THE DESIGN AND USE OF TROMMEL SCREENS FOR PROCESSING MUNICIPAL SOLID WASTE

JOHN C. GLAUB, DANIEL B. JONES, and GEORGE M. SAVAGE Cal Recovery Systems, Inc.

Richmond, California

ABSTRACT

Theoretical and practical considerations influ· encing the design and use of trommel screens as devices for segregating various components from municipal solid waste have been identified and

'

studied. Reported are design considerations de· rived from the results of a comprehensive analyti· cal modeling effort of the dynamics and screening phenomena occurring during trommel screening and from the results of experiments carried out on unprocessed and processed municipal solid waste under various operating conditions and using var· ious screen configurations.

NOMENCLATURE

D Diameter Da Aperture size Dp Particle size fa Open area fraction F Ferrous Fc Centrifugal force Ffz Component of frictional force in the axial

direction

Ffe Component of frictional force in the tan-gential direction

Fgz Component of gravitational force in the axial direction

Fge Component of gravitational force in the tangential direction

447

g G

L m rh rnf n

NF 0 P PI r t tr V

W Wr

x,y,z ex � 1] B Bp Bs

Gravitational acceleration Glass Inorganic other Length Mass Mass flowrate Mass fraction Revolutions per minute; number of con­tacts between a particle and the screen sur­face Normal force Number of cycles for a particle in a given radius as it travels the entire length of the screen Non·ferrous Organic other Probability of passage; Paper Plastic Radius Time Residence time Fraction of a given class of material that is retained in the screen Weight Gravitational force component in the radial direction Cartesian coordinates Departure angle Inclination angle Screening efficiency Angular position Angular position of particle Angular position of s.lippage point

Angle yielding reflected passage of a particle Kinetic coefficient of friction

lis Static coefficient of friction

at Number of contacts between a particle and the screen surface per unit time

W Rotational velocity Wcrit Critical rotational velocity Wt Rotational velocity of trommel

INTRODUCTION

This paper presents the design considerations .derived from a research and development effort to characterize the operation and perfonnance of trommel screens used in the processing of munici­pal solid waste (MSW). The study had a twofold aspect, namely, the development of a theory to describe particle dynamics and screening perfor­mance, and simultaneously the acquisition of field test data through use of two trommel screens that differed from each other in dimensions and design. The primary objective of the work was the defIni­tion and quantification of the parameters that gov­ern and describe the trommeling process as applied to the recovery of a high quality fuel fraction from solid waste. Consequently the research addressed the two solid waste materials commonly considered for usage as fuel and hence for trommel screening, namely raw (unprocessed) MSW and the air-classi­fied light fraction (ACLF) separated from shredded MSW. From the analytical theory developed during the research as well as from the data acquired in the conduct of the field tests, a methodology was developed for defining and characterizing the opera­tion and perfonnance of trommel screens used in the processing of raw MSW and ACLF.

In terms of work effort, the emphasis placed on the processing of MSW was the same as that placed on processing ACLF. The rationale for the equality of emphasis was the realization of the valuable roles trommel screens can serve in a variety of applica­tions. Although in new resource recovery facilities trommels may be used prior to the size reduction step (Le., "pre-trommels"), the use of trommels (teImed "post-trommels") or other types of screens for improving the quality of the processed fuel frac­tion (viz. ACLF) is being considered for a signifi­cantly large number of plants already in existence. Consequently, retrofits in the form of the installa­tion of a post-trommel will likely be common occurrences.

448

Post-trommels are used to remove fine inorganic materials from ACLF. This removal can be achieved because the particle sizes of the combustible mate­rials (e.g., paper and plastic) tend to be relatively large. Removal of the fine inorganic materials de­creases the ash and moisture contents of the ACLF and thereby increases the heating value of the ma­terial. Thus, the net effect of post-trommeling is an increase in the quality of the refuse derived fuel.

Enhancement of the quality of the refuse de­rived fuel may also be accomplished through the use of a pre-trommel. Through such a use, most of the inorganic materials are removed prior to size reduction. The advantages following from the use of the pre-trommeling option instead of the post­trommeling option are: (1) glass is removed prior to size reduction, thus avoiding the difficulty of removing it by screening after it has been pulver­ized and impregnated into other materials; and (2) the portion of raw MSW that has a particle size small enough not to warrant size reduction is re­moved. Inasmuch as this portion may constitute as much as one-half of the incoming MSW, the amount of MSW requiring size reduction is correspondingly reduced, and the overall cost of size reduction becomes substantially lower. The two options generally are readily available only while the plant is in the design stage. Unfortunately, for some of the existing plants, the only practical recourse is the use of a post-trommel for the improvement of the quality of refuse derived fuel produced by the plants.

