LIBRARY
MEChiQan State
HN@NMy

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINE return on or before date due.

 

DATE DUE

MTE DUE

DATE DUE

 

lfigfifi

9‘3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/” WWW“

LIGHT MEDIUM SEPARATION OF HIGH
DENSITY POLYETHYLENE AND
POLYPROPYLENE IN A HYDROCYCLON E

By

David Charles Carlson

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Chemical Engineering

1995

ABSTRACT

LIGHT MEDIUM SEPARATION OF HIGH DENSITY POLYETHYLENE AND
POLYPROPYLENE IN A HYDROCYCLONE

By

David Charles Carlson

Experiments were conducted to study the use of hydrocyclones to separate high
density polyethylene (HDPE) and polypropylene (PP) from glass microbubbles
and water. The size distribution of the microbubbles is much smaller than the size
of the two plastic constituents. Experiments show that the recovery of I-IDPE in
the underflow stream increases as the efl‘ective density of the feed stream
decreases below 900 kg/m3. The purity of the underflow stream, as measured by
the mass ratio of HDPE/PP, remains above 9 for feed densities above 800 kg/m3.
The high microbubble loading in the feed stream is attributed to the concomitant
separation of the low density microbubbles in the flow field The critical factors
which control the I-IDPE/PP separation performance include the concentration of
microbubbles in the underflow stream, the geometry of the hydrocyclone, and the
flow split ratio. The recovery coeficient E for I-IDPE can be correlated with
py/pso, where or is the density of the feed stream and p50 is an intrinsic cut-density

of the separator.

To my wife and my parents...

Acknowledgments

I wish to thank Dr. Charles A. Petty and Dr. Syed K. Ali for their guidance in this
research, and the undergraduate employees of the hydrocyclone research lab,
especially John Burke, who assisted me with the experiments. Funding for this
research was provided by the American Plastics Council, the Department of
Energy Innovative Concepts Program, the Michigan State University Foundation,
the Michigan Materials and Processing Institute, and the State of Michigan
Research Excellence Fund. Technical assistance and materials were provided by
Robert Hill of 3M, Gerald Kelton and Mark Hoyack of Krebs Engineers, Paul

Marsh of Michigan Polymer Reclaim, and PlastiPak.

TABLE OF CONTENTS

Section Page
Number

List of Tables ................................................................................................ vii
List of Figures ................................................................................................ ix
Notation ......................................................................................................... xi

1. Introduction .................................................................................................... 1
1.1. Motivation ............................................................................................ 1
1.2. Background .......................................................................................... 3
1.3. Objectives ............................................................................................ 8
1.4. Methodology ........................................................................................ 8

2. Experimental Design ...................................................................................... 10
2.1. Glass Microbubbles ............................................................................ 10
2.2. High Density Polyethylene and Polypropylene ................................... 18
2.3. Test Hydrocyclones ............................................................................ 22
2.4. Flow Circuit and Sampling Protocol ................................................... 27
2.5. Definition of Separation Performance ................................................. 31
2.6. Scope of the Study ............................................................................. 39

Section Page

Number
3. Experimental Results ..................................................................................... 41
3.1 Hydrodynamics .................................................................................. 41
3.2 Stability of Microbubble Suspension .................................................. 44
3.3 Separation Performance ...................................................................... 51
4. Discussion of Results ..................................................................................... 63
4.1. Hydrodynamics ...................................................... ‘ ............................. 63
4.2. Stability of Microbubble Suspension ................................................... 68
4.3. Separation Performance .................... ‘ ................. ‘ .................................. 72
5. Conclusions and Engineering Significance ..................................................... 78
6. Recommendations for Further Study .............................................................. 86
Appendices
Appendix A. Hydraulic Data ........................................................................ 89
Appendix B. Medium Stability Data ............................................................ 91
Appendix C. Separation Performance Data .................................................. 97
List of References ............................................... ............................................ 107

Table

A. 1.
A2.
B. l.
8.2.
3.3.
3.4.
8.5.
3.6.
C. l.
C.2.

C.3.

LIST OF TABLES

Page
Number
Growth of Thermoplastics Recycling ......................................................... 2
Properties of HDPE and PP ........................................................................ 4
Options for Microsorting Mixed Thermoplastics ........................................ 5
Options for Lowering the Density of the Continuous Phase ..................... 11
Comparison of Light Medium Hydrocyclone and Dense Medium
Hydrocyclones .......................................................................................... 81
Flow Rate/Pressure Drop Data for the 20° Hydrocyclone ......................... 89
Flow Rate/Pressure Drop Data for the 10° Hydrocyclone ......................... 90
Average Density of Glass Microbubbles .................................................. 91
Size Distribution for the K20 Microbubbles ............................................. 92
Size Distribution for the K46 Microbubbles ............................................. 93
Medium Separation Data .......................................................................... 94
Glass Microbubble Breakage in Different Pumps ..................................... 95
Glass Microbubble Breakage ................................................................... 96
HDPE/PP Particle Size Distribution ......................................................... 97
Data for the 20°-22 Hydrocyclone at 5 psi ............................................... 98
Data for the 20°-22 Hydrocyclone at 10 psi .............................................. 99

“Oi

Table Page

Number
C.4. Data for the 20°-16 Hydrocyclone at 5 psi ............................................. 100
C5. Data for the 20°-16 Hydrocyclone at 10 psi ........................................... 101
C6. Data for the 20°-10 Hydrocyclone at 5 psi ............................................. 102
C7. Data for the 20°-10 Hydrocyclone at 10 psi ........................................... 103
C8. Data for the 20°-22 Hydrocyclone at Spsi with K46 Microbubbles ........ 104
C9. Data for the 20°-22 Hydrocyclone at 7psi with K46 Microbubbles ........ 105
C. 10. Data for the 10°-22 Hydrocyclone ......................................................... 106

Figure

10.
ll.

12.

13.

14.

15.

LIST OF FIGURES

Page

Number

Separation Concept .................................................................................... 7
Size Distributions of Glass Microbubbles ................................................ 16
Size Distributions of PP and HDPE ......................................................... 19
SEM Micrographs of HDPE and PP with Microbubbles .......................... 21
Photographs of the Krebs DB4-12 Hydrocyclone .................................... 23
Dimensions of Krebs Hydrocyclone and Fittings ..................................... 24
Underflow Withdrawal Schemes .............................................................. 26
Schematic of Light Medium Flow Circuit ................................................ 28
Separation Performance Measures for the Light Medium Hydrocyclone 37

Pressure Drop vs. Flow Rate for the 20° Hydrocyclone ........................... 42
Pressure Drop vs. Flow Rate for the 10° Hydrocyclone ........................... 43
Split Ratio as a Function of Feed Flow Rate for the

20° Hydrocyclone .................................................................................... 45
Split Ratio as a Function of Feed Flow Rate for the

10° Hydrocyclone .................................................................................... 46
Underflow Density as a Function of Pressure Drop ................................. 47

Comparison of Microbubble Break-Up in Centrifugal
and Progressive Cavity Pumps ................................................................. 50

ix

Figure Page

Number

16. Microbubble Break-Up with Time in a Progressive

Cavity Pump ............................................................................................. 52
17. Photograph of Separation of PP and HDPE in a Light

Medium Hydrocyclone ............................................................................ 53
18. The Effect of Split Ratio on the Performance of a

Light Medium Hydrocyclone .................................................................. 55
19. The Efi‘ect of Inlet Flow Rate on the Performance of

a Light Medium Hydrocyclone ................................................................ 57
20. The Effect of Microbubble Size on the Performance of a Light

Medium Hydrocyclone ............................................................................ 59
21. The Effect of Cone Angle on the Performance of a Light

Medium Hydrocyclone ............................................................................ 60
22. Percent Yield of HDPE in a Light Medium Hydrocyclone

as a Function of HDPE Feed Concentration ............................................. 62
23. Pressure Loss Coeficient vs. Reynolds Number for

the 20° Hydrocyclone .............................................................................. 66
24. Pressure Loss Coefficient vs. Reynolds Number for

the 10° Hydrocyclone .............................................................................. 67
25. Yield and Underflow Density at Difi‘erent Inlet Densities ........................ 69
26. 950 as a Function of Split Ratio ................................................................ 74
27. Flow Diagram for a Light Medium Hydrocyclone ................................... 83

IAPI

<Ua>

C9
DWI
Br
152

HDPE

PP

Notation

Absolute value of the pressure drop over the hydrocyclone
Average tangential velocity

Parameter used to fit an exponential curve to HDPE yield
data

Pressure loss coefficient

Dense medium hydrocyclone

Recovery of HDPE in the underflow stream
Recovery of PP in the overflow stream
High Density Polyethylene

Constant relating pressure drop to flow rate
Light medium hydrocyclone

HDPE stream purity coemcient in stream x
PP stream purity coefficient in stream it
constant relating pressure drop to flowrate

Hydrodynamic parameter relating tangential velocity to
location in the hydrocyclone

Power (watts)

Polypropylene

Qx

Re

TRCZ

Wix

Yix

Greek

Sr

82

Pso

Subscripts

Volumetric flowrate of stream x
Radial distance from vortex axis
Reynolds number

Split ratio

Toroidal Recirculation Zone

Bulk feed velocity to the hydrocyclone

Mass fraction of constituent i in stream x

Volume fraction of constituent i in stream x

Density ratio

Major cone angle
Underflow cone angle
Overflow purity coefficient
Underflow purity coeflicient
Diameter of a particle
Density

Feed density at which 50 % of the HDPE reports to the
UP

HDPE
PP

Microbubble

XII

'11

CO:

Water

Feed
Hydrocyclone
Overflow

Underflow

xiii

CHAPTER 1

INTRODUCTION

1.1 Motivation

The rapid increase in the use of thermoplastics have led to many improvements
and advancements in the automotive and packaging industries. These materials
offer the advantages of relatively low weight/cost and durability over traditional
materials such as paper and steel. However, these same properties make
thermoplastics a problem when their useful life cycle has ended inasmuch as they
comprise a disproportionate volume of landfill space when compared to other

materials such as glass and steel.

Like many other materials presently going to landfills, thermoplastics have the
option of being recycled and reused. Table 1 shows that the rate of thermoplastic
recycling is growing rapidly. With this expansion, it has become increasingly
important to obtain clean feed stocks from this resource. This study is concerned
with the removal of contaminate thermoplastics from HDPE. Most thermoplastics
are heavier than water and are easily removed from HDPE, which has a density
lower than that of water. PP, however, also has a density lower than water and is

difiicult to separate from HDPE.

Table 1: Growth in Plastic
Packaging Recycling (APC,1993)

 

Percent Change from

 

 

 

 

 

 

Resin Type
1990 to 1991

PET 29.2
HDPE 75.1
PVC 6.7
LDPE/LLDPE 10.4
_PP 1200

PS 85.3

 

 

 

 

3
Unfortunately, the presence of PP in the recycle waste stream may affect the

physical properties of the reclaimed HDPE. For instance, PP has a much lower
izod impact strength than HDPE (see Table 2). Also, PP may have a different
color than HDPE as is the case with a milk jug and its PP cap. Even in small
amounts, the PP may change the hue of the reclaimed HDPE, making it more
difi'icult to obtain a consistent product (Carlson et al., 1993). The demand for
uncolored material is much greater than the demand for a colored product. In
1993, the estimated amounts of recycled HDPE in the United States were 240
million pounds for natural HDPE and 125 million pounds for the colored material.
This represented approximately 3% of the virgin HDPE sales for that same year
(Carlson et al., 1993). The national goal for recycling HDPE is 10% of virgin

sales, so there is tremendous opportunity for separating PP from HDPE.

1.2 Background

There are several difierent methods being evaluated for microsorting mixed
streams of thermoplastics. These include float/sink technology, optical techniques,
chemical salvation, and hydrocyclones. Table 3 briefly summarizes the
advantages and disadvantages of the listed methods. The simplest method for
separating PP and HDPE is the float/sink process in which PP is the light
component and HDPE is the heavy component. This type of process may employ

either a homogeneous fluid or a fine suspension with an effective density

8a— .wemamoooi new mam—canon oEouobu ..¢.Z esfim .
82 32.2.2 .3. «32%. Sam 5082 .

 

ed - to 8 v 8.38; 8. 2e - 8e .5

 

 

 

3 - wd en T 2 £3an 8 “—355 mm? mom - :5 name.—

6.»
A532,: . Gov .dEom. eds—oh
Lamp—ohm 8.3:: we: . .95? 0.55 2:0on0 3 2308.; .5528... man—O

 

 

 

 

 

 

 

 

.3 2: an: He seems.— & es:

catfish—um

 

3.8 3:98 32

 

xEmbuo—h— 3 25am mace—052v»:
82.3.8 05an 8530 such“: 389—80
cot—Sq
28° 2258: 233 $2: 0528 :8 ”Bog—om ecu—xx
:wE wig—5:0 Esau:

.«e omen. 023 .23 etc?

 

 

ozmeoaxm Each :3:

awry—8m:— vo=c :23 x83

Mohamewfi 5:53 2.3.5

 

 

88: ooeoEmoe 9:5

”8829’ owes:—

mews—aficama

3:828 31.60 £ bacon

 

moms=a>c<

 

 

a2: a

£590 60.33822
«.25 .856 2%

Beacon—coo

gag—ow Eat—.26

 

 

Bags—ooh 3230

 

€558...—

 

 

uofio—z eaten—«now

 

 

me_.ma_._eE._u__.—. 9.3:: ”5:89.32 3.. 2.82.6 "m 93:.

6
in-between HDPE and PP (Afland et al.,1994; Nugent, 1991). The same materials

which are used in float/sink operations can also be used in a hydrocyclone. The
hydrocyclone offers the advantages of short residence times, easy adaptability into
existing plants, and relatively low capital cost. However, complete separation

does not usually occur in hydrocyclones.

Figure 1 illustrates the concept of separating HDPE and PP in a hydrocyclone
using a suspension of glass microbubbles and water. Separation occurs because
the size distribution of the microbubbles is much smaller than the size of the two
plastics and makes the suspension appear as an effective medium. In this
environment, PP migrates towards the axis of the flow field and is removed with
the overflow stream. The HDPE migrates towards the outer portion of the flow
field and is removed in the underflow. This practical application of an effective
light medium in a hydrocyclone, albeit challenging, has much commercial

potential.

