (2) - (1) - 7.8528 gms.
(5) Density of liquid (2) - (l) (3) - .8025 gms./cc
(6) Wt. of bottle and powder - 11.7869 gms.
(7) Wt. of powder (6) - (1) - 3.3691 gms.
(8) Wt. of bottle, powder and liquid - 17.7656 grns.
(9) . (7~x~5~ 2.70 1.439· gms./cc - 1.874 Denslty of powder (2 + 7 -(8 )
STOKES' DIAMETER
Suspending liquid kerosene Viscosity , 0176 poise
Height of fall 10.45 em, Shape factor_..;;;;;;l~,6;;.::1;.:;:2=--------
D2 _ 18xl08 n h S _ 18xl08 X 10.45 X ,0176 (1,612)2 22900 - (d - d0 ) g t K- (1,439 - ,8025) 980,6 X t X 60 t
MOISTURE CONTENT
(1) Tare wt. of dish - 27.2956 gm.s.
(2) Dish and sample - 29.3940 gms,
(3) Dry sample and dish - 29.1988 gms.
I
Per cent moisture f~1 - f31 = .1952 - 10,30 ~ (4) - per cent 1 1,9032 -
Figure 16, Miscellaneous data sheet for sedimentation analysis.
22 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
After the suspending medium has been prepared, its viscosity and d~nsity should be determined. Then using the pycnometer bo_ttle the density should be determined. Figure 16 illustrates a form for this determination that has been found convenient.
rhe concentration of particles in the suspending medium is important. Figure 17 shows a series of runs that was made on a carefully mixed sample of rye dust. This series of runs .lrid1",;,;. · cates that below a concentration of about 2 gm of dust per 100 ml of suspending medium, concentration has- little effect on the results. For this reason, a concentration of 2 gm per 100 ml has been chosen for all of the work reported in this paper. Stairmand (ll) indicates that free falling conditions are not approached until the concentration is less than 0~2 per cent by volume, but since these very low concentrations decrease the-accuracy it is believed that nothing much is gained by using concentrations below about 2 per cent.
1&1 N (I)
~ :z: ... Ill: 1&1 z ii: ... z 1&1 u a: 1&1 L
00 I I ~f-T .--- -_.., --~ ~~~
-9, 5 GM/100 ML.
I I r') '/ v ~ b><~ _,...-:::: ec 3 GM/100 ML. ~
~ / ~ " I GM/100 ML.
· I I 7n J '-... 2 GM /100 ML. - !') v 1/ ~
~t'l 1 I ;/ I) l'~ r
4l'
.. ,., tfj w - I v /;~ ot'l
I"' /~ v """
~~ , ~ 0 2 4 6 8 l()o 12 14 18 18 20 22 24 26 28 30 32 34 36
SIZE • MICRONS
FIG. 17
PARTICLE SIZE BY SEDIMENTATION DIRECT WEIGHT METHOD - RYE FLOUR DUST
EFFECT OF PARTICLE CONCENTRATION
Preparation of samp~--When the apparatus is first set up, the volume of sample necessa~ to bring ~he liquid level in the flask 1 or 2 mm above the lip of the cylinder was determined. Also, the approximate weight on the balancing pan, with a blank
TIME-WEIGHT DaTA 23
sample in the flask,. was dete~ned. Next, the.weight of dust required for· the desired concentration was placed in the prede~er.mined volume of liquid and stirred from five to ten minutes
--with an electr;l.c stirrer. With the liquids used, no trouble was , experienced in obtaining a thorough dispersion. .
Time-weight data--While the sample was being stirred, the apparatus was checked, the thermome~er put in place and the. data sheets prepared. Then the dispersion wa& poured from the beaker into the flask,meanwhile continuously stirring the suspension in the beaker to prevent settling. As.soon as all of th~ suspension was in the flask; the stop watch was started and a weight determined as rapidly as po~sible by one· of two methods.