The two approaches followed in the course of the research, that is, the analytical approach and the empirical approach, were conducted in parallel. The analytical approach culminated in the develop­ment of a computer model that simulates the dy­namics and screening characteristics of refuse ma­terials tumbling in a trommel and serves as a means of calculating various screening parameters such as screening efficiency, component mass fraction, etc.

In the empirical approach, field test data were collected with the use of an existing trommel 3 ft

(0.9 m) in diameter and 16 ft (4.9 m) in length at the University of California (Berkeley) solid waste processing laboratory, and of a trommel 7.7 ft (2.3 m) in diameter and 20 ft (6.1 m) in length designed and fabricated as a part of the study. The research plan called for the use of both trommels in the part of the study that involved ACLF, and only of the fabricated trommel in the part that dealt with raw MSW. A variety of aperture sizes and shapes and of other operating parameters were ap-

plied in order to accumulate for evaluation and analysis, as much field test data as was possible. The field test data were used to corroborate the predictions, certain of the parameters in the com­puter code, and in some cases to calibrate the parameters.

ANALYTICAL MODELS Analytical models aid in the design and opera­

tion of trommel screens by providing a means for the prediction of undersize and oversize flow-rates as well as composition of the flow streams as a func­tion of various design and operating parameters. The models reported herein describe the two basic aspects of particle behavior that must be analyzed, namely, the particle dynamics and the rate at which particles pass through apertures in the screen. As a result of space limitations, only key relations of the various models are presented.

The major inputs to a trommel screening model can be categorized according to screen construc­tion parameters, operating conditions, feed charac­teristics, and computer simulation parameters. Screen construction parameters include diameter, total length, screening surface length, inclination angle, aperture size, aperture shape, and open area fraction. In most cases, the only operating condi­tions that can be altered are mass feed rate and rotational speed, although on some screens inclina­tion angle also can be adjusted. Feed characteristics include the size and material distribution of the feed, as well as material properties of the feed such as bulk density, moisture content, and coefficients of friction. Computer simulation parameters. refer to solution algorithm requirements such as time­step increments, distance increments, and conver­gence criteria.

The major outputs that are desired from a trom­mel screening model vary with the application of the model. In general, this information includes the location of the material in the screen, the residence time of the material in the screen, the number of cycles the material goes through while in the screen, the screening efficiency, the mass splits, and the size and material distributions of the undersize and oversize fractions.

Material action inside a rotating trommel screen may take several:different forms depending upon the rotational spyed of the screen, the loading of the screen, material properties of the feed, and the presence of lifters. The emphasis in this report will be on "centrifugal action" in which a particle is

449

carried above the horizontal plane that passes through the center of the screen until at some point it detaches from the screen and falls until it contacts either the screen surface or material that is already in contact with the screen surface. The particle rises to its point of detachment as the re­sult of centrifugal force. Modeling of another type of material action, referred to as kiln action, is dealt with elsewhere [1]. In kiln action, a particle remains stationary with respect to the screen sur­face as the screen turns until the particle reaches a position at the surface of the bed. At this point, the particle tumbles down along the surface of the bed until it comes to rest at some lower position. The cycle is then repeated. Under certain condi­tions, both centrifugal action and kiln action may occur simultaneously with respect to material at different radii in the screen.

In the case of centrifugal action, axial advance of a particle occurs during periods of falling through the ai!. and during periods of slippage on the screen. In both instances, axial advance is due to the gravi­tational component in the axial direction. Slippage may further affect particle dynamics by determin­ing the height to which a particle can be carried up the screen surface before it detaches. Unfortunately, accounting for slippage significantly complicates the model. For this reason, two particle dynamics models are presented. In the first model, slippage is neglected; whereas in the second model it is taken into account.

NO SLIP-FREE FALL MODEL

The No Slip-Free Fall Model is patterned after a model originally presented by Davis [2]. In this report, it has been modified to account for inclina­tion of the screen and to include three dimensions. Slippage of material on the screen is neglected and free fall is assumed.

A particle remains in contact with the screen until the force component of gravity in the radial direction exceeds the centrifugal force. A free body diagram illustrating this condition is shown in Fig. l. A force balance leads to the relation for the depar­ture angle (x,

(X = Cos-1 __ t_ ( rw 2 ) g Cos �

(1)

where r refer$ to radius, Wt refers to the rotational velocity in radians per second, g refers to gravity, and � refers to the inclination angle. In terms of

revolutions per minute, n, Eq. (1) can be re-ex­pressed as

(2)

The value of the rotational speed that renders a

equal to zero is commonly referred to as the "criti­cal speed," Wedt.

FIG. 1 FREE BODY DIAGRAM FOR FORCES ON PARTICLE AT DEPARTURE POINT

It can be shown that the total number of cycles, Nc, that a particle in a given radius goes through if it travels the entire length of the screen is given by

N = L

c 8r Tan(3 Cosa Sinla (3)

where L refers to the total length of the screen. The number of cycles that a particle goes through is directly related to its probability of passing through an aperture in the screen.

