ANAL YSIS OF A PROPOSED DEVICE FOR THE CLEANUP OF RECOVERED MAGNETIC MATERIAL

MARC L. RENARD National Center for Resource Recovery, Inc.

Washington, D.C.

ABSTRACT

A method is proposed for cleaning magnetic materials recovered from mixed municipal waste (mostly canstock) from their "loose" organic contaminants to enhance product quality and conform to specifications. The scheme combines a tumbling motion with the action of a steam jet in a rotary screen. The device is analyzed in terms of feed characteristics, geometry, jet dynamic pressures and product exposure time. Expressions are derived for the quantities needed for a prelim­inary design of the unit. An illustrative example is given for a prototype installation rated at 0.6 tons/ hr maximum throughput.

INTRODUCTION AND STATEMENT OF THE

PROBLEM

In the U. S., a potential exists for the recovery of 6.4 x 106 tons/year of magnetic metals from municipal solid wastes [1] .

The energy savings resulting from recovery and reuse of magnetic metals have been estimated at 42.2 x 106 Btu/ton [2] . Hence, the savings potential is 270 x 1012 Btu/year, or the arith­metic equivalent [1] of 119,000 barrels of oil per day.

As a separation process, magnetic separation is widely used by the scrap industry, in ore process­ing plants and in resource recovery. The market value or acceptability of recovered magnetic metals from MSW is strongly dependent on its

351

grade, form and no less importantly, on its cleanli­ness. The latter is expressed as a maximum accept­able value of "total combustibles" in the recovered material. This target value is typically of the order of 4 percent for iron and steel foundries or iron and steel production users, and much less (about 0.5 percent) for other users such as the copper precipitation or ferroalloys industries. Measured contamination by loose nonmetallic materials of recovered magnetic metals ranged from 0 percent to 9.3 percent in a series of twenty-six runs per­formed by NCRR on recovered magnetic materials from the heavy fraction of air classified shredded MSW [3] . Weight fractions between 10 and 20 percent for all nonmetallic components have been reported on recovered magnetic metals from coarsely shredded refuse [4] .

The question of organic contaminants as a barrier to the reuse of magnetic metals recovered from solid waste has recently been examined at the National Center for Resource Recovery by E. J. Duckett [5] . Briefly stated, the potential negative effects of organic contaminants include yield losses, added particulate loadings on air pollution control equipment, sanitation problems, odors, poor aesthetics and disturbances in the induction furnace melting process. Of the organic contaminants, a part might be described as "bound." Such is the case for coatings and lacquers on beverage cans. The remainder are "loose" and are added during use or mixing with the other components of the solid waste: food residues, paper labels, plastics or fabrics mechanically snag-

ged or trapped by the magnetic materials would fit into this category. In the above-referenced study [5] , it was concluded that organic coatings and lacquers (about 2 percent by weight in steel cans) did not appear as a barrier, in view of the techno­logical steps already taken by reprocessors in those areas sensitive to organic coatings, namely detin­ning and copper precipitation. A removal of at least "some" of the loose contaminants, on the other hand, appeared desirable, and in some cases mandatory, to satisfy existing or proposed specifi­cations regarding "cleanliness."

Thus it seems a worthwhile objective to investi­gate simple and economical means to reduce the weight fraction of loose organic contaminants in magnetically recovered metals.

OBJECTIVES IN THE CLEANUP PROCESS

The definition of a "clean" state, as regards organic contaminants (or other contaminants, for that matter) is often imperative. As pointed out in [6] , it is user's process-oriented, depending on whether the recovered metal is for reuse in detin­ning, ferroalloys, casting alloys, copper precipita­tion or further upgrading by the scrap processor. Alter and Reeves [7] developed target specifica­tions and the base for these targets for most of these possible markets.

To strike a sensible balance between a perfectly clean product (no organic contaminants) and a high level of contamination (> 10 percent), tenta­tive standards, at this date of writing, are being proposed by ASTM Resource Recovery Subcom­mittee E-38.02 (ferrous metals). For total "com­bustibles," the maximum admissible percentages, by weight, would be 0.2 percent for precipitation, 0.5 percent for ferro-alloys, 4.0 percent for iron and steel foundries, or iron and steel production, and 3.0 percent for detinning.

The objective of the cleanup process could thus be stated as follows: reduce the loose contaminants' weight fraction from their initial level (5 or 6 per­cent, say) to 2 percent, approximately. If 2 percent is allowed for bound contaminants (lacquers and coatings), the 4 percent remaining contaminants would represent a total level acceptable for iron and steel production, or iron and steel foundries applications under the proposed standards.

352

DESCRIPTION OF PROPOSED CLEANING

DEVICE

In a 1972 NCRR study [8] , a "tumbling washer" was mentioned as a possible means to clean the light ferrous fraction (can stock). The need to study the cost, effectiveness and water requirements was recognized.

The present study is an analysis of the device to determine the technical feasibility of the proposed cleanup method, provide insight into the Significant parameters, and guide design trade­offs when building and testing a prototype unit.

