Particle size spectrum and compressibility of raw and shredded municipal solid waste


Material Information

Particle size spectrum and compressibility of raw and shredded municipal solid waste
Physical Description:
xix, 353 leaves. : illus. ; 28 cm.
Ruf, John Adam, 1939-
Publication Date:


Subjects / Keywords:
Refuse and refuse disposal -- Mathematical models   ( lcsh )
Factory and trade waste -- Analysis   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Bibliography: leaves 340-351.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 14198653
System ID:

Full Text







To my parents




I would like to express my sincere appreciation to my

committee chairman, Dr. James P. Heaney, for his guidance,

assistance and inspiration, which were necessary for the com-

pletion of this dissertation. The support and contributions

made by other members of the committee, namely, Dr. Russell

II. Susag, Dr. James H. Schaub, Dr. Charles C. Hortenstine,

Dr. David T. Knuth, and Professor Thomas deS. Furman, also

are gratefully acknowledged.

I am deeply indebted to Mr. Ken Fess, who collaborated

with and assisted me in the early stages of the study. Deep

thanks are extended to Bill Maloney, Don Sloan, and John

Townsend, who assisted in the laboratory analyses. Appreci-

ation is extended to Mr. Herb Houston of the-Gainesville

Municipal Waste Conversion Authority and Mr. Carlton Wiles

of the Johnson City, Tennessee, composting plant, who assisted

in the collection of shredded waste samples from their plants.

Appreciation and thanks also go to Mrs. Carol Brody (editor),

Mrs. Elizabeth Godey (typist), and Mr. Ashley Wood (draftsman).

A special thank you is extended to my wife, Maureen, for

her continuous help and encouragement throughout the long and

tedious project.


Recognition and appreciation are extended to the Depart-

ments of Environmental Engineering and Mechanical Engineering

of the University of Florida for supplying the equipment,

materials and facilities which were used during the course of

the project.

Deep appreciation is extended to the U.S. Public Health

Service and the U.S. Environmental Protection Agency, whose

financial support and active duty assignments at the Univer-

sity of Florida for this advanced degree made this research






ABSTRACT . . ... .xvii



Objectives of Investigation ... 1
Rationale of the Approach . 5


Introduction . . 8
Methods of Previous Investigators .. 9
Sampling ... . 9
Sample Reduction . ... 12
Size of Gross Samples . 13
Sample Drying and Storage ... 15
Experimental Methods and Equipment ....... 19
Sampling . ... 19
Sample Preparation . 23


Introduction . . 29
Methods of Previous Investigators 33
Introduction . . 33
Refuse Particle Size Analysis 37
Particle Size Distributions of Refuse
Components . ..... 60
Refuse Composition Classification
Techniques . ... 73




III (continued)

Commercial Application of Screens ... .78
Particle Size Requirements for Various
Industrial Processes . 83
Experimental Methods and Equipment .. 85
Introduction .. . 85
Equipment . . 88
Size of Test Samples . ... 96
Screening Procedures . .. 104
Duration of Sieving ........ .107
Refuse Composition Classification
Techniques . . 112


Introduction . . .117
Methods of Previous Investigators .. .128
Density . . 128
Compressibility . 139
Experimental Methods and Equipment 159
Introduction . .. 159
Equipment . . 162
Analytical Techniques . .166

ANALYSIS . . 173

Introduction . . 173
Analytical Techniques . ... 174
Frequency Distributions . .. 181
Sample Statistical Parameters ... .196
Model Fitting. .... . .. .198
Analysis of Variance . 202




Introduction .
Analytical Techniques
Display of Results .
Analysis of Variance



. 215

. 215
. 216
. .. .218
. 219

. 231

Particle Size Analysis . .
Frequency Distributions and Statistical
Parameters .
Analysis of Variance .
Predicting Particle Size Distributions
and Statistical Parameters .
Density and Compression Analysis .
Loose Bulk Density Analysis .
Compression Curves . .
Rebound Analysis . .
Model Fitting and Correlation with
Particle Size Statistical Parameters


Conclusions. . .
Recommendations . .









* 231

. 231
* 251

* 252
* 258
* 258
. 261
* 265

* 267

* 277

. 277
. 281

. 285

. 290

. 299

S. 302


. 311

. 314










Table Page

1 Relative Effect of Temperature on Drying
Soil . . 16

2 Summary of Samples. . .. .24

3 Parameters of Selected Components of
Shredded Waste . . 69

4 Parameters of Glass Component from Shredded
Waste . .. .. .... 71

5 Gilson Screen Data Used in Research 94

6 Size of Partial Openings in the Gilson
Screens .. . ... 95

7 ASTM Suggested Bulk Volumes and Weights of
Test Samples . ... 100

8 Test Sample Sizes Utilized by Previous
Investigators . . 101

9 Sizes of Sieve Test Samples Used in Research 103

10 Approximate Densities of Common Solids .. .121

11 Calculated True Density of Refuse .. 123

12 Reported Densities of Various Types of
Refuse .. ... 134

13 Density of Sized Shredded Refuse ... .137

14 Wet Weight Densities of Various Components
of Refuse . . 138

15 Sizes of Density and Compression Test
Samples . . .. .. 168

16 Summary of Difference in Weights of Sieve
Test Samples and Sum of Sieve Fractions 175

17 Estimation of Population by Subsamples ... 178


Table Page

18 Recommended Subsample Weights for Various
Sieve Fractions . . 179

19 Size of Subsamples Used . ... .180

20 Frequency Distribution of Paper Component
of Raw Waste . . 184

21 Reproducibility of Sieving Refuse with the
Gilson Sieve . ... .204

22 Reproducibility of Classification of Refuse 207

23 Analysis of Variance Data of Twin Refuse
Samples . . .. .209

24 Analysis of Variance for Twin Refuse Sample
Data . . 209

25 Analysis of Variance Data of Refuse Sieve
Samples . . 213

26 Analysis of Variance of Refuse Sieve Sample
Data . . ... .213

27 Analysis of Variance of the Density-Compression
Data for Different Waste Types . 224

28 Sources of Variation of Density and
Compression Data . ... 228

29 Statistical Parameters of Various Types of
Raw and Processed Waste . 241

30 Effect of Shredder Grate Size on Shredded
Waste . . 244

31 Size Reduction of Waste Components by
Different Types of Shredders . 246

32 Statistical Parameters of Different Types
of Raw and Shredded Waste Components 249

33 Comparison of Theoretical with Actual
Composite Distributions . .. 254

34 Weighted Average Composition of Waste from
Gainesville, Florida . ... .255


Table Page

35 Comparison of Standardized (Same Composition)
and Empirical Composite Distributions .. .256

36 Loose Bulk Density of Various Waste Types 258

37 Expansion Characteristics of Compressed
Waste .. .. . .265

A-1 Raw Refuse Samples Collected from the
Gainesville, Florida, Landfill ... 285

A-2 Primary Shredded Samples Collected from the
Conveyor Belt after the Primary Shredder at
the Gainesville, Florida, Compost Plant .. 286

A-3 Secondary Shredded Samples Collected from
the "Tripper" after the Secondary Shredder
at the Gainesville, Florida, Compost Plant 287

A-4 Rasp Shredder Samples Collected after the
Rasp Shredder at the Johnson City, Tennessee,
Compost Plant . . 288

B-l Summary of Refuse Composition at Gainesville,
Florida . . 292

B-2 Gainesville, Florida, Shredder Specifi-
cations .. . 295

D-l Gilson Screen Sizes . ... 304

H-1 Raw Waste Particle Size Statistical Data 318

1-2 Primary Waste Particle Size Statistical Data 319

H-3 Secondary Waste Particle Size Statistical
Data . . 320

H-4 Rasp Waste Particle Size Statistical Data 321

1-1 Table of Common Probability Distributions 323

1-2 Fitted Distributions to the Particle Size
Data . . ... 324


Figure Page

1 Particle size distributions of various types
of waste . . 38

2 Particle size distributions of various types
of shredded waste .. ..... 49

3 Particle size distributions of compost and
shredded waste at Gainesville, Florida 50

4 Comparison of moisture content on mini-shred
setting on an Eidal Shredder ... 50

5 Cumulative number of objects per ton of raw
refuse with longest dimension greater than
given lengths . ... 53

6 Cumulative weight of objects with longest
dimension greater than given lengths .. 53

7 Cumulative number of objects per ton of raw
refuse which will pass through a square mesh
plotted vs. mesh size . ... 54

8 Cumulative weight of objects which are greater
than a square mesh, plotted vs. mesh size 54

9 Effect of grate size on particle size dis-
tribution at Madison, Wisconsin ... 56

10 Comparison of sieving and direct particle
measurement at Madison, Wisconsin .. 56

11 Replicate sieve tests of refuse sample from
the Tollemache Mill at Madison, Wisconsin 57

12 Particle size distributions of waste shredded
by the Gondard and Tollemache Mills at
Madison, Wisconsin . 57

13 Effect of moisture content on particle size
distribution at Madison, Wisconsin .. 61






. 61

. 64

. 66

14 Changes in particle size distribution due to
cumulative hammer wear at Madison, Wisconsin

15 Number of objects per ton of raw refuse vs.
length for principal categories of refuse

16 Total weight per ton of raw refuse of objects
in given length ranges for principal cate-
gories of refuse . .

17 Particle size distributions of metal and
glass waste . .

18 Particle size distributions of miscellaneous
components of waste . .

19 Photographs of the Gilson mechanical testing
screen . . .

20 Cumulative sieving vs. period of sieving .

21 Compaction of size reduced waste at Texas

22 Compaction of artificial waste at West
Virginia . . .

23 Compressibility of solid waste .

24 Energy versus volume reduction for several
refuse components . .

25 Compression equipment . .

26 Cross-sectional diagram of the compression
cylinder . . .

27 Schematic of subsampling procedure .

28 Frequency distributions of paper component
of raw waste .. . .

29 Particle size distribution of raw waste from
Gainesville, Florida . .

30 Particle size frequency distribution of
primary shredded waste from Gainesville,
Florida . . .

S. 93

. 109

. 146

S. 146

. 149

. 156

. 164

. 164

. 176

. 186

. 187

S. 188

. .

. .


Figure Page

31 Particle size distribution of secondary
shredded waste from Gainesville, Florida 189

32 Particle size distribution of rasp shredded
waste from Johnson City, Tennessee 190

33 Cumulative distributions of raw waste 192

34 Cumulative distributionsof primary waste 193

35 Cumulative distributions of secondary waste 194

36 Cumulative distributions of rasp waste 195

37 Reproducibility of sieving . ... 206

38 Variation of particle size of twin samples 211

39 Particle size distribution of modified raw
waste from Gainesville, Florida . 217

40 Compressibility of modified raw solid waste .. 220

41 Compressibility of solid waste after primary
shredding ... . 221

42 Compressibility of solid waste after secondary
shredding ... . 222

43 Compressibility of solid waste passed through
a rasp . . ... 223

44 Compressibility of test samples from same
gross samples . ... 226

45 Particle size distributions of composite
wastes from Gainesville, Florida, and
Johnson City, Tennessee . 232

46 Particle size distributions of raw waste
and components from Gainesville, Florida 234

47 Particle size distributions of primary
shredded waste and components from
Gainesville, Florida . 235



Figure Page

48 Particle size distributions of secondary
shredded waste and components from
Gainesville, Florida. .. . .236

49 Particle size distributions of rasp shredded
waste and components from Johnson City,
Tennessee . . 237

50 Compressibility of various types of waste 262

51 Mathematical model of compression data 270

52 Mathematical model of rebound compressive
data . . 272

53 Comparison of compression mathematical
models to the empirical curves .. .274

I-1 Particle size distributions of food waste
from Gainesville, Florida, and Johnson
City, Tennessee . . 328

I-2 Particle size distributions of garden waste
from Gainesville, Florida, and Johnson
City, Tennessee . .. 329

1-3 Particle size distributions of cardboard
waste from Gainesville, Florida, and
Johnson City, Tennessee .. . .330

I-4 Particle size distributions of paper waste
from Gainesville, Florida, and Johnson
City, Tennessee ...... .. .331

I-5 Particle size distributions of plastic
waste from Gainesville, Florida, and
Johnson City, Tennessee . .. .332

1-6 Particle size distributions of textile
waste from Gainesville, Florida, and
Johnson City, Tennessee. . .333

1-7 Particle size distributions of wood waste
from Gainesville, Florida, and Johnson
City, Tennessee . .334


Figure Page

I-8 Particle size distributions of ferrous waste
from Gainesville, Florida, and Johnson City,
Tennessee . . .335

I-9 Particle size distributions of non-ferrous
metal waste from Gainesville, Florida, and
Johnson City, Tennessee . ... .336

1-10 Particle size distributions of glass waste
from Gainesville, Florida, and Johnson City,
Tennessee . .. . 337

1-11 Particle size distributions of sand and
rock waste from Gainesville, Florida, and
Johnson City, Tennessee . ... .338

1-12 Particle size distributions of miscellaneous
waste from Gainesville, Florida, and Johnson
City, Tennessee .. . 339


Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



John Adam Ruf

December, 1974

Chairman: James P. Heaney
Major Department: Environmental Engineering Sciences

The particle size spectrum and density-compressibility

of raw and shredded municipal solid waste were investigated.