The residence time in the screen, tr> for a par­ticle in a given radius is equal to

t = L (360 - 4a + 229.2 Cosa Sina) (4) r 48nr Tan(3 Cosa Sin2a

where a is stated in units of degrees.

SLIP MODEL

If slippage is considered, a force balance on a particle leads to the following governing equations of motion for the tangential and axial directions, respectively (see Fig. 2).

450

dd

2

t

()lP

= IJ.k (d()dt

P_V -(C--

2 ()_C

_O

_S()

T-' P�2-(3)

-O-::-s J os P + an .

IJ.k gCos(3 +----

r

+ gCos(3 C () -r- os p

Cos()p Sin()p

(5)

where () p refers to the angular position of the par­ticle, z refers to axial distance, t refers to time, and IJ.k refers to the kinetic coefficient of friction.

y

x

mgCosjJ F ge = mgCos f3 cose

{l H."".",

- fJ .,z

FIG. 2 FREE BODY DIAGRAMS FOR THE ANALYSIS OF SLIPPAGE

by The angle at which slippage begins, ()s, is given

IJ.s rWtl . -C (3 + IJ.s Sm()s = (Cos 2 ()s + Tan2(3)o.s (7)

g os

where IJ.s refers to the static coefficient of friction.

Equation (7) may be solved by an iterative ap­proach. However, the governing equations of mo­tion [i.e., Eqs. (5) and (6)] are nonlinear ordinary differential equations that are best solved by a numerical approach. The authors employ a Runge­Kutta scheme for this purpose. Alternatively, the •

governing equations can be simplified and an anal­ytical solu tion achieved [1].

SCREENING RATE MODELS

The screening rate is directly related to the probability of particles passing through apertures in the screen. The probability of passage through an aperture by a single particle at impact with the plane of the screen is given by the ratio of the area of the surface yielding passage to the total area on which the particle can fall.

The "simple probability model" states that the area available to passage per aperture is the aper­ture area itself. Thus, the simple probability model only accounts for particles that pass through the aperture without contacting the sides of the aper­ture. Assuming spherical particles falling perpendi­cular to the screen surface, the simple probability of passage can be stated as

P = (1 - Dp/Da)2 fa (8)

where P refers to the probability of passage, fa refers to the open area fraction of the screen, Dp refers to the particle size, and Da refers to the aperture size (length or diameter). 1b.is relation holds for either circular or square apertures. Ob­viously, for all Dp greater than Da there is no pas­sage through the screen.

In addition to the area available for passage due to the aperture area itself, a particle may also pass through the aperture following reflection off the side of the aperture. Accordingly, a "reflected probability of passage" model was developed for punched-plate screens to account for the total area available for passage, i.e., the aperture area and the area of contact yielding reflected passage. For nor­mal incidence, the reflected probability of passage model can be stated as

where

451

The reflected probability of passage also applies both to circular and to square apertures. Use of the reflected model given by Eq. (9) leads to greater screening rate predictions than does use of the simple model given by Eq. (8).

The fraction of a given size class of particles re­maining in the screen, V, after n contacts is

V (n) = (1 _ P) n (11)

. Assuming an approximately constant number of contacts per unit time, at> Eq. (11) becomes

(12)

The fraction of particles remaining in the screen can also be expressed in terms of the weight of the particles as

_ W(t) V(t) -

W(O) (13)

where W(O) is the initial weight of a given size class of particles entering the screen and W(t) is the weight remaining in the screen at time t.

The screening rate is given by

d W(t) = at In (1 - P) W(t) (14)

dt

The effect of bed depth on screening rate was modeled by analyzing the material in multiple layers. In addition to the size of the particle, the probability of passage is a function of the layer that the material occupies. One approach to speci­fying the probability of passage associated with dif­ferent layers is to assume that passage of particles through the screen only takes place in a finite "screening layer" next to the screen. Thus, par­ticles in layers that are within the screening layer have a chance to pass through the screen; whereas particles in layers that are radially inward from the screening layer do not pass through the screen.

COMPARISON OF PREDICTED MODELS AND EXPERIMENTAL RESULTS

The analytical models described previously were computer-coded, and simulations were run for the purpose of comparing the predicted results with the experimental results. The Slip Model was used to predict the particle dynamics and drag was neg­lected. The screening rate model was based on the

probability of passage model for punched-plate screens [Eq. (9)] and a specified screening layer thickness. The latter was determined by a compari­son between the'predicted screening efficiencies and those that were measured.

sented in Table 1. The table presents comparisons for a screen 7.7 ft (2.3 m) in diameter having 3.5 in. (90 mm) square apertures. A comparison of the simulation results and the corresponding experi­mental measurements for air-classified light fraction is presented in Table 2. The screen' studied for this set of ACLF tests was 3 ft (0.9 m) in diameter and had 0.50 in. (12 mm) circular apertures.