When us~ng the analytical balance, it was found best to preset the weights and observe the times of b~lance but with the beam balance, Figure 13, it was found best to observe the weight at specified times. The time interval may be varied from 20 seconds at. the beginning to 20 minutes at the end of 2 hours. It is important, however, to obta'in as many readings during the first several minutes as is possible. Usually time-weight da~a was taken until the weight remained constant for two readings, but on some runs where the percentage of fines was unimportant, the run was terminated earlier. The sample data sheet, Figure 18, shows
1 part of the data taken for run 1 of Figure 22.
Calculation of results--The first step in the calculation of the results was the plotting of the time-weight data as shown in Figure 19. The use of two time scales was found to increase accuracy and to reduce the size of the graph paper necessary ..
Earlier in this discussion it was mentioned that it is important to obtain several weights as soon as possible after the start of the run, becaus~ the first several points should fo~ a
1 line of constant slope until all of the largest size particles have reached the bottom of the.can. Therefore, the straight line through the first several points was extrapolated to ~ero time in order to determine-- the weight ·a.t zero watc_h time. If it were ob-vious that the points did not define a straight line, then the run should be rejected. The final weight was determined by the horizontal portion of the curve or by the weight when the run was discontinued. The difference between the extrapolated weight at zero time and the final weight was then taken as the total weight settled.
Next, tangents to the curve at various times were drawn. Then the percentage of the.total weight o~ particles settling out between any two given times is the ratio of the distance between
24 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
SEDIMENTATION ANALYSIS
Sample Run 1 From Poor Rgl1l Su~:tiQD Date Mal 62 1949 Remarks Pa~e 1 Observer Legg 2 g/100 m/concentration
Elapsed Weight Correspond Per Cent Time Settled TemE• - OF Size - U Finer
Min. 1 1.65 79.5 1.5 123.5 98.5 2
1 3.70 79.5 2 107 89.8 1~ 5.35 79.5 2.5 96 80.2 2 6.50 79.5 3 87.5 72.0 2~ 7.40 79.5 3.5 81 63.2 3 8.10 79.5 4.5 71.4 57.4 3~ 8.60 79.5 5.5 64.5 52.1 4 9.00 79.5 7 57.2 44.0 4~ 9.40 79.5 9 50.5 37.5 5 10.00 79.5 11 45.7 30.9 5~ 10.20 79.5 15 39.0 25.7 6 10.20 79.5 18 35.7 19.8 7 10.60 79.5 20 33.8 16.5 8 10.90 79.5 25 30.2 14.2 9 11.15 79.5 35 25.6 10.2
10 11.35 79.5 40 23.9 6.6 12 11.65 79.5 80 16.9 4.4 15~ 12.00 79.5 90 16.0 2.2 18 12.30 79.5 20 12.45 79.5 25 12.60 79.5 30 12.75 79.5 40 12.90 79.5 50 12.95 79.5
Figure 18. Data sheet for sedimentation analysis.
PROBLEM 25
the intercepts of the two tangents on the weight scale to the total change in weight. By calculating the size of particle which will have completely fallen out at the given times, the percentage by weight in the given size interval was determined. However, it is more desirable to express the results in cumulative percentages finer than a given size. Referring to Figure 19 and the data sheet, Figure 18, the following calculation will serve to show how the percentage finer than a given size is calculated.
Problem; To find the percentage of particles in the sample finer than the Stokes' diameter corresponding to a time of 3 minutes.
1. Draw a tangent to the weight-time curve at 3 minutes intercepting the weight axis. (point C)
2. Calculate the percentage of particles finer than size.
% = AC 8 6 AB X 100 - 13: 6 x 100 - 63.2%
where AC = distance between intercept at 3 minutes and final weight
AB = total weight settled out.