Figure 1 illustrates the proposed cleanup device. A cylindrical drum, the walls of which are made of perforated sheet metal or woven, fence-type wires, rotates about an axis very slightly inclined on the horizontal towards the discharge end. The product (mostly canstock) is regularly fed through the inlet section labeled "0." At some point down the axis of the cylinder, a powerful jet of steam impacts on the product, which has already gone through a phase of mechanical tumbling in order to loosen snagged fabrics, food wastes, etc. The steam jet will impart on the loose organics mechanical pressure forces and the condensed water simultaneously helps to rinse off deposits on the surfaces of the feedstock elements. Water and organic particles are drained through the woven wire walls. Downstream of the jet section, a second phase of mechanical shaking takes place for further cleaning of the material. The cleaned product exits through sec­tion "1". If need be, a single magnetic head pulley could be used for added separation from remaining loose paper or dirt at the end of a short discharge conveyor belt.

Note that the design will be evaluated and sized for one single steam jet. The same arrangement for more than one jet, or with compressed air instead of steam, could be used. In the latter case, the fluid inlet pressure, nozzle characteristics and distances would have to be adjusted accordingly and the collection of loose particles might have to be effected by induced draft pickup.ducts. It is ex­pected, however, that the lack of rinsing action would make this method much less efficient. As an example, we may note that a steam jet is the rinsing agent used in the food industry when cleaning poultry-eviscerating tables [9] .

To define the parameters of the proposed process, we shall characterize succeSSively:

(a) The typical feedstock element to be cleaned.

-

(b) The cleaning device. 1. The rotating cylindrical tumbler. 2. The steam jet and its action on the

feed. (c) The output variable(s) in relation to the

input variable(s), for fixed (but adjustable) design parameter values to be determined by prototype testing. The most significant relationship would express the efficiency, 1/ (in percent), defined as the ratio of the percentages of pickable contami­nants removed in the device and pickable contami­nents present in the feed product, as a function of • the inlet mass flow rate, M (in tons/hr) (see Fig. I):

Section "0"

---f - . I

,

• 1/ = f (M)

Section "1"

/ /

-----

To I.IAste­.... ·ater treatment

FIG. 1

xagr:etic Daterials

• • .' \ l'> •

Others

TYPICAL FEEDSTOCK ELEMENT

GEOMETRY AND MASS

It is evident that the mean diameter or shape factor of constitutive elements of the removable magnetic metals stream entering the device strong­ly depends on previous processing step6 (shredding, crushing, trommeling . . . ). Most of the magnetic metals in MSW are often canstock-upward of 90 percent in an NCRR experiment on refuse from three separate cities [6] . Therefore, for the purpose of this study, it will be assumed that the typical element of feedstock is represented in Fig. 2. Given the characteristic linear dimension of the element, say its length "£," the other dimensions are taken to be:

353

• IS:

width = w = aw£, with aw a "width to length" aspect ratio;

height = h = ahQ, with ah a "height to length" aspect ratio.

The total area of the exposed sides (one-sided)

• ,

�" -- . - . �,",,,,,-o;," • -� - ' -:-

�, � I ) . .. '

d' ... " " '1 ."

. � .. , � , " , . L--�':'-'_"";;"" __ �--'7", __

r---- o. .. .t ---.. � '

FIG.2

To define the "spacing" between adjacent ele­ments in the device, one could possibly compute the arithmetic average £/3 (I + aw + ah) = 0.58Q, if aw = 0.5, ah = 0.25, considered to be typical values. The average "upper" surface exposed to the steam jet would be Au = 1/3 £2 (0.5 + 0.25 + 0.125) = 0.29 £2. To account for asperities in crumpled cans, the spacing distance adopted is slightly larger and taken to be the diameter of the sphere having the same volume,

6a ah 1/3 w Q rr

or for the above values of aw, ah:

deq = 0.62£ Au = 0.30£2

(I)

The mass of the element is obtained from the surface in the undeformed state, a cylinder of diameter d and length £

d2 mass of tops = 2rr4 at

mass of side = rrdQos

in which at, as are the masses per unit area of top sheets and side sheets, respectively.

Example

The values:

Q "" 4 in.; w "" 2 in.; h "" 1 in.

could be considered as typical of lightly crumpled and flattened beverage or food cans. Then

aw = 0.5; ah = 0.25

deq = 0.620 Q = 2.48 in.

A = 28 sq. in.

A "base box" corresponds to 112 rectangles of sheet metal, 14 x 20 in. in dimension or 31360 sq in. in total surface [10]. It weighs 55 lb for sides, 95 lb for tops [11]. Thus

as = 55/31360 = 1 .754 x 10-3 lb/sq in.

at = 95/31360 = 3 .029 x 10-3 lb/sq in.

Mass of 1 can, 12 oz, length = 4 �� in., diameter (11) .

= 216m.

m = 0.0713 + 0.0344 = 0.10571b

Number of cans per short ton (2000 lb)

nc = 18,915

PACKING

For reasons which will appear more clearly in discussing steam cleaning, it appears preferable to deal with single layers of feedstock (Fig. 3). Suffice it to say that the impact pressure of a jet diffusing in an atmosphere at rest decreases rapidly with the distance away from the nozzle, and. that it could be reduced by a factor of 3 to 4, typically, over a depth equal to 2 element lengths, Q. In a cylindrical section of the mechanical shell and for moderate values of the intercepted arc on which elements are resting ("" 60 deg. at most), the area packing density, as seen by the jet, is expressed as the ratio of the areas of solids to the total area.

The void fraction, fv, is 1 minus the latter quan­tity. An average between "compact" (case A) and "sparse" (Case B) patterns would lead to an estimate of approximately 45 percent for the void fraction. Since some elements might not always be in contact with their neighbors, a conservative value for the area packing density would be 50 per­cent. A maximum value is 79 percent (case A).