A comparison of the simulation results and the corresponding experimental measurements for the pre-trommeling of raw municipal solid waste is pre-

TA BLE 1 COMPARISON OF SIMULATION RESULTS AND EXPERIMENTAL MEASUREMENTS FOR RAW MUNICIPAL SOLID WASTE

Screen Design Parameters Simulation Parameters

Diameter = 7.7 ft (2.3 m) Initial Bulk Density = 7.6 lb/fe (122 kg/m3) Total Length = 20 ft (6.10 m) Coefficient of Static Friction Between Material and Screen = '0.8 Perforated Length = 16 ft (4.88 m) Coefficient of Kinetic Friction Between Material and Screen = 0.7 Inclination Angle = 2 degrees and 6 degrees Coefficient of Static Friction Between Material and Itself = 0.7 Aperture Size = 3.5 in. (90 mm) Coefficient of Kinet'ic Friction Between Material and Itself = 0.6 Aperture Shape = Square Screening Layer Thickness = 2 in. (51 mm) Open Area Fraction = 0.509

Residence Time Inclination Percent Feed rate Measured Computed Screening Efficiency

Angle Rpm Critical tons/hr (Mg/hr) Leading Edge Carton Paper Measured Computed

2 degrees 8 29 3.2 (2.9) 64 68 68 79 88 84 2 degrees 8 29 14.2 (12.9) 55 66 58 47 82 67 2 degrees 20 73 6.2 (5.6) 46 32 31 63 98 99 2 degrees 20 73 14.2 (12.9) 61 38 29 58 97 64

6 degrees 8 29 6.4 (5.8) 30 31 25 29 70 85 6 degrees 8 29 14.5 (13.2) 32 29 24 24 72 81 6 degrees 20 73 9.3 (8.5) 25 18 15 22 90 88 6 degrees 20 73 19.5 (17.7) 21 32 26 21 70 76

TA BLE 2 COMPARISON OF SIMULATION RESULTS AND EXPERIMENTAL MEASUREMENTS FOR AIR-CLASSIFIED LIGHT FRACTION

Screen Design Parameters

Diameter = 3 ft (0.91 m) Total Length = 16 ft (4.88 m) Perforated Length = 15 ft (4.57 m) Inclination Angle = 15 degrees Aperture Size = 0.5 in. (12 mm) Aperture Shape = Circular Open Area Fraction = 0.422

Percent Feed rate Rpm Critical tons/hr (Mg/hr)

20.7 47 0.18 (0.16) 20.7 47 2.27 (2.06)

26.0 59 0.31 (0.28) 26.0 59 3.86 (3.51 )

36.6 83 0.19 (0.17) 36.6 83 3.32 (3.02)

Simulation Parameters

Initial Bulk Density .. '6.0 Ib/fe (96 kg/m3) Coefficient of Static Friction Between Material and Screen = 1.3 Coefficient of Kinetic Friction Between Material and Screen = 1.2 Coefficient of Static Friction Between Material and Itself = 1.0 Coefficient of Kinetic Friction Between Material and Itself = 0.9 Screening Layer Thickness = 0.5 in. (12 mm)

Residence Time Screening Efficiency Measured Computed Measured Computed

13.8 16.4 84.6 87.2 10.5 14.6 70.6 64.8

12.5 13.9 80.2 85.8

11.9 11.6 59.8 60.0

8.2 12.6 81.5 72.5

8.1 13.7 57.6 67.2

452

The screen design parameters and the simulation parameter values that were used are shown at the top of each table. Inconsistencies in the measured residence times make it difficult to draw firm con­clusions about the reasonableness of the predicted residence times. (The leading edge and various tracer times were measured.) However, residence time is one of the only measured parameters that can be used to evaluate a particle dynamics model. The computed screening efficiencies shown in the tables generally follow expected trends with var­ious operating parameters. In some cases, predicted values agree closely with measured values, while in other cases there are significant discrepancies.

The presence of short lifters in the screen, which was not taken into account in the model, probably exerted an influence on the particle dynamics and consequently contributed to some of the discrep­ancies between the predicted and measured results.

The limitations of the "screening layer thick­ness" model and of the probability of passage model are partially responsible for the variations between the predicted and the measured screening efficien­cies. The mechanism by which undersize particles pass through the screen and the mechanism by which apertures are blocked by oversize particies are extremely complex phYSical phenomena to model. Despite substantial analytical and experi­mental efforts by various researchers [3 -7], much work remains to be done on the development of screening rate models.