The use of suspensions to separate materials of different specific gravities is not
new. The first commercial use of a heavy-medium separation in the US. was in
1936 by the American Zinc Company for separating ZnS and limestone (American
Cyanamid, 1951). The dense medium hydrocyclone (DMH) for coal beneficiation

was developed by the Dutch State Mines in the late 1930’s (Driessen, 193 9). The

 
  
   
   
  
   

 

 

 

 

 

 

 

+ OVERFLOW
FEED
(tangential wall jet)
' Zia. . . a - . ‘
”:33 . . a '0
g‘? 2 .0 .' '
b': ~— Vortex Finder
Air Core / Air/Liquid Interface
Toroidal Recirculation
Zone
Region of High
Microbubble .
f Low
ggflzgg) milieu
Concentration
(higher density)
0 HDPE
@ PP
' “WWW" UNDERFLOW
' (umbrella discharge)

Figure 1: Separation Concept

8
further development of technology for dense media separations continues as an

active area of research. Coal and shale mixtures with particle sizes between 1-6
mm are easily separated in hydrocyclones using magnetite suspensions. Presently,
hydrocyclone technology is being investigated to separate coal particles which are

finer than 150 um (Miller, 1991). The analog of the dense medium hydrocyclone

for a light medium separation will be designated LMH.

1.3 Objectives

The goal of this study is to explore the separation performance of a commercial
hydrocyclone for separating HDPE and PP using a fine suspension of glass
microbubbles. The split ratio, inlet density, flow rate, and cone angle significantly
afi‘ect the separation performance of a DMH, so these same factors should also be

important for a LMH operation.

1.4 Methodology

This thesis examines the separation of HDPE and PP using an effective light
medium of glass microbubbles and water in a hydrocyclone. The plastics used
were obtained from PlastiPak, Inc. (Plymouth, MI) and the microbubbles were
obtained from 3M (St Paul, MN). The plastics were shredded to obtain a size
appropriate for testing. A dense medium hydrocyclone was supplied by Krebs

Engineers (Menlo Park, CA) for testing the LMH concept. Initial tests were

9
conducted to determine if the available flow loop in the hydrocyclone laboratory

could handle the thermoplastic/glass microbubble/water mixture and to determine
if PP and HDPE could be separated. This study explores in more detail the earlier

proof of concept study of Petty et a1. (1993). .

Chapter 2 summarizes the physical properties of the glass nricrobubbles, PP, and
HDPE used. in this study. The light medium flow circuit, test hydrocyclones,
experimental procedures, and the theoretical basis for evaluating the separation
performance are also developed in Chapter 2. In Chapter 3, the experimental
results are presented and then discussed in Chapter 4. The conclusions and
engineering significance of the results are presented in Chapter 5. A material
balance flow sheet for the use of the LMH concept to a specific HDPE/PP process
stream illustrates the potential of the proposed separation strategy. Chapter 6
identifies further research and development needed to commercialize the LMH

technology.

CHAPTER 2

EXPERIMENTAL DESIGN

2.1 Glass Microbubbles

Selection of Light Medium

The choice of an appropriate suspending medium for an LMH is critical.
Although there are many options for controlling the specific gravity for DMH,
there are few naturally occurring substances available with densities lower than
water. Table 4 lists possible materials for LMH. These options are: organic
liquids, air, fly ash, and hollow glass spheres (microbubbles). This sections
describes the advantages and disadvantages of these materials, and why glass

microbubbles were chosen for this study.

Organic liquids, such as ethanol and methanol, ofi‘er the advantage of a completely
stable medium in that the suspending medium is a homogeneous continuous phase.
This greatly simplifies the control of the separation because the density of the
continuous phase is uniform throughout the system, and can be tightly controlled
by mixing two fluids. However, the physical properties of HDPE may be afi‘ected
by the presence of an organic liquid. For instance, the organic liquids may absorb

into the HDPE, swell the polymer and, thereby, alter the physical

lO

11

Table 4: Options for Lowering the Density of the

 

 

 

 

 

 

Continuous Phase
Material Advan es Disadvan es

Organic Liquids Easily controlled density; May interact chemically with
uniform density plastics; high recovery costs

AirBublis Low cost; no need for Possible problems with

. recovery stabiliy, coalescence

Fly Ash Great abundance; little or no Small amount of floaters;
cost dependent on feed coal stock
Microbubbles No interactions with Cost; losses due to breakage;

thermoplastics; easily unstable medium

 

recovered

 

 

 

12
properties (Encyclopedia of Polymers, 1988). For this reason, organic liquids

were not chosen as the method for lowering the density of water between HDPE

and PP.

Another option for lowering the density of the continuous phase is sparged air. By
passing an air stream through a porous medium, small air bubbles can be formed
which have a suitable size distribution for light medium cycloning. An air sparged
hydrocyclone is a very attractive prospect because of the relatively low cost of
generating the bubbles, and because the air would not have to be recovered. Also,
unlike an organic liquid, the air bubbles would not interact chemically with the
polymers, or change the physical properties of the HDPE and PP. The air sparged
hydrocyclone is not without its own difiictrlties. There may be possible stability
problems with the air bubbles. Although it may be possible to obtain a suitable
feed size distribution, the turbulent environment within a hydrocyclone may break
the air bubbles or allow them to coalesce forming a distribution quite difi‘erent
from that of the feed In either case, the resulting size distribution may not be

suitable for this particular application.

Fly ash was studied as another option for lowering the density of the continuous
phase. Fly ash is a waste material fi'om coal burning power plants which, if not

captured, would fly out the top of the smokestack. The composition of the fly ash

13
depends on the source of the coal and the operating conditions of the bumer. A

portion of fly ash has a density which is less than that of water, so it was
conjectured that this material may also be suitable for light medium cycloning.

Also, since this is a waste product, the fly ash would be available at a low cost.

Samples of fly ash were obtained from the MSU Power Plant and the Lansing
Board of Water and Light. The samples were tested for particle size distribution
and density distribution (floaters/sinkers). The particle size was determined to be
approximately 30 um and the average density was estimated to be about 2000
kg/m3 for both samples. It was noted that there were floaters in each sample, but
the amounts were too small for any analysis. Although fly ash may be an
economically appealing option, the samples examined were inadequate for this

study.

Glass microbubbles were chosen as the suspended material for the light medium
hydrocyclone. The microbubbles are comprised of a glass shell with an air core,
and are commercially available from the 3M Company. The microbubbles were
chosen because they offer the advantages of low density and small size. Also, the
glass microbubbles do not interact with either HDPE or PP, and they are easily
removed and recovered from the thermoplastics. However, the microbubbles used

in this investigation have two disadvantages: (1) they quickly migrate to the vortex

14
core and are removed with the overflow stream (see Figure 1); and, (2) they break

in high shear flows. These issues will be addressed in detail later in this thesis.

Density of Microbubbles

The density of microbubbles is an important parameter in the light medium
separation. The density determines the concentration of microbubbles necessary
to produce a given inlet density and affects the stability of the water/microbubble
suspension. The average density of the glass microbubbles was determined using
the following procedure. A small amount of glass microbubbles of known mass
was placed into a graduated cylinder partially filled with a known volume and
mass of water. The cylinder was covered and agitated to suspend the
microbubbles. The total volume of the suspension was quickly determined before
a significant portion of the microbubbles came out of suspension, thus skewing the
reading. Knowing the mass and volume of the microbubbles allowed the density
to be calculated. The average density of the K20 distribution was determined to be
210 kg/m3 and that of the K46 distribution to be 440 kg/m3. The manufacturer’s
values for these distributions are 200 kg/m3 and 460 kg/m’, respectively. Although
the density of a microbubble is inversely related to its diameter (Ali et al., 1992),
this study did not attempt to determine a mathematical expression for the density

as a function of diameter.

l 5
Microbubble Size Distributions

The size distribution was determined via light scattering using a Malvem
MasterSizer X. A dilute suspension of glass microbubbles was placed in a small
flow cell and pumped through the MasterSizer. The MasterSizer requires the
refractive indices of the continuous and dispersed phases, and the absorption of the
dispersed phase. A refractive index of 1.54 and an absorption of 0.01 were used
' for the glass microbubbles, and a refractive index of 1.33 was used for the water.
The MasterSizer was programmed to take 50 sweeps of the light intensity within
its cavity, and then calculate a size distribution from these readings. This
procedure was repeated three times and the results were averaged. A mean size of
52 pm and 32 um was determined for the K20 and K46 microbubble distributions,
respectively. The manufacturer’s numbers for these distributions are 62 um and
44 pm,- respectively. The cumulative distributions for these nricrobubbles are
shown in Figure 2. The actual data from the Malvem MasterSizer are presented in

Tables B2 and BB.

The size of the microbubbles is a major factor controlling the migration of the
microbubbles towards the core of the vortex. For a particle Reynolds number less
than 0.1, Stokes’ law provides a good approximation for the drift velocity

(Svarovsky, 1984):

l6

83:39.22 $20 he 333....sz SE "n «Eu...—

-0--
d—
.0-

n.53. 2... u ....

.53. § u man:

.53 2,... u 8.59.35 e3.
.53 2 N u 8:33.... .5.
8288

a... a u .3.
a... Na u .5.
new :82

8.33222 as. . . .

£33.53 email

2:

‘—

1-

«I.

di-

~r—

uh

uee..o_... .35

d-

P
1

‘-

         
      

b p .
u H

 

o—

cN

cm

cw

on

ch

3

oo-

°ZIS ”Pull %

17

u = 3’ (9.: -pa) <ue >2 (1)
D 18% r

 

In the above equation, the following definitions apply:

Up = drift velocity of the particle

t = microbubble diameter

pc = density of the carrier fluid (water)

on = density of the microbubble

tic = viscosity of the carrier fluid (water)

I = radial distance from the vortex axis (see Figure l)
<ue> = swirl component of the mean velocity

For t = 32 um, pc = 1000 kg/m3, pH = 440 kg/m3, ’4de = 0.01 cm2/s, <ue> = 2
m/s, and r = 5 cm, the drifi velocity given by Eq. 1 is approximately 0.25 cm/s for

Re = lpcub/uc = 0.08. As the microbubble migrates towards the core (i.e. r -> O),

the acceleration of the particle, <u9>2/r, may increase from 8g near the wall to 80g
near the air core (see Figure 1). The magnitude of this efl‘ect obviously depends

on the internal flow patterns and the behavior of <u9>.

Clearly Eq. (1) shows that reducing the size of the microbubbles will significantly
decrease the drift velocity and may, thereby, increase the stability of the
suspension. For the two microbubble products used in this investigation, the ratio

of drifi velocities for the mean particle size is

name) _ [l’(pc -p.)1.... = (5:) (0.79) - 3 .,

up (K46) - [(20% “panic“ 3— 6-3; -

18
2.2 High Density Polyethylene and Polypropylene

Size Distributions of HDPE and PP

The HDPE and PP were obtained from PlastiPak, Inc. in Plymouth, Michigan.
The HDPE was received in the form of detergent bottles with PP being the cap.
These sources had not been in contact with any chemicals and were considered to
be pure thermoplastics. Due to losses while conducting the experiments, these
plastics were augmented with more HDPE and PP. Both the HDPE and PP from
this secondary source were unpigrnented and could not be distinguished from each
other after shredding so the materials had to be run separately. The sizes of HDPE
and PP were reduced using a shredder which is housed in the School of Packaging.
Shredding is distinguished fi'om grinding in that shredding entails the slicing of a
material with a sharp edge, while grinding causes size reduction by impacting a
material with a blunt device and actually shattering the piece into smaller

fractions.

The size distributions of HDPE and PP were determined on a mass basis using an
automated shaker and standardized sieves of known size. The sieves were placed
in descending order of size into the shaker and then the plastics were allowed to
separate for thirty minutes. The weight of thermoplastic on each tray was
determined, giving the cumulative size distributions for HDPE and PP shown in

Figure 3. The sizes of the sieves and data for determining the cumulative size

l9

 

 

 

 

 

2:: E... .E 2 2.2.3.5... new 2.. 2:»...

E... Jim 52%

mad

 

  

 

 

.N

8N

 

9..—

.0-

mad

0—

 

 

on

em

 

 

ow

 

 

 

 

 

2.2.2.52. 3.8.5 gm... 3

05.32.50.— 0

 

cm

 

ow

 

 

 

2..

cm

 

 

 

 

 

co—

%

pun ssew

OZISJO

20
distributions are shown in Table C.l. Figure 3 shows that HDPE and PP have

approximately the same size distribution and that 90 % of the material has a screen
size between 1 mm and 3 mm. These values are much larger than the average
sizes of the microbubbles (52 and 32 mn) and allows the suspension to act as an
effective medium. Figure 4 shows the microbubbles compared to the HDPE and

PP.

Density of HDPE and PP

The densities and density difference between two materials is important for
separation in a hydrocyclone (see Eq. (1)). These factors influence the choice of
medium and the density control of the efl‘ective continuous phase (microbubble
and water). The larger the density difl‘erence, the less stringent the control of the
feed density. The densities of both HDPE and PP are less than that of water (z
1000 kg/m3). As shown in Table 2 (see p. 4), HDPE and PP have density ranges

which are distinct from each other.

The densities of HDPE and PP were determined using a gravimetiic method. A
known weight of thermoplastic was placed in a graduated cylinder partially filled
with a known volume and weight of methanol. The volume change was

determined and the density of the material was calculated. Using this method, the

 

21

00001?

n
u.
0
0
5

 

hs of HDPE and PP with

: SEM Micrograp

Figure 4

Microbubbles

22
density of HDPE was determined to be 960 kg/m3, and that of PP to be 910 kg/m3.

These numbers are within the range of tabulated values shown in Table 2.

2.3 Test Hydrocyclones

A commercial 100 mm dense media hydrocyclone was used for this study. The
hydrocyclone is constructed of a metal outer shell and a polyurethane inner lining.
The polyurethane protects the. metal shell from the abrasive conditions of dense
media separations and is easily replaced. The hydrocyclone is comprised of four
major sections with difl‘erent interchangeable parts for the cone and underflow.
This modular design allows for a wide range of geometries to be studied. Figures

5 and 6 show the hydrocyclone and dimensions of each of its associated sections.

The hydrocyclone has an involute feed (see Figure 5) which begins as a circular
opening, but becomes a slit entry into the upper swirl chamber. The swirl chamber
consists of two cylindrical sections which are connected to each other and to a

metal plate which is used to support the hydrocyclone on the scaffolding.