The particle size spectrum was determined using large soil-

aggregate sieves and large size samples. The dried samples

varied in size from 2.17 to 16.55 lbs. Four different types

of waste were analyzed: raw refuse and three different types

of shredded waste. A total of 63 test samples were sieved.

The sieve fractions of each test sample were separated into

their component parts, e.g., paper, glass, plastic, etc.

Particle size distributions were obtained for the composite

wastes and their constituents.

The density and compressibility of "twin samples" of the

particle size analysis were determined using a 12.094 in. ID

x 10.969 in. test cylinder. Some 79 test samples were analyzed

by the combined density-compression test. The average sample

size was 3.23 lbs. The raw and shredded waste materials were

gently placed into the cylinder and the loose bulk density


determined for each test. The samples were statically loaded

using a hand-pumped hydraulic cylinder with large steel pis-

ton. Increased pressures from 0 to 174.7 psi were incremen-

tally applied to the test samples and the corresponding densi-

ties measured. The total loading time averaged 29.3 minutes.

Three rebound measurements were made during the analysis.

The particle size and the combined density-compression

analytical techniques were found to be dependable and repro-

ducible. The major variation resulted from the different

compositions of test samples.

Particle size distributions were obtained for the refuse

components. Mathematical functions were fitted to each dis-

tribution and methodologies were developed for predicting

particle size distributions of different raw and processed

composite wastes. Hammermills tended to reduce the waste to

log-normal type distributions while rasp shredders tended to

produce beta distributions. Hammermill shredders increased

the coefficient of variation, while it was slightly reduced

by the rasp.

The mean particle size of shredded waste was found to be

directly proportional to the size of the shredder grate, as

measured by the minimum width of the largest grate opening.

Each waste component reacted differently to shredding

and to the different types of shredders. Glass was the

easiest to reduce by the hammermills, while textiles, plastic

and metals were difficult to break down. The rasp produced

considerably coarser glass than the hammermills.


The dry bulk density from the compression curves was

found to vary with the logarithm of applied pressures. The

waste particle size distribution (as defined by the product

of the mean and standard deviation) and the waste composition

also were important variables. Mathematical functions were

fitted to the data. The models were very accurate in predict-

ing the data obtained from the study and those reported by

others. The equations were modified to account for moisture





Objectives of Investigation

Mixed refuse or solid waste is a heterogeneous mass of

particles of different materials, e.g., paper, glass, plas-

tic, metal, etc., of different sizes, configurations, rigidity,

densities, and other characteristics. Its composition varies

considerably depending on where it is collected (residence,

stores, etc.), the season of the year, the day of the week,

the size and type of community, and the geographic location.

Its physical and chemical characteristics are directly affected

by the various materials or components of the mixture.

Individual particles of raw refuse are almost always

irregular and may range from small items, such as soil, sand,

and grass, to bulky items, such as discarded furniture and

appliances. The configuration of the particles may vary from

long cylinders (e.g., tree branches), thin sheets (e.g.,

paper, plastics, fabrics), short cylinders (e.g., cans and

bottles), spheroids (e.g., sand, rock, and food), large rec-

tangular items (e.g., boxes and cartons), to small jagged

particles. Because of its many components, the individual

particles of municipal refuse can be large or small, organic

or inorganic, hard or soft, flexible or brittle, fragile or

tough, smooth or rough, dry or wet, dense or light, or


One of the most important physical characteristics of

solid waste is the size of its individual particles. The

spectra of particle sizes of raw and shredded waste uniquely

characterize the components of refuse and the performance of

any processing machinery, e.g., shredders. The particle

sizes of municipal refuse also have an important role in the

efficiency of many solid waste systems, such as milled land-

fills, pyrolysis, suspension furnaces, composting, pipeline

transportation, air classifiers, and other solid waste sepa-

ration and bio-chemical processes. The need to measure par-

ticle sizes of solid waste in order to optimize the efficiency

of these systems is obvious.

Unfortunately, although analytical techniques have been

developed to evaluate and measure various chemical, physical,

and microbiological characteristics of air, water, and waste

water, researchers have not yet successfully applied these

same efforts to define and characterize solid waste. The

literature review revealed that knowledge of the physical and

chemical composition of refuse is surprisingly meager. At a

time when the field of analytical chemistry is directed

toward methods and equipment of micro-samples and micro-sensi-

tivities, we are faced with the need to physically analyze

large-sized samples in order to arrive at reliable character-

izations of solid waste.

Most of the analytical procedures presently used to

determine chemical characteristics of municipal solid waste

were not developed for solid waste, but for the agricultural

and mineral industries, and the air, water, and waste water

sectors. Almost all of these methods generally use a very

small, relatively homogeneous sample with small particle

sizes, unlike the very large particles and heterogeneous mate-

rial found in municipal waste. Subsequently, these methods

required modification when applied to solid waste [1].

Although no reliable quantitative method has been suffi-

ciently developed yet for measuring particle sizes of raw and

shredded solid waste, various attempts have been made using

screens, hand-sorting, and visual judgment. Only marginal

information is available regarding the efficiency and success

of these methods. Various researchers have mentioned specific

refuse particle sizes but have not described their techniques.

It has also been common to report particle sizes from shredders

by the size of the shredder grates.

Since the density of raw and shredded refuse is one of

the critical factors affecting its transportation, processing,

and disposal, an effort also must be made to correlate par-

ticle size distributions with density. Only marginal informa-

tion is available on changes in refuse density caused by

shredding. It also appears that the compressibility of

shredded waste varies with respect to refuse particle size.

The compressibilityy" of a material can be defined as
the volume decrease due to a unit load.

The objectives of this research are listed below:

(1) to develop standard laboratory techniques for determining

(a) particle size, (b) density, and (c) compressibility,

specifically for raw and shredded municipal refuse;

(2) to determine the particle size distributions, densities,

and compressibility of raw municipal refuse, and of primary

and secondary shredded refuse from the Gainesville, Florida,

compost plant;

(3) to determine the particle size distributions for the

various refuse components (e.g., paper, plastic, metal) of

the above composite wastes and define the particle size rela-

tionships between the components and the composite;

(4) to determine the particle size distributions, densities,

and compressibility of shredded refuse from the Johnson City,

Tennessee, compost plant;

(5) to determine the particle size distributions for the

various refuse components of the above composite wastes and

define the relationships between the components and the com-


(6) to determine the relationships between refuse particle

size, density, and compressibility;

(7) to determine the characteristics of each respective par-

ticle size distribution, how to express the distributions,

and how the distributions are affected by additional shred-


(8) to develop a technique for predicting particle size dis-

tributions for various raw municipal refuse, knowing only

the proportion of components; and

(9) to develop a technique of predicting particle size dis-

tributions for various processed wastes, knowing the proces-

sing technique and the proportion of components.

Rationale of the Approach

The first step in this research effort was to review the

literature to determine the appropriate methods for (1) sam-

pling, (2) sample preparation, (3) particle size analysis,

(4) component classification, and (5) density and compressi-

bility analysis for municipal solid waste. The literature was

also reviewed to determine appropriate methods for expressing

the data and defining the relationships among the various


In deciding which analytical techniques to use, an

analyst must first consider the purpose of the analysis. For

example, what is generally required in industrial processes

for the particle spectrum of solid waste is not the size of

the particles, but the value of some property of the particles

that is size dependent, e.g., chemical reactivity. In such

circumstances it is important, whenever possible, to measure

the desired property, rather than measure the size of the

particles by some other method and then deduce the required

property. Also, the analytical techniques chosen must not

destroy or alter certain characteristics of the samples which

may be analyzed at a later time. Nevertheless, the final

criterion for choosing an analytical technique is that it

measure the appropriate property with accuracy and repro-

ducibility sufficient for the particular application, yet

still be reliable, simple, inexpensive, and timely [2].

The most promising analytical techniques discovered in

the literature were chosen for further research and develop-

ment. The preferred methods were those which appeared reason-

ably adaptable to municipal solid waste. The techniques

found in the industrial, soil, and mineral sectors were usu-

ally not directly applicable because solid waste is consider-

ably more heterogeneous and contains larger amounts of organ-

ics and larger sized particles than are commonly found in

materials in other sectors. These considerations required

development and modification of the techniques before they

could be utilized. A significant portion of the total

research was devoted to the development of analytical tech-

niques suitable for use with municipal solid waste.

Several solid waste samples were analyzed using the pro-

posed methods, with alterations made as necessary. The relia-

bility of the methods were determined on multi-samples and

evaluated in conjunction with particle size distributions,

densities, and compressibilities for raw and shredded municipal

refuse samples.

After the sensitivity and reliability of the analytical

techniques were statistically evaluated and the analyses com-

pleted on various types of municipal and processed refuse,

the best methods of expressing the data were determined and

presented. The relationships between the components and the

composite were evaluated and presented. A comparison was

made of different types of samples. Statistical testing was

used to determine if there was a difference between the

samples. Simple and multiple regression techniques and the

analysis of variance were used to determine the relationships

among the various variables.

Previous observations and investigations indicated each

component of raw municipal refuse had a distinct particle

size distribution and that further processing would alter

this distribution in a predictable manner. Thus, a method

was proposed to determine the particle size distributions of

raw and processed refuse, knowing only the mechanical process

and the mixture of various components in the refuse.




Accurate sampling and sample preparation techniques are

important and basic requirements for further analysis. The

reliability of the analysis depends on the reliability of

sampling and sample preparation techniques. Great care must

be taken to obtain samples which are truly representative of

the material to be tested, especially with regard to the

parameters which will be measured. Numerous inconsistencies

result from sampling which does not truly represent the mate-

rial in question. Therefore, a standardized method should

be adopted to increase both accuracy and reproducibility [2,3].

This is true not only for sampling and sample preparation,

but also for further analysis.

Because municipal solid waste consists of so many differ-

ent materials which vary in size, shape, density, hardness,

degradability, etc., a problem exists in determining a repre-

sentative sample. The composition of solid waste varies

considerably throughout the day as well as the week and year.

Each truckload of material arriving at the disposal site

represents a different sample of waste. Indeed, the same

holds true for different parts of the refuse in a collection


truck, since the refuse may be collected from different types

of waste producers. Therefore, it should be clear that a

representative sample of refuse defines only the solid waste

from which the sample was prepared; to construe otherwise

could be foolish. Only when several representative samples

are collected and analyzed can any kind of average character-

istics be attributed to solid waste from a given location.

One must be careful during the sampling and sample

preparation techniques not to alter the characteristics which

will be measured later. For example, many recognized refer-

ences on analyzing solid waste recommended the solid waste

samples be finely ground, mixed and successively quartered, in

order to obtain representative samples and to increase the

accuracy and reproducibility [4-8]. Needless to say, these

methods were not developed for most physical analyses, and

cannot be used in this research.

Methods of Previous Investigators


Sampling techniques vary because of variations in the

characteristics of materials sampled, the forms in which it

is available and the specific methods of analyzing the mate-

rial. For example, the material may have particles which are

fine, medium or coarse, and it may be in a pile, truck, or

flowing on a conveyor belt. The American Society for Testing

Materials (ASTM) standards [3] describe various sampling

techniques, discussed in the following paragraphs.

One of the most accurate places to sample is where the

material drops from a chute or conveyor belt. If the waste

stream is small enough, a pail or other suitable container

can be swung completely across the flow in a brief interval

of time with a uniform movement to collect the sample. Under

no circumstances should the sampling container be allowed to

overflow, because this tends to reject a higher proportion of

the larger particles.