Inaccuracies in experimental measurements are also partially responsible for the variations between

the predicted and the measured screening efficien­cies. The assumption of steady-state operation is a critical factor in experimental accuracy. Inasmuch as the undersize and oversize flow streams collected during the sampling period are assumed to be origi­nating from a common feed source; variations in the composition and mass flowrate of the feed dur­ing the course of an experiment can cause signifi­cant errors.

RESULTS AND DESIGN ASPECTS

APERTURE SIZE

The results of the experimental program can be used to form judgments as to the effect of changing various parameters that influence the properties of the trommel fractions. For example, the effect of changing one key parameter, aperture size, may be examined by assuming a screening efficiency of

453

100 percent, a given feedstock composition and size distribution, and the requisite fundamental material properties. Similarly, a reduced efficiency (e.g., 90 percent) could be used for the analysis; however, unequal screening rates of different com­ponents also may introduce discrepancies into such an analysis. Limiting the discussion to an examina­tion of the fuel quality of pre-trommel oversize fraction and post-trommel light fraction in terms of heating value and ash content, the following serves to illustrate the utility of the approach when applied to the trommeling of MSW and air classi­fied light fractions. By deftnition, post-trommel light fraction refers to the oversize material from the trommeling of shredded and air classified MSW.

As a starting point, the composition and size distribution of the" aforementioned feedstocks were evaluated for MSW generated within the City of Richmond, California, and air classified light frac­tion produced at the University of California (Berkeley) solid waste processing laboratory. The results are presented in Figs. 3 and 4, respectively.

A�rture Size (inches)

l00Qr __ � __ � __ ���4 ����6�� __ -.8

eo go 80 'Ii : 70 i 60 �

11. 50 ,� 40, :; E a

50 100 Aperture Size (mml

150 200

FIG. 3 RAW MSW COMPONENT SIZE DISTRI BUTION

Ap.rtu r, 811.. (Inch •• ) r--�·�"2��·�"·==��·'�·��·�" ·====�1. �·--4' "2 100 -

•

.. c

.

. ..

, .

O-o,;.nlc, High Air Flow O-In organlc, High AI, Flow .�OrD.nlc:. Low Air Flow __ Inorganic, low Air Flow

2. Ap.,t"",. Siu ,,,,m)

s.

FIG. 4 ACL.F COMPONENT SIZE OISTRI!3UTIONS

The component size distributions and material properties (heating value and ash content) were subsequently used in the following presentation.

RawMSW

The ash content and heating value of each com­ponent of MSW shown in fig. 3 (as determined by size class during the conduct of the research) was used to estimate the fuel quality of pre-trommel oversize fraction as a function of aperture size. As shown in Fig. 5, the heating value of the Richmond MSW is about 6,000 Btu/lb (14,000 kJ/kg) Le., the heating value corresponding to an aperture size of zero. If the aperture size of the pre-trommel was 4 in. (100 mm), the heating value of the oversize fraction would be almost 8,200 Btu/lb (19,000 kJ/kg). IncreaSing the aperture size beyond 4 in. (100 mm) would be accompanied by diminishing marginal increases in the heating value of the over­size and a substantial decrease in energy recovery as shown in Fig. 6. For example, if the ash content of the MSW was about 39 percent, material having a particle size larger than 4 in. (100 mm) would have an ash content of approximately 18 percent. Increasing the aperture size to 8 in. (200 mm) would yield an oversize fraction with an ash con­tent of about 11 percent.

The results of the research also indicate that there would be a substantial gain in fuel quality of the oversize fraction when all glass is broken and passes through the screen openings and all ferrous is subsequently removed from the oversize fraction by magnetic separation. For the purpose of this study, a fuel quality parameter has been defIned in terms of the ratio of the caloric value of the fuel to its ash content, i.e., (kJ Heat/g Fuel)/(g Ash/g Fuel), and is abbreviated energy/ash. The degree of improvement is visible by examining the curves for unadjusted energy/ash and ferrous and glass adjusted energy/ash ShoWIi in Fig. 6.

Air-Classified Light Fraction

With respect to the post-trommeling of the Uni­versity of California ACLF j an analysis indicates that the heating value and ash content would not change SignifIcantly if the aperture size was larger than 0.6 in. (15 mm), as shown in Fig. 7. If the aperture size was less than 0.6 in. (15 mm), the ash content of the screened light fraction would climb substantially as the aperture size approached the range of zero to 0.2 in. (5 mm). It may also be

454

40 20 cuo.,

.0 u: aooo � A

�o , .. 20 It ·

1000

-; i m > c m c ·

10 ,4 � eooo . . %

00

12 (1182' 50 100 ,"0 .00 2.0

Apallur. Siz. (mm)

FIG. 5 PREDICTED ASH, MOISTURE CONTENT, AND HEATING VALUE OF SCREENED RAW MSW ASSUM­

ING 100 PERCENT SCREENING EFFICIENCY (OVEN-DRY BASIS)

100 120 (US)

120

7' UOa A

! Q-Eurn/Aal'l (unadJn,ad) ,

. ob eo � 0

.0 110 -: . '" o -Enargy/A."