3. Calculate the Stokes' diameter corresponding to the time of 3 minutes.
2 8 _ 10 X 10· J,( h
d- (1'-~)gtxso where:
)-! = .0176 poise
h - 10.45 em
;Oo = .8025 gm/cc (density of liquid)
fJ = 1.439 gm/cc (density of dust)
SK - 1.612 (shape factor)
g = 980.6
t = 3 minutes
~-
0-
B, I I I I I I I I 0 2 3 4 5 6 7 8 9 10 II
EXPANDED SCALE I I I I I I I I I I I I 0 10 20 30 40 50 60 70 80 90 100 110
CONDENSED SCALE
--TIME IN MINUTES--
WEIGHT VS TIME PLOT PARTICLE SIZE BY SEDIMENTATION ---- D1RECT WEIGHT METHOD
POOR ROLL SUCTION -RUN I
CONCENTRATION I 2 G /100M L. KEROSENE
FIG. 19
I I
I I
PROBLEM
100
90
1&.180 N
c;;70 z C(
~60
a: ~so ii:
40 ... z ~!0 a: 1&.1 CL20
"" N
10
100
90
80
Cii 70
z I ~ 60
.... a: ~50 La: ~ 40
"" l ffi 30 I a..
20
10
2 4 6
27
-~ ~ e::::-~
~ ~ t;::?O
~ v ~ tr
~ v It r7"
.A r
" A~ C RUN NO. I
~'~ t::t. RUN NO.2 -
0 RUN NO.3
8 10 12 14 16 18 20 22 24 26 28 !0 32 34 36 38 40 SIZE IN MICRONS
FIG. 20 PARTICLE SIZE BY SEDIMENTATION
DIRECT WEIGHT METHOD- RYE FLOUR DUST
~ ~ t---/
~ ~ v .--a
y ~ ~ ~
7'. VA v ) ~ ~ ~ P"'
~
Jed ~ o -RUN NO. I
~~ 'ex c -RUN NO. 2
l:J. -RUN NO.3
J~ x -RUN NO. 4
B -RUN NO. 5
I I I
f-a
-~ o 10 20 30 40 50 so 10 eo 90 100 110 120 1'30 140 150 1eo 110
SIZE - MICRONS
FIG. 21- PARTICLE SIZE BY SEDIMENTATION
DIRECT WELGHT METHOD- FAMILY PASTRY FLOUR
28 APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
d - diameter of particle - microns
d 18 X 108 X 0.0176 X 10.45 (1.439 - .8025) 980.6 X 60 X t X l.bl2
=J2'3~oo =J23~00 = 87.5 microns
Steps 1, 2, and 3 above, were repeated for each time at which it was desired to obtain a size and percentage; the resulting values were plotted to give the complete particle size distribution curve as shown in Figure 22.
Markley (9) gives an analytical method for obtaining the distribution curve from the time-weight data but it is believed that the graphical method is more aGcurate and rapid.
100
90
80
~70 N (f)
z60 cr I 1-50 a:: ~
~40 La.
~ 30 ~ 0 a:: ~20
10
RU~ 4-~
~ ~r ~ ~
I
~ If ~ RUN 5/
~ ~ ~ L RUN I
J ~ ~RUN 2
.~ v ~,
~ r--- RUN 3
~ I 2 ~M. I 1100 MT.
~ ~
sQI 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
SIZE - MICRONS
FIG. 22
PARTICLE SIZE BY SEDIMENTATION
DIRECT WEIGHT METHOD- POOR ROLL SUCTION
RESULTS I 29 1
Results
Figures 20, 21, and 22 show the results of a number of runs on different mill products. It is to be noted that the largest deviations are at the coarse end of the curve. This is due to the close spacing of points in the initial straight portion of the time-weight curve. Increasing the viscosity of the suspending medium prolongs the period during which the largest particles are settling permitting the wider spacing of points in this segment of the time-weight curve and hence more accurate plotting.