It is extremely difficult to secure truly representative

samples from a pile. This is especially true when the mate-

rial contains large particles, since the particles tend to

segregate during the placement of materials in the pile. It

is unlikely this exists for raw solid waste after it is

unloaded from a collection truck, because the method of

unloading the trucks does not produce this type of segrega-

tion. The ASTM procedures recommended small random samples

be taken from as many parts of the pile as accessible, and

collected so the composite has the same grading as the larger


The ASTM or similar sampling procedures are generally

used and recommended by the various investigators or institu-

tions working with solid waste. For example, in sampling

compost, the Swiss Federal Institute for Water Supply, Sewage

Purification and Water Pollution (EAWAG) recommended the

samples be taken from as many parts of the window pile as

possible and then mixed together [6]; the methods proposed by

EAWAG are routinely used in Western Europe. LD Groote [9]

recommended equal amounts of compost (about 5 lbs.) be col-

lected from different locations in a compost pile at a depth

of about 20 to 30 in. Graduate students at the University of

Florida obtained representative samples from a large pile at

the Gainesville, Florida, composting plant by random sampling

at four points of the pile with the bucket of a front-end

loader [10]. The four samples, each about one cubic yard in

volume, were then mixed together by turning the mixture with

the bucket, being careful not to damage the particles. Approx-

imately one-quarter of this mixed sample (328 lbs.) was removed

with the bucket and set aside for composition analysis. Multi-

samples were also collected by the project personnel at the

Gainesville composting plant for chemical and composition

analyses [11]. The samples were randomly collected from the

conveyor belt. Winkler and Wilson [12] used various sampling

techniques at Middleburg, Vermont, and Cambridge, Massachu-

setts. They found the most satisfactory method of obtaining

a representative sample of manageable size was to take the

material from a pile just dumped by a packer truck. They

recommended that it be taken from the lengthwise side of the

pile rather than from one end.

Instead of randomly choosing samples from a large refuse

pile, a few investigators have used a quartering technique to

obtain their initial samples from the complete pile. For

example, Bell used this method on refuse piles of 1,600 lbs.

or greater to obtain manageable samples for composition

studies [13]. A few investigators have used complicated and

time-consuming grid and quartering techniques to obtain

samples from a truck load of refuse [14,15].

Representative samples also have been collected directly

from the refuse source, e.g., homes and businesses [8,12,

16-18]. Bell [8] developed a method of sampling refuse in

several cities (2,400 samples were collected during his

study). Sample areas were selected by a stratified random

sampling procedure after the city was classified into high,

medium, and low socio-economic areas, on the basis of housing

and/or property market value. Weekly samples of refuse were

then obtained from the sample areas.

Sample Reduction

Since only small samples are required for laboratory

testing, the usual procedure is to collect relatively large

initial samples (gross samples) and reduce them to a suitable

size for analysis (test sample), without impairing them in

any way. The most common method described in the ASTM pro-

cedures [3] and by Allen [2] is to use the "coning and quar-

tering" procedure in which the gross sample is placed in a

cone by depositing each shovel-full of material at the apex

of the cone, allowing it to run down the sides equally in all

directions. This procedure allows the material to mix. The

cone is then spread into a circle of uniform thickness and

quartered. Two opposite quarters are rejected and the

remainder is again mixed into a conical pile taking alternate

shovels-full from the two quarters. This procedure is repeated

until the sample is reduced to the required size.

Gross samples, if not too large, may also be reduced by

one or more passes through a sample splitter or riffler which

divides the sample in half (via a metal compartment contain-

ing many chutes) while maintaining the characteristics of the

original sample [2,3]. By repeated passes, the sample can be

split into quarters, eighths, etc., until the desired sample

size is obtained. Unfortunately, the sample splitter is not

reasonably adapted to material with large particles, because

the passages in the splitter (chutes) should be at least three

times as wide as the largest particle in the sample for it to

work properly.

The ASTM sample reduction procedures appeared to be the

standard way to reduce solid waste sample sizes. Many inves-

tigators have used the quartering technique [5,7,16,19]; one

investigator used the sample splitter on finely ground refuse

[11]. The American Public Works Association (APWA) recommended

a modified quartering technique in which all of the initial

quarters are rejected except for a shovel-full from each [4].

The four shovels-full are then combined, mixed, and quartered

again. A 0.11 to 0.22-1b. sample is then collected from each

quarter for repetitive analysis.

Size of Gross Samples

The size of the gross sample depends not only on the

characteristics of the material, the form in which it is

available and the requirements of the analysis which will be

run, but also on whether the sample reduction methods will be

used later. Thus the size of the gross sample can vary widely.

Needless o say, thbeisize should be sufficient to allow one
to ran' all of his analyses, including repetitions. Also,
in ordeT' tb obtain' the needed accuracy and reproducibility,
it is necessary to analyze larger test samples for materials
with large particles than for finely ground materials. Thus
larger gross samples were required for coarse materials, such

as raw solid'waste. Many investigators have drastically
reduced the size requirements of their samples by fine grind-

ing and thoroughly mixing'the material. This can be done for
chemical analysis, usually reducing the test sample require-

ments to less than a few grams, but it is not suitable for
physical analysis.

The literature review indicated various investigators
have used widely varying sample sizes for gross samples of
solid waste. APWA [4] recommended relatively large initial

samples (gross samples) be collected for most mixed refuse; a
size of not less than 500 lbs. is suggested. The sample was

then reduced by shredding and quartering until four samples
were obtained, each about 0.11 to 0.22 lbs. in weight. Bell

[8] used a gross sample size from 2.2 to 6.6 lbs. for chemical
analysis. Higginson [7] recommended that samples submitted
for chemical analysis (gross sample) be between 0.022 to 0.044
lbs. in size. A sample size (gross sample) of approximately
0.5 cu. yd. (100 lbs.) was collected for chemical and biologi-

cal analysis at the Gainesville, Florida, composting plant,
while sample sizes of 37 to 328 lbs. were collected for com-
position analysis [11,20]. EAWAG [6] recommended a gross

sample size of 2.2 to 4.4 lbs. Gawalpanchi et al. [21] col-

lected 7-lb. gross samples of shredded waste for particle

size analysis; the size of the test samples was only 0.13 to

0.22 lbs. Klee and Carruth [22] found there was no signifi-

cant difference (at the 5 percent risk level) between the

precision obtained from a 200-to 300-1b. sample and the pre-

cision obtained from much larger samples in estimating refuse

composition. The sizes of their samples for estimating com-

position varied from 200 to 1,700 lbs. Winkler and Wilson

[12] used 175-and 600-lb. samples in their studies of refuse

size characteristics.

Sample Drying and Storage

Samples should be immediately dried, not only to stan-

dardize the results, but also to prevent partial particle

degradation due to biological action (if the samples are to

be stored for any length of time). The method of choice is

the oven-drying technique, preferably in a large convection

oven. The ASTM standards [3] recommend all samples be dried

to constant weight, usually at a temperature of 1000 5C,

except in those cases where such a temperature might have an

adverse effect on the material.

Soils are usually dried overnight in 105 or 1100C ovens.

Lambe [23] felt this difference of 5C is of little practical

importance in drying. He also presented curves to show the

effect of temperature in drying five extremely different soils

to indicate that soil is far from dry even at 1100C.

Pertinent data from his curves are provided in Table 1 to show

the relative effect of temperature in drying compared to the

1100C drying oven. The effect of drying temperature was

found to be less than 3 percent (1 percent of moisture-wet

weight) over the 400C range. As expected, colloidal soils

showed the most variation in drying temperature, while sand

showed almost no variation. Lambe also cautioned that ele-

vated drying temperatures (1050C) may be high enough to decom-

pose some types of organic matter in soils.

Table 1

Relative Effect of Temperature on Drying Soil [23]

Soil Type

Ottawa sand

Boston Blue

Leda clay

Mexico City


aThe "dry
weight of the
of the solids
The "wet
weight of the
weight of the

Relative Moisture Content (Dry
Weight Basis)a

Drying Temperature (OC)
70 80 90 100 105 110

1.000 1.000 1.000 1.000 1.000 1.000

0.994 0.997 0.997 1.000 1.000 1.000



0.980 0.987 0.993 0.996 1.000

0.980 0.985 0.994 0.997 1.000

Range of
(Wet Weight





0.972 0.981 0.980 0.992 0.995 1.000 85.92-86.25

weight" moisture content is defined as the
moisture removed by drying divided by the weight
retained after drying.

weight" moisture content is defined as the
moisture removed by drying divided by the net
material (solids plus moisture).


Lambe pointed out that drying time should depend on the

type and amount of sample and the shape of the soil specimen.

For example, he stated that a few grams of sand can be dried

in an hour or less, whereas the same weight of fine-grained

clay may require many hours. He also urged that the samples

be cooled before weighing because hot containers encourage

spillage and hot material can disturb the accuracy of a beam

balance. A non-plastic soil may be cooled normally at room

humidity, Lambe said, and a plastic soil can be treated in

the same way if weighed within an hour or two after removal

from the oven. He recommended that samples be kept in a

desiccator for a longer period of time.

The methods outlined in APWA [4] also state that high

temperatures (105C) have an adverse effect on municipal

refuse because they permit volatile fractions (other than

water) to evaporate. Therefore, it recommended the refuse be

dried to constant weight at 750C in a forced-air circulation

drying oven. It is also recommended that the material not be

packed in containers during drying and the lids be cocked or

removed. For samples with less than 60 percent moisture, it

stated that drying for 24 hours in the forced-air oven was

usually sufficient, but 48 hours was preferred. It was recom-

mended that the material be cooled either in desiccator jars

or cabinets with desiccator containers, prior to weighing.

Personnel at the National Environmental Research Center

of EPA [5] apparently concur with this temperature because

they recommended samples be dried in an oven 70 to 75C

overnight (or for 24 hours) or until the weight loss for an

additional hour of drying is less than one percent of the

total previous weight loss. They concluded that if the

sample is small (3 to 5 lbs. for moist compost, or less for

a bulkier material), the drying can be carried out conveni-

ently in a large laboratory oven, but for large samples an

industrial-size oven should be used. They recommended the

material be cooled in a desiccator or covered with an alumi-

num foil prior to weighing.

Several other researchers also dried their refuse samples

within this temperature range. Bell [8] found his drying time

varied from 12 to 48 hours (depending upon the type of mate-

rial) when drying refuse to constant weight at 700C; the time

was further reduced when the material was periodically

stirred. Researchers at the Gainesville, Florida, composting

plant, the University of Florida, and the University of Louis-

ville dried their refuse samples for 48, 72 and 24-48 hours

at 75, 70 and 700C, respectively [10,11,16].

Many investigators, especially those in Europe, recom-

mended a higher temperature for drying. Higginson [7] felt

the samples submitted for chemical analysis should be dried

to equilibrium at 800C. The procedures of EAWAG [6] recommend

that samples be spread thinly and dried at 105C in a drying

oven (preferably ventilated) until their weight becomes con-

stant (no drying time was recommended). The samples are

then allowed to cool in a turned-off drying oven and weighed

immediately afterwards. Hafeli [17] recommended samples be

dried in an oven at 103 to 1050 for 24 to 26 hours. Gawal-

panchi et al. [21] dried their refuse samples to constant

weight for 18 to 24 hours at 1040C; only 0.44-1b. samples

were dried by spreading them on glass pans and covering them

with perforated aluminum foil.

It is often necessary to store samples for later analy-

sis. When storing, it is important that the samples be pro-

tected from contamination so they do not absorb moisture; it

is also important that parts of the samples are not lost.

Various types of containers have been recommended. The pro-

cedures outlined in APWA [4] recommend dried solid waste be

stored in wide-mouth screw top jars until needed. Bell [8]

stored his material in air-tight cans. Air-tight plastic

bags were proposed by other researchers as storage containers


It is desirable to redry the samples when ready to use

to eliminate any possible moisture the samples may have picked

up. The APWA procedures [4] recommend the samples be redried

at 750C for at least two hours before using them.

Experimental Methods and Equipment


In order to accomplish the objectives of this investiga-

tion, it was necessary to collect solid waste samples from

four different sources, three from Gainesville, Florida, and

one from Johnson City, Tennessee. The specific sources of

waste sampled and their designation are the Gainesville,

Florida, raw waste (R), discharges from the primary shredder

(P) and secondary shredder (S) at the Gainesville compost

plant, and the discharge from the rasp shredder at the John-

son City, Tennessee, compost plant (J).

Shredded refuse from the Gainesville compost plant was

chosen basically because of the close proximity of the facil-

ity to the University of Florida (about a mile). In reality

there were very few alternatives, because at the time of the

research there were only a few operating shredders in the

United States. Because of the earlier extensive research

done at the Gainesville compost plant [11,20], it also was

expected that each investigation would complement the other.

Raw refuse was chosen from the City of Gainesville mainly

because of convenience; it also was felt that the Gainesville

waste represented typical municipal refuse. Shredded refuse

from the Johnson City compost plant was chosen because it was

from a completely different type of shredder, and was one of

the very few shredder plants in operation in this country

during the time of the research.