1 �

� (farrowa ,dJuat.d) � m � 40

m .. . 2. .0 c ..

0 0 0 '0 100 '"0 200 "0

"pertll" Slzl (Mm)

FIG. 6 PREDICTED ENERGY RECOVERY AND ENERGY PER ASH OF SCREENED RAW MSW

ASSUMING 100 PERCENT SCREENING EFFICIENCY

40 20 (8.04)

30 U m .ooo�

� � ! ..

"0 ·

! , · 7000 ..

20 " ;

> � m

0 c m c · · · %

10 14 · 1000 %

00

12 U1l2) 10 '" .0 "

FIG. 7 PREDICTED ASH AND HEATING VALUE OF SCREENED ACLF ASSUMING 100 PERCENT SCREEN­

ING EFFICIENCY (OVEN-DRY BASIS)

noted that for aperture sizes within the range of 0.05 to 0.6 in. (1 to 15 mm), the percentage of energy recovered in the post-trommel oversize frac­tion would decline at a rate of approximately 4 per­cent per mm, as evidenced by the curve given in Fig. 8. Also shown in Fig. 8 is the predicted values of the energy/ash ratio for a spectrum of aperture sizes.

100 200 (I,') 6 10 6

6

� ,. tlO -; · .0 �

.. 6 ·0 "0

.0 6 100 .... .. o�

c � � G · . .. .0 · 20 w "

0 0 0 10 , . 20 2.

FIG. 8 PREDICTED ENERGY RECOVERY AND ENERGY PER ASH OF SCREENED ACLF

ASSUMING 100 PERCENT SCREENING EFFICIENCY

An examination of Figs. 5 through 8 indicates that there are tradeoffs involved in the selection of the optimum aperture size. For example, increasing the apertm:e size to increase RDF quality (as de­fined by the ratio of energy content to ash content) results in a decrease in the energy yield. Consequent­ly, a detailed examination of the screening objectives and priorities for each situation is required in order to determine the screen opening size which is the optimum. For example, in some communities, there is an abundance of solid waste and a limited market for RDF. In such an instance, RDF yield could be sacrificed in order to produce a quality RDF.

APERTURE SHAPE

Inasmuch as there is little cost differenoe be­tween square and round aperture punched plate, the only major consideration in the selection of aperture shape is standardization with laboratory sample analysis equipment. From a production per­spective, there is little difference. Where feaSible, it is recommended that the geometry of the screen apertures should be limited to square openings due to the fact that square apertures are consistent with the screen geometry used in commercially available laboratory sieving equipment. Consequently, size distribution measurements performed with labora­tory equipment can be compared directly to data collected in the field. Currently, off-the-shelf labora­tory sieve equipment cannot be obtained with non­square screen openings (e.g., circular apertures).

It should be noted that the equivalent square and round apertures that pass the same size par­ticles must be considered in the selection of aper­ture size [1].

OPEN AREA FRACTION

For a given set of operating conditions, the rate of screening (Le., flow of particles through the

455

screen openings) is proportional to the open area fraction, fa. Consequently, a large open area frac­tion is desirable provided there are no deleterious effects, such as blinding of material on the screen surface or wrapping of articles about the screen openings. In this study, blinding and wrapping were not experienced when a screen surface of perforated sheet was used and ACLF was the feedstock. [The screen openings were 0.5 in. (12 mm) in diameter and open area fraction was 42 percent.] In a pre­vious study, blinding did occur when using wire mesh with 0.5 in. (12 mm) square openings. The open area fraction of the wire mesh (estimated as exceeding 90 percent) was considerably larger than the 42 percent.

No blinding or wrapping problems were encoun­tered while trommeling raw MSW with screen sur­faces having 4.75 in. (120 mm) diameter openings and an open area fraction of 55 percent. However, it is believed that at some larger value of open area fraction, wrapping may become a Significant prob­lem for pre-trommels.

SIZING OF TROMMEl SCREENS

The computer simulation program was used to develop relationships between overall screening effiCiency (1]) and the length (L) and diameter (D) of a pre-trommel with the intent of assessing the sensitivity of 1] to the geometriC variables L and D. I� addition, the effects of loading [i.e., throughput (m)] were also studied.

There are two significant findings stemming from the investigation. First of all, tradeoffs in trommel performance (measured in termS of 71) may be characterized in terms of L, D, and m. Secondly, the results of the exercise suggest that praoticallirnits exist for selecting the length of a screen operating under a given set of conditions.