354

!:ote: ::-=:.':. to sea lc . CrO$s-�cction nor:r.al to the axis of rt!volution

FIG. 3. PACKING OF ELEMENTS IN SINGLE LAYER.

ROTATING DRUM

For purposes of discussion and analysis, the device is essentially a hollow cylindrical, wire-mesh drum rotating at speed n about its axis (Fig. 4). Its characteristic dimensions are diameter D, length L, and slope s = tan 'Y. Along the direction x of the flow of material through the device, L could be divided in

x = 0 to x = LI "Pre tumbler." To effect mechanical tumbling and shake off or loosen organics prior to cleaning.

x = LI to X = � "Steam Cleaning Length," in which actual spraying of the feedstock with steam occurs.

x = Lz to x = L "Aftertumbler." Shakes off organics loosened by the cleaning.

Obviously, many variations on or additions to the above basic design are possible. One item of special interest appears to be an on-line fmal sepa­ration by a magnetic head pulley, downstream of the device, to reject loose paper and dirt from the recovered magnetic materials (Fig. 1).

Steam

In Feeder� �

PretUl'IIbler -\ . .. -

\ ... --- - ..

I

Cleaning length

FIG.4

STEAM JET

AttertUmbler

The steam jet is directed at the lower portion of the "cleaning section." Its axis can be directed away, by an angle \{I, from the descending vertical, so that the jet can impact the bed of material normally to its tilted surface during rotation (Fig. 5).

The steam supply is characterized by:

Po, pressure of the steam, assumed to be saturated.

msteam, mass flow rate of steam. Qj' height of the steam nozzle exit section

measured normally to the product bed (Fig. 5).

The use of a manifold of steam jets or multiple nozzles is obviously possible and although it will

Steam jet

•

/ Axis of • steam Jet

•

!

/ •

I !stearn I

' nozzle irO" '/ ' � -------" I .1't---.... .!/

/, ,/ I, , I

..n..

d surface t· 6

I ,

I I

FIG. 5

355

not be considered in the calculations which follow, it should be kept in mind as a possible design al ternative.

The amount of condensates, mc, shall be estimated from

mc = msteam (1 - fs.m.> + mo.c. in which fs.m. is the estimated mass fraction of the steam (condensed) carried away by the discharge product and mo.c. the mass flow of cleaned up organic materials present in the condensates.

EFFICIENCY VS THROUGHPUT: PARAM­

ETERS TO BE VARIED

For a given geometric layout of the drum, the parameters which could be varied most easily to run a series of tests would be:

n, drum angular speed Q j. height of jet above the bed

and to a lesser extent Po, upstream steam pressure (Fig. 6) msteam, steam mass flow rate (using a set of

geometrically similar nozzles of varying throat diameters)

The weight fraction of pickable organics is measured before (as Yo) and after cleaning (as y), as a weight fraction, dry basis. The efficiency 11 (percent) is defmed as

11 = (1 - :o)X 100

It should be measured as a function of throughput in the steady state, for Qj, Po and msteam fIxed_

Supply

po Ste-a-", "jet

, axis

I I

,

, , " ' , , '

, : I '

"""P:I"I�I I' " '1\, Throat ' I

FIG. 6

, , I I - -

... , II \ I -,--__ � _I P2 .. }Yi __

I '--�-.,

I I

- -I

.. ,

ANAL YSIS OF PROPOSED DEVICE

GEOMETRY OF DIFFUSING STEAM JET

It is assumed in the development that Po, steam pressure (supply side) and msteam are con-

i Primary I Core I I �'

I- - '-.J J'4ain Flo'"

, -",,­I 1 I 1 19.6 DZ

���_�1=��J�_��� ____ ��1D�2 _t o __ l o�o� , _______ '/ _____ �

FIG. 7

stants, given in any run, but possibly adjustable from one test to the next one. The steam nozzle is of the convergent-divergent type (Fig. 6) and of circular cross-section, exactly adapted to the ex­pansion ratio, E, from the supply pressure to the atmospheric pressure:

E=� Patm

in which P2 stands for the pressure at the exit section of diameter D2, of the nozzle. p I is the pressure at the nozzle throat.

For saturated steam, the ratio of the critical flow pressure, PI *, to Po is 0.575 [ 12], or PO/PI * = 1.739. Thus, the flow downstream of the throat will be supersonic, since in practical cases the supply pressure is in the range of from 80 to 100 psig, leading to expansion ratios always larger than 1.739.

For all practical purposes, it is expected that the. jet Reynolds number will be of the order of lOS, for nozzle diameters of 3/8 in. or larger, i.e., much larger than 2 x 103. The submerged jet flow will thus be turbulent. Downstream of exit section 2, in which the velocity is V 2, the turbulent jet flow can be subdivided as follows [13-19] (Fig. 7):

1. A primary "core" region, of length L'c, in which the velocity along the axis is constant. The relationship between L�, D2 (diameter of the ex­haust section) and Mj Get exhaust Mach number) is given in appendix AI .

2. A "transition" region, of length Ls, a func­tion of M j and D2 (see Appendix A2)' Here the velocity decreases, the flow becomes subsonic and the fully developed turbulent jet is being estab­lished.