A noticeable discontinuity occurs at the junction of the swirl chamber and the
conical section. The major conical diameter is approximately 3 mm larger than the
diameter of the cylindrical section. This discontinuity was designed into the

hydrocyclone to accommodate interchangeable components. The tolerances on the

23

 

.

Figure 5: Photographs of the Krebs DB4-14 Hydrocyclone

24

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. “~- -
IOWO'Q a? 1k :15" >4——"’ —><——" —~

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dimensions of Main Section
Ratio 20° Cone 10° Cone
DH 100 mm 100 mm
DC/DH 1.03 1.08
Do/DH 0.19 0.19
1313/1)H 0.52 0.55
lv/DH 0.74 0.74
IIIDH 1.29 1.29
12/911 1.15 1.15
I3/DH 1.29 2.58
Ap/AH 0.04 0.04
VH/DH3 2.65 2.99
LH/DH 5.57 6.86
Dimensions of Underflow Fittings
Ratio DU = 10 mm 16 mm 22 mm
W 0.5? 0.55
DU/DH 0.10 0.16 0.23
28 20° 173° 143°
I4/DH 1.29 1.29 1.29
15/1)“ ' 0.55 0.55 0.55
Scale 1:3.3
20°-22 Hydrocyclone Drawn

Figure 6: Dimensions of the Krebs Hydrocyclone and Fittings

 

25
polyurethane inserts are not as stringent as the machined parts, and the

discontinuity avoids a reverse shelf which would adversely affect the flow patterns

within the hydrocyclone.

The apex diameter of the hydrocyclone was controlled by using one of three
interchangeable polyurethane socks. This allowed for three different underflow
withdrawal schemes (see Figure 7). The withdrawal schemes are classified as

conical, hyperbolic, and parabolic. These are defined according to the relative
sizes of the major cone angle, 201, and the angle of the underflow fitting, 20. If20r
is less than 23, then the underflow is parabolic. A hyperbolic scheme is produced
when 201 is greater than 28, and a conical withdrawal is formed when the two

angles are equal.

The difl‘erent combinations of or and B give six different withdrawal geometries.
To distinguish between the different configurations, each geometry will be denoted
by its major cone angle (i.e. 20:) followed by the size of the underflow diameter.
For instance, the 20° cone in conjunction with the 16 mm underflow fitting gives a
hyperbolic design and will be designated as the 20°-16 hydrocyclone. This style
of designation was chosen because it allows for quick and easy recognition

between the different configurations.

3.2—om 33.2333 Beuaouab ”b 953%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

26

 

 

 

 

 

 

0V~ oON NNroON
ow; 0A: NN-oc— ohm oGN OmuoON OON OON OuroON
m: S 2.8 an an 2.8 na .8 2.5
andm nNAGN nNfldN
3.2.5:? 3.2.2.... .3233? 3.33%: assess» .835
an .

an
5N

 

 

27
2.4 Flow Circuit and Sampling Protocol

Light Medium Flow Circuit

The light medium flow circuit is shown in Figure 8. A 200 liter rigid HDPE tank,
100 mm hydrocyclone, and a centrifugal pump (Myers QP 30-3; 3 hp, 3450 rpm)
are the major components of the light medium circuit. A two inch copper tube
exits the bottom of the tank and feeds into the centrifugal pump. The copper
tubing then extends from the pump and connects to a section of high pressure

flexible hose. This hose is then attached to the feed inlet of the hydrocyclone.

A recycle flow stream through a 3/4 inch line provided additional agitation in the
tank. A wide range of pressures could be tested by applying back pressure on the
pump with the ball valve in the recycle line. Another ball valve located after the
bypass was used to set the inlet pressure and flow rate to the hydrocyclone. The

inlet pressure was measured with a 0-60 psi pressure gauge.

Valves located on the low pressure side of the pump allowed for the system to be
drained. The placement of the valves was such that either the entire system

including the tank or just the pumps could be drained.

Because of the abrasive nature of the microbubbles it was essential that the pumps

be flushed at least every other day. A gate valve located just above the bypass

28

l Air Vent

F... ,9—1—
— Y

 

 

 

Hydrocyclone

 

 

_%_

Flush Line Overflow

l nderflow
Mixer '

Recycle Line

, i l

Tank

 

 

 

Centrifugal - ._..
Pump ( )

 

v
A

To Drain

 

 

Figure 8: Schematic of Light Medium Flow Circuit

29
allowed clean water to be brought into the flow circuit. When the valves on the

bypass, hydrocyclone feed, and tank outlet were closed, the valve on the flush line
and the drain valve were opened. Water was then allowed to flow through the
open pipes to remove any particles. which were wedged in the pumps.
Backpressure was periodically applied to ensure that the entire pump cavity was

rinsed.

Density Measurements of the Light Medium

The density of the feed stream was an important parameter in this study. This was
controlled by the amount of microbubbles added to the system. A gravimetn'c
method was employed to determine the density of the system. This section

describes the method to sample the feed and underflow densities.

To ensure proper mixing, the centrifugal pump was started and the suspension was
allowed to recycle through the system for a minimum of five minutes at a flow rate
of 40 1pm or greater. The overflow and underflow streams were then combined
and allowed to collect in a three gallon container. The collected material was
mixed to ensure homogeneity, and then a portion of the material in the container
was poured into a graduated cylinder of known weight. The volrnne and weight of
the suspension were measured, and then the density was calculated. This

procedure was repeated five times and depending upon the average, the system

30
density was adjusted to the desired value by adding or removing microbubbles.

This same procedure was used to determine the density of the underflow stream.
Special care was taken to determine the volume of the suspension in the graduated
cylinder before the microbubbles began to come out of suspension and form a
froth at the liquid/air interface. When the microbubbles come out of suspension,
the volume increases due to the packing of the microbubbles. If the volume of the
suspension was measured after this occurred, then the calculated density would be

less than its true value.

Sampling of HDPE and PP

Once the density of the glass microbubble/water suspension was known, the
volume of the material in the tank was determined. This allowed for an estimate
of the total mass flowing throughout the system. The material entrained in the
piping and pump was considered insignificant and not taken into account when
determining the total mass in the flow loop and the amount of thermoplastics to

add to the system.

The suspension concentration of HDPE used in this study was approximately 0.1
wt.%. The separation of PP was also studied at feed concentrations of 0.1 wt.%.
This low concentration was necessary because the centrifugal pumps would

overheat when too many plastics were added to the system. The maximum amount

 

31
of thermOplastics used in this study was 0.5 wt.%. At this loading, the prunp

began to overheat and began to make uncharacteristic whining noises. Future

testing will employ progressive cavity pumps designed for multiphase flows.

Since some of the PP and HDPE were uncolored, each plastic was tested
separately. Once either thermoplastic had been added to the tanlg the pump was
started. The valve between the bypass line and the feed line was adjusted to obtain
the desired flow rate (see Figure 8). The system was allowed to run for a couple
of minutes, and then the samples were collected. Samples were collected using a
Number 20 standard sieve (850 um). Two or three samples were collected for
each inlet density. The samples were allowed to dry overnight and then weighed.
Afier completing the separation runs for HDPE, the plastic was removed from the
system using the sieves and replaced with PP. The separation experiments were
repeated for PP using the same inlet pressures and hydrocyclone configuration as

with HDPE.

2.5 Definition of Separation Performance
The following overall and component steady-state material balances for
immiscible mixtures are used to evaluate the separation performance of the light

medium hydrocyclone:
Qr = Qo + Qu (2)

32
YarQF = yioQo + Yiqu , i = 1, 2. 3, 4 (3)

where
Q: = total volumetric flow rate. of the feed stream
Q0 = total volumetric flow rate of the overflow stream
Q; = total volumetric flow rate of the underflow steam

yix = volume fraction of constituent i (i = l, 2, 3, 4) in stream X (X=F,O,U)

Each component of the process stream is identified by the index i (i= 1, HDPE; i =
2, PP; i = 3, glass microbubbles; and, i = 4, water). The mass density of each
stream can be calculated in terms of the pure constituent densities, p°i and the

volume fractions, ygx:

px =Zyixpf , X=F, O, U. (4)

The mass fraction of constituent i in stream X is Wix = p°i yix / px.

An important, and useful, performance measure is the stream purity eoefi’icient

Mlx for HDPE:

 

Mm. Z'xp' . ,x= F, O,U. (5)
er91 + erpz

A similar coefficient can be defined for PP:

M2x=1-M1x. (6)

33

The goal of the LMH separation process is to produce a high purity HDPE stream
from a feed stream contaminated with PP. Current commercial recycling
technology can produce a relatively clean HDPE/PP mixed stream for which
Mgr/M": = 0.01. Unfortunately, this level of contamination causes the melt
extruded product to have a grayish tint. If the light medium hydrocyclone could
yield an underflow stream with Mzu/Mlu = 0.001, then this product may be an

acceptable alternative to virgin HDPE.

The recovery coeflieients for HDPE and for PP provide additional performance

measures:

E E Y“) QU
l

, E, 5 M. (7 a,b)
Yerr

Y2? Q.

E and E; are also referred to as the yield of HDPE and PP, respectively. The
ability of the separator to remove HDPE from the overflow stream and PP from
the underflow stream can be evaluated in terms of the overflow and under/low

purity coeflicients defined by

YiF - YIO YZF - YZU (8 a’b)

3'2— , 325

Yrr Y2?

The overall and component material balances can be used to relate the above

measures to the split ratio S I Qu/Qo:

34

 

 

a +8
E = ' 9
' 1+8 U
l+Se2
= 10
2 1+8 ( )

Note that if 81 = 0 (i.e., ylo = yly) and S > 0, the recovery or yield of HDPE in the
underflow stream is larger than zero (E1 = S / (1 + 8)). Under these conditions the
hydrocyclone acts as a flow splitter. For positive values of 81, Eq. (9) implies that

——S-—SE,SI forOSSISI. (ll)
1+8 _ .

Similar observations can be made regarding the separation of PP, viz.,

—-1—SE251 forOSezsl. (12)

HS

/

The utility of 81 and a; as intrinsic separation performance measures stems from

their insensitivity to the split ratio S.

- Ifthe LMH objective is to maximize the separation of HDPE and PP then a useful

definition of separation efficiency is (see p. 166 in Bradley, 1952)

E a YIUQU +Y20Qo
Yerr +Y2PQ1=

E = Y1? Er + (1" Yrs )E2 (13)

35

 

 

a +8 1+Ss
E=Y ' +1—Y 2
" 1+s ( 'F) 1+s
where
Y..=--——yi—. (14)
ylF+YZF

Note that for E1=E2, the above definition for the separation efliciency reduces to
E=E1=F4. For a given hydrocyclone design and for a specific HDPE/PP ratio in

the feed steam, the separation efficiency E is determined by the Reynolds number

(Re; a 4Qp/ (nDpvc)) and the density ratio (by:

0<¢Fa%%E:—<l (15)
4 3

where p°4 and d3 represent, respectively, the densities of water and microbubbles.
The split ratio S (5 Qu/Qo), which also affects E1 and E2, is determined by Rep for

a hydrocyclone operating with an air core.

The feed density pp can be changed by the addition of nricrobubbles to the mixed
plastic suspension (see Eq.(4)). As m. -+ 0 (i.e., pF —> p°3), both HDPE and PP
will be separated to the underflow (e, -) l and a; —> 0). As <11; —) 1 (i.e., n.- -)
p3), both HDPE and PP will be separated to the overflow (a, —-> 0 and a; a 1).

For these limiting conditions, the separation efficiency becomes

36
_1+Y,FS

 

lr:l=l.r:2=0 — 1+5 (16)
__ (1 - Y.) + S
Elel=o.22=1 - lit-S (17)

Figure 9 illustates the anticipated behavior of E as (Dr: changes. The density ratio
(D, (= 0.923) on Figure 9 corresponds to a feed steam having a density equal to
HDPE (i.e., D]: = p°1). Likewise, (D2 (= 0.839) corresponds to pp = p°2. The
hypothetical example shown indicates that the maximum efficiency (E a 0.9)
occurs for (PM < (D; < (1),. The value of (1);.- for which E1 = 0.5 is defined as
4,0)”. (Dmso is defined as the density ratio for which E; (@350) = 0.5. Figure 9
also illustates the limiting values of E for e; = 0 (Eq. (16)) and for a; = 0 (Eq.

(17)).

Eq. (13) shows that E depends on Ylp, 8, an, and 82. The overflow and underflow
purity coefficients, a; and 32, are stongly influenced by (Dr: and Rep. For low
plastic loadings (ylp and 3’21: less than 0.05), e. and a; are expected to be
independent of Y”. The split ratio, S, for a specific hydrocyclone operating with
an air core is often a weak function of Rep; therefore, the efiect of Q; on E is
primarily through a; and 82, not S. This observation partially motivates the use of
a characteristic density related to the intrinsic performance of the hydrocyclone to

scale the feed density. For instance, a cut-density characteristic of a; could be

37

9:29.896»: .5552 :3..— o... .5.— mo..=m.32 355.83.— ..e..a..a._om & 9...»...—
..
3—6 9

r .. .5\1/......

 

m...—

 

 

 

 

 

 

mu.
m
m:
magi
m: A
9.1;-..

 

 

 

 

 

 

 

 

38
defined as follows: al(¢(”50) = 0.5. For small split ratios, this cut-density would

be comparable to the (Dmso associated with E1 inasmuch as E =-. a; for S << 1.
Using a cut-density related to a], rather than to E1, would provide an intrinsic
measure of separation performance independent of the split ratio (Svarovsky,
1984). In this thesis, however, the cut-density associated with E1 will be used to
scale the feed density. Therefore, one of the objectives of this study is to test the
following similarity hypothesis
E. (9..Re.)—-> E. (p./p§:3 . (18)

The validity of Eq. (18) would clearly be an important simplification for
engineering design and process development. In this study, experimental recovery
data for HDPE were correlated with the dimensionless group pp/pso (pm I pmso as

defined previously) by using the following empirical equation:
p 15
E1 =1- exp[ 40') 112)] (19)

The parameters on and b were determined by using a least squares fit of the above

equations to the data (13,, pp).