The specific case addressed during the investiga­tion consisted of an evaluation of a pre-trommel equipped with 4.75 in. (120 mm) round openings and processing municipal solid waste. A No-Slip Model was used (to simulate the use of lifters), and the rotational velocity was chosen as 80 percent of the critical value. Throughputs used in the modeling simulation were 22, 44, and 88 tons/hr (20, 40, and 80 Mg/hr). The diameters chosen for consideration were 9.8, 14.8, and 19.7 ft (3.0, 4.5, and 6.0 m), and length was modeled from 6.6 to 98 ft (2 to 30 m).

The results of the modeling simulations are shown in Fig. 9. For a particular screen length, the consequences of a low throughput (Le., 22 tons/hr)

and concomitantly low rate of loading are apparent when the results are compared to those for a high rate of loading (Le., 88 tons/hr). Specifically, the curves for low rates of loading approach a horizon­tal asymptote sooner than those for high rates of loading, i.e., the efficiency curve for low through­puts approaches a horizontal asymptote at a screen length which is less than that required for high throughputs. The modeling data suggest that the screening efficiency of pre-trommels levels off (Le., approaches a point of diminishing returns) between 50 and 65 ft (15 and 20 m) when the screens are designed to process 22 to 44 tons/hr (20 to 40 Mg/hr). However, for throughputs of 88 tons/hr (80 Mg/hr), the point of diminishing re­turns does not occur until the screen length ex­ceeds 65 to 80 ft (20 to 25 m).

Using the trommel model, it is also possible to establish the trade·offs with respect to whether or not the diameter Qr length should be manipulated in order to optimize the overall screening efficiency. For example, the model predicts that a pre-trom-

Length (ft) '00;..0 -...,--'i'--....,--.'i'-0-,----:i'-----.-¥---,--'-i

g> • c: • 41 " 41

i i i

S,,,,bOI.;

03111. 20WeIPH 0 ..... 20",,11., <> '.0, :rOM,/flt Q3f1'1, 40 'I •.•. 40 � •• o, 40 .3M. 10 .4.', .0 • '.0, '0

eX oL-_� __ � __ �_��_� __ � o

Length (m)'

FIG. 9 PREDICTED PRE·TROMMEL SCREENING EFFICIENCY AS A FUNCTION OF MSW

THROUGHPUT AND LENGTH OF SCREEN

mel, equipped with lifters and 4.75 in. (120 mm) openings and operating at 0.8 Wcrit, will have a screening efficiency of approximately 87 percent if the diameter and length are 9.8 ft (3.0 m) and 33 ft (10 m), respectively, or 15 ft (4.5 m) and 26 ft (8 m), respectively, as shown by the data in Table 3. Such information would be of use, for example, if the addition of a pre-trommel was being considered for a plant where height and length restrictions would have to be applied to the pre-trommel design. Comparisons of the consequences of geometries dif­ferent than those of.the preceding example are also possible from the data .shown in the table.

An additional consideration in the design of a pre-trommel is that of glass breakage. Our test re­sults have shown that a dropping height of 6.5 ft (2 m) is sufficient to achieve satisfactory glass breakage if the screen surface is rigid and loadings are moderate. Creater heights are required for flex­ible surfaces. Similarly, the greater the loadings, the greater the diameter required for satisfactory glass breakage.

INCLINATION ANGLE

The research performed by Cal Recovery Systems has indicated that screening efficiency increases with decreasing inclination angle (within the range of 2 to 8 degrees) when all other parameters are held constant. Since screening efficiency is directly proportional to the length of the trommel, to achieve a desired screening efficiency, the use of a small inclination angle (Le., 2 degrees) would allow the use of a shorter trommel than would be required for a larger inclination angle.

The experiments have shown that use of a 2 de­gree inclination angle results in higher screening

TAElLE:3 CALCULATED SCREENING EFFICIENCY FOR VARIOUS PRE-TROMMEl LENGTHS AND DIAMETERS

(4.75 in. ( 120 mm) ROUND OPENINGS, W = 0.8 Wcrit)

Diameter of Feedrate Pre-trommel tons/hr l=5 ft L = 10 ft L=15 ft L = 20 ft L = 25 ft L = 30 ft L = 50 ft L= 100 ft

ft (m) (Mg/hr) ( 1.5 m) (3 nil (4.5 m) (6 m) (7.5 m) (9 m) (15 m) (30 m)

9.a. (3.0) 22 (20) 65.3 77.3 8 1.8 84.2 85.7 86.7 88.9 9 1.9 44 (40) 44.2 62.6 71.4 76.2 79.2 8 1.2 85.4 88.7 88 (80) 25. 1 41.6 52.7 60.3 65.8 69.7 78.3 85.0