3. A "main flow" region, in which the velocity decays similarly to that of subsonic jets [22] .

For an incompressible, axially symmetric flow [23] , the velocity along the axis varies with distance x measured from the nozzle exit section as

V D2 > X --"," K-- for 100 ;---D � 7 V2 X 2

(8)

If the boundary layer thickness in the exit section is negligible,

K","6.2

The functional form (8) is retained. It is valid for x � Lg, K being the best possible value obtained from theory or from experimental data such as reported in [21] in the applicable range of Mach number. The basis for this approach is given in appendix A3. For the cone angle, a, of the outer boundary of the jet, an average accepted value is adopted [13,14,17,26].

a"'" 20°

Also, this is apparently the angle observed on Fig. 11, [19] , for a convergent-divergent nozzle and an expansion ratio of 2.43.

Finally, at distances larger than about 100 nozzle diameters, a "terminal" region where the jet dissipates very rapidly is of no interest in this application.

Let p be the specific mass of the jet, at a temper­ature and pressure essentially equal to those of the surrounding space. Using the best available value of K in Eq. (8), as explained previously, the dynamical pressure on the axis, which is a measure of the jet impact force on a unit area normal to it, can be ex­pressed as a function of x

Its decrease along the jet axis is measured, if Pdyn (0) = 1/2 p V�, by

Pdynt� - K2 (10)

Pdyn 0 - (x/ D2)2

and is represented on Fig. 8, for the assumed value K = 6.2. It is seen that for the ratio of dynamic pressures to be equal to 10 percent, the distance away from the origin of the jet is 19.6 nozzle diameters of D2 .

The volume flow rate across section "x," qx' for constant fluid density is given by Abramovitch [25] as

356

x for 7$ D2 �;too (11) Normal to surface

The numerical value of the coefficient 0.31 should be used with the same limitations as apply Steam to the numerical value K, as discussed above. These ve loc i ty

.--

numerical values are thus not meant to be defini­tive, but indicative of those used by practitioners in the field of industrial jet cleaning [14] and sub­ject to refinements.

ratio of dynanic preslu:e on axis at -x" to oriC?inal dynamic preS5\!Ce (at x-O)

I •

.8

.6

.2

�'l' (,,) �.,.(�.o)

11,1'- 10.0\ 'X

o. L----,S'6�7"8'5'IO--���--��O�--� Dz

1<.: 6.2

FIG. 8

I ratio of distan::e fron nozzle ::0

19.' nozzle dh:::,.ete:-

DYNAMIC PRESSURE REQUIREMENTS

It is obviously difficult to predict what kind of impact (for normal incidences) or erosive forces (for incidences other than 90 deg.), combined with the "rinsing" effect of condensed water will be needed to remove a significant fraction (one-half) of the pickable organic contaminants on the magnetic material surface. This will be dependent on a large number of factors, illustrated in Fig. 9, such as:

1. The nature and size, Dc, of the containment (food, paper, etc.).

2. The nature of the cohesive force (adhesion, mechanical "snag," etc.).

3. The nature of the substrate (coating or bare metal, etc.).

4. The local angle of impact of the stream flow, �s·

5. The residency time under the steam jet, etc. Needless to say, only experimentation will make

it possible to get a handle on this problem, in a specific situation in which these many variables would have some degree of regularity, at least in a statistical sense.

At this stage, however, assumptions have to be

Contaminant

ting

Magnetic �aterial

FIG. 9

made to obtain figures likely to be meaningful. Consider the dynamic pressures recommended as rules of good industrial practice [14] for industrial cleaning by air jets at high pressure (a few atmos­pheres). The fluid is assumed to be dry air with a density of 0.075 Ib/ft3. The velocities and dynamic pressures listed are those needed to dynamically "detach" particles from their underlying clean, planar support.

TABLE 1

Nature and/or Shape of Particle Diarete.r

Aluninun shavings

vbod chips

100 microns 500 micruls

Area 2 15 an 2 1.2 an 0.3 �

weight 14311'9

2 4 11'9 911'9

Velocity Pequired, FPS

23 33

11 18 39

26

Dynamic Pressure Required. PSI

0.0042 0.0087

0.0011 0.0027 0.0125

0.0056

However, practi tioners recommend to increase these values of the velocity very significantly to allow for local adhesion (greasy particles), surface roughness, etc. Consequently, it was decided at\this stage that a conservative design should allow veloc­ities on the axis (which are maximum values) tOI be approximately eight times higher than the largest of the velocities given above. The analytical or design basis for this coefficient is given in Appendix A4• Then

Vmin :::::: 312 fps rounded up to 100 m/sec (328 fps) (12)

Thus, the minimum dynamic pressure should be

357

(Pdyn) min = 0.87 psi (13)

This condition, introduced into Eq. (10), yields the equation (with Pdyn (0) in psi)

x2 _ K2 Pdyn (0)

D2 max- 0.87 (14)

This equation specifies the maximum distance (in number of nozzle diameters) at which the material should be located away from the nozzle to main­tain effective "jet" cleaning properties. The dia­meter of the circular section of the cone of the �pex angle of 20 deg. (see Fig. 7) at xmax' (Dj)max' IS

. _

(1 + 2 xmax tan 10°) Djmax - D2 D2

or if criterion (13) is adopted,

DJ· = D2 1 + 0.353K Pdyn (0) � max (15) 0.87

with the pressure expressed in psi.