The steam purity coeflicient Mm (see Eq. (5)) is related to E, and E; by the
following equation

M = MIFEI
ru MrrEr +(l-Merl-Ez)

 

(20)

39
where M 1;: represents the steam purity coefficients for HDPE in the feed. For the

special case of M1 p = 0.5, the above equation reduces to

E.

M'” = E, +(1-E,)' (21)

 

At some value of pp/pso, B, will be equal to 5;. It follows from Eq. (21) that at this
point M w=E1=Ez. Therefore, the experimental data presented hereinafter will be
interpreted in terms of E1(pp/pso) and M1u(pF/pso) for a 50:50 mixture of HDPE
and PP in the feed steam (i.e., Mlp/Mzfil). The cross-over point (i.e., E;=M.u)

gives the value of pp/pso for which E=E1=E2=M1u.

2.6 Scope of Study
This study was designed to determine the feasibility of separating HDPE and PP
using light medium technology, and to determine which design and operating
parameters are important to the separation. The following factors were
considered:
hydrocyclone design and operating conditions

0 two cone configtuations (20° and 10°)

0 feed ratios between 48 lpm and 81 lpm

a split ratios between 0.1 and 2.0
light medium designs and feed densities

- two microbubble products (3M: K20 and K46)

40
a feed suspension densities from 1000 kg/m3 down to 700 kg/m3

HDPE and PP characteristics
0 particle sieve size from 2-3 mm
0 feed concentations less than 0.5 wt.%

0 separation and grinding experiments conducted separately

CHAPTER 3

EXPERIMENTAL RESULTS

3.1 Hydrodynamics

Hydrocyclone Flow Rates

The hydrocyclones were allowed to directly discharge to the atmosphere and,
consequently, operated with an air core (see Figure 1). The air core is caused by
the formation of a low pressure region in the center of the hydrocyclone created by
the swirling motion of the fluid. The formation of the air core could have been
avoided by placing backpressure on the overflow and underflow outlets with
valves, but it was decided to allow for the free discharge of fluid back into the
tank. This meant that the hydrocyclone was operated with low inlet pressures and
pressure drops. Figures 10 and 11 show the flow rate/pressure drop curves for the
20° and 10° hydrocyclones. The curves show that the flow rates are related to
each other by the equation: I AP I = KQ". For the 20° cone, the values of K range
from 8x10" to 1x104 bar/lpm“, and it varies from 2.0 to 2.1. The K and 11
parameters for the 10° cone varied from 1x10" to 2x10'4 bar/lpm“, and 1.8 to 2.5,
respectively The values for 11 agree with typical values of 2 to 2.4 from the

literature (see p. 91 Svarovsky, 1984).

41

42

 

 

 

10

“ Pp' Po 3 KQ”

‘" Du = 100mm

_. By K n
m bar/lpm”

‘" 10 8‘10" 2.1

,_ 16 8‘10" 2.1
22 1910* 2.0
Krebs 1510‘ 2.1

“ 0., is notapeeified

 

   

 

 

a 10 mm UF Orifice
-'K— 16 mm UF Orifice

 

 

 

 

 

 

 

1; ' ° " 22 mm UF Orifice
e, Krebs Engineers (1992) o
G- 3 I
pp = 1000 kym (
E 1 -. .1 .
2 _ I
g -.
9.. .
0.1 'r tiiiiiii : :;;;;H
10 100 1000
Feed Capacity (lpm)

Figure 10: Pressure Drop vs Flow Rate
for the 20° Hydrocyclone

43

 

 

 

10
L Pr Pa = KQ"
‘ Du= 10011:!)
By K n
" m int/1pm"

-- 10 22104 1.9
16 mo‘ 2.0
“ 22 mo" 2.5
Krebs mo“ 1.8
.. Du knot-wit?“

 

 

  

 

 

D 10 mm UF Orifice
-K - 16 mm UF Orifice
" 0 ' 22 mm UF Orifice

 

 

 

 

 

 

 

A
= Krebs Engineers (1992)
e 3 o
E ' .
G 1 ~- '°
2
= -
Q? 2.
m .
J.-
a
0,1' i :::%:::§ % :i:::::
10 100

Feed Capacity (lpm)

Figure 11: Pressure Drop vs Flow Rate
for the 10° Hydrocyclone

1000

44
Hydrocyclone Split Ratios

The split ratio for this study is defined by

 

S _ 9£ _ Volumetric Flow Rate UF‘ (22)
Q0 Volumetric Flow Rate OF

This operating parameter is greatly affected by the relative size of the underflow
diameter to the overflow diameter, Du/Do. Figures 12 and 13 show how the split
ratio changes as a function of underflow diameter and flow rate for the 20° and
10° hydrocyclones. The 22 mm underflow diameter has a much higher split ratio
than either the 16 mm or the 10 mm underflow fittings. To a lesser extent, the
flow rate also influences the Split ratio. The higher flow rates have slightly lower

split ratios.

3.2 Stability of Microbubble Suspension

Migration of Microbubbles

The migration of microbubbles has an important impact on the separation of PP
and HDPE. Although microbubbles have the lowest density of any constituent in
the LMH, their drift velocity is small because of their small size (see Section 2.1).
Eq. (1) shows that the drift velocity increases as the acceleration <u9>2/r increases.
Therefore, the migration of the microbubbles toward the core of the vortex
increases as the pressure drop increases (i.e. as the flow rate increases). Figure 14
shows that microbubble migration makes the underflow suspension density larger

than the feed density. Note that the underflow density initially rises quickly

45

9:29.39.ch can 2: .5.—

83— 32,.— .25 .3 552::— a 3 83% 5am "mu 9...”:—

 

 

 

 

 

 

Ea ~ .83— BoE zoom

oi o: 2: oo 3 2. cc an ac
.. . . . . . . . . .o..-rn..-ww..-wn.u-wM-..u...........!.:....r.r. .3

41
X ................ X. ................ X ........................ X i md

855 a: 5&9 nu... ..

a

85555282....

855 a: as all m. m.—

o H

O

o :3

.?
nd

 

 

00/00 ‘oncu ands

46

 

9:29.395»: 03 a... .8.—
83— Barn :3..— .S .8225..— a ma 8:5— 3:5 ”2 «Sur—

 

 

 

 

 

E: ~ 6.3— 3:3 18..—
ov. on. on. c__ oo— c& cm as ow on cc
9 w- .“nunanunnnw.n.nuruu.uwv.nu.r..“0wrvwn. o6
OIIIIIATIIIIIII I w
A? IIIIIIIIId u
Had
m to
x ............... a .......... i m. 2
....... .........x a
m 3
.5382 «a m c.—
855 “5 552 not w
flu;
8555552..t: n
HQ;
855 “5 .5: 3+ n”
m 3
o u .
0 Ha—
. 3

 

 

 

oo/no forum was

9:: 2:39...— .E .5325..— a 3 3.239 Becca—=5 3.— 9...»...—

.:3 595 9.38...— 23.9»0

 

47

 

 

    
 

 

 

 

 

.1 N._ _ ad ed v6 Nd o
i n n . a . q u . n _ . n . m3
. .
_ .2
. _
2.20622: ~38 " n .- 25
338-5232 GNU— . -
«Eh,— § I E95 . .
use. so a ...... u u .. m3 0
u
call . n 4* .. W
8» I m .. . . m.
8s a e .. u m .. 8a m
nEEx ...—Q . ? a m
. a... . 3a m.
u ..u ..n
. . _ l.
_ \.\\ . m
" \\. u .7 8% m2
\.
a . \\\. _
"II—II‘IIIIIIp‘e‘ — 1w
rrrrrrrrrr . . . I . . l. .4 _
a lllllllllllll . _ -. mom
... .. . Allie...» Illv.
. 3.532.. he once: _

 

48
with I AP I and then tapers off as the density of water is approached. For the

conditions shown, the inlet velocity changes from 2.7 m/s forIAP I = 0.4 bar, to
5.2 m/s forIAP I = 1.38 bar. Also, for the feed conditions shown, it should be
noted that the underflow density is much higher than either HDPE or PP even for

feed suspension densities approaching 830 kg/m3.

The hydrocyclone cannot be operated at pressure drops lower than 0.3 bar because
of flow instabilities. However, because of adverse microbubble separation for
IAPI> 0.7 bar, the separation experiments were conducted for 0.3 <IAP I < 0.7
bar. The vertical dashed lines in Figure 14 show the range of operating pressures
used in the experiments described later in this chapter. Clearly, the 20°-22
hydrocyclone does not favor the development of an underflow suspension density
which would support the separation of HDPE and PP for feed densities larger than

830 kg/m’.

Microbubble Break- Up

While conducting the performance studies, it was noted that the microbubbles
were breaking and, consequently, changing the density of the underflow because
broken glass would preferentially report to the underflow. The inability of the
microbubbles to withstand the stresses within the flow circuit would clearly limit

the practical utility of the LMH. To counter the break-up, the system was

 

49
periodically cleaned, but some broken material was always present due to the

continuous recycle arrangement of the flow circuit. This section attempts to

quantify the amount of microbubble breakage in the LMH.

The break-up of microbubbles was studied using a centrifugal pump and a low
shear progressive cavity pump. The progressive cavity pump was used to
continuously recycle the suspension through a flow loop consisting of a 17 ft
section of 1” diameter pipe. The centrifugal pump was part of the light medium

flow circuit (see Figure 8).

Figure 15 shows the break-up of the K20 microbubbles in the two flow circuits.
Initially, there is approximately 6% microbubbles broken in a fresh batch. The
manufacturer’s number for the amount broken is 4%. The amount of
microbubbles broken in the centrifugal pump circuit rises very quickly and then
tapers 03‘. About 40 wt.% of the microbubbles were broken after four hours of
continuous operation. This is in contrast to the break-up in the low shear pump
circuit which showed a steady increase in the amount of broken microbubbles.
The amount broken, however, is well below the amount of breakage experienced
in the centrifugal pump circuit. It is not know if the amount of broken material in

the two systems will reach the same asymptotic value.

50

RN

_aw=.=.==oo E .— 9.1.3.5 03:39.82 he gut-2.590 "mu 9.53%

3:5.— .523 023959:— ...—a

325.:— 68:.

 

 

 

 

 

com c2 2: cm
59 co n .5
3.3.8.2: as.
n n
n
n
a .
n 9:?— Soam 3.5 £ maze—520:2 L
4 -
4
4
4
4 a8...— Ewétaoo E 8.33322 -.
4

 

o—

om

cm

3.

an

ce

mama samnqwrw %'1M

51
Figure 16 shows the break-up of microbubbles in the progressive cavity flow

circuit for a longer period of time. On this chart, the y-axis is the volume percent
of broken material instead of the weight percent of broken microbubbles as was
used on the previous graph. This procedure was chosen because it provided a
quick measure to determine any trends in the break-up process. The second
experiment was performed to determine if the breakage would reach a plateau, and
to ascertain the effect of restarting the flow system. As shown, the amount of
microbubbles broken tapered off around four hours into the study. Once this
plateau was achieved, the system was shut down and the microbubbles were
allowed to settle out of suspension overnight. The loop was then restarted and
samples collected. The amount of broken microbubbles began to increase again
and did not approach a new plateau druing this four hour study. Also, the figure
shows that increasing the flow rate did not influence the amount of microbubbles
broken in the system. This would have been noted by an increase in the rate of

microbubble break-up.

3.3 Separation Performance

As stated in Chapter 1, the objective of this study was to determine the separation
performance of the LMH. This section describes the efi‘ect of the design and
operating variables on the performance of the light medium hydrocyclone. Figure

17 is a representative sample of the type of separation obtained.

52

E5..— hgau 0229.»?5
a E 25,—. :33 abs—3.;— oSaaaeaofiz "3 Bang
3.552 .oEE. .

cow :3 see con ooN cc.

 

» h
a d

New Day

 

 

0 8.33202 as.

 

-o—

 

-N—

mun-Ia samnqwrw mm

53

 

Operating Conditions
20°-22 Hydrocyclone
| AP | = 0.7 bar
Q: = 81 lpm
pF = 300 kg/m3

Figure 17: Photograph of Separation of PP and HDPE in a
Light Medium Hydrcyclone

54
The initial feed composition was approximately 60% HDPE (blue) and 40% PP

(orange). The figure shows that a relatively clean stream of HDPE (>90%) may be
obtained using a light medium hydrocyclone. Note that the feed density is lower

than 830 kg/m3, as recommended by the results shown in Figure 14.

Eject of Split Ratio

For this study, the inlet pressure was fixed at 1.4 bar, and the K20 microbubbles
were used to lower the density of the continuous phase. The 20°-22, 20°-16, and
20°-10 hydrocyclones were used for these runs. The split ratios for these
configiuations were 2.0, 0.6, and 0.1, respectively and the feed rates were 58, 49,

and 48 lpm.

Figure 18 shows the effect of increasing the split ratio on E1 and Mm. The data

fall on the same similarity curves when correlated with the reduced density, p/pso.

The values of p50 were determined to be 830, 750, and 660 kg/m3 for S equal to
2.0, 0.6, and 0.1, respectively. The long tails on the right side of the graph for the
two curves are associated with the 20°-10 hydrocyclone configuration with a very
low cut-density of 660 kg/m’. Also note that at high yields the inlet densities are

much lower than the densities of either HDPE or PP.

3293963— 52—52
E»: a .3 8553.29.— 2: .5 e23— :Em he 89...:— 2; £— 9...»...—

 

 

55

        

 

 

 

 

 

 

 

 

 

 

 

 

Sufi—Q

m._ .1 n._ N._ ... _ ad «6 «no
«.rflw. ..nflnn “...“nnnawnnnqro

5:35am 0.332 e9. ..
v8". 58.35822 $25 2...-.. c—
.- om
.. cm
-. cv
-. on
2 08 _.c we v.9 263 o\ co

2 0mm 06 av v.9 2-08 \ ..
2 can QN an vd 3-0g \ -- on

A. as} ...a m as .o .2 2..-... 25.98 Beam ..
b -. cm

tam XI
a

o o r \ .. co

I. d. ..

|II a... ‘ O O
IIIIIiuIIOiIIIIIQl o .