Figure 23 is a comparison of the average of the micro- . scopic and sedimentation analysis made on the ~e dust. In this figure the sedimentation results have been plotted both with.and without the shape factor of 1.612. Figures 24 and 25 also show a comparison of microscope and sedimentation results on a flour
.and a poor roll suction sample. A comparison of these curves shows that the microscope results do not correspond to the sedimentation results even with the factor, SK· Further study of these and other sets of data indicate that there is a different shape factor for each size range. The factor for any given size range, however, appears to be· common for all mill products covered by this study.
100
90
80 1&1 N
1
:70 ~60
a:: ~50 i&:
1.,_ 40 z 1&1
~30
20
10
SEDIMENTATION : SPHERICAL -PARTICLES
I I
) I
I /v v_L ~
MICROSCOPE-----...... lvJv / I I I
II"
/~ ~
~ ~ ~)
~
lY ~ 'F-
rv .......1 v d f"
L ~ L_ rf/· /
/ .I
~ /I SEDIMENTATION : CUBICAL PARTICLES
}/
;' L
~ I 0 0 2 4' • 6 8 10 12 14 18 18 20 22 24 26 28 30 32 34 36 38
SIZE IN MICRONS
FIG. 23
MICROSCOPE VS. SED I MENTATION
AVERAGE OF RUNS - RYE FLOUR DUST
30
100
90
eo
LLI70 N (/)
z60 <t :I: 1-50 a:: lLI
~40 LL.
1-z 30 lLI 0 a:: ~20
10 ~ v v
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
,.~ ~ 7 /
~ v v~
v / K-sEOIMENTATION AVERAGE I
~ v r\__ I I I I
MICROSCOPIC AVERAGE
~ v ~v
~ ~ ~ v
J.,::::: 0 o 10 20 30 40 50 so 10 eo 90 100 110 120 130 140 150 160 110 1eo
lLI N
100
90
eo
u; 70
z ~ 60 1-
ffi 50 z LL.
40 1-z lLI ~ 30 lLI 0..
20
10
SIZE - MICRONS
FIG. 24
AVERAGES OF SEDIMENTATION a MICROSCOPIC DETERMINATION
OF PARTICLE SIZE - FAMILY PASTRY FLOUR
~ J,....--v v-
./
v~ / /
/v ~v / X v f\-MICROSCOPIC AVERAGE
/ l I I _l_
/ v I I I I
/ f\- SEDIMENTATION AVERAGE
/ v v
/
L_ / v
~ v v v
~ ~ 0
o 10 20 30 40 50 60 10 eo 90 100 110 120 130 140 150 160 110 1eo SIZE - MICRONS
FIG. 25
AVERAGES OF SEDIMENTATION a MICROSCOPIC DETERMINATIONS
OF PARTICLE SIZE - POOR ROLL SUCTION
OTHER OBSERVATIONS 31
The Andre~son sedimentation technique has been considered as a. ~tandard by many investigators. A comparison of results on the rye dust as determined by the direct weight and Andreason methods is presented in Figure 26. Agreement is very good in the middle range of the distribution; but deviates at the end points.
A number of conclusions can be dr~wn from Figure 27. ihe fact is, that SKis greater than 1.612 above 30 microns and nearly constant. The microscope measurement is high due to particles having their least dimension perpendicular. We might assume that density and shape of the particles are nearly constant above that size. However, below about 30 microns the particles are predominantly spherical in shape and of higher density thus reducing ~· It must be understood that this is a coefficient that accounts for several variables such as density variations with particle size, changes in shape, and the tendency for particles to lie flat on a microscope slide. Thus it would be possible to correct observed sedimentation results from the SK values for each size range as selected from Figure 27 giving very close agreement with microscopic observatio~s.
Other Observations
Following are some useful observations made while developing the sedimentation apparatus:
1. The lip of the cup must be no more than 2 mm below the surface of the suspension for consistent results.
2. A concentration of no more than 2 gms of particulate material per 100 mls of liquid should be used. In any case the concentration should be kept constant to make runs comparative.