14.8 (4.5) 22 (20) 71.9 80.4 83.9 85.8 86.9 87.7 89.7 93.0 44 (40) 53.9 69.5 76.0 79.6 8 1.8 83.4 86.6 89.5 88 (80) 34.3 52.2 62.4 68.6 72.7 75.5 8 1.6 86.6

19.7 (6.0) 22 (20) 68.4 78.2 82.2 84.4 85.8 86.7 88.8 91.9 44 (40) 6 1.5 74.1 79. 1 82.0 83.8 85. 1 87.7 90.4 88 (80) 42.1 59.7 68.4 73.4 76.6 78.8 83.7 87.6

456

efficiencies than 4 degree, 6 degree, or 8 degree inclination angles. Extrapolation of the above trend to angles less than 2 degrees has been assessed qualitatively with the judgment that further re­search is required for trommel inclinations less than 2 degrees; For a given set of operating conditions, decreasing the inclination angle will increase the residence time, which in turn results in an increase in the screening efficiency, and concomitantly will

increase the bed depth which in turn results in a de­crease in the screening efficiency. It is unclear which phenomena (increased bed depth or increased resi­dence time) would be more dominant at inclination angles less than 2 degrees.

The lower limit for the inclination angle may also be governed by a practical problem, namely, at 2 degree inclination angle it was noted that material fell from the feed opening of the trom­mel. It is reasonable to assume that the problem of spillage from the feed opening will be more serious at inclination angles less than 2 degree unless suit­able precautions are taken, e.g., restraining the ma­terial through the use of built-up sides around the periphery of the trommel feed opening or discharg­ing the feedstock well into the trommel.

ROTATIONAL SPEED

An analysis of the experimental results have indicated that screening efficiency increases with increasing rotational speed and the rate of change of screening efficiency with respect to rotational speed decreases with increasing rotational speed (within the range of 29 to 73 percent of Wcritical). Therefore, since screening efficiency is directly proportional to the length of the trommel and in­versely proportional to the mass throughout, a poor screening efficiency resulting from a high throughput rate or an insufficient length of screen can be remedied, at least in part, by an increase in the rotational speed.

The physical stresses on the rims and ribs of the trommel also increase as the speed increases. Metal fatigue was not exhibited in the pilot-scale trom­mel, but cases of metal fatigue have been encoun­tered in commercial-scale trommels which have operated over long periods of time.

CONCLUSIONS

The results of the research reported herein and in the fmal report for the DOE study [1] should

prove of considerable interest and utility to the solid waste industry both from the standpoint of trommel design and overall system design for re­source recovery systems. The trommel computer model and the empirical results of the research program provide a powerful tool and data base for analyzing the trommel screening of raw MSW and air classified light fraction under a wide range of operating conditions. In particular, the model pro­vides the capability of determining the size distri­bution and properties of oversize and undersize fractions as a function of operating parameters and component size distribution of the feedstock.

The paper presents only a few selected examples illustrative of the versatility and applicability of the trommeling model and the results of the research with emphasis being placed on design aspects. More examples, along with a more detailed analytical pre­sentation of particle screening behavior and the pre­sentation of design gUidelines, are covered in the fmal report.

ACKNOWLEDGMENTS

This work was supported by the Department of Energy Office of Energy From Municipal Waste under contract no. DEAC0379CS20490. Mr. Wil­liam Lambert, San Francisco Operations Office, served as the DOE Technical Program Manager.

REFERENCES

[1] Glaub, J. C., Jones, D. B., Tleimat, J. U. and Sav­

age, G. M., Trommel Screen Research and Development

for Applications in Resource Recovery, Final Report under

U.S. DOE Contract No. DE-AC63-79CS20490, October

1981.

[2] Davis, E. W., "Fine Crushing in Ball-mills," Trans­

actions, American Institute of Mining and Metallurgical

Engineers, Vol. 61, pp. 250-296,1919.

[3] Gaudin, A. M., PrinCiples of Mineral Dressing,

McGraw-Hili, New York, 1939.

[4] Kaye, B. H., "Investigation Into the Possibilities

of Developing a Rate Method of Sieve Analysis," Powder

Metallurgy, Vol. 10, pp. 199-217, 1962.

[5] Baldwin, P. L., "The Continuous Separation of

Solid Particles by Flat Deck Screens," Transactions, Insti­

tution of Chemical Engineers, Vol. 41, pp. 255-263, 1963.

[6] Jansen, M. L., and Glastonbury, J. R., "The Size

Separation of Particles by Screening," Powder Technology,

Vol. 1, pp. 334-343,1967/68.

[7] Vorstmann, M. A. G., and Tels, M., "Segregation

in Trommelsieving," Recycling Berlin 79, Volume II, 1979.

Key Words

Materials Recovery, Mathematical Model, Refuse Derived Fuel. Rotating Drum, S creening, Separating, Trommel

457