-8—

56
Efi'ect of Pressure Drop

The feed velocity is an important parameter in that the tangential component of the
velocity, <u9>, is directly related to this variable. With an inlet of constant cross-
sectional area, the feed velocity is directly proportional to flow rate. The 20°-22
hydrocyclone was operated at IAP I of 0.4 and 0.7 bar with a corresponding inlet.
velocity of 2.7 m/s to 3.7 m/s, respectively. As IAPI increases from 0.4 to 0.7
bar, the Split ratio decreases from 2.0 to 1.7. The effect of this change in pressure
drop on the recovery coefficient E. and the HDPE purity, Mm, is shown in Figure
19. Both the yield and purity curves seem to correlate with pp/pso. The value of
p... for the higher inlet velocity, 780 kg/m3, is well below the value of 830 leg/m3

for the smaller inlet velocity.

Microbubble Distribution

As concluded from Figure 13, the segregation of microbubbles in the LMH is
significant. The migration is caused by the large density difference between the
K20 microbubbles and water (pc - p3 = 790 kg/m3). To reduce this migration, a
smaller, more dense distribution (K46) was used (Dc - p3 = 560 kg/m’). The K46
distribution has approximately 1/4 the Stokes’ drift velocity of the K20 distribution
for a given acceleration (see Eq.(1)). For this comparison, the 20°-22

hydrocyclone was operated at a pressure drop of 0.4 bar. These parameters were

57

T

0:29.396»: ...—:32

Ewi a .... San—....abm 2: .5 83— .33..— ..c 89...:— 23. .3 9...»:—

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

gain.
n. N. ... _ a... a...
.hhulllfll. . ... ... ......Tlo
c 8:35.... 2332...: .5. ..
e8. ... 83885.... .\....> a...
a .. c.
.- cm
L. OM
.. 9.
.. om
\. .. e
2 Sb .... ... ...c . «9.8
... on. 3 an to ~98 ..e .- E
e .535... m a... .o .2. 2...... 2.2.5 3.2%
_. as... .\. .. 8
.7
.\ .. 8
his
III.‘ ...
III II- .II- III III ..II C I L

 

58
chosen because the combination of high split ratios and low inlet pressures

(velocities) produced the highest yield for a given inlet density.

Figure 20 shows that the recovery .of HDPE in the underflow stream (yield) as
well as the % purity (M m x 100%) for two different microbubble distributions can
be correlated with pp/pso. Again, the difference between the performance of the
two distributions is p50. The value of p50 for the K46 (smaller) distribution is 870
kg/m3 compared to 830 kg/m3 for the K20 (larger) distribution. As previously
noted, the feed density of the suspension at high yields are well below the

densities of HDPE and PP.

Cone Angle .

Figure 21 shows the performance of the LMH with hydrocyclones of different
angles. The 20°-22 and 10°-22 hydrocyclones were used for this comparison. The
10°-22 hydrocyclone was operated at a pressure drop of 0.4 bar while the 20°-22
hydrocyclone was run at 0.5 bar. The difi‘erent operating pressures ensured that
the two hydrocyclones had the same volumetric flow rate and, consequently, the
same feed velocity and split ratio. The K46 microbubble distribution was used in
these trials to vary the inlet density from the density of pure water (z 1000 kg/m’)

down to 750 kg/m’. The curves show approximately the same performance when

plotted as a function of the reduced density pp/pso. The 10°-22 hydrocyclone has a

S9

vac—232?»: 53:92 E»: a .3 355.829.—

e... .... 5.2.5.2.: ohm 2.33.5.2 ... ea... 2.... ..." 9...»...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.3....
m. 2.. N. ... ... 3.. a... 3... a...
Ur. . .. ......» ¢ ..............c
o ..8. ... 8.23.5.2 is 2....
8203...... 2...... -- o.
-- on
.- cm
-- S.
-- on
o\. :8
mm 2. ed mm to .3. -- 2.
t on. 3 ..m to .5. ..e
.. .53....e m ...... .o .... 2..-... 8.32.222 .2. m K .- 8
..u 5.5.3. -.....
III III III. III- .III \ ‘I. III. 0 ;
OI I o o - 8—

 

6O

9:29.325»: EBB—Z E»:
a .3 355.83.— 2: .5 o_w=< 250 .... tot”.— 2; "—N 9...»..—

 

      

 

 

 

 

 

 

 

 

 

 

 

 

. . 2...... .
mm . N . m... ... 8.. . 3... a... 3o a o
T. « . . . . . . . . . . . .j . . . fl . J . . w . . . w c
8.25.3.0 9.33.5.2 o3. w
8.... =.8..........E.2.\....> 8...“, o.
.. on
m, a
$5.3 .
m- e.
o .
O \ .
\ O “a On
\ w
. . . \ “I 8
ma 2... o .. S 2. ~92 . o \ .
cm 8.. .... S m... 3-08 .... ... 2.
a .EBxéq m E... .0 .3.— .<._Lm 0:296 .358 \ \d/IL.
\ .
0 .MI II. .\ \ ”u 8
II I I I r I III I I . . 1
o 0 H 8—

 

61
higher pm (880 kg/m3) than the 20°-22 hydrocyclone (860 kg/m3). The difference

for these two cone angles is smaller than the other comparisons.

Eject of HDPE Concentration

The concentration of HDPE in the feed (i.e., ylp) may greatly afl‘ect the economic
feasibility of the LMH. Experiments were conducted to determine the efl‘ect of y":
on E for the 20°-22 hydrocyclone using the K20 microbubbles with a feed density
of 780 kg/m3. Figure 22 shows that the yield (i.e., E1) is independent of HDPE
concentration over the range studied (loo-5000 ppm). The data at a pressure drop
of 0.4 bar (Q? = 58 lpm) agree well with the same operating point shown on Figure
16. For HDPE feed concentrations above 1000 ppm, the feed pressure was
increased to 0.5 bar (QF = 64 lpm) to avoid clogging. The experiment was
terminated at an HDPE concentration of 5000 ppm because the centrifugal pump

could no longer handle the solids loading.

 

gush—.855 coo..— mmnl ..e 5.22:..— a ma
9:29.395»: E232 Em: a ... man: .3 22> 2.8.5.— .NN tawE

535m too...— 05 :_ BEG: e\o...$
36 Yo mad nd 36 N6 26 _.o 36 o

_ _ b _ p _ h — p
u a q q u _ u _ d

8.3.88.2 .5.

..8... ... 8.3.88.2 .\. ...> a .r
.23.. ...... a 2.2.8 8....
0:00 NNIOON i

62

 

 

 

an

3.

cm

2.

cm

8—

EIJGH 1° PPM %

CHAPTER 4

DISCUSSION OF RESULTS

4.1 Hydrodynamics

Pressure Drop/Flow Rates

A small incremental increase in the cost of processing HDPE could limit the
practical utility of the LMH because of the low profit margins associated with the
recycled material. Consequently, the economics of the process is very important.
One variable which may be used to estimate operating costs is the pressure drop

over the hydrocyclone. Energy consumption in the LMH can be estimated by

 

calculating the power (0’ = Q- AP I ) required to operate the hydrocyclone. For the
hydrocyclone configurations and flow rates studied, the energy consumed per unit
time (i.e. power) required to operate the LMH was estimated to be 300 W for Q].- =
100 lpm and I AP I = 1.4 bar. This corresponds to 0.3 kW-h for one hour of
continuous operation. Thus, a lower bound on the operating cost to process 200

kg/h of HDPE is 2¢/h, based on an energy cost of 7.6¢/kW-h and 4 wt%iplastics

loadings.

63

64
Split Ratio

Figures 12 and 13 show the strong dependence of the split ratio (Qu/Qo) on the
underflow diameter. This strong dependence is actually a function of the ratio of
the underflow diameter to that of the overflow which was fixed for the Krebs
hydrocyclone. Svarovsky (see p. 100) states that the split ratio is proportional to
(Du/Do)”'. Using the data fiom Figures 12 and 13, the exponent on Du/Do was

determined to be 3.75 for the 10° hydrocyclone and 3.26 for the 20° hydrocyclone.

It was also noted in Section 3.1 that the split ratio was a weak function of IAP I.

This dependence is given by Svarovsky as S ac IAP I '0" (see p. 100). Values of
this exponent were determined to be -0.20 and -0.32 for the 10° and 20°
hydrocyclones, respectively. These values correlate well with the literature value

of -0.24.

Pressure Loss Coeflicient
Bradley states (see p. 90) that the major source of pressure loss in a hydrocyclone
is from the centrifugal head, and that other sources of pressure loss are negligible.

The pressure loss coefiicient is defined as:

M (22)

C= ,
" my
2

 

65
where v represents the bulk average velocity of the feed and r is the density of

water (no microbubbles).

Figures 23 and 24 show the pressure loss coefficient as a function of Reynolds
number. It appears that Cp is independent of the Reynolds number, indicating that
the viscous losses are small compared to the losses associated with the centrifugal

head. The values for Cp range fi'om 10 to 17 for the 20° hydrocyclone, and are

between 7 and 15 for the 10° hydrocyclone.

Knowledge of the pressure loss coefiicient allows an estimation of the values for at

and N from the following equation (see p.90 Bradley, 1965):

finer-1]

For Dn/Du = 4.5, a value of C9 = 19 results when N and a are set equal to one.
This should be an upper bound on the pressure loss coefficient. The curves on
Figures 23 and 24 show this to be valid. Also, by solving Eq. (23) for a, and
requiring that a S 1 places an upper bound on the value of N. This corresponding

value of N was determined to be approximately 0.75 for C, = 11, a = l, and

DH/Du = 4.5.

66

9:29.395»: .03 2: ...... con—E. Z 3.2.23—
.8 552:...— a ma «5.3.:er mm...— o..=mmo..._ "mu ...—aura

 

 

 

 

 

 

.oafiaz 3.323—
ooooop oooom oooom .oooon ooooo oooom oooov cocoa
. w . . u u . . w . o. 0 u . w . . x. . u a o +1 . . .\¢. 0 . . d .t m
:m
o O lATl :o—
. ..z
..Np
-1 m—
x ............................... .. ...................... x
X Y
85.5 ...: ...... ... in! \
11111 m—
8....o.5......2:.x: \\t\\\\
D \“
\t\\\ :o.
beacon—DEENNIT n...\\\\
D

 

 

 

 

 

 

h—

;uapgpog 5501 arnssard

67

9:29.396»: ...: 2: ..8 ..2...=. Z 3.2—mom
.... 252:5 a 2. .565er m8..— o+=mmu§ 3." «...—u...—

 

 

 

 

 

 

 

._2_E=Z 922.23— .
coco. . 882 88.. 88¢ 822 88m 88m 88¢ 88m
. . . b . v . L. .. .1 . m
0
.... Q
..
: 2.
o lo ..
o ., m
a x i
........ it a
........... X h o—
x . ......
...... t ——
.... X
.5} 8... u... .. 2
neuron—DEESIWI .
mttltltmr u :n—
82.535572: ii...:.:...:..
tttttttt :v.
85.0.55... -|o| a ..
....

 

 

rucpgpog ssoq arnssua

68
4.2 Stability of Microbubble Suspension

Migration of Microbubbles

A density gradient of microbubbles is established in the hydrocyclone because of
microbubble migration relative to the continuous phase. Depending on the local
density of the PP/HDPE separation zone, the HDPE and/or PP may appear either
heavy or light and, thereby, report to either the underflow or the overflow. Figure
25 shows how the yield and underflow density change as a function of inlet
density. The yield does not increase significantly until the underflow density
approaches the density of HDPE. When the underflow density equals 960 kg/m3,
the HDPE becomes mutually buoyant and has a 50-50 chance of reporting to either
the underflow or overflow. As the underflow density decreases further, the yield
of HDPE increases significantly, however, the amount of PP reporting to the
underflow stream begins to increase as pu approaches 910 kg/m3. The purity of
the underflow stream drops sharply as pu decreases below the density of PP.
Clearly, high yields of HDPE at relatively high purities can be achieved in the
LMH for underflow densities which satisfy the following inequality:

PF < PP? < PU < pHDPE-

This inequality explains why such low feed densities are needed for the Krebs
hydrocyclone. It is noteworthy that for a dense medium operation, the separation
occurs at a higher density than the feed density (see p.171 Bradley, 1965).

Apparently this occurs because the migration of the medium is favorable to the

69

ERIC]! 10 PPM °/o

83.2.9: 3:.— Eobta 2. 5.2.2— .5....2253 .25 22> ”mu 9...“...—

 

 

 

 

 

 

 

.....u.. 5.2.5 8....
on. 8.. or... 85
T . . .1 J. . . L. P. w + n . . . L. can
8:35.... 2353.2 So. H
8.35.5.2 23> 3.... ..
....uma.....nuee.6...:.8.. ..
8228...... 3.2 -- on»
m .. n
x -. 8.. m.
. ..
x n
V .. a
\ .. 8. a
x H ......
8 . . .W
. .52.. 8. u m......... o \ \ .
on L $me ceaua \ . 8a m
2...; ..\..cm 3. .. ...
2oz ..
ON 11 \ \ ...
.... , .-....
CO 4.. \U\ \ q l.
\ \ \ .H
8. ...... \ u. P So.

 

 

 

7O
separation. In the LMH, however, the migration of the medium is apparently

unfavorable to the separation inasmuch as the underflow density is always higher
than the feed density and the overflow density is always lower than the feed

density.

The fundamental distinction between a dense medium and a light medium cyclone
is the relative motion of the medium in the swirling flow field. Glass
microbubbles tend to migrate towards the air core in a LMH whereas magnetite
particles tend to migrate towards the conical wall in a DMI-I. Perhaps an improved
LMH design could be developed based on the unique features of the light medium
with the result that
9?? < pHDPE < PF < PC-
The overflow density, p0, can be related to the feed density, underflow density,
and split ratio by using a steady-state material balance (see Eq.(2)):
' p.=(p.-p.).s+p. (24)

For p. = 830 kg/m’, pa = 970 kg/m’, and s = 2.0, Eq. (24) implies that p() = 550
kg/m3. This density is low and poses a large barrier for the migration of PP toward
the vortex core. An overflow density of 550 lth3 corresponds to a microbubble
concentration of 57 vol.%. Thus, approximately 87% of the microbubbles in the

feed are reporting to the overflow. This is a significant separation of the

71
microbubbles, and implies that only a small fiaction of the microbubbles (z 13%)

may be participating in the separation of HDPE and PP. Clearly, further control
of the migration of rrricrobubbles within the LMH could provide improvements in

performance.