3. Vertical oscillations of the cylinder must be kept as low as possible, preferably less than 0.5 mm.
·4. When using a volatile suspending medium a fairly tight cover should be fitted on the flask to prevent evaporation. evaporation is serious because the cooling at the surface causes convection currents to form in the suspension.
32
100
90
80
70 ILl N <nso z 4 X 50 1-a:: ~40
i:i: 1-30 z ILl 0 a:: 20 ILl n.
10
APPAMTUS AND TECHNIQUES FOR FLOUR MilL DUST
I l I l /:~~to-DIRECT WEIGHT~ v
I j ~ANDREASON
f I I
I -
I I
AND~EASON-\y
+1 10 20 30 40 50 60
PARTICLE SIZE - MICRONS
FIG. 26
ANDREASON VS. DIRECT WEIGHT SEDIMENTATION
RYE DUST
70
2.2r----r-~-r---,--~-r-----r---r--r-----r---r--r----r--r--r-----r--r----,
2.011-----t---t--t----t---t-----'<,......_--t--~( "---t--+----1--+--+----1--t---+---lf---1
+ RYE OUST
0 REG. PAT. FLOUR
c FAMILY PASTRY FLOUR
A POOR ROLL SUCTION -
STOKES' DIAMETER - MICRONS
FIG. 27
RELATION BETWEEN STO~ES' DIAMETER
AND MICROSCOPIC DIAMETER
WICROSCOPE DIA, = ~· ITOKES' DIA.
CONCLUSIONS 33
5. Temperature duringarun should not vary more than 0.2 F. If the temperature of the suspension is &t room temperature before it is put in the flask, this is easily maintained.
6. Graphical differentiation of the time-weight curve should be performed carefully, but errors tend to average out except at the end points.
Conclusions
The main purpose of this investigation was to select and correlate methods for determining subsieve particle size distribution that could be adapted to the study of mill dusts. Methods were desired which would be rapid, accurate, adapted to a wide range of measurement, reasonably simple to operate and relatively inexpensive. The purpose of this preliminary investigation was not to study dusts but to develop the methods of stuqy. The achievenB nt of this purpose is ref:iected in the following conclusions:
l. Twc methods of measurement suitable for the subsieve range have been selected, namely direct measurement by rnicroprojector and indirect measurement by means of the direct-weight sedimentation apparatus.
2. Satisf&ctory procedures for sample selection, sample dispersion, counting, and expression of results have been developed for the rnicroprojection method.
3. Satisfactory apparatus has been developed, factors relating to its operation have been evaluated, and procedures for operation have been established for the direct-weight sedimentation.
4. Ivlicroprojection and sedimentation .ua.v e oeen compared and the shape factor SK has been shown to varY with particle size.
34
1.
2.
3.
4.
.5.
6.
7.
8.
9.
10.
11.
APPARATUS AND TECHNIQUES FOR FLOUR MILL DUST
BIBLIOGRAPHY
Allen, H. S., "On the Motion of a Sphere in a Viscous Fluid," Philosophy, Magazine, Vol. 1, pp. 323-519. 1900.
Andreason, A. H. M., "Uber die Gultigheit des Stokes' Schen Ge setzes fUr Nicht Kugelformige Teilchen, 11 Kolloid z, Vol. 48, ' pp. 175-179. 1929.
Dalla Valle, J. M., Micromeritics--The TechnologY. of Fine Particles, 2nd Edition, New York, Pitman Publishing Company. 1948.
Dotts, W. M., "Measuring tne Particle Size Distribution of Dispersed Systems," Analy:tic Chemistr:r-, Vol. 18, p. 326. 1946.
Fairs, G. L., "The Use of the Microscope in Particle Size Analysis, 11 Chemistry and Industn, Vol. 62, pp. 374-378. October, 1943.
Hildebrand, F. C., Ferrari, C. G., Borchardt, L. F., and Anker, C. A., "The Determination of Flour Particle Size," Cereal Chemistry:, Vol. 19, pp. 805-818. 1942.