The ASTM sampling procedures [3] appeared to be the best

or most suitable method and were followed in the collection

of all waste samples at Gainesville; single randomly chosen

samples were collected from the plant discharge pile at

Johnson City for each sample from this source. The raw

refuse samples at Gainesville were randomly chosen from a pile

and the shredded wastes were collected from conveyor belts.

Ten samples of raw refuse were collected at the Gaines-

ville land disposal site from September 2, 1971, to October

25, 1971. No two samples were collected on the same day.

A description of the samples is found in Appendix A. The

samples were obtained immediately after the refuse was emptied

from the city's packer trucks and just prior to compaction by

the landfill dozer. An attempt was made to obtain represen-

tative samples by randomly choosing three to five subsamples

from the pile. Each was about three to five gallons in size

and large and bulky items were avoided. The subsamples were

combined to form the gross samples. Each was about two cubic

feet in size. The samples were then transported to the labo-

ratory in a 10-gallon plastic garbage container. They were

immediately dried for several days in the oven. These raw

refuse samples will be designated as R-l through R-10.

The packer trucks apparently provided a very minimal com-

paction of the raw refuse, as most bottles were unbroken,

cans and heavy plastic containers appeared to be only slightly


Initially, it was planned to collect the primary and

secondary samples at the Gainesville compost plant on a ran-

dom schedule. There was no intention of collecting two or

more samples on the same day. Unfortunately, due to a hurried,

unscheduled closing of the plant, it became necessary to col-

lect nine primary and eight secondary samples all on the same

day (August 10, 1971). This was just a few hours before the

plant closed. These samples were collected only a few minutes

apart. A short description of the Gainesville compost plant

is found in Appendix B.

Fourteen samples were collected after primary shredding

from April 29, 1971, to August 10, 1971. A description of

these samples is found in Appendix A. These samples were

collected from the conveyor belt just ahead of the first

magnet. In order not to stop the plant operation, it was

necessary to collect almost all of these samples by scraping

the refuse directly from the moving conveyor belt. Three of

the samples were collected when the conveyor was stopped by

taking a swath across the full depth of the belt. Negligible

quantities of paper were being removed at the picking belt

during sampling, so these samples had essentially the same

composition as the raw refuse. Each sample was approximately

one to two cubic feet in volume and was transported to the

laboratory in 20-gallon sealed plastic bags. The samples

were immediately dried and stored in sealed plastic bags

until used. These primary shredded samples will be designated

as P-l through P-14.

Fifteen samples were collected after secondary shredding
from April 29, 1971, to August 10, 1971. A description of

these samples is found in Appendix A. The samples were col-

lected from the discharge of the tripper just prior to placing

the shredded refuse in digesters. A cardboard box was placed

for several seconds under the free fall of the waste to

sample from the complete flow of the falling material. Both

magnets and the sludge pump were turned off and paper was

not salvaged during most of the sampling period, so the

samples essentially had the same composition as the raw

refuse. Each sample was approximately one to two cubic feet

in volume. The samples were transported to the laboratory

in sealed 20-gallon plastic bags where they were removed,

dried and stored in sealed plastic bags until used. These

secondary samples will be designated as S-1 through S-15.

During mid-June, 1971, Carlton Wiles, Project Engineer

of the Johnson City compost plant, collected four samples of

shredded refuse from his plant. The samples, all collected

on the same day, were obtained immediately following the rasp.

Sludge had not been added but metals and bulky materials had

been removed. The 1-1/2 cu. ft. samples were each air-dried

prior to sealing in dual plastic bags and shipped to Gaines-

ville by air express. Upon arrival, each sample was immedi-

ately dried in the oven for several days, and placed in

sealed plastic bags until they were analyzed. A short

description of the Johnson City compost plant is found in

Appendix C. These rasp shredded samples will be designated

as J-1 through J-4. A summary of the samples utilized in

this research is found in Table 2.

Sample Preparation

Nearly all of the previously cited investigators used two

ranges of temperatures for drying refuse, 70 to 750C and 100

to 1100C. Since several of them cautioned against utilizing

the higher temperatures (because it might volatilize or

Table 2

Summary of Samples

Source of Samples Location Identification

Raw refuse Gainesville Landfill R-l to R-10

Primary shredded
refuse Gainesville Compost Plant P-l to P-14

Secondary shredded
refuse Gainesville Compost Plant S-1 to S-15

Rasp shredded
refuse Johnson City Compost Plant J-1 to J-4

decompose some of the organic matter), the lower temperature

range was considered appropriate and was generally followed.

Two gravity convection drying ovens were utilized for drying

the samples. The primary oven, a large Napco Model 1430, was

purchased in order to readily dry the large, moist gross

samples. A smaller Thelco oven was used for redrying and for

standby service. A description of this equipment is found in

Appendix D.

The drying procedures, as outlined in APWA [4], were con-

sidered appropriate for the samples and were generally fol-

lowed. The only exception was that the samples were cooled

on the laboratory table and not in a desiccator cabinet as

recommended. This was done because no such cabinet was read-

ily available (due to the large sizes of the samples). It

also seemed unnecessary since the laboratory was dehumidified

and the samples were analyzed immediately after cooling.

The gross samples were dried immediately after they were

received at the laboratory. Since most of the samples were

not completely analyzed until quite a while later, it was

important to remove the moisture so the refuse would not

degrade during storage. All of the samples were heated at

least two days, and usually longer, until they were dry.

Most of them were first dried in the Napco oven which was

vented to an exhaust hood. The temperature was maintained at

70 to 750C in the oven throughout the drying period. The

refuse was heaped (up to 4 in. high at the center) on 18 in.

x 26 in. x 1 in. deep metal trays. The drying was enhanced

by gently mixing the material periodically during the drying


Emergency drying procedures were adopted after August

10, 1971, when 17 primary and secondary shredder samples were

collected on the same day. Normally, the two ovens were ade-

quate for drying one to three gross samples at the same time.

In order to prevent any degradation to the particles, it was

decided to pre-dry the samples before placing them in the

ovens. This was done on the roof of the laboratory during a

very hot period. The samples were dried for about four hours

and were periodically mixed gently to improve the efficiency

of drying.

The samples were removed from the roof and sealed in

plastic bags until the ovens became available again (up to

one week later). At that time they were dried in the ovens

for about one-half day, replaced in plastic bags and sealed

until all of the samples received the initial oven drying.

They were then redried in the oven for about one day, placed

back in the bag and sealed until it was time for the analysis.

All raw samples were immediately analyzed after the

initial drying period which varied from two to four days.

The shredded samples were placed in large tagged plastic bags

and sealed until a later date when they they were analyzed

further. All of the shredded samples were further dried at

70 to 750C for another day just prior to their analysis.

Thus, almost all of the samples received two to four days of

drying time.

The material from the oven was emptied in a pile on a

clean floor just prior to analysis. Usually, the pile was

quartered, being careful that all of the fines were swept to

the proper quarter. Randomly (by flipping a coin), a quarter

of the sample was chosen for the particle size analysis. Two

of the other quarters were randomly chosen for the compressi-

bility tests. The last quarter was kept in reserve or for

"make-up," in case one of the quarters did not have a suffi-

cient sample size. Sometimes the original sample size was

too small to permit this procedure. In those instances, the

sample was divided into only two or three fractions, using

the same general procedures, rather than being quartered.

The preparation of the raw samples did not completely

follow the above procedure. Since the complete raw gross

sample was analyzed for particle size and composition, it was

necessary to wait until those analyses were completed before

the compression tests could begin. It was necessary to reuse

the sieve analysis sample for the compression test.

The quartering procedure of the raw refuse was somewhat

different because of the overly large sized particles in these

samples. First, since the particle size and composition analy-

ses were completed, one quarter was not needed for this analy-

sis. Secondly, because the particles above one inch in size

were always small in number, but large in weight, it was felt

the quartering procedure should be modified. It appeared a

more representative sample could be obtained by using the

quartering procedure only for those fractions less than one


Those fractions of the sieve analysis less than one inch

were emptied into a pile on a clean floor, gently mixed, and

then quartered. Those fractions greater than one inch were

divided by hand, by placing into each of the above quarters a

similar number of particles of each component for each size

range. For example, the tin cans in the 2 in. to 4 in. size

fraction were manually divided so the same number of cans

were placed in each quarter. In effect, an attempt was made

to obtain the same percentage of weight for each component of

each size range in each quarter to insure the same total

weight of each component. Each quarter was then gently mixed

prior to use in the compression test.

It was necessary to modify the raw samples (except for

R-10) in order to use them in the density and the compression

analysis. This was because the large particles (8 to 16 in.)

were either too large for the compression test cylinder or

only one or two particles existed in this size range, thus

producing test samples which were not representative of the

gross sample. It was felt these large particles would dis-

proportionately influence the compression tests. In order to

overcome this problem, the large particles were physically

removed from samples R-l through R-7. In samples R-8 and

R-9, the large particles in the 8-to 16-in. range consisted

only of cardboard and paper. Thus these particles were cut

to form 4-to 8-in. particles. The composition and screen

analysis data were modified accordingly when used with the

compression data. With respect to sample R-10, it was felt

the particles in the 8-to 16-in. range were sufficient in

number and small enough not to overly influence the compres-

sion test results. Thus, this sample was not modified. For

clarity, the particle size and composition analysis data for

the raw samples, which will be used with the compression data,

were designated R-1M through R-10M.




Prior to the beginning of this research (1971), there

were very little published data on the particle size analysis

of solid waste. Many articles mentioned the particle sizes

of certain types of solid waste, but it was obvious most of

the authors used subjective measurements, based either on the

maximum grate size of the shredders or a "visual measurement"

of the larger pieces of refuse. At any rate, this practice

has changed considerably within recent years and several

investigators have reported good quantitative research on the

particle size analysis of municipal refuse. Unfortunately,

the published data are not sufficient for information on

analytical techniques, precision, and accuracy for a proposed

standard method.

Before we can define particle size analysis techniques,

the terms "particle" and "particle size" must be unambiguously

defined. In a classical sense, an individual particle is

defined as a minute unit of matter whose size and shape depend

on the forces of cohesion. A group of two or more individual

particles held together by strong chemical forces is defined

as an "aggregate." Aggregates are commonly found in chemical

precipitation processes; they are stable with normal handling,

stirring and shaking, but are broken up more or less readily

by shear stresses. An "agglomerate" represents larger groups

of individual particles or aggregates. Agglomerates commonly

occur in sieving or drying operations [24].

In the case of many materials, the term "individual

particle" is rather arbitrary since the particle size depends

on the degree of disaggregation imparted on it during testing.

A more rational definition is desirable for solid waste.

Winkler and Wilson [12] defined a particle (or object), in

measuring the sizes of solid waste particles, as those items

which would be expected to remain intact, even after a reason-

able amount of mechanical handling. They tested "border-line"

particles by lifting the questionable part, and if it broke

apart it was considered separate particles. They also emptied

bags and inverted containers for a second and the portion of

contents which fell out also was considered separate parti-

cles. Newspapers consisting of several sections, folded one

within another, were considered as single particles. In prac-

tice, they seldom found this definition as ambiguous.

Particle sizes are defined in terms of the method by

which they are measured [25]. Ideally, the size of a particle

is that representative dimension which best describes its

state of subdivision. For a spherical, symmetric particle,

the diameter is obviously that dimension and thus its size.

Only in the case of a cube or a sphere can the size of a

particle be completely described by one parameter.

Unfortunately, the particles the analyst must measure are

rarely these shapes; in addition, the size range of the par-

ticles in any one system may be too wide to measure with any

one measuring device [2,26].
In small particle analysis, all of the analytical tech-

niques, except one (microscopy), measure only one dimension.

However, the mean ratio of significant dimensions for a par-

ticulate system may be determined by using two methods of

analyses and finding the ratio of the two mean sizes. This

proliferation of measuring techniques is due to the wide

range of sizes and size dependent properties which have to be

measured [2].

Derived diameters are determined by measuring a size

dependent property of the particle (such as volume or par-

ticle settling rate) and relating it to a linear dimension.

Thus, a unit cube has the same volume as a sphere with diam-

eter of 1.24 units, hence this is its derived volume diam-

eter [2].

Derived diameters can also be provided for irregular

particles, usually found in solid waste. The diameter of a

particle deviating from spherical symmetry, may be defined as

any one dimensional distance between two points on the exter-

nal surface of the particle which intersects the center of

gravity of the particle. A large number of non-equivalent

diameters satisfying this definition are possible for an

irregular particle. The distribution of these diameters is

continuous between an upper and lower limit. Therefore, the

size of any irregular particle is a statistical average of

all those non-equivalent diameters. Consequently, when this

method is used the size of a given particle depends on the

averaging method [26].