Microbubble Break- Up

Figure 15 shows that the high shear centrifugal pump environment breaks the
microbubbles to a larger extent than the low shear progressive cavity pump. The
sharp rise and tapering off of the cumulative breakage curve suggests that the
centrifugal pump quickly breaks the weak microbubbles. The stronger
microbubbles are more resistant to the high shear environment and take a longer
exposure time to break in the shear pump. In contrast, the low shear environment
of the progressive cavity pump gradually breaks the weak microbubbles resulting

in the lower break-up rate shown in Figure 15.

The short time breakage data summarized by Figure 15 does not show a plateau
for either pump. However, the long time experiment in the progressive cavity
pump (see Figure 16) seems to suggest that a residual amount of microbubbles are
resistant to breakage. This suggests that the nricrobubbles could be preprocessed
to remove the weak ones and that the remaining strong microbubbles could be

used in a process indefinitely. Therefore, flesh microbubbles would have to be

72
added to make up only for losses from the nricrobubble recovery system, not from

breakage. This observation may influence the practical application of the

proposed LMH for separating HDPE and PP.

Another aspect of the strength of the microbubbles is their use for short-time
processing. The behavior of the microbubbles to withstand periods in which they
are not suspended and allowed to dry is important. When the microbubbles float
to the water/air interface, they tend to dry out. It is thought that this drying
process may cause stress fractures on the microbubble surface resulting in a
weaker product. These weakened microbubbles are then more susceptible to
breaking up in the pump. The linear breakage rate portrayed by the data of Figure
16 suggests this efl'ect. The microbubbles on the second day showed a higher
amount of breakage, and consequently, may imply that the drying process
damaged the microbubbles. Unlike the earlier portion of the curve, the breakage
curve does not taper ofl' after four hours of operation. This phenomena suggests

that the microbubbles should remain wet whenever the flow is interrupted.

4.3 Separation Performance
The separation performance curves presented in Chapter 3 show that the yield and

purity could be correlated with the dimensionless inlet density, pp/pso. The cut-

density p50 is the feed density for which the recovery of HDPE in the underflow is

73
50% (i.e., 51(1) = 0.50). Eq.(l9) was found to be a useful empirical representation

of E1(pF/p50) with b z 20 for all the variations in design and operating conditions

examined.

It is noteworthy that at p50, pu = pm)”; for the specific cyclone design and
operating conditions studied. A desirable goal, albeit not attainable with the

current design, is for p50 > puma.

Efl'ect of Split Ratio on p50

The split ratio afi‘ects the amount of medium needed for a given separation. For
dense medium hydrocyclones, less material is needed when the underflow to
overflow volumetric ratio is high. This same trend was observed in the light
medium hydrocyclone. The values of pm for S=0.1 and 0.6 were 660 and 750
kglm’, respectively. This means that a large amount of nricrobubbles are needed
to achieve the proper underflow density for separation. Apparently, when the
microbubbles enter the system and begin to separate, a large portion are caught in
the strong upward flow and never participate in the separation. To counter this
effect, more microbubbles are needed to achieve the proper underflow
concentration for separation. Conversely, at higher split ratios the migration of
microbubbles to the overflow decreases. Figure 26 shows graphically how p50

changes for the 20° hydrocyclone with three difl‘erent underflow fittings. The

74

m.N

3...: ...am .... 252:...— a 2. S... ”on 9...»...—

..o...o 5...... .....m

m.—

P

9 m6

1+

 

 

 

 

N
8.32.222 ..9.
.... .... u _..< _
...N u m

88.2.2.2: «9.8

b n n
‘

.... n m
25.9822: c . ..ou

   

o... n m
2.2.2.56... 2...."
O

   

 

 

coo

one

005

can

con

Ono

cur/Br] “rid

7S
figure shows that p50 increases significantly for split ratios below one, but then

increases less rapidly for split ratios above one. This implies that there may be a

practical upper limit on the ratio of Du/Do.

Eflect of Feed Flow Rate on p50

p50 increases with the feed flow rate, Qp, because the centrifugal acceleration,
<ue>2/r, is proportional to Q13. If the tangential velocity is proportional to the feed
velocity, then the centrifugal acceleration for the higher flow rate is clearly larger
than it is for the lower feed flow rate at any given radius. This increases the
migration of the rrricrobubbles to the core of the hydrocyclone, resulting in a lower
concentration in the underflow. Again, more microbubbles are needed to attain the
proper underflow density. As stated earlier, the split ratio also slightly decreased
from 2.0 to 1.7 at the higher flow rate. Using Figure 26, the differences of pso’s
due to the change in split ratio is 15 kg/m3 (830 down to 815 kg/m3). The
experimental values for the two flow rates are 830 kg/m3 at 58 lpm and 780 kg/m3
at 81 lpm. This is a difference of 50 kg/m3 and shows that the increase in
centrifugal acceleration is more dominant than the decrease in split ratio for these
conditions. Consequently, feed flow rate (velocity) is an important parameter

because of its influence on the centrifugal acceleration.

76
Efect of Microbubble Distribution on p50

The smaller size microbubble distribution (K46) provided better performance than
the K20 distribution inasmuch as the yield of HDPE was higher for a given feed
density at the same flow rates and split ratios. The values of p50 were determined
to be 830 kg/m3 for the K20 distribution and 870 kg/m3 for the K46 distribution. It
appears that the K46 microbubbles provide a more stable medium. This was
anticipated because of the lower drift velocity for the K46 distribution. Which is
approximately one-fourth of the drift velocity for the K20 microbubble distribution

(see Section 2.1).

Eflect of C one Angle on p50

The cone angle seems to have the smallest effect on the yield of HDPE in the
rmderflow of any of the parameters discussed The 10°-22 hydrocyclone has a
higher pg), 880 kg/m’, than the 20°-22 hydrocyclone, 860 kg/m3, at the same inlet
flow rate (67 lpm) and split ratio (1.8). This result was anticipated from the
literature on dense medium separations and is attributed to a more active toroidal
recirculation zone (TRCZ) in the 20°-22 hydrocyclone (Moder, 1952). In the 20°
configuration, the microbubbles which are in the lower portion of the
hydrocyclone can be caught in the TRCZ which would increase the microbubbles

chances of reporting to the overflow. This is less likely to occur in the 10°

77
hydrocyclone because of the less active TRCZ. The small difference in the pso’s

for the two hydrocyclones is due the fact that the cone angles are not that great.

Efi’ect of HDPE Concentration

In dense medium separations, the amount of solids does not affect the performance
of the hydrocyclone up to concentrations of 4 wt.% (Moder, 1952). This also
appears to be true in the light medium hydrocyclone, but the concentrations
studied are well below the DMH concentrations. As stated in Chapter 2, the
centrifugal pump would not function with plastic concentrations above 0.5 wt.%.

Even though the concentrations are low, the results are still encouraging.

The fact that the concentration of HDPE did not have an effect on the recovery of
HDPE in the underflow is very important to the application of this process. If this
fact holds for high plastic loadings then this would greatly reduce the amount of
medium necessary for separation. However, the efl'ect of plastic concentration
needs to be further researched to determine how the purity of the underflow stream

is afl'ected, and to determine at what concentration the yield begins to decrease.

CHAPTER 5

CONCLUSIONS AND ENGINEERING SIGNIFICANCE

Polypropylene and High Density Polyethylene can be separated in a hydrocyclone
using a suspension of glass microbubbles in water. However, the separation of
these two materials in a hydrocyclone is much more complicated than a float/sink
tank. The migration of the microbubbles to the core is not advantageous to the
separation, so more microbubbles are needed than the amount 'which would be
calculated for a float/sink operation. Consequently, the suspension density of the
feed stream to the hydrocyclone must be lower than either plastics for an effective
separation (high yield, high purity). This study also shows that the underflow

density, not the feed density, is the important factor in the separation of PP and
HDPE, and that the inequality, ppp < on < plum, must be satisfied for an efi‘ective

separation of PP and HDPE.

The operating variables such as split ratio (QB/Q0), flow rate (Qp), cone angle (or),
and microbubble size distribution afi'ect the yield of HDPE in the light medium.
These parameters influence the separation through concentration of microbubbles
in the underflow stream. The higher this concentration, the higher the yield (i.e.

E1). The best combination of these variables for the operation of the reverse flow

78

79
hydrocyclone is: high split ratios, low inlet velocities, small cone angle, and small

microbubbles. These conditions will provide the highest yield at the lowest inlet
density, but may not be the best in terms of productivity. Figure 27 shows that for
the 20°-22 hydrocyclone the separation efficiency, E, is a maximum for pp/p50 =

0.96.

Also, it was shown that the similarity hypothesis, E, (¢F,Re,,) -> El (pp/p2,) ,
appears to be valid, and that the efl‘ect of the hydrocyclone design and operating
parameters are expressed in Eq. (19) through dependence of the recovery
coefiicient on p50. Furthermore, it appears that the parameter b, which reflects the
sharpness-of-separation, is insensitive to the operating conditions and -

hydrocyclone geometry with an approximate value of 20.

Finally, the reverse flow hydrocyclone may not be the best design for separating
I-IDPE and PP. The 100 mm hydrocyclone which was used in these experiments
was designed for dense medium separators. Table 5 is a comparison of the light
medium and dense medium separations. It shows that the behavior of the two
processes are similar in some respects and opposite in others. The most important
difl‘erence being the amount of suspended material needed for separation. The

dense medium hydrocyclone takes advantage of the medium’s migration and,

80

:21— 2: ... 85:28.5.— =e_.a..2_om "hm «Sufi

 

 

 

SEQ
m... ... 3.. . mad . ad mad ad
flu. .- . 1 q u] n 1 q < u u o
I../. Q \\ -1 o—
o/ n \
./.. \ -- ON
1. H \\ ..
_ |\/. . \ .. on
m..\.. 0 / n \M
/ . \ (... .\. .- ....
.. H \
/ . \
o .. u x
o x n x .- 2
o H
m .. 8
u ..
\H it Oh
\ . -. 8
8...........8.2 63. x ..
Ea_wnu.0 A \\ I.”
.....vdu_..<_ xx .. .. 8
2.29.8.2: 3.2 \ x x z . .
or IIIII m antt I..l- oo—

 

 

 

81

Table 5: Comparison of Light Medium Hydrocyclone and
Dense Medium Hydrocyclones

 

      

 

 

 

 

 

 

 

 

Effect on DMH Efl‘ect on LMH
Amount of material in Less than amount Higher than amount
system calculated for float/sink calculated for float/sink
tank tank
Decrease amount of Decrease amount of
Increase split ratio medium needed for medium needed for
separation separation
Decrease amount of . Increase amount of
Increase flow rate medium needed for medium needed for
. separation separation
Decrease amount of Decrease amount of
Decrease cone angle medium needed for medium needed for
separation separation

82
consequently, uses less material than the amount calculated for a float/sink

operation. In the LMH, however, the migration of the microbubbles is not
advantageous, resulting in the use of more nricrobubbles than what is calculated in
order that the density be between the densities of PP and HDPE. It is for this
reason that the reverse flow hydrocyclone may not be suitable for separating

HDPE and PP.

Engineering Significance

The yield/purity curves (see Figures 18-21) allows for an estimation of the flows
in the process on the basis of a specific operating point. Figure 28 is a schematic
for a process to produce a clean stream of HDPE. This flow diagram was
constructed to process a feed stream from a typical reclamation facility. The flow
rate of the feed stock was quoted from Michigan Polymer Reclaim (MPR) which is
a HDPE recycling facility in Lansing, MI which uses float/sink technology to
separate heavy contaminants fi'om light thermoplastic materials. MPR produces
approximately 1000 lbs. of HDPE per hour (7.6 kg/min) containing about 1 wt.%
PP. The following hypothetical example calculation identifies an LMH process

which reclaims an HDPE product containing 0.2 wt.% PP.

In order to calculate the stream variables, it was assumed that the performance of

the LMH was independent of thermOplastic concentration (up to 4 wt.%), and that

83

 

nee—939...»: .5582 EM: a ...... 5233 Be..— .wu 9...“...—

55.9. a... u a...

55.9. .... "......
as... N «new.

5...... ...... u ......
55.8. ...n u ......

 

 

.8... u ...2
......9. 2.. u ....
...... .... u no

as... 2 u 9..

 

......E ... u 9...
55.9. 2 u on...
58.». 8... u R...
55.». .... u ......

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.... n 22
.....9. .... ... ....
...... .... a co

 

 

 

25. name:
. .. .. ....9. ... .
3222...: .73....2 3.32.222
58.9. .... u h....
5...... 3.. u ......
as... ... u ......
as... .... u a...
as... 2 .. h......
a
........ 80° . an:
:2.— a... o... r .0 too..—
82 a ....E.
8... u 2...: i... ll...
S ..
BeEo>O

84
the loss of medium for the process was 2 kg per ton of treated material as is the

case in dense medium separations (see p. 21-33, Perry’s, 1984). Also, it was

assumed that the microbubbles did not break in the flow circuit.

For these calculations, the 20°-22 hydrocyclone configuration was chosen because
it had the highest flow rate of the hydrocyclones which were studied. The flow
rate for this configuration is 81 lpm at a pressure drop of 0.7 bar. A reduced inlet
density of 0.96 was chosen because of the high yield and high purity (see Figure
19). This reduced inlet density corresponds to an actual feed density of 750 kg/m3.
At the specified flow rates of the thermoplastics, 240 lpm of feed suspension is
required. To meet the flow requirements of the facility, 3 hydrocyclones operating
in parallel are needed. The performance data from Figure 19 shows a HDPE
recovery in the underflow of 75% and a corresponding stream purity of 85%. This

coincides to approximately 87% PP recovery in the overflow stream.

Figure 28 shows the conditions of all process streams. The important factors to
note are the HDPE stream purity coeflicients, M 1x, the stream densities, and the
relative amount of PP to HDPE. The stream purity coefl'rcient is 0.99 for the inlet
stream to the hydrocyclone assembly. It is increased to 0.998 in the underflow
stream by passing it through the LMH. The overflow stream purity, however, has

been decreased to 0.964. The overflow stream may be further processed to

85
improve its purity and also to increase the yield of HDPE for the entire process.

The ratios of PP to HDPE in every stream are quite low. However, the flow
diagram shows that the grade of HDPE in the underflow is improved significantly.
Figure 28 shows an anticipated decrease in the mass ratio of PP to HDPE by a

factor of seven.