EXPOSURE TIME REQUIREMENT

We define as "exposure" time, Te, the period of time during which, under conservative assomp­tions, an "average-size" particle of contaminant should be exposed to the steam jet on either side of the jet axis after it becomes loose and detached from its substrate. The "residency" time, T r, is the time spent by an element under the jet at each pass­age: Tr = 2Te.

Let F be the average propulsive force, acting steadily in the direction of motion, T e be the time durin.g which it act� and M{', vp the mass and velocity of the typical particle after time Te. Applying the momentum theorem

-(16)

The value of F is estimated as follows:

F = kO kmv Pdyn S (17)

in which kO: is a reduction factor taking into accoun t

that only the component of F (total) along the direction of motion is pro­pulsive.

kmv: takes into account the fact that the momentum obtained in section x is only a fairly small fraction of the one which would exist if the velocity along the axis was held constant across the jet section.

358

S: is the area of the particle normal to the jet.

Details on the calculation of kO, kmv are given in Appendix As .

.

The criterion on vp is that it should be higher than any likely pick-up velocity, taken to be

vp:;;;' 50 fps (15 m/sec) (18) Thus, in Eq. (16), the required exposure time T e can be computed.

DESIGN PARAMETERS OF ROTATING DRUM:

JET CLEANING LENGTH

Going back to Fig. 4, it is seen that the "jet cleaning length" can be determined from the above analysis, if Dj is realtively small compared to D, as

The maximum cleaning length is computed as:

Lc = D· = D [1 + 0.353 K (Pdyn(O) psi)1/2) Jmax 2 0.87 if the criterion (13) is adopted to determine the maximum length away from the nozzle exit section.

PRETUMBLER (Lp) AN D AFTERTUMBLER

(LA) LENGTHS

In a typical trommel screen [27) , it is estimated that most of the screening, or removal of fines, is effected in the first 72 in. of 36 in. diameter trom­mels. Therefore, a minimum length for the tumbl­ing or mechanical shaking of the feedstock prior to entering the settling length (pretumbler) or after leaving the cleaning section can be reasonably estimated to be:

Lp = LA � 2D (minimum) (21) Therefore, the rotating drum would have a total

length/diameter of

L� Dj +4 to 5 D for Dj �D (22) which appears to be within normal ore processing practice (L � 2 to 6D).

PERIPHERAL (Vt) AND AXIAL (Va)

VELOCITIES

Since the velocity of the jet on the axis de-

creases with the distance to the nozzle, up to a distance of about 100 nozzle diameters, and much more rapidly further downstream, it follows that a single layer of elements is preferable. A typical length of 4 in. corresponds to an increase of 5.7 diameters of a 0.7 in. diameter nozzle in the ex­ample treated below. At the limit of "effective jet cleaning," i.e., 34 D2 or at a 24 in. distance from the nozzle, the dynamic pressure on the axis would drop 27 percent over the added 4 in. distance away from the nozzle. Furthermore, only those surfaces subject to the direct impact of the jet are likely to be effectively "scrubbed" clean from their loose contaminants.

Therefore, the dimensions will be limited to a one-layer thick distribution of the product. The trace of the jet on this layer is approximately a circle C; of diameter Dj in a horizontal plane (Fig. 10) if Dj is small compared to D.

, / // . �/

c· I

.. '

FIG.l0 TANGENTIAL (Vt) AND AXIAL (Va) VELOCITIES

Knowing de ' the eqUivalent "diameter" of the element, ana the area packing density, 0a (see the section on packing), the number of ele­ments, nj, present in Cj, is:

Dj 2 nj = oa -­

deq (23)

-

Let 8 be the arc on the circular cross-section of the drum of diameter D, having chord AA' of length Dj: Let i be the time needed for an element to cover 8. Obviously, if n is the drum angular speed,

- - - -

8 8 8 n =-::- D' = D tan-� D-t' J 2 2 � small compared toj> On the average, as the one-dimensional layers pro-

359

•

gress along the axis, a chord aa has length c=1I" Djl4, and the average residency time in circle Cj at any one rotation cycle of those elements subject to the action of the jet is:

11" D· T1I" =4 vJ , where Vt is the tangential velocity.

t -

Let 8 I, DI be the arc and chord along the whole transverse length of the layer, nl be the number of elements in a one-layer bed over a length deq along x, and nj,1 the number of elements exposed to the jet:

-

81 nl kl - -- --8 nj,1 The actual time for any element to have an average residency time Tr in circle Cj, T r, is

-11" e I

Tr = kl Tr = - -4 8

Let S I be the (one-sided) lateral surface of the ele­ment

where the width and height aspect ratios are aw = w/Q, ah = h/Q, respectively. The projected area normal to the stream jet is 11"/4 deq2 on the aver­age.

To expose all sides of the surface of the element to the jet, as the product progresses along the axis, the number of exposures should be:

8 Q 2 f.le = - ( ) (aw + ah + awah) 11" deq

(24)

The axial velocity, Va, should be such that any element is exposed ne times to the jet, as this element progresses axially, for a total average time

Ttotal = (25)

Hence the ratio of axial to tangential velocity is

4 (26)

DETERMINATION OF DRUM DIAMETER D

Since the nozzle has to be installed IIIslde the

drum, the minimum value of D is Qj, distance away from the nozzle corresponding to intercept Dj on the drum surface (Fig. II). On the other hand, large values of D would better guarantee that the bed be shallow (one-layer). Call 4>max (see Fig . 11) the angle at which the uppermost element, in the

FIG.11

£lidina i\n91e .

direction of rotation, starts sliding towards the vertical. This angle, for crumpled cans on a smooth, wet surface, was determined to be about 30 deg. Then the maximum number of elements, nl, max' which could be contained in a one-layer bed over an axial length of deq, is related to the number of elemen ts exposed to the jet, nj,1 , by:

4>max D n - n· = 4> I, max - (Dj/D) J ,I max �q

(27)

Now:

=

and .