Jacobsen, A. E. and Sullivan, W. F., "Method of Particle Size Distribution for the Entire Sub sieve Range,"" AnalY-tical ChemistfY-, Vol. 19, pp. 855-860. November, 1947.
Kent Jones, D. w., "Studies in Flour Granularity," Cereal ChemistrY-, Vol. 18, pp. 358-369. 1941.
Markley, M. C., "Flour Particle Size by Sedimentation Method," Cereal Chemistry~ Vol. 11, pp. 654-660. 1934.
Oden, s. A. "A New Method for Determining the Distribution with Respect to their Diameters of Particles in ·suspension," Kollo'id Z., Vol. 18, pp. 33-48. 1916.
Stairm.and, C. J., "Some Practical Aspects of Particle Size Analysis," Transactions Institute Chemical Epgineers, (London), pp. 58-68. February, 1947.
1.
BIBLIOGRAPHY 35
12. Wichser, F. W. and Shellenberger, J. A., "Methods for Determining Flour Particle Size Distribution," Cereal Chemistry, Vol. 25, pp. 155-167. 1948.
13. Work, L. T., "Methods of Particle Size Detennination," ~ ceedings of the American 3ocietx' for Testing Materials, Vol. 28, pp. 771-812. 1928.
36 ENGINEERING EXPERIMENT STATIO
The, Engineering Experiment Station of the University of Minne· sota was established by an act of the Board of Regents on Decembe: 13, 1921.
The purpose of the Station is to advance research and graduat~ study in the Institute of Technology, to conduct scientific and in· dustrial investigations, and to cooperate with governmental bodies technical societies, associations, industries, or public utili tie: in the solution of technical problems. The results of scientifi1 investigations will be published in the form of bulletins and tech· nical papers. Infonnation which is of general interest and yet no1 the result of original rese~rch may be distributed in the form o~ circulars.
For a complete list of publications or other information con~ cerning the work of the Station, address the Director of the Engi~ neering Experiment Station.
BULLETINS AVAILABLE
12. Thermal Conductivity of Building Materials, by Frank B. and A. B. Algren. x +134 pages, 109 illustrations. $1.50. (Purchased through University Press.)
14. Square Sections of Reinforced Concrete under Thrust and Nonsym metrical Bending, by Paul Andersen. vi+ 42 pages, 8 figures 23 diagrams. 1939.
15. Laborato~ Studies of Asphalt Cements, by Fred C. Lang and T. Thomas. x + 96 pages, 43 illustrations. 1939.
16. Factors Affecting the Performance and Rating of Air Filters, Frank B. Rowley and Richard C. Jordan. viii + 54 pages, illustrations. 1939.
18. Condensation of Moisture and Its Relation to Building Construe tion and Operation, by Frank B. Rowley, A. B. Algren, ~n C. E. Lund. vi+ 69 pages, 28 illustrations. 1941.
19. Pulp, Paper, and Ins~ation Mill Waste Analysis, by Frank B. Rowley, Richard C. Jorrl.an, Reuben M. Olson, and Richard F. Huettl. vi +55 pages, 34 illustrations. 1942.
20. Conservation of Fuel, by Frank B. Rowley, aichard C. and C. E. Lund. vi+ 61 pages, 22 illustrations, 17 1943.
21. Aids to Technical Writing, by R. C. ·Jordan and M. J. Edwards. viii+ 112 pages, 60 illustrations, 13 tables. May, 1944.
24. Factors Affecting Heat Transmission through Insulated Walls, b Frank B. Rowley arid C. E. Lund. iv+ 25 pages, 8 illustrations, 8 tables. April, 1946.
25. Vapor Resistant Coatings for Structural Insulating Board, by Frank B. Rowley, M. H. LaJoy and E. T. Erickson. vi+ 3 pages, 9 illustrations, 10 tables. September, 1946.