As stated earlier, the underflow density controls the recovery of HDPE in the
LMH. Therefore, because of similarity in the yield/purity curves, the yield for
difl‘erent operating parameters can be predicted for a given underflow density.
Following this logic, the underflow density was determined fi'om Figure 24 to be
950 kg/m3. Using the feed and underflow densities, along with a split ratio of 1.7,
the overflow density was calculated to be 410 kg/m’. To further utilize the
overflow in the separation process, it would be necessary to add water to increase

the overflow density.

CHAPTER 6

RECOMMENDATIONS FOR FURTHER STUDY

The following recommendations for further study with the current light medium
hydrocyclone are made:

1. Perform flow visualization experiments on the LMH. The setup of the LMH
would not allow the separation of PP and HDPE to be viewed. An
understanding of where the separation is occurring in the hydrocyclone may
provide insight into the design of new hydrocyclones for this application. This
entails more than just making a clear hydrocyclone and viewing the HDPE and
PP. The microbubbles make the suspension opaque, even in small
concentrations. Consequently, the continuous phase will have to be made
optically homogeneous by adding a soluble constituent to the water in order to

match the refractive indices of the water and the microbubbles.

2. Perform more experiments with the current light medium flow circuit to fill in
the areas which were not covered in this research. It is suggested that
experiments be conducted at higher flow rates using lower feed densities.

Also, experiment with difi'erent cone angles and microbubbles size

distributions. Since p50 is a function of the design and operating parameters, a

86

87
more detailed study of the dependence of 950 on the hydrocyclone geometry

and feed conditions is needed in order to perform any significant engineering
calculations. Also, further study may indicate if b is a function of any of the

experimental conditions.

. A more detailed study of the microbubble breakup may be helpful. This may
indicate how the microbubbles break in the system and how to manufacture
microbubbles which are more resistant to breaking. Also, from an economic
point of view, it would be important to know if there is an upper limit to the

amount of breakage.

. Try to effect the size and shape differences in HDPE and PP via grinding
protocol. Table 2 showed that there are significant temperature differences in
the thermal transitions of PP and HDPE. Also, HDPE is difficult to fracture at
cryogenic temperatures while PP is easier. Depending on the conditions
(temperature, impact rate, residence time), it may be possible to obtain
difl‘erences in size and or shape of the particles. Dreissen et a1. (1963) have
shown that shape can be a major factor in the separation of particles with
similar densities. This may lead to an autogeneous (water only) design, or at
least decrease the amount of microbubbles necessary for separating PP and

HDPE.

88
5. Experiment with completely different hydrocyclones. As stated in Chapter 5,

the reverse flow hydrocyclone may not be the best design for this separation
because the rrricrobubbles are quickly removed in the overflow stream. A
hydrocyclone with a totally different flow pattern, such as a forward flow
hydrocyclone, may decrease the migration to the overflow. Alternatively, the
migration of the microbubbles may be decreased by increasing the length of the
vortex finder (See Figure 1). This would provide a physical barrier to the
migration of the microbubbles to the vortex, thereby increasing the

concentration of microbubbles in the apex region of the hydrocyclone.

APPENDIX A

HYDRAULIC DATA

89

9...:
2.5.
0......
ovd.

Go... 2:...
3528... :82

3.6.
0......
3...:
3...:

.83 22F
3:628... :82

$6.
36.
3.2
9.6.

Go... 2......
8528... :32

....
......
2....
8..
3....

....> .88..

......
3..
......
n2
......5

..o> coo...

N. ..
S...
......
8..
3....

._o> too.—

8....
2....
a...
......
......

......
S...
......
2..
......

......
a...
2....
.2
......

$.—
2...
Q...
8N

3.6
9.6
.nd
:6

..o> woo“. ..ORD

mod
86
86
mod

._o> too”. main—D

had
had
had
and

._o> too... ...OED "5.0

wed
no...
cod
and

KEG
no...
36
wad
mad

......3

36
Med
36
bed

"<3

and
and
cm...
and

ES
3...
mod
cod
mod

...”...

02...... 3.2.... 23.5.32. .3 505.522. c

5.96
55mm
mn..~
«Nd.
€53
swoon

mmdu
v. .NN
NM.»—
cad.
€53
$.00...

......"
“......
2.2
8.2
2......
...8.

a . ...
and
3.5
36

€53
.5

3.5.
cad.
m . .N.
am.»
€53
".0

«can
of...
3......
we...
€53
......

25...... n u ..>
6:29.255: can 2.. ...... 8.5 5...: 9:83.55..."— BeE ...< 93:.

9...... ...
8.... m .,
a... ...
...... m
€53 2.5.
up: 0.530.... .2:—
ooctO 30:50ch 55 NN
...... on
as n.
t... ...
a... n
€53 an...
up: 0.59.0.5 .2:—
8....o 32.82.. ...... ...
2 .. ..N
8.. n.
8.. ...
.3... n
......s ......
a: sedan «0.:—
8....o 8.8.x... ...... ...

9O

3...:
ovd.
3...:
wvd.

A8... 2......
3528.. :83.

.36.
med.
0......
2.6.

Go... 2......
352.3. :33.

0......
ovd.
0v...—
0......

A8... 2......
3:023... :82

3..
8....
N...
8...
3.5

._o> ..oo...

8..
a...
8..
....
Es.

..o> .5»...

on...
8...
.N.
a a
.....E.

._o> .8".

.2.
a...
......
2.2
3...

6.2
8....
S...
2....
we.

3....
......
8.2
2 ..
......

em.—
va.
2...
3..

no...
.0...
3...
on...

..o> to... 50...:

cc...
bod
mod
N. d

..o> ..uo... m0»...

and
mm...
3.6
mm...

..o> .60.. 50...... "5.0

.9:
No...
.0...
on...

"5.0
3...
no...
Na...
and

"5.6

.9...
no...
we...
no...

.55

and
an...
an...
.v...

"5.3
cc...
2....
ac...
. ..c

......D

92...... .....o..... 8... 5...... .... 93.5.8.2. c

. ....n
.93..
Ewen
mo...—
€53
...—.00..—

8.3
N93
o. . . N
Nvé.
€53
...—.00.—

... 6N
3.2
an...
cmd.
€53
«too...

at...
.... .N.
a.»
v. .o

€53
...O

3.:
and.
RN.
mod
€53
".0

......
8...
8.:
.... _
.5...
no

as...» .... u =>

3.8 on
3.3 m.
8.2 c.

.n... n
€53 an...

...: 2:30... ...—...
8.....0 Backus: ...... «N
a. ... cm

and m.

an.» o.

2.... ...

Ea... ......

.5 0.38... 8....
3.....0 3050...... E... e.

an. an

mm. . n.

n... 2

ms... n
€53 an...

..5 0.38.... 3....
8.....0 Becki... ...... c.

329396.: a... o... ...... ...... 5...... 2:82....5. ~......— .N.<_o_......

APPENDIX B

MEDIUM STABILITY DATA

91

Table B.1: Average Density of Glass Microbubbles

 

 

K - 20 MICROBUBBLES K-46 MICROBUBBLES
BULK BULK
TRIAL# DENSIT ENSITY TRIAL# DENSIT ENSIT
(kg/m3) (kg/m3) (kg/m3) (kg/m’)

1 317 70 1 500 250

2 173 51 2 455 221

3 225 88 3 337 212

4 187 98 4 413 207

s 167 84 s 467 240
AVE 214 78 AVE 444 226

ST. DEV 62 18 ST. DEV 45 18

92

Table B.2: Size Distribution for the K20 Microbubbles

Determined Using a Malvem MasterSizer

1... 51.57
Size % Under Size % Under
microns microns

0.5 0.50 9.94 2.90
0.55 0.60 10.9 3.30
0.6 0.70 12 ‘ 3.70
0.66 0.80 13.2 4.20
0.73 0.90 14.4 4.70
0.8 1.00 15.9 5.40
0.88 1.10 17.4 6.30
0.96 1.10 19.1 7.40
1.06 1.10 21 8.70
1.16 1.20 23 10.30
1.27 1.20 25.3 12.40
1.4 1.20 27.8 14.90
1.53 1.20 30.5 17.90
1.68 1.20 33.5 21.60
1.85 1.20 36.8 26.20
2.03 1.20 40.4 31.60
2.23 1.20 44.3 37.80
2.45 1.20 48.7 45.00
2.69 1.30 53.4 53.10
2.95 1.30 58.7 61.60
3.24 1.30 ' 64.4 69.70
3.56 1.30 70.7 77.20
3.91 1.40 77.6 83.60
4.29 1.40 85.2 88.30
4.71 1.50 93.6 92.00
5.17 1.60 103 94.90
5.67 1.70 113 96.80
6.23 1.80 124 98.00
6.84 2.00 136 . 98.80
7.51 2.20 149 99.30
8.25 2.40 164 99.70

9.05 2.70 180 100.00

93
Table B.3: Size Distribution for the K46 Microbubble

Determined Using a Malvem MasterSizer

(so: 32.]
Size % Under Size % Under
microns microns

0.5 1.1 9.94 7.1
0.55 1.4 10.9 8.5
0.6 1.7 12 , 10.1 _
0.66 1.9 13.2 11.9
0.73 2.1 14.4 14.1
0.8 2.3 15.9 16.6
0.88 2.4 17.4 19.4
0.96 2.6 19.1 22.6
1.06 2.6 21 26.3
1.16 2.7 23 . 30.5
1.27 2.7 25.3 35.3
1.4 2.8 27.8 40.7
1.53 2.8 30.5 46.6
1.68 2.8 33.5 53
1.85 2.8 36.8 59.7
2.03 2.8 40.4 66.5
2.23 2.8 44.3 72.9
2.45 2.8 48.7 78.9
2.69 2.9 53.4 84
2.95 2.9 58.7 88.3
3.24 2.9 64.4 91.6
3.56 2.9 70.7 94
3.91 3 77.6 95.8
4.29 3.1 85.2 97.1
4.71 3.2 93.6 98
5.17 3.3 103 98.6
5.67 3.5 113 99.1
6.23 3.8 124 99.4
6.84 4.1 136 99.6
7.51 4.6 . 149 99.7
8.25 5.3 164 99.9

9.05 6.1 180 100

94

Table 8.4: Medium Separation Data

K20 Microbubble Distribution
Cone = 20°-22 20°-22 20°-22 20°-22
5 psi 7.5 psi 10 psi 20 psi

Inlet Underflow Underflow Underflow Underflow
Dens1ty' Density Density Density Density
$3111.32 gkg/m’1 Qg/m’) (kg/6131 (11316132

 

 

1000 100 100 1000 1000
900 980 990 990
880 980 990 990 1000
830 970 980 980 990
K46 Microbubble Distribution
Cone = 20°-22 20°-22 10°-22
5 psi 7 psi 5 psi

Inlet Under-flow Underflow Underflow
wm’) (ks/m3) (kLLm’

 

1000 1000 1000 1000
930 970 980 970
900 970 970 960
870 950 960 960

830 930 950 940

95

Table B.5: Glass Microbubble Break
Different Pumps

Progressive Cavity Pump
(6% breakup new)

Weight Weight
Time Broken Unbroken % Broken
(Minutes) (grams) (grams)

2 0.02 0.13 13.96
30 0.06 0.65 7.88
60 0.07 0.40 15.18
90 0.07 0.67 10.07
120 0.07 0.53 11.29
150 0.06 0.64 9.09
180 0.14 0.51 21.55
210 0.17 0.80 17.70
240 0.18 0.46 27.98
K20 Microbubbles
90 lpm
Centrifugal Pump
(6% breahtp new)
Weight Weight

Time Broken Unbroken % Broken
(Minutes) (grams) fi(_gnms)

 

2 0.09 0.24 27.60
30 0.24 0.22 52.41
60 0.16 0.60 20.69
90 0.32 0.31 50.36
120 0.30 0.59 33.86
150 0.27 0.48 35.73
180 0.38 0.53 41.37
210 0.27 0.39 41.13
240 0.31 0.53 36.55
K20 1416160015016-
20°-22 Hydrocyclone

9011M

96

Table B.6: Glass Microbubble Breakage

K20 Microbubble Volumetric Breakup

 

(2.6 “/0 breakup new)
Time Flowrat Temp Vol % Broke

(1pm) (C) sample A sample B Avg

30 66 24.00 3.40 3.90 3.65
60 66 23.50 3.40 3.40 3.40
90 66 23.50 4.30 4.50 4.40
120 66 23.00 4.20 4.20 4.20
150 80 21.00 4.50 4.80 4.65
180 80 22.00 5.00 5.00 5.00
210 80 23.50 5.20 4.50 4.85
240 80 25.00 5.30 4.80 5.05
270 80 22.00 6.40 4.70 5.55

K20 Microbubble Volumetric Breakup

(Next Day)

Time low rat Temp Vol '/o Broke

 

(lpm) (C) sample A sample B Avg
270 66 23.00 6.40 6.40 6.40
300 66 23.00 5.56 6.34 5.95
330 66 23.00 6.47 6.59 6.53
360 66 22.00 7.37 7.14 7.26
390 66 23.00 7.19 7.98 7.58
420 66 23.00 7.51 7.19 7.35
450 94 24.00 8.57 7.61 8.09
480 94 22.00 8.82 9.06 8.94
510 94 22.00 8.54 8.98 8.76
540 94 22.00 9.09 10.98 10.04

APPENDIX C

SEPARATION PERFORMANCE DATA

 

97

Table C.1: HDPE/PP Particle Size Distribution

Sieve specifications:
ASTM opening opening tare wt.
tray no. mesh (mm) (in) (g)
20 20 0.85 0.033 428.3
14 12 1.40 0.056 453.8
10 9 2.00 0.078 451.0
8 8 2.36 0.094 470.0
6 6 3.35 0.132 514.1
5 5 4.00 0.157 519.9
Size distribution of HDPE
Total Tray Total Wt. Sieve ' Wt.HDPE Mass
Plastics Number Sieve Tare Wt. Tray Undersize
(s) (8) (s) (s) (8)
242.2 20 454.1 428.3 25.8 0.25
242.2 14 579.5 453.8 125.7 10.90
242.2 10 521.6 451.0 70.6 62.80
242.2 8 489.5 470.0 19.5 91.95
242.2 6 514.1 514.1 0.0 100.00
242.2 5 519.9 519.9 0.0 100.00
Size distribution of PP
Total Tray Total Wt. Sieve Wt.PP Mass
Plastics Number Sieve Tare Wt. Tray Undersize
(s) (s) (8) (s) (s)
318.6 20 444.0 428.3 15.7 0.53
318.6 14 542.8 453.8 89.0 5.46
318.6 10 596.9 451.0 145.9 33.40
318.6 8 536.3 470.0 66.3 79.19
318.6 6 514.1 514.1 0.0 100.00
318.6 5 519.9 519.9 0.0 100.00