Dj D2 j 2ne Tekl, max 2ne TeD4>max

Dj Dj D 2ne T e4>max

(D·)2 n h, max= 3600 _J _ _ _

deq 2neTe

(30)

(31)

Equation (31) can be used to relate the required throughput to the nozzle system distance Dj, and to derive the drum diameter D. Let us examine two special cases.

The minimum diameter of the drum cor­responds to the case illustrated in Fig. 11. In this case Dmin = Qj or

D "'=' _D",-j/_2 _-_D_2 /_2 tan (a/2)

If a = 20 deg., and D2 /Dj � 1 is neglected,

D-D"'=' _J_ = 2.84 D·

0.353 J

The maximum capaci ty, in number of elemen ts per Hence hour, nh max' is: ,

3600 Va D nh max = - 4>max (elements/hr) , (deq) deq

(28)

In the linear range, let kl, max be the value of kl corresponding to maximum capacity

4>max D kl max = ---, D.

(29) J

For4>max =30 deg., 3/2 �kl,max �3 since Qp:� D ::;;; 2 Q j

Indeed, it is not considered practical that Qj be smaller than D/2, i.e., that the jet be much closer to the bed than one radius away from the bed, since this would lead to a poor utilization of the space, the elements being neither tumbling nor exposed to the jet for about two-thirds of the time.

360

(32)

The maximum diameter would correspond to

Hence

D = 2 Q. max J

D· J D"'=' 0.176

nh,max 111.9

Angular Speed, n

5.67 Dj

(33)

n is derived straightforwardly from n = 2Vt/D

Slope, s, of the Drum Axis

If the formula for rotary kilns is adopted (no lifters) and Tt is the transit time [28] , the required slope s, in in./ft, is

_ L 601T _ Va s - 12x 0.19 x nOT 30 -7.16 V t t (34)

This completes the analysis and the derivations of formulae required in the design of the cleanup device. An illustrative example follows.

APPLICATION TO AN EXAMPLE:

PROTOTYPE TEST UN I T AT 0.6 TPH

THROUGHPUT

1. Given a value of throughput (nh), or number of elements/hr to be cleaned, and a description of the feedback (Q, aw, ah), the designer's first constraints are generally the characteristics of available steam cleaners, which in turn will deter­mine nozzle and jet parameters. In the example, the following data are given:

(a) Throughput rate: 0.60 tons/hr (II ,390 elements/hr) Diameter of drum: to be minimized

(b) Feedstock element: Q = 4 in., aw = 0.5, ah = 0.25. Thus, from equation (I); deq = 0.62 Q = 2.48 in.

(c) Nozzle characteristics: the values for a commercially available portable steam cleaner are Po = 80 psig = 95 psi a (supply pressure) m = 2 gpm = 1,000 Ib/hr

The maximum mass of condensates is estimated, for 1180 Ib/hr, canstock from Eq. (2), using the values mo.c = 50 Ib/hr, fs.m = 10 percent, to mc :::::: 950 Ib/hr.

2. Dimensioning of nozzle sections and dynam-• IC pressure po.

Allowing for a 15 percent enthalphy loss in the nozzle [12] , starting with saturated steam at 95 psia, the following values are calculated: Pthroat = 55 psia Velocity at throat = 1484 fps. Section at throat "" 0.20 sq in. Dz = 0.71 in. For 9 percent initial moisture, P"" 0.041Ib/ft3. Pdyn (0) = 25.6 psi. The·mini­mum dynamic pressure being 0.87 psi, the maxi­mum diameter Dj of the intersection of the jet

with the feedstock layer is computed from Eq. (IS), where K = 6.2: Dj = 0.71 x 12.87 = 9.14 in. The layer should be placed at distance Qj from the nozzle exit section: Qj = 23.9 in., or 33.7 nozzle diameters

The minimum drum diameter is D = Qj. To pro­vide a 2 in. clearance for installing the nozzle, take D = 26 in.

3. The exposure time, Te, on the basis of criterion (I 8), is found to be 0.36 sec for Mp = 0.0578 oz., S = 1 sq in., ke = 0.1, kmv = 0.175. -Hence Tr = 0.72 sec, Tr = 1.08 sec.

The number of exposures required, ne, is computed from eq. (24):

8 4 ne = - ( )2 x 0.88 = 5.83, taken to be 6.

1T 2.48 The "axial time interval" is, since kl = 1.5, Taxial = 6.48 sec.

4. Hence the axial and tangential velocities, angular rate and slope of the drum are Va = 1.41 in./sec, Vt = 9.97 in./sec, n = 7.32 rpm, s = 1.01 in./ft.

5. The length of the drum is, from Eq. (22), L = 113 in., rounded up to 9.5 ft.

6. Summarizing: the prototype will be a 9.5 ft long, 2.2 ft diameter trommel, rotating at 7.3 rpm, inclined 1.01 in./ft on the horizontal and handling 0.60 tons/hr of typical canstock. The steam noz­zle, supplied at 95 psia with saturated steam at the rate of 2 gpm, should be located 2 ft above the feedstock in the trommel.