The diameter of an irregular particle in some preferred

direction is meaningless when applied to a single particle,

although it is valuable when applied to a large number of

particles in random orientation. Such diameters are known

as statistical diameters. Mean diameters may be derived from

graphical or statistical treatment of size distribution data

or statistical treatment of secondary size characteristics

such as surface area [2].

When the sieve analysis is used the size of the particle

is defined as the minimum aperture through which the particle

passes. In reality, because it is impossible to have an

infinite number of screens, the particle size is specified as

the smallest sieve aperture through which the particle passes

and the size of the following aperture through which it fails

to pass [27,28]. Fractionation by sieving is a function of

two dimensions of the particle only, maximum breath and maximum

thickness; for unless the particles are excessively elongated,

the length does not hinder the passage of the particles

through the sieve aperture.

Sieves classify particles according to geometric similar-

ity, regardless of the density. Both the shape of the par-

ticle and the shape of the openings affect the size of the

particles which can pass through a given opening. For

example, a particle in the form of a thin square plate can,

in principle, pass through a square hole of side "L" if the

length of the edge of the plate is less than L/T. However,

the probability of this plate passing through such a hole

(when it is adjacent) is quite small because of the small

probability that it will be properly oriented [25].

Methods of Previous Investigators

The trend of particle size research in the past was

directed toward micro-analytical techniques. There is a

wealth of information on measuring the sizes of small parti-

cles and their subsequent interpretation [2,24,27,29-31].

Although the measuring techniques appear to have limited value

for most forms of solid waste, the statistical and graphical

methods of evaluating and presenting the data were found

valuable. Methods commonly used in small particle analysis

are microscopy, wet and dry sieving, sedimentation, elutria-

tion, centrifugation, Coulter counter, light-scattering,

permeametry and gas-absorption. Several of these methods

indirectly measure the size of the particles by suspending

them in a liquid medium; this is unacceptable for municipal

refuse if we want to know the dry size of the particles (since

the liquid medium would alter the particles). Many of the

methods are also time-consuming since they directly measure

each particle separately. The only method in small particle

research which appears applicable to municipal refuse is


Sieving is probably the easiest and certainly the most

popular method of size analysis in the industrial field. In

fact, it is a standard procedure for soil analysis [23]. It

is restricted to materials having the greatest proportion

coarser than 60 to 75 microns and is used almost exclusively

for materials above this size. Sieve analysis can be com-

pleted in a relatively short time, requires little operator

skill, is a reasonably accurate technique, and involves rela-

tively inexpensive equipment [2,3,25,29].

A sieve is an open container, usually cylindrical, having

definitely spaced and uniform openings in the base. The open-

ings are usually square and constructed from wire mesh. The

sizes of the screen openings have been standardized, and two

standard series are used in the United States, the Tyler

Standard Scale and the United States Sieve Series. By stack-

ing the sieves in order of ascending aperture size and placing

the material on the top sieve and shaking or agitating for a

predetermined time, the material is classified into fractions.

The particles having a dimension smaller than the sieve open-

ings are allowed to pass through while larger particles are

retained on that respective screen. A pan is placed at the

bottom of the stack to collect the fines and a lid is usually

placed on top of the screens to prevent loss of material.

The method of agitation may be either manual or mechanical

but mechanical shakers generally give more dependable results

because of their reproducible action. The portion of the

sample retained on each sieve is weighed [2,3,23,25,26,29]

and commonly expressed as the percent by weight of the sample

retained on that screen or the percent by weight passing the

sieve in question.

The apertures of a sieve may be regarded as a series of

gauges which reject or pass particles as they are presented

at the aperture. The probability that a particle will present

itself at an aperture depends on the following factors [2,32]:

(1) particle size distribution of the material;

(2) number of particles on the sieve;

(3) physical properties of the particles;
(4) method of shaking the sieve;

(5) dimension and shape of the particles; and

(6) geometry of the sieving surface (e.g., open area/
total area).

Whether or not the particle will pass the sieve when pre-

sented at the sieving surface depends on its dimension and

the angle at which it is presented.

The distribution given by a sieving operation also

depends on the following variables:

(1) duration of sieving;

(2) variation of sieve aperture;

(3) wear;

(4) errors of observation and experiment;

(5) errors of sampling; and

(6) effect of different equipment and operation.

Because of the many variables which can influence the

particle size distribution, it is imperative the results be

standardized; this is especially true for the finer sizes.

Not only should equipment be standardized, but also the size

of the test sample, method of shaking, time of sieving, etc.

ASTM has adopted a number of standards using sieves for par-

ticle size analyses on specific materials, and some appear

to be readily adaptable for solid waste [27,28,33,34].

ASTM also published a manual titled, Manual on Test

Sieving Methods [3], which was intended as a supplement (not

a substitute) for the many published ASTM standards relating

to sieve analysis of materials. This manual brings together,

from many sources, proven methods of sieve analysis. It


because of the widely different properties of
the various materials to be sieved, such as size of
particles, density, moisture, hygroscopic properties,
particle shape, friability, abrasiveness, cohesive-
ness, etc., it is not possible to specify a single
procedure to follow in making all sieve tests. For-
tunately, standard sieve test procedures have been
established for many important materials and groups
of similar materials, and, whenever such standard
procedures exist, it is important that they be
followed to the letter by all laboratories or indi-
viduals making sieve tests of the materials the
standards cover. In spite of the considerable
standardization work that has been done, there are
hundreds of granular materials for which sieve
analysis data are desired but for which standard
test procedures have not been established or pub-
lished (p. 1).

There are two ASTM standards which appeared to be reason-

ably adaptable for solid waste: ASTM Designation C 136, "Test

for Sieve or Screen Analysis of Fine and Coarse Aggregates"

[33], and ASTM Designation E 389, "Test for Particle Size or
Screen Analysis at No. 4 Sieve and Coarser for Metal Bearing

Ores and Related Materials" [28]. Both have very similar

methods but the latter accounts for large particles and

highly friable materials. Since screens were not available

for particles over 4 in., it was necessary to classify each

large particle by direct measurement.

Refuse Particle Size Analysis

Prior to 1970, most of the work in defining the size dis-

tribution of municipal waste was done in Europe. Although

many researchers have published their results, there is very

little description of their analytical procedures. All

appeared to use some form of screens.

One of the earliest analysis of refuse was done by Dezso

and Jeno [35] at Budapest, Hungary. The results provided in

Figure 1 were obtained on both winter and summer raw refuse.

The results represented seasonal trends in refuse where com-

bustibles were burned at homes in the winter and large quanti-

ties of ashes were found in the municipal waste. The analyti-

cal procedures were not described.

A quantitative and qualitative program for refuse was
established in Great Britain in 1961 in an effort to define

the yield and composition of raw refuse [7,18]. Large hand

and vibrating screens were used to classify the refuse dust

(e.g., the material which passed the 1/2-in. screen). The
results are provided in Figure 1.



$-l 0 Ok
k 0 0 bo

.r as o0 o -- a i-4 c 4o
k E A c: a s H f
0 pa caDH H

'r <> ^ O at /


0 4

0 cd

P CIS 4- 0 4

V) 0

o -o
) ( D)f)


0 4)U 4)

0 )c t C

0 O*
O 4 Op a s

(b w) N
(Iqbn;.19U $3 Je 9 0J
30 / ~ 0
^~U a)^/ r

4-ra^L r
b)E= 3
kO~re b
-- I I I 1 I f-
0~~9 0 00 00
0/ 00- t <-
(^6j9M^9)^uij ^uaj

Patrick [36] reported a large rotary screen (7 ft. 8 in.

diameter) was tested in London to grade refuse for later

shredding in hammermills, thereby eliminating the need for

manual picking. The screen was formerly a part of a Volund

drum-type shredder with the beater arms removed. Two sets of

screen plates were installed in the drum; the upstream half

was provided with 2-in. diameter holes, while the second

section consisted of 144 holes, each 7-1/8 in. x 5-7/8 in.

in size. The screening results on typical domestic refuse

are presented in Figure 1.

The particle size distribution of shredded waste was

determined by Rena [37] at Dundee, England. The refuse,

shredded by a 200-horsepower Jeffrey-Diamond hammermill rated

at 20 tons per hour, was fed to compactors. The results of

the analysis are provided in Figure 2.

| Several other Europeans have also provided size analysis

of solid waste. Finnie and others [38,39] reported a 50-ton

per hour hammermill shredder at the Newham barge transfer

depot in London was capable of reducing raw refuse so that 90

percent passed a 2-in. screen and 25 percent was below 1/2 in.

Finnie [38] also stated the product from a rotating cylinder

type shredder produced a finer product than that of a hammer-

mill. The 50-ton per hour hammermills at Cringle Dock in

London were specified to insure at least 80 percent of the

material passed a 1.75-in. screen and the remaining 20 percent

measured no more than 6 in. [40].

Several European researchers have recommended screens be

used in the laboratory for analyzing refuse. Hafeli [17]

recommended refuse be screened through 0.197-and 0.984-in.

screens and material greater than 0.984 in. be separated by

hand into combustible and non-combustible fractions. Four

subfractions were thus provided on which individual analysis

can be run. Hafeli felt the division of waste into these four

subdivisions produced more reproducible results. Novak et al.

[41] screened refuse through a 0.787-in. rotary sieve and

then through 0.197-and 0.079-in. screens. Various chemical

and physical analyses were then determined on each fraction,

except for the fraction greater than 0.787 in. Screens also

have been proposed to determine the amount of foreign matter

(glass, ceramics, metals, etc.) in compost [42]. A 6.6-1b.

sample was screened through a standard sieve with 0.268-in.

round holes. The residue was hand-sorted to determine the

weight of each classification. A larger sample (22 pounds)

was recommended for compost used in agriculture. Sifting

was done through screens having 0.748-and 0.535-in. holes.

The permissible amounts of foreign matter depended on the

purpose for which the compost is used. Horstmam [43] proposed

screens be used to determine the rapidity or degree of decom-

position of compost. He reported the screenings through

0.236-and 0.984-in. mesh sieves showed definite differences

in the quality of compost. One study showed green compost,

having a particle size greater than 0.236 in., constituted

about 75 percent of the total mass of processed waste.

Following decomposition, this amount was decreased to about

12 percent.

Very little research was done prior to 1970 in the United

States in developing methods or defining the particle size

distribution of solid waste. One of the earliest research

studies published was on food waste ground in garbage grinders.

Baumannet al. [44] proposed a procedure for determining the

particle size of ground garbage, which involved the suspension

of the material in a volume of water sufficient to facilitate

the passage of particles through a set of sieves. Detergents

were used to cut the grease in the garbage. More than 300

tests were conducted to determine the particle-size distribu-

tion of ground garbage. Cohn [45] also researched garbage

grinders and recommended the home garbage grinders shred the

food waste sufficiently so that 100 percent would pass a 1/2-

in. screen, at least 90 percent pass a 1/4-in. screen, and

not over 5 percent pass a No. 40 sieve.

A sieve analysis was run on various types of heat dried

sewage sludge in Baltimore to determine the efficiency of

processing equipment [46]. Due to the excessive amount of

dust and finely divided particles in its dried sludge, the

City found it difficult to sell its product or to encourage

its employees to work around the material. The City was inter-

ested, therefore, in equipment which produced granules from

the material (compactor, granulating mill, and vibrating

screens). The screen analysis of the tests is found in

Figure 1. Their analytical procedures are not known.

Oberaker [47,48] experimented with various types of

screens to determine the size distribution of shredded waste.

To help overcome the diversity in flexibility and rigidity

and the agglomeration of various size particles, he attempted

to keep the refuse mixed with jets of air. He proposed this

would keep the material suspended in a gentle stream of air

and also would allow for sufficient tumbling and mixing to

break up the clumps. The gentle air flow, he theorized,

would allow all of the lighter particles to float upwards

while, at the same time, not force larger particles through

smaller screens.

The initial design was a tall, thin box with six screens

having openings ranging from 1/16 to 1-1/2 in. The larger

screens were placed on the bottom, several inches above a

small blower. The refuse was placed between the blower and

the screen. The air flow blew the lighter fractions (paper,

plastics, cardboard) upward through the screens. Ports were

provided on the side of the box to insert an air-jet to keep

the material mixed. After several test runs, however, the

results were found to vary widely. Many of the lighter, more

flexible particles were blown through screens smaller than

their actual size. When the air-flow was reduced to prevent

this, the air-stream was insufficient to suspend the heavier

pieces, including cardboard and wood fragments. Glass, metal,

and rocks were completely unaffected by the blower, regardless

of the air-flow.