~UBLICATIONS 37
26. Moisture and Temperature Control in Buildings Utilizing Structural Insulating Board, hy Frank B. Rowley, Millard H. LaJoy, and Einar T. Erickson. Vl. ·+38 pages, 16 illustrations, 8 tables. July, 1947.
27. Water Permeability of Structural Clay Tile Facing Walls, by J. A. Wise. iv + 32 pages, 26 illustrations, 4 tables. August, 1948.
28. Thermal Properties of Soils, by Miles S. Kersten. xiv + 227 pages, 138 figures, 15 t~bles, 5 plates. June, l949.
29. Pro.ceedings of the Symposium on Engineering Research, edited by C. E. Lund. x + 110 pages. August, 1949.
~0. Some Causes of Paint Peeling, by Frank B. Rowley and Millard H. LaJoy. vi+ 34 pages, 23 illustrations, 11 tables. September, 1949.
TECHNICAL PAPERS AVAILABLE
1. Condensation within Walls, by Frank B. Rowl~, A. B. Algren, and C. E. Lund. 12 pages. Janua~, 1938.
3. Air Filter Performance as Affected by Kind of Dust, Rate of Dust Feed, and Air Velocity through Filter, by Frank B. Rowley and R. C. Jordan. 10 page::. June, 1938.
32. Construction and Operation of a 15-inch Cupola, by Fulton Holtby. 4 pages. August, 1941.
~2. Abnormal Currents in Distribution Transformers Due to Lightning, by J. M·. Bryant and M. Newman. 5 pages. September, 1942.
~6. Discoloration Method of Rating Air Filter, by Frank B. Rowley and R. C. Jordan. 10 pages. September, 1943.
~8. Valve Guide Leakage in an Automotive Engine, by M. A. Lindeman I and B. J. Robertson. 22 pages. May, 1944. 52. Carbon Dioxide Variation in a Vented Stack, by :t-1. H. LaJoy. 27
pages. May, 1945. 53. Thermal Conductivity of Insulating Material at Low Mean Tempera
tures, by Frank B. Rowl~, R. C. Jordan, and R. M. Lander. 6 pages. December, 1945.
Calculation of Bearing Capacities of Footings by Circular Arcs, by Paul Andersen. 3 pages. June, 1946.
Comfort Reactions of 275 Wor~en during Occupancy of.Air-Conditioned Offices, by Frank'B. Rowley, R. C. Jordan, and W. E. Snyder. 4 pages. June, 1947.
A Statistical Analysis of Water Works Data for 1945, by G. J. Schroepfer, A. S. Johnson, H. F. Seid·el, and M. B. Al-Hakim. 32 pages. October, 1948.
Theory and Use of Capilla~ Tube Expansion Device, by M. M. Bolstad and R. C. Jordan. 6 pages. December, 1948.
Heating .Panel Time Response Study, by A. B. Algren and Ben Ciseel. 4 pages. lvfarch, 1949.
38 ENGINEERING EXPERIMENT STATIOl
65. L~pact Strength Testing Machine, by Frank B. Rowley and M. H. LaJoy. 16 pages. June, 1949.
66. Ground Temperatures as Affected by Weather Conditions, by A. B, Algren. 6 pages. June, 1949.
67. Theory and Use of Capillary Tube Expansion Device, Part II, Nonadiabatic Flow, by M. M. Bolstad and R. C. Jordan. 7 pages, June, 1949.
68. Thermal Conductivity of Soils, by M. S. Kersten. July, 1949. 69. Specific heat Tests on Soils, by M. S. Kersten. 5 pages. May:
1949. 72. rlesistance Gradients through Viscous Coated Air Filters, by Franl
B. Rowley and R. C. Jordan. 8 pages. December, 1949.
CIRCULARS AVAILABLE
3. A Five-Year l-iectanical Engineering Curriculum, by R. C. Jordan, 5 pages. Januars, 1947.