CONCLUSIONS

An analysis has been given of the design of a device combining a tumbling action in a trommel and the effect of a steam jet to lower the level of loose organic contaminants to specified levels in recovered magnetic materials. It is hoped that the procedure outlined will prove helpful in the rational choice and assembly of components of a prototype unit on which the efficiency of con­taminants removal, energy consumption, waste­water treatment requirements and economic trade­offs will be determined in an extensive series of tests.

ACKNOWLEDGEMENTS

The author wishes to thank Dr. Harvey Alter of the National Center for Resource Recovery for his guidance and helpful suggestions.

361

APPENDIX

AI. Length of the core region, L 'c. For sub­sonic exhaust velocities, [20-21] tc R< 4Dz. For supersonic exhaust velocities, the relationship be­tween core length and jet exhaust Mach number Mj is given by:

LcR«5.22Mt9 + O.22) Dz [21].

A2. Length of the transition region Ls. The distance Ls, measured along the jet axis from the exhaust section to the sonic poin t, is given by:

Ls = (5MjZ + 0.8) Dz [18].

A3. Compressible, subsonic region. In [23], the assumption that the static pressure in the stock-free flow downstream of the observed stock structure was verified. The stagnation temperature was also measured to be equal to the temperature of the surrounding medium. Abramovitch [24] , in the case of high subsonic jet speed with small heat transfer, found that that effect of compres­sibility upon the fundamental properties of jet mixing for Mj ::s; 1 is negligible (also see [25]).

A4. Determination of minimum velocity, vp. The factor 8 by which the highest pickup velocity appearing in the table given in the body of the text (40 fps) has been multiplied to obtain vp was derived as follows:

a. A factor of 2.5 on the velocities (to about 100 fps) has been included to take into considera­tion the recommendation of industrial air jet clean­ing practi tioners [14] .

b. Another factor corresponding to the square root of the ratio or the densities of steam and standard air or 1.34 accounts for the dependence of pick-up velocities on fluid density.

c. The velocity across the jet section is maxi­mum on the axis, and decays away from it in Gaussian fashion. Using a Gaussian curve, having a maximum equal to V on the jet axis, and a para­meter computed from equation (11), the follow­ing relation was obtained between the average and maximum dynamic pressures at x:

p = 0.18 P (x).

Hence a factor of (1/0.18YI2 = 2.38 on the veloci ties.

d. The final factor used is from a., b., and c.:

2.5 x 1.34 x 2.38 R< 8

AS. Estimating ke and kmv. It is assumed that the typical situation is one in which the jet impacts

on a particle of density 1, cross section normal to the jet 1 sq in., thickness 0.1 in. Over an arc of about 10 deg. (Fig. 3), the average tangential component of the force is of the order of 0.1 times the normal force (ke R< 0.1). kmv is comput­ed as explained in I, c (kmv = 0.176).

REFERENCES

[1] Alter, H., "Energy Conservation and Fuel Pro·

duction by Processing Solid Wastes," Environmental Con·

servation, Vol. 4, 1977, pp. 11-19.

[2] Franklin, W. E., Bendersky, D., Park, W. R.,

and Hunt, R. G., in The Energy Conservation Papers,

R. H. Williams, Ed., Ballinger, Cambridge, 1975, Chap. 5.

[3] Alter, H. and Crawford, B., Material Recovery

Processing Research, National Center for Resource Re·

covery, Washington, D.C., Contractor's Report to EPA

No. 67-01·2944, Oct. 1976, p. 61.

[4] Sullivan, P. M. and Makar, H. V., "Quality of

Products from Bureau of Mines Resource Recovery Sys·

tems and Suitability for Recycling," Proceedings of the

Fifth Minerals Waste Utilization Symposium, Chicago,

Illinois, April 13·14,1976, p. 228.

[5] Duckett, E. J., "Contaminants of Magnetic

Metals Recovered from Municipal Solid Waste," National

Center for Resource Recovery, Washington, D.C., Novem·

ber, 1977.

[6] Alter, H., Natof, S., and Woodruff, K. L., "The

Recovery of Magnetic Metals from Municipal Solid Waste,"

National Center for Resource Recovery, Washington, D.C., April 1977.

[7] Alter, H. and Reeves, W. R., Specification for

Materials Recovered from Municipal Refuse, EPA 670/

2·75-034, U. S. Environmental Protection Agency,

Washington, D.C., 1975, 109 pp.

[8] National Center for Resource Recovery,

Materials Recovery System Engineering Feasibility Study,

Washington, D.C., pp. 4·87.

[9] Greze, J. P., Industrial Detergency, Ed. by

William W. Niven, Jr., Reinhold Publishing Corp., New

York, 1955, p. 138.

[10] Brick, R. M., Daly, J. J., Koehler, E. L., and

Skibbe, A. G., "The Selection of Tin Coatings for Steel

Containers," in Metals Handbook, Vol. 1: Properties

and Selection, American Society for Metals, 1961, p.

1133·1141.

[11] "Steel in Packaging, Committee of Tin Mill

Products Producers," American Iron and Steel Institute,

Publication TM650-676·20M·AP, p. 5.