After modifying the original design to correct these

deficiencies, the final design consisted of a wooden box

6-3/4 in. x 4-1/2 in. x 5 ft., with a series of six screens

with openings from 1/16 to 1-1/2 in. The largest screen was

placed at the top for this design and the smallest at the

bottom. A blower was attached at the top of the box to pro-

vide a very gentle downward air-flow. The flow kept the

small fibrous dust from clinging to the larger particles.

Ports were again provided, but this time they were positioned

below the screens and aimed upward so that the air-jet would

strike the screen from the bottom and keep it from becoming

clogged. The front side was hinged to allow access to the

screens and had windows for observation.

Test runs were conducted with residential refuse which

had been ground in a horizontal hammermill and dried. The

dimensions of the box limited the sample size to approximately

14 gms. The samples were placed in the top of the box and

the blower was adjusted to provide a very small flow. The

samples were then hit with a jet of air and allowed to fall

downward through the screens. Jets of air were used on each

screen until it appeared that only particles larger than the

screen openings on which they rested were present on each

screen. The total time for one run was between five and

eight minutes. The average results of seven tests runs are

presented in Figure 2.

Oberaker suggested several modifications to improve the

performance of the screens:

(1) Increase the unit screen area from approximately 28 sq.

in. to 100 to 150 sq. in., in order to give increased repro-

ducibility and accuracy.

(2) Provide additional screens to reduce the need of inter-


(3) Provide smoother walls to prevent the lighter particles

from clinging.

(4) Reduce the air-jet pressure in order to prevent the

breaking-up of the particles.

Oberaker [48] also experimented with a rotary drum screen

to determine the particle sizes of shredded refuse and certain

components of the refuse. A 55-gallon drum with both ends

removed was used. Screens were clamped to one end in.succes-

sive order. The drum was constructed so that its axis could

be adjusted from 10 to 450 from the horizontal. During opera-

tion, it was rotated slowly about the chosen axis angle. An

angle of 25 to 30* was found to be optimum. A good screening

action was provided at this angle and the best self-cleaning

of the screens was obtained. Coarse screens were used first

and the material which passed through was then re-screened

with a smaller screen. This was done consecutively until the

smallest screen was used. Five different screens were used

(1/8 to 2 in.) and the samples were placed through the open-

end of the drum directly on each screen. The particle size

distribution results were not provided.

Fess [49] attempted to use visual and photographic
methods to determine the particle size of processed refuse

from the Gainesville.compost plant. During the visual trials,

shredded refuse was spread on a sheet of paper alongside a

rule, and an estimate was made of the sizes of the pieces.

The middle dimension of each particle was considered to be

the size of the particle, e.g., the width. Fess then attempted

to count the particles in the various interval widths, e.g.,

1 to 2 in., 1/2 to 1 in., etc. This method presented certain

problems because: (1) particles below the top layer of mate-

rial could not be measured; (2) there were too many paper

particles which were on their sides, thus their widths could

not be measured; and (3) the smaller particles were too small

and numerous to be counted. This method was soon abandoned.

During the photographic trials, Fess spread the refuse

in thin layers so that many of the particles could be observed.

As before, the larger particles hid the smaller ones and those

below 1/2 in. were too small and numerous to count. Also,

the small size of the sample introduced accuracy and repro-

ducibility errors. Even though hand sorting of the larger

particles allowed the smaller particles to become visible,

they still were too small and numerous to count. Fess did

not obtain a usable particle size distribution of any sample

using this method and it was soon abandoned.

The author collaborated with Fess in a preliminary inves-

tigation using screens to classify processed refuse [49].

All screenings were made on the Gilson Mechanical Screen,

Model CL 325, which was designed primarily for coarse aggre-

gate from No. 4 to 4 in. in size. It had a capacity of 1 cu.

ft. of material, and was capable of classifying seven frac-

tions at one time (see Appendix D for a description of similar

equipment). Each screen had interior dimensions of 23 in. x

15 in. x 3 in. and the equipment provided vibrations primarily

in the vertical direction with an amplitude of about 1/2 in.

Fess used 8 Gilson sieves in his research, ranging from

No. 16 to 1-1/2 in. Processed samples from the Gainesville

compost plant were collected after the primary and secondary

shredders, immediately after the digesters, and from a mature

compost pile. The latter two samples represented the effect

of biological action on degradation of waste, as the digested

waste was allowed to decompose about six days and the mature

compost had been allowed to decompose two years. The mature

compost also received a final shredding. The five different

types of waste were air dried and screened, basically using

ASTM Method C-136 [33]. The volume of each sample was approx-

imately 0.6 cu. ft., and the samples were placed incrementally

on the top screen during the screening process. Each sample

was screened for five minutes. It was found that most of the

screening was done after one to two minutes. Considerable

fluffy material on the No. 4 screen resisted breakup and

undoubtedly more material was left there than should have

been. On several occasions certain smaller screens became

too full and it was necessary to divide that fraction and run

each half separately. The particle size distribution of the

samples is provided in Figure 3. In summary, Fess concluded

the screen was the best method for determining the particle

size distribution of shredded waste. Although the Gilson was

designed for 1-cu. ft. samples, he felt this was too large,

and recommended samples no greater than one-half this size

be used.

Fess [50] also worked with the author in analyzing and

evaluating early samples collected for this research. The

samples which he evaluated (P-l to P-4 and S-1 to S-6) aver-

aged 3.448 and 4.224 lbs., respectively. Since the samples,

sample processing, analytical techniques, and results of the

individual samples will be reported later, only his statisti-

cal evaluation will be presented here. Fess calculated the

median particle size for each sample and the population

medians. The results were:

Number of Median Deviation
Type of Sample Samples (in.) (in.)

Primary 4 0.804 0.273

Secondary 7 0.422 0.162

He concluded (using the statistical t-test) the two medians

were different at the 95 percent confidence level; thus the

data showed the secondary shredder further reduced the par-

ticle size.

Jones [51] determined the particle size distribution of

waste shredded by a Roto-Shredder Model 12. This mobile self-

propelled machine straddled and traveled the length of the

windows of unsorted refuse. A revolving drum (with cleats on

its periphery) ripped and shredded the material. This machine

was very similar to the Cobey window turner used at the

Johnson City compost plant (see Appendix C). Four metal

frame sieves (each 18 in. x 24 in. and with effective openings

from 3/4 to 7 in.) were used to screen the waste. Representa-

tive samples of 8 to 10 lbs. were analyzed. It was not

stated whether the samples were dried. Fourteen samples were

analyzed at two different drum speeds and two different num-

ber of passes over the windows. The results are provided in

Figure 2.

Trezek [52] discussed the size distribution of various

types of shredded waste, including the data collected at

Madison, Wisconsin [19]. He presented the size distributions

of waste produced by two additional shredders (hammermill and

a vertical shaft shredder), as well as the size distribution

of waste shredded with different moisture contents (Figures

2 and 4). Both of the shredders were small in size as the

hammermill was rated at 0.5 tons per hour and the Eidal verti-

cal shredder was 2.5 tons per hour. The analytical techniques

were not described.

Researchers at Sulfur Springs, Texas [53], screened

shredded residential waste using both single milled and double

milled refuse samples. The single milled samples represented

refuse which was passed once through a flail mill while the

double milled samples were passed through twice. Flail mills

are a type of shredder in which chains (or flails) are

attached to the rotor. The swinging chains are responsible

(qb6!9M A8)!j ua3J9d

10 1.0 0.1 0.01

Figure 3. Particle size distributions of compost and
shredded waste at Gainesville, Florida [49].

| !I iI

10- grind



O.I I I I i .1 1 III 1 1- ,


0.01 0.1

Figure 4.

Comparison of moisture content on mini-shred
setting on an Eidal Shredder [52].

for reducing the refuse. The shredder used in this research

had two rotors, each powered by a 40-horsepower motor. Four

sets of articulated flails were provided for each rotor at

900 spacing. There was only a 1 in. clearance between the

swinging flails at the closest point of travel. The shredder

also was provided with breaker bars and the samples were

milled at about 24 tons per hour.

Three screens (3 in., 1 in., and 1/2 in. openings) were

used to determine the particle size of the two samples (126

lbs. for single milled and 162 lbs. for the double milled).

The screens were each 3 ft. x 3 ft. and were used in decreas-

ing order. A drop cloth was used as the pan. Further details

of the screening procedures were not provided. The results

are shown in Figure 2.

Winkler and Wilson [12,54] determined the size of indi-

vidual particles of raw solid waste by actually measuring

them by hand with a ruler. All three dimensions were recorded,

as well as the material's composition and weight. Since a

great many objects were not rigid, they noted the dimensions

were somewhat elusive to define. Flexible or collapsible

objects were simply dropped on the measuring table as found

and measured. They felt this was appropriate since it probably

approximated a general mechanical handling system better than

any other measuring scheme. The samples which they examined

were collected from Cambridge, Massachusetts, and Middlebury,

Vermont. Most of the samples were collected during the winter

and were not dried. The composition of the wastes was expected

to be somewhat different than others since both communities

had very active newspaper recycling programs and Cambridge

used a separate collection for food waste. In the course of

the study, approximately one ton of refuse was examined by

hand on an item-by-item basis. Only objects with at least

one dimension greater than 1 in. were examined; the rest were

not categorized and were termed "Miscellaneous and Uncate-

gorized." This category was found to be 15 to 20 percent of

the total weight.
The data were classified into nine major categories and

fifty minor categories; the size data were classified in

1-in. increments from 2 to over 20 in. Cumulative number and

weight distributions of the particle lengths were developed

(Figures 5 and 6). Figures 7 and 8 are similar to the pre-

ceding pair of figures, but here the number and weight of

objects, which will pass through a wire screen of various

sizes, are shown. These figures are based on the criterion

that in order to pass through a given mesh, an object must

have at least two dimensions smaller than the mesh size.

This amounts to 100 percent screening efficiency without any

particle deformation. Actually, vibrating screens might pass

somewhat fewer objects, except for flexible items such as

paper and textiles, which might pass through a smaller mesh.
Sieve analyses were made on shredded refuse at Madison,

Wisconsin, using the techniques and sieves utilized for aggre-

gate analysis [19]. The refuse was shredded through an 8 ton

per hour Gondard horizontal hammermill using a grate spacing

1H913M -V.101 J O IN30U3d



a % Oi-o
o 0 .a
P.4-4 -H4 -

*- 0 .0
P O *!

r c1-i e
r 4) 0 Q:

HU 0 >
l 0) r-4 M



) 0 00

4C rq
t.0 b* 0

r> 4) ta
$4. 4 n-
*4 >
. O4 I 0
an (
rt 0 t/i bf
i- ol *->(
3 bC (-

B ^ (3





1H913M -IV101 O 1N3303d



c a
.0 o a
^ 0 C >
4) (05

0 4-' ( D

00< 4 r0
4) 0 a)

Ir 3 (i 6 N

S0 .1-1
Nq r= t4

.)- 4) C-

4 ) 3n

r *.-i u
C. a5 +4
So o


0y al -i
N ) U

cd o >



varying from 3-1/2 to 6-1/2 in. No other details were pro-

vided on this experiment. The results are shown in Figure 9.

Gawalpanchi et al. [21] also reported the particle size

gradation of shredded waste at Madison, Wisconsin. They used

both sieves and direct measurements of the longer dimension

of each particle. A set of thirteen sieves ranging in size

from 0.0049 to 2.0 in. were used on previously dried refuse.

The 0.132-to 0.220-lb. test samples were accurately weighed,

sieved, and the material retained on the sieves accurately

weighed. Figure 10 shows the results obtained by measuring

the particles of a similar sample by hand and with sieves.

Note that the curves become nearly identical for particles

less than approximately 1/2 in., so the error in sieve analy-

sis is primarily at the larger particle end. Apparently, for

such sizes, particles larger than the mesh in question can

work their way through the sieve.

The researchers also explored the method of shaking the

screens. Comparative tests indicated that the method of shak-

ing, whether by horizontal or vertical motion, was unimportant.

They found that approximately 100 to 125 shakes were suffi-

cient to segregate the material.

The sieve tests were replicated to evaluate sampling and
experimental errors. Four 0.176-lb. test samples of dried

shredded refuse were randomly collected from a larger gross

sample. Each replicate sample was sieved and the size dis-

tribution determined (Figure 11). The four distributions

indicated the sieve analysis technique was very reproducible.


60 61
z 40


CL 0
W -

Figure 9.





1.0 0.5 0.1 C05 0.01

Effect of grate size on particle size
distribution at Madison, Wisconsin [19].

I 0.3 0.1

Figure 10. Comparison of sieving and direct particle
measurement at Madison, Wisconsin [21].



Figure 11.
Figure 11.