[12] Marks, L., "Standard Handbook for Mechanical

Engineers," 7th Edition, McGraw·Hill, New York, 1967,

pp. 4·62.

[13] Tuve, Heating, Piping and Air Conditioning,

Vol. 25, No. 1, 1953, pp. 181-191.

[14] Bouilliez, L., "The Use of Air Jets in Dust

Control (in FrenCh)," Le Courrier des Etablissements

Neu, No. 72., December 1976, pp. 31·39.

[15] Albertson, Dai, Jensen and Rouse, Trans.

American Society of Civil Engineers, Vol. 145,1950,

pp. 639·664.

362

(16) Pai, S. I., Fluid Dynamic of Jets, D. Van

Nostrand Company, New York, 1954.

(17) Abramovich, G. N., The Theory of Turbulent

Jets, M.I.T. Press, Cambridge, 1963.

(18) Nagamatsu, H. T., Sheer, R. E., and Harvey,

G., "Supersonic Jet Noise Theory and Experiments," in

Basic Aerodynamic Noise Research, I. R. Schwarz, ed.,

NASA SP 207, 1969, pp. 17-51.

(19) Dosanjh, D. D., Abdelhamid, A. N., and Yu,

J. C., "Noise Reduction from I nteracting Coaxial Super­

sonic Jet Flows," in Basic Aerodynamic Research, op. cit.,

pp. 63-101.

(20) Nagamatsu, H. T., Sheer, R. E. and Harvey, G., "Supersonic Jet Noise Theory and Experiments," op.

cit., p. 22.

(21) Nagamatsu, H. T. and Harvey, G., "Supersonic

Jet Noise," Report 69-C-161, GE Research and Develop­

ment Center, 1969.

(22) Nagamatsu, H. T., Sheer, R. E. and Harvey, G.,

"Supersonic Jet Noise Theory and Experiments," op.

cit., p. 24.

(23) Dosanjh, D. D., Abdelhamid, A. N. and Yu,

J. E., "Noise Reduction from Interacting Coaxial Super­

sonic Jet Flows," op. cit., p. 68.

(24) Abramovich, G. N., ".The Theory of a Free Jet

of a Compressible Gas," NASA TM 1058, 1944.

(25) Pai. S. I., Fluid Dynamics of Jets, op. cit., p.

143.

(26) Donald, M. B. and Singer, H., "Entrainment

in Turbulent Fluid Jets," Trans. Institution of Chemical

Engineers. Vol. 37, 1959, pp. 255-267.

(27) Taggart, A. F., Handbook of Mineral Dressing,

J. Wiley and Sons, New York, 1945, p. 7-34.

(28) Perry, R. H. and Chilton, C. H., Chemical

Engineer's Handbook, 5th Ed., McGraw-Hili, New

York, 1973, pp. 20-40.

Key Words

Analysis

Ferrous

Organic

Refuse

Rotating Drum

Separating

363

Discussion by

J. P. Dodt Rust Engineering Company

Birmingham, Alabama

Mr. Renard's description of the cleaning mech­anism, the subject of his paper, led me to believe that his described theoretical approach could be utilized for other purposes than purely cleaning the magnetic material in the front end recovery of "can stock" of the incoming waste stream.

As may be known, the Wheelabrator-Rust installation at Saugus, Massachusetts, does recover magnetic material. But since this installation is a mass-burning steam generating plant where the prime emphasis on reclaiming the various available "recyclables" from the municipal waste stream is on the energy portion, the attention (and money spent) on achieving a high grade of metallic mag­netics from the ash residue is minimal. As a result, it can be easily stated, as in most mass-burning systems, the value of the recycled magnetics from a system of this sort, especially with today's scrap prices, is doing well to merely pay for the cost of the installation and operation of the equipment. Most of the time at Saugus, we are only able to justify the removal of the magnetic material by taking into account the savings for the cost of handling and land filling this amount of the system's ash residue.

To improve the above economics we have con­sidered upgrading the recovered magnetic material by washing, steam-cleaning, etc., but to do this would entail simple, low first cost, low operating cost technology to be able to take advantage of the incremental improvement on return. It was thought the technology presented in Mr. Renard's paper would accomplish this, but would appear this is so highly theoretical and is specifically

364

applied to the removal of clean (untouched by a combustion process) putrescible from the incoming waste stream.

It would appear a full scale demonstration sys­tem must be installed to obtain test results of this same application to the removal of ash residue im­purities from the magnetic material on the ash handling end of a mass-burning system. Additional­ly, it would appear the magnetic material must receive more than just a single steam jet cleaning treatment but also must be exposed to some form of impacting, massaging, size classifying, etc., to loosen the sintered ash from the desirable magnetic metals.

In applying this precarious and extensive test program, backed up only by some high level the­oretical mathematics, to an existing operation which is already marginal from an economic view­point, would appear to be throwing good money after bad.

It appears the good design considerations brought out in this paper can only be applied to the front end system and not to the more complex requirements established by the type of material requiring cleanup subsequent to mass-burning com­bustion of the waste stream.

AUTHOR'S REPLY

ToJ. P. Dodt Mr. Dodt is quite right. This paper does not

deal with post-incineration ferrous scrap, nor was it ever intended to. That the upgrading of inciner­ator magnetic residue should be a matter of con­cern is well understandable. Even in the latter case, it is our belief that some analysis should precede, not follow, the building of "full scale demonstra­tion systems."