10 I 0.1
Replicate sieve tests of
from the Tollemache Mill
Wisconsin [21].


refuse sample
at Madison,

1.0 0.5 0.1 0.05

Figure 12.

Particle size distributions of waste shredded
by the Gondard and Tollemache Mills at
Madison, Wisconsin [21].









I a



This and other experiments indicated the distributions tended

to diverge at the larger sizes and to overlap at smaller

sizes. Therefore, they decided that, rather than working with

a complex function describing the entire size distribution,

it was sufficient to check for experimental error only at cer-

tain portions of the size distribution curve. For example,

the quantity of material passing a 1-in. sieve could be com-

pared. If the quantities were found to be unequal, as con-

firmed by statistical tests, the particle size distribution

could be different. The precision of the replicated sieve

tests was calculated at the 1, 0.5 and 0.185 in. sizes. The

plotted data suggested the experimental error was nearly con-

stant at all points along the curves, and calculations con-

firmed this finding. The estimated variance due to sampling

experimental error was 3.82 percent. The estimated mean per-

centage of material passing a sieve was accurate within

approximately 2 percent.

Gawalpanchi et al. [21] also compared the performance of

two different types of shredders and evaluated the effect of

operational variables, specifically moisture content and

hammer wear. Shredded waste was evaluated from the Gondard

and Tollemache mills. The Gondard mill was a horizontal shaft

hammermill with grates at the bottom of the milling chamber.

It was rated at 8 tons per hour and had a 150 horsepower,

1,200 rpm motor. The Tollemache mill was a vertical shaft

mill without grates and was rated at 15 tons per hour. Its

motor was 200 horsepower and 1,300 rpm. The particle size

distributions for both shredders are shown in Figure 12. The

distributions were found to be highly skewed with about equal

mean particle size. The distributions of all samples tended

to converge asymptotically at small particle sizes. The

region of greatest difference was found to be the 1-in. size.

Statistical t-tests were made to compare the distributions at

the 1.0-and 0.185-in.sizes. At these sizes and the 95 percent

level of significance, there was no reason to reject the

hypothesis that the two shredders produced milled refuse having

an equivalent size distribution.

It was noted large amounts of variability existed beyond

the experimental errors. This natural variability tended to

overshadow any difference arising from physical differences

between the mills. It was concluded the variation in refuse

composition contributed significantly to the differences in

size distributions.

It was further concluded that factors other than moisture

content affected the particle sizes since increased moisture

content did not always correspond with a finer grind. The

correlation coefficient between moisture content and the per-

cent finer than 1-in. mesh was only 0.213. In an effort to

avoid variation because of refuse composition, a controlled

experiment was run to evaluate the importance of moisture

content. A large pile of unprocessed refuse was mixed and

divided into three piles. Different quantities of water were

sprinkled on two piles so that each of the three piles had

different moisture contents (46.3, 66.2 and 91.5 percent

moisture on a wet weight basis). Each pile was shredded

separately and replicate samples obtained. The particle size

distributions (measured in triplicate for each sample) are

shown in Figure 13. The graphic presentation very definitely

indicates that different distributions resulted from the

three different moisture contents. The statistical t-test

confirmed the observation.

Tests also were made to determine the importance of

hammer wear in shredding refuse. A large amount of wear

caused the particles to be less finely ground and the size

distribution to be less uniform (Figure 14). The trend was

well described by an exponential decay curve with time. Mul-

tiple correlation analyses using a linear second order model

confirmed hammer wear was a highly significant variable (95

percent confidence limit), but moisture content was not sig-

nificant at even the 50 percent test level.

Particle Size Distributions of Refuse Components

Several investigators reported refuse components have

different particle size distributions not only for raw refuse,

but also for shredded waste [12,36,49,50,53,54]. They

reported the various components react differently by shred-

ding [53,55]. Patrick [36] was one of the first to recognize

different particle size distributions for various raw refuse

components. He screened raw refuse utilizing a rotary screen

with two sieves (2-in. square and 7-1/8 in. x 5-7/8 in.),

thus providing three fractions of sized refuse. The largest





I 0.1

Effect of moisture content on particle size
distribution at Madison, Wisconsin [21].


Figure 14. Changes in particle size distribution due
to cumulative hammer wear at Madison,
Wisconsin [21].

fraction reportedly contained considerable amounts of large

textiles and carpeting with the remainder consisting of car-

tons and large paper. Plastic bottles usually passed through

the large screen and were retained on the medium screen.

As previously reported, Winkler and Wilson [12,54]

manually measured the size of individual components of raw

refuse. Each object was classified into one of nine major

component categories and 50 minor categories. Since they

found a great many objects were composites of two or more

materials, they decided to classify these materials by secon-

dary and tertiary material composition as well, e.g., a clear

glass container (primary material) with steel lid (secondary)

and paper label (tertiary). Objects which could not be ade-

quately described by the use of the three categories were

classified as miscellaneous. They defined an "object" as an

item which remained intact after a reasonable amount of

mechanical handling.

A special problem presented itself with large plastic

and paper bags of waste. Since they assumed a reclamation

system will be provided with mechanical devices to open such

bags, they decided those bags which could be ripped by grasp-

ing two points by hand and giving a good tug were mechanically

breakable and thus the contents were considered as separate

objects. They found little ambiguity in this criterion.

Measurements of flexible objects were somewhat arbitrary, but

were taken with the object in an "as-received" condition with-

out either crumbling or straightening the object.

Figures 15 and 16 show the particle size histogram for

principal refuse constituents. The height of each column

represents the number (or weight) of objects per ton of raw

refuse, in 1-in. intervals from 2 to 20 in. The ">20" block

at the right of each histogram represents the number (or

weight) of objects greater than 20 in. in length. The area

of each block represents the same number of pieces (or weight)

as an equal area on the histogram.

They concluded each component has different size dis-

tributions and interesting patterns emerged. For example,

almost half of the total weight of metallics consisted of

objects between 3 to 5 in. in length (mostly cans); further-

more, metallic objects made up 40 percent of the total weight

of refuse in the 3-to 5-in. size range, as opposed to only 9

percent for refuse of all sizes. Similarly, there was a

"bottle peak" in the glass category between 5 to 10 in., and

glass objects accounted for 38 percent by weight of the refuse

in this range. The researchers noticed very few glass bottles

were broken in the collection and dumping process (in a sepa-

rate study, some 94 percent of the bottles had remained intact).

The objects greater than 12 in. in length were predominantly

paper products (except the curious peak at 16 to 17 in. in

the metal histogram which resulted from a single baby car-

riage), 67 percent by weight. Of all the paper products, just

over 50 percent by weight fell in the over 12-in. range, yet

the number of objects here were rather small; only about 1,000

pieces per ton of refuse. Furthermore, they found an



I .I'

!2 0 In o


441 -i $- tH4-
o 00
A) 4-1

0 0
0) 0 CUI W
Qp, v

4-4 4-) -t
4-1 0 Cd Lfn
0G bO C

So a) be
r C V r- r-4
03 0 0 1d -

) *rH 0)
Cd a9C

0 CO *q ?-# 4)
o c -i b o
0 .4 $4



0,4 0C U

(3 C *P -
too L*

r C4-4 r-

4) ) 0 Ua

ti) 0 u
U> F )-r4

0 H -o
4-410 0
4-4 ( 444-4
0- 44J 0

z 0 %' 0 ba





overwhelming majority of paper products with high recycle

value, such as bundles of newsprint and corrugated cartons,

fell in this size range.

They concluded sorting by size can be an important pro-

cess in any automated recovery process. Although sorting

alone cannot give a pure product, sorting by size can be used

to drastically increase the percentages of sized waste streams

of certain components or, to state it differently, to "enrich

the ore" from which the products are to be eventually recovered.

Other researchers have also determined the particle size

distribution for various components of shredded waste. Stir-

rup [32,56] reported the distributions of glass, tin cans and

wood when these components were fed separately through a 10

ton per hour Jeffrey-Diamond horizontal shaft hammermill. The

results are provided in Figure 17. The details of the sieve

analysis were not provided. Stirrup mentioned the glass par-

ticles would have been reduced further if they had been mixed

with refuse and thus retained in the machine for a longer

period of time. He noted pieces of shredded wood were as

large as 8 in. x 1-1/2 in.

Fess [49] reported the particle size distributions for

various components of shredded waste. He visually estimated

the percent weight of the major components of each fraction

size of several samples. His results have been adjusted and

are presented in Figures 17 and 18. The composite samples

are the same as presented earlier.


I i I--i d0, -

( r u r--i n r ? >,--, 3"y2

E 0 0l ,0 0 CU d U 0

Cd4 -,-I m 'I -, Q ,-.
rI j H ., _. i, a I 0 ,-0 0O
r to c 0 // 0 r--.
e4d n 10 Ce c u *rT

0 to 0

\ I k t
S .,-- / ^ ,

0 0o 00 0 t

0 c6 -, o

-- (i e e c U uN u l
tn E// 0 C
Cd r=; U i 1. c u M

/IA I O ^'
r-4 in So no t

to a toO0- H
^ / cd0 .


r-4 tn-
UC' Oc z
*M n c A m

0 0) V- 0 0 .
O- O O O >

0 co wn Q 1
1.--, ---- ,-- ^- ^- -,-i 3


0 i
0 r,-4
4- r4- 0 *i )o
V) *-I +1 >M-q 4J
.sn U) .f 1

n- 1 "-i (0 0 1
(0 o< as i as *> a

c -4 co u 0
( $:44 rt ca > E 6

0)## O
.i < E t* o

,- e r
l 0 0 :

d u
,- / d I ^ 0

$-q 4 0 0

o0 N N -'

r o
S 0

,* I
0 c /n / r ) 4-4


Q o

\s a9

4 P .*

", u


0 co (D '- NC

Fess [50] collaborated with the author in analyzing and

evaluating the particle size distributions of selected com-

ponents (paper, cardboard and glass) of shredded waste samples

used early in this research. The distributions of the com-

ponents were determined by manually classifying each sieve

fraction into its various components. For example, the 2-to

4-in. fraction was subdivided by hand into 13 categories

(e.g., paper, cardboard, glass, ferrous metal, non-ferrous

metal, plastic, etc.). This was done for each fraction. The

sorted components were weighed and the subsequent distributions

obtained. It was found that sorting the very fine material

(less than 3/8 in.) was a time-consuming effort (1 to 5

hours). His procedure was to separate the easily distin-

guished and separable items from the fines, i.e., ferrous by

magnet, etc., and then estimate by partitioning the propor-

tion of the remaining items (this technique was altered later

in this research). Estimates were independently made by two

or three individuals and averaged.

The median particle size of each component of each sample

was calculated and the population medians estimated. The t-

test was used to determine whether the medians of the secon-

dary shredded components were significantly less than the

medians of the primary shredded components (95 percent confi-

dence level). The results are provided in Table 3.

Fess concluded that of the three solid waste components,

the cardboard fraction showed a less significant difference

between the primary and the secondary samples.


0o Ln0

.c .io o
0 4*J 0 r H

c0 0 E
in 4J 0 n E
-c00 ) I n 40
f-4 LI n tn C

H.1 eJ in '(4-4
0 r' 0 0
-4 tD LH
Do C >Fi *1

o ,
Cd 0

o 4 i- *-C 4-4 U
(0 C Cd

) c4 3 a)

En CCd oo 0

a) *r
m 04 -4
M *

4- 0 tH Cd
S Ca v.i i c
C Cd +j O H C
S O. n c c M -4
U C *H )

() *4 Q r* *P
ci) p 0 M (-

SEH o 4r
4) S ri s 0 '


a) ,0 a)

(0 3 c0

k Z 0


4 0
( (d rl v Hm

SCv 0 -4 0 Un
U) *H U CU 11
.4 o u 1 0
d. co +j + u ( I f

Fess also found the distributions of the paper components

were highly skewed to the right when plotted on arithmetic

graph paper. When the cumulative finer distributions were

plotted on semi-log paper, he found that the curves were

definitely "S" shaped; a straight line was obtained when

plotted on log-normal graph paper. Similar results were

obtained for the cardboard and glass components. Fess also

evaluated the "goodness of fit" using the Kolmogorov-Smirnov

test for maximum deviations.

Oberaker and Schomaker [57] analyzed the particle size

distribution of glass from different types of shredded waste.

Three of the samples were obtained from the Johnson City com-

post plant (sample Nos. 1 to 3) and the other from the Gaines-

ville compost plant (sample No. 4). A description of the

samples is provided in Table 4.

The glass was separated from the compost using liquid

gravity separation techniques and successive acid treatment.

The samples were then washed and dried. The sieve analysis

was performed in accordance with the procedures outlined in

ASTM D 422-63 [58]. A mechanical sieve shaker was used and

the samples sieved for five minutes. The results are shown

in Figure 17. The shredders at Gainesville produced the

finest glass particles, while the rasp at Johnson City pro-

duced the coarsest glass.

Several other researchers have reported the particle

size of various components of shredded waste. The Hazemag

Novorator hammermill is reportedly capable of shredding glass






o 0







41i CP .rl4



c co -d

a) A
0U .
cn C

So r-H

-4 r-4


c n





3 a

P< (

U 3
0 d -
; U

i-I M t-














* *

cd -H

-4 .

rLi U)

and porcelain, so that only 1.6 percent remained on a 5-mm

screen [59]. Other researchers also concluded that glass,

porcelain and stone are reduced to very small particles by

hammermills [48,50,53,55,60] or high speed Roto-Shredders


Elastic or fibrous material, such as rubber, nylon,

textiles or leather, are not efficiently shredded and fre-

quently pass through shredders virtually unaffected [49,50,

55,60-64]. Rugs may come out of shredders in pieces up to a

square foot or more in size. Leutelt [55] easily separated

these items from shredded waste utilizing screens with 0.984-

and 3.150-in. mesh. Steel and aluminum cans were found to be

mangled and rolled into balls [53,61,65]. Large pieces of

wood were not shredded in small mills [53] but were ground to

finger size in large shredders [65]. Large plastic bottles

were mangled but not substantially torn or shredded [53,62].

Sheet plastic tended to come out in long pieces [53].

Large particles resulted when special hammermills for

large metal objects and bulky objects were used. The typical

particle size of shredded metal from automobile shredders is

2 by 4 in. [66]. The Hazemag Impact Crusher used at Buffalo,

New York, to shred bulky materials provided the following

percentages of crushed particles for shredded metals, measured

according to the greatest dimension [63].

Over 12 in. -- 93 percent

4 to 12 in. -- 6 percent

Under 4 in. -- 1 percent

This machine had a horizontal rotor and suspended impact

arms; it had no grates. Sheet metal from appliances were

reported torn off in large pieces and came through the

crusher as crumbled sheets from 1/2 to 12 sq. ft. in area.

The larger sheets came through in unpredictable random dimen-

sions, e.g., 4 by 3 ft., 6 by 2 ft., etc. It was necessary

to recycle the heavier gauge metals (e.g., hot water heaters)

so these particles were sufficiently reduced such that one

dimension was approximately 8 in. Carpets and tires were

observed to frequently pass through the crusher virtually


Refuse Composition Classification Techniques

In order to determine the particle size distributions

of various compositions of refuse, the components of the

refuse should be classified into categories. The sample must

be physically separated so that each particle is defined to

some mutually exclusive category, e.g., paper, food, plastic,

metal, etc. Determining the physical and quantitative com-

position of solid waste is a necessity for solid waste systems,

e.g., incineration, composting, resource recovery. Thus, many

researchers in the past have classified refuse [10-12, 14-16,

19,22,32,49-51,53,54,56,67-72], almost exclusively for raw

refuse. Separation techniques have even been recommended by

APWA [4] for use in chemical analysis.

One of the initial decisions in classifying refuse is to

determine the subdivisions of the material. A classification

scheme should have the following properties [22]:

(1) categories which lend well to visual recognition
and hand separation;

(2) categories composed of materials of similar
nature; and

(3) categories selected to permit relatively direct
comparison with previous studies and existing

Unfortunately, the individual requirements of a classifi-

cation scheme are often contradictory. A research may require

narrowly defined categories. When hand separation is utilized,

however, such a requirement may multiply the costs and prob-

lems, thus prohibiting the scheme.

The literature review indicated that, although no stan-

dard set of categories for refuse constituents has yet been

adopted, many researchers have used similar categories. The

overwhelming choices appear to be:

(1) paper;

(2) glass and ceramics;

(3) metals;

(4) plastics, leather and rubber;

(5) textiles;

(6) wood;

(7) food;

(8) garden and yard waste; and

(9) ashes, dirt, bricks and stone.

Many researchers have subdivided several categories into

additional units, e.g., metals have been replaced with fer-

rous and non-ferrous metals, plastics have been made a

separate category, and paper has been divided into different

types of paper. Many classification schemes have a "miscel-

laneous" category.

The classification scheme proposed above has been recom-

mended by personnel of the federal solid waste program [15,

22]. They felt that it essentially incorporated mutually

exclusive components, each of which was similar in composi-

tion. In addition, the personnel felt the scheme was adapt-

able to the analysis of solid waste received at any disposal

facility or generated at any source. They estimated four men

could separate and weigh approximately 2,000 lbs. of raw

refuse in an eight-hour day utilizing this scheme.

The selection of a classification system does not guaran-

tee the accuracy of the results [12,22,55]. Frequently sub-

jective decisions are required with inherent ambiguities.

The technician who, for example, encounters a glass jar filled

with food wastes, a toy constructed of both wood and plastic,

or a residue of unrecognizable fines, is faced with a number

of critical decisions. As with everything else, an excellent

classification scheme with adequately trained personnel is

required, but the technician's best judgment is the final

arbiter in these matters. If the decisions are consistent,

then results will be comparable and reproducible, a primary

requirement for any classification plan.

Any proposed classification scheme presents several poten-

tial ambiguities which must be solved prior to separation.

These problem areas are:

(1) Items constructed of several materials. Winkler and

Wilson [12,54] felt they overcame these problems by adding

secondary and tertiary classifications.1 Winkler and Wilson

emptied boxes and bags prior to analysis to reduce the bi-

composition problem. The investigators at Sulfur Springs,

Texas, emptied the contents from bottles, cans, boxes and

bags, as well as removed aluminum tops from bottles [53].

(2) Differentiation of wood in "Garden and Yard Waste"

from "Wood." Fess [50] considered a tree branch (less than

1/4 in. in diameter) as "Garden and Yard Waste," while inves-

tigators at the Gainesville compost plant [10,20,73] used

1 in. as the critical diameter.

(3) Classification of Fines. Winkler and Wilson [12,54]

did not classify objects less than 1 in. in diameter, but

instead categorized them as "Miscellaneous and Uncategorized."

This represented 15 to 20 percent of the total waste. Higgin-

son [7] and Stirrup [56] had difficulties in classifying

fines passing 1/2-in. mesh screen; they provided a separate

category for the fines. Britton [15] also used an additional

category for the fines. In defining the end-point of the

fines, he used two screens (1 in. and 1/2 in. openings). He

found the manpower required to separate a 100-lb. sample to

fines which would pass a 1-in. screen ranged from one to five

man-hours (averaging 2.25 man-hours). Separation to this

point left an average of 5.6 percent fines. The additional

This appears to complicate the problem for it requires
the addition of many more categories.

manpower requirement for continuous separation to the point

the fines would pass a 1/2-in. screen ranged from 0.33 to

2.75 man-hours, and averaged 0.79 man-hours. For this addi-

tional labor, the fines were only reduced to 3.8 percent of

the original sample.

Fess [50] expended considerable time in classifying the

fines from shredded waste. About one to five hours were

required to classify the screen fractions less than 3/8 in.

from samples with a total weight of 2,000 grams. He found it

necessary to separate easily distinguishable and separable

items (e.g., ferrous metals by magnets), and then estimate

the other components.

Oberaker and Schomaker [57] utilized liquid gravity and

acid treatment separation techniques to remove glass from


(4) Defining Ferrous and Non-Ferrous metals. Magnets

were used by many researchers [10,11,20,50,73] to define

ferrous metals.

(5) Salvageable vs. Unsalvageable Paper. This category

was arbitrarily defined by researchers at Gainesville as

large pieces (over 12 in.) of relatively clean and dry paper


(6) Cardboard vs. Other Paper. Cardboard was arbitrarily

defined as corrugated and other thick paper by researchers at

the Gainesville compost plant [10,11,20,50,73].

(7) Rock and Gravel vs. Sand and Dust. Fess [50]

defined rock and gravel as particles greater than the No. 8

screen; all other particles were assumed to be sand and


(8) Synthetic Clothing. Woven synthetics were considered
as textiles instead of plastics by Niessen and Chansky [68].

Moisture introduces potential errors. The majority of

researchers conducting refuse composition studies have

reported their results on a wet weight basis [67]. In order

to accurately compare the results from several studies, it is

important the moisture content of each study be specified.

Certain materials are hygroscopic and thus moisture transfers

can occur between one constituent to another (such as paper

coming in contact with wet food wastes), or certain components

can absorb rain when the refuse container lids are left off.

Several researchers have attempted to solve the problem by

either drying the samples or reporting the results on a dry

weight basis. Britton [15] mentioned the separation data of

his studies were recorded on a wet weight basis, but during

the separation study the moisture content of each component

was determined and the components were later adjusted to a

dry weight basis.

Commercial Application of Screens

The sizing of material is not a new industrial process.

It has been used for ages, especially in the mineral-oriented

fields. It has even been used for some time with respect to

solid waste. A combination of shredders and screens has been

found to constitute a unit separation process because one or

more of the refuse constituents tend to grind more easily

than another, thus producing various sizes of fractions [32,

50,74]. For example, during shredding, brittle materials

(e.g., glass) usually end up in smaller fragments than paper,

while metals, textiles, plastics and heavy items end up

larger. This also is true when screens are used in conjunc-

tion with digestion, where the biological action attacks and

reduces the paper and food fractions, while the metals, glass

and plastics remain untouched. Screens have been proposed

for use with raw refuse to remove the smaller particle-size

fractions of solid waste which are too small to be feasibly

reclaimed and interfere with the reclamation of the larger

particle fractions [15,25,75].

Screening is relatively simple and inexpensive with very

low power requirements. The only significant maintenance cost

has been the screen cloth replacement. Screens are also very

versatile and can be used as stationary or moving, flat,

inclined, rectangular or circular. The screen movement may

be brought about by shaking, rotating, or vibrating. Screen

openings vary from several inches or more to fine screens

with openings of 0.0015 in. in diameter or finer. Coarse

grizzly screens have even been constructed with railroad

tracks. The screen openings may be round, square, rectangu-

lar or slotted, and the wires may be round, flat, and wedge-

shaped [74].

The greatest use of screens with solid waste has been

at composting plants in conjunction with a combination of

rotary and vibrating screens [32,76-80]. They were provided

for two main purposes: (1) to remove the fines (sand, earth,

and ashes) from the raw refuse prior to shredding and digest-

ing; and (2) to remove contaminants and oversize items (rub-

ber, leather, plastic, glass, metals) from the finished com-

post. They also have been used in intermediate processes to

remove oversize items immediately after shredding and prior

to digestion [79]. One investigator mentioned the use of a

vibrating screen at the front of a composting plant to remove

bulky items [81].

Screens have been used ahead of shredders without subse-

quent composting. Stirrup [56] mentioned the City of Salford,

Great Britain, provided two large rotary screens ahead of

their transfer station to extract dust and fine cinders. The

screen plates were provided with 1-1/2 in. long and 3/8 in.

wide slots. Patrick [36] felt that: (1) screens can be used

to separate undesirable items (ashes, dust, bulky items) from

refuse prior to shredding; (2) large screen openings can be

used to remove most of the rejects and small screens can be

used to remove fines; (3) the hammer life of shredders can be

lengthened considerably if the ashes and other fines are

removed (one location reported 50 percent increase of hammer

life when screens were used to remove 30 percent ash); and

(4) the screen could serve the dual purpose of a screen and

a feeder, thus eliminating the need of one conveyor.

Screens are used in many reclamation systems including

the wet pulping process at Franklin, Ohio [82], and the

Bureau of Mines incinerator ash recovery system at College

Park, Maryland [83]. They may be provided directly ahead of

air classifiers in order to size shredded waste for further

processing [32,74,84].

Several rotating drum shredders (which are common in

Europe) are also provided with screens [36]. The machines

shred waste utilizing the principle of attrition. Water in

controlled quantities is added to the refuse to reduce the

fiber strength of materials, such as paper, food and garden

waste, though it has little effect on reducing the size of

metals, plastics, wood, textiles, etc. Thus a screen can be

used to separate the two fractions of wastes. The shredding

is achieved by the churning effect of the rotating drum with

the amount of reduction being a function of rotation speed

and retention time within the drum.

The Fermascreen is an octagon or hexagon shaped rotating

drum which performs the multiple duty of mixer, screen, and

shredder in one unit. The drum rotates at 1.6 rpm and the

process is completed in approximately 2-1/2 hours. Since the

shredded material will not pass through the screen until it

is reduced to the required size, a uniform product is ob-

tained. Rejects may be discharged at the same time as the

shredded material or retained in the drum for later discharge,

whichever is convenient [36].