98
Table C.2: Data for the 20°-22 Hydrocyclone at 5 psi

K20 microbubbles
IAPI =5 psi(QF=581pm)
sample time = 5 see

‘debrmdbystadystatematefldbdmee

111161 thDPE “4mm; thDPE“ w: PP thP «11181M
Density UF or Feed UF or Feed
(s/mL) (8) (8) (8) (8) (8) (s)

1 0.026 1.208 1.234 0.000 3.057 3.057

0.061 2.992 3.053 0.000 3.672 3.672
0.025 0.451 0.475 0.000 2225 2.225
0.95 0.069 2.681 2749 0.082 2.736 2.818
0.226 2.632 2858 0.000 2.966 2.966
0.300 2447 2747 0.000 2.741 2741
0.925 0.405 2.482 2887 0.085 1.850 1.935
0.459 2534 2993 0.130 1.454 1.585
0.322 2450 2772 0.079 1.897 1.976
0.9 0.324 3.511 3.835 0.027 1.300 1.327
- 0.379 3.178 3.557 0.016 1.666 1.682
0.390 3.169 3.559 0.031 1.410 1.441
0.875 0.374 3.043 3.416 0.075 2855 2930
0.313 3.172 3.485 0.072 2891 2963
0.375 2741 3.116 0.065 2671 2.736
0.875 0.448 2656 3.104 0.147 3.013 3.160
0.475 2876 3.351 0.166 3.059 3.225
0.607 3.106 3.713 0.184 3.003 3.187
0.85 1.059 2318 3.377 0.014 2404 2417
0.611 2959 3.570 0.036 2549 2585
0.567 2649 3.216 0.009 2522 2531
0.825 1.884 2005 3.889 0.410 3.454 3.864
2018 2138 4.156 0.349 3.095 3.444
2.798 3.004 5.803 0.242 3.056 3.299
0.825 2616 2.271 4.886 0.287 4.353 4.640
2656 2230 4.886 0.303 4.143 4.446
2105 2195 4.300 0.426 5.382 5.808
0.825 3.452 2440 5.892
2916 2402 5.318
3.272 2427 5.699
0.8 1.688 2422 4.110 0.082 1.788 1.870
1.884 2294 4.178 0.147 3.521 3.668
1.476 1.840 3.316 0.159 3.909 4.068
0.8 2461 3.040 5.501 0.758 4.420 5.178
2260 2710 4.970 0.690 4.030 4.720
2430 2930 5.360 0.670 4.160 4.830
0.775 3.187 1.030 4.217 0.232 2290 2522
3.120 1.000 4.120 0.244 4.000 4.244
2790 0.980 3.770 0.167 2560 2727
0.75 4.169 0.076 4.245 0.718 3.335 4.053
3.687 0.153 3.839 0.409 2971 3.380
3.582 0.113 3.695 0.368 2988 3.356
0.7 8.297 0.072 8.369 5.613 0.932 6.545
7.785 0.060 7.845 5.382 1.434 6.816
6.953 0.070 7.023 5.492 2019 7.511

 

99

Table C.3: Data for the 20°-22 Hydrocyclone at 10 psi
K20 microbubbles

IAPI =10psi(Qz=581pm)

sample time = 5 sec

Inlet wt HDPE wt HDPE wt HDPE" Wt PP wt PP wt PP“
Density UF OF Feed UF OF Feed
(OIML) (0) (0) (0) (0) (0) (0)

1 0.044 2.587 2.631 0.000 2.952 2.952
0.95 0.138 4.178 4.316 0.000 4.935 4.935
0.925 0.267 4.358 4.625 0.017 2.31 1 2.327

0.9 0.250 4.241 4.490 0.032 1.785 1.817
0.875 0.224 3.588 3.81 1 0.026 4.107 4.132
0. 875 0.575 4.353 4.928 0. 136 3.752 3.888

0.85 0.356 3.947 4.3% 0.013 4.136 4.148
0.825 2.077 4.454 6.531 0.131 4.428 4.560
0.825 1.832 4.251 6.083 0.120 6.192 6.312
0.825 2.603 5.338 7.941

0.8 1.720 4.290 6.010 0.360 5.630 5.990

0.8 1.386 3.394 4.780 0.111 3.880 3.991
0.775 2.330 3.970 6.300 0.065 4.030 4.095

0.75 3.927 0.671 4.598 0.408 5.050 5.457

0.7 8.861 0.550 9.41 1 3.246 5.700 8.946

‘determined by steady state material balance

100

Table C.4: Data for the 20°-16 Hydrocyclone at 5 psi
K20 microbubbles

IAPI =5 psi(Qp=581pm)

sample time = 5 sec

'determined by steady state material balance

Inlet wt 1mm: wt 1mm; wt HDPE Wt PP wt PP wt PP“
Density UF 0F Feed UF OF Feed
(glmL) (8) (8) (8) (8) (8) (g)

1 0.027 2.428 2.455 0.000 3.150 3.150
0.95 0.072 2.861 2.933 0.094 2.771 2.866
0.925 0.130 1.741 1.871 0.072 2.308 2.379
0.9 0.130 3.462 3.592 0.024 1.064 1.088
0.875 0.167 2.801 2.968 0.133 3.100 3.233
0.875 0.319 4.236 4.555 0.029 2.072 2.101
0.85 0.185 2.442 2.628 0.007 3.076 3.083
0.825 0.971 3.615 4.586 0.105 4.279 4.384
0.8 1.000 4.440 5.440 0.090 3.110 3.200
0.8 0.913 3.666 4.579 0.034 2.562 2.596
0.775 1.035 2.940 3.975 0.070 4.080 4.150
0.75 1.064 1.556 2.620 0.174 4.301 4.475

101

Table C.5: Data for the 20°-l6 Hydrocycloneat 10 psi

K20 microbubbles

IAP|=10psi(QP=581pm)
sample time = 5 sec

‘determined by steady state material balance

Inlet wt HDPE thDPEthDPE Wt PP thP thP‘
Density UF or Feed UF or Feed
(g/mL) (8) (8) (8) (8) (8) (8)
1 0.041 3.067 3.108 0.000 2.707 2.707

0.95 0.174 5.048 5.222 0.000 3.788 3.788
0.925 0.138 3.260 3.398 0.065 2.052 2.117
0.9 0.163 2.668 2.830 0.058 1.396 1.454
0.875 0.082 3.707 3.790 0.031 3.310 3.342
0.875 0.193 4.364 4.557 0.008 2.992 3.000
0.85 0.151 3.685 3.836 0.000 4.296 4.296
0.825 0.473 3.673 4.146 0.119 5.865 5.984
0.8 0.650 6.130 6.780 0.063 '5.200 5.263
0.8 0.652 5.760 6.412 0.014 3.158 3.172
0.775 0.613 4.650 5.263 0.030 5.270 5.300
0.75 0.778 3.313 4.091 0.049 5.609 5.657

 

102

Table C.6: Data for the 20°-10 Hydrocyclone at 5 psi
K20 microbubbles

lAPl=5psi(Q.=581pm)
sampletime=SSec

'determined by steady state materid balm

Inlet wt HDPE wt HDPE wt 1mm: Wt PP wt PP wt PP“
Density UF OF Feed UF OF Feed
(8/mL) (8) (8) (8) (8) (8) (8)
1 0.006 1.483 1.489 0.000 2.426 2.426
0.95 0.000 4.722 4.722 0.000 2.209 2.209
0.925 0.000 2.424 2.424 0.046 2.198 2.244
0.9 0.018 2.222 2.239 0.019 1.414 1.433
0.875 0.000 3.341 3.341 0.000 2.239 2.239
0.875 0.014 4.471 4.485 0.000 1.754 1.754
0.85 0.000 2.891 2.891 0.000 2.893 2.893
0.825 0.071 3.596 3.666 0.022 3.242 3.264

103

Table C.7: Data for the 20°-10 Hydrocyclone at 10 psi

K20 microbubbles

IAPI =10psi(Qp=58lpm)
sample time = 5 sec

Inlet thDPE thDPEthDPE Wt PP thP thP"
Density UF 0F Feed UF OF Feed

(g/mL) (8) (8) (8) (8) (8) (8)

1 0.000 2.848 2.848 0.000 3.406 3.406

0.95 0.000 6.301 6.301 0.000 3.082 3.082
0.925 0.027 3.847 3.873 0.020 2. 172 2. 192
0.9 0.023 2.602 2.625 0. 024 1. 113 1.137
0.875 0.000 2.514 2.514 0.000 3.265 3.265
0.87 5 0.000 6.012 6.012 0.000 2.733 2.733
0.85 0.000 2.900 2.900 0.000 4.000 4.000
0.825 0.042 3.573 3.614 0.229 3.953 4.182

'determined by steady state material bdmee

104

Table C.8: Data for the 20°-22 Hydrocyclone at
5 psi with K46 Microbubbles

Inlet wt HDPE wt HDPE wt HDPE"I

Density UF
(slmL) (8)
1.00 0.110
0.151
0.93 0.742
0.762
0.90 2.223
2.116
0.83 7.735
7.114
0.75 12.290
11.050

01'
(8)
4.705
5.491
6.072
6.820
4.342
4.239
0.977
0.984
0.000
0.000

Feed
(8)
4.815
5.642
6.814
7.582
6.565
6.355
8.712
8.098

12.290
1 1.050
'determlned by steady state materlal balance

Wt PP

UF
(8)
0.000
0.000
0.031
0.055
0.115
0.087
1.655
2.384

10.120

7.450

wt PP
or
(8)
5.597
5.070
6.365
5.592
7.280
6.579
2.967
5.365
0.011
0.024

wt 1318*
Feed
(8)
5.597
5.070
6.396
5.647
7.395
6.666
4.622
7.749
10.131
7.474

105

Table C.9: Data for the 20°-22 Hydrocyclone at
7 psi with K46 Microbubbles

Inlet wt HDPE wt HDPE wt HDPE"

Density
(s/mL)
1.00
0.93
0.90
0.83

0.75

UF
(8)
0.111
0.116
0.575
0.543
1.749
1.657
8.236
6.621
9.510
7.890

OF
(8)
6.061
5.802
7.317
6.851
6.192
5.554
1.761
1.245
0.024
0.013

Feed
(8)
6.172
5.918
7.892
7.394
7.941
7.211
9.997
7.866
9.534
7.903

Wt PP

UF
(8)
0.015
0.007
0.005
0.000
0.076
0.073
2.095
1.673

10.180

9.140

‘determined by steady state material balance

wt pp wt 1:1"
or Feed
(8) (8)

6.052 6.067

5.895 5.902

5.250 5.255

4.923 4.923

10.310 10.386

10.880 10.953

6.572 8.667

6.711 8.384

0.030 10.210

0.028 9.168

106

Table C.10: Data for the 10°-22 Hydrocyclone

Inlet wt HDPE wt HDPE wt HDPE" Wt PP wt pp wt PP‘
Density UF or Feed UF or Peed
(B/mL) (8) (8) (8) (8) (8) (8)

1.00 0.107 6.790 6.897 0.008 5.255 5.263

0.160 5.244 5.404 0.009 5.395 5.404

0.93 1.137 5.502 6.639 0.079 5.998 6.077

0.947 4.970 5.917 0.122 5.500 5.622
0.91 6.448 4.790 11.238 0.177 7.500 7.677
2.564 5.190 7.754 0.063 7.240 7.303
0.90 2.500 5.402 7.902 0.138 7.335 7.473
2.214 4.324 6.538 0.082 6.845 6.927
0.83 8.503 1.507 10.010 void 5.187 void
7.482 0.996 8.478 1.884 4.920 6.804
0.75 12.910 0.000 12.910 12.683 0.000 12.683
11.800 0.012 11.812 . 11.510 0.000 11.510

‘determlned by steady state material balance

LIST OF REFERENCES

LIST OF REFERENCES

Ali, S.K., Z.C. Yang, J .F. Foss, C.A. Petty, “The Separation Performance of Long
H‘ydrocyclones with an Annular Feed Entry”, Paper presented by Syed Ali at the
4 International Conference on Hydrocyclones (September 23-25, 1992).

American Plastics Council, 1993, Post Consumer Plastics Recycling Rate Study -
Executive Summary.

American Cyanamid Corp., 1951, “Heavy-Media Separation”, Ceramic Bulletin,
30(3), 63 (1951).

Altand, B.L., RM. Enick, EJ. Beckrnan, and S. Chang, “Optimization of Batch
Thermoplastic Microsortation Using Near-Critical Liquids”, Paper presented at the
1994 AIChE Annual Meeting, San Fransisco, CA (November 13-18).

Battle K.E., A.P. Moore, J .C. Lynch, and EB. Nauman, “Plastics Recycling by
Selective Dissolution”, Transcripts from presentation by Ken Battle at The Dewitt
Conference (March 25-27, 1992).

Bradley, D., 1965, The Hydrogclone, Pergamon Press.

Carlson, D. C. and M. Kakabalia, 1993, “The Use of Hydrocyclone Technology for
HDPE Recycling- A Market Study”, report from a market survey conducted for
the MTA 890 graduate marketing course.

Dahlstrom, D.A., 1949, “Cyclone Operating Factors and Capacities on Coal and
Refuse Slurries”, Mining Transactions, 184, 331 (Sept).

Dreissen, H.H., and RI. Fontein, 1963, “Applications of Hydrocyclones and Sieve
Bends in Wet Treatment of Coal, Minerals and Mineral Products”, Society of
Mining Engineers, 101 (March).

Dreissen, M.G., 1939, “Cleaning by Heavy Liquids, with Special Reference to the

Staatsminjnen-Loess Process”, Journal of The Institute of Fuel, 329 (August,
193 9) '

107

108
Hegberg B.A., G.R Brenniman, and W.H. Hallenbeck, 1992, Mixed Plastics
Recycling Technology, Noyes Data Corp.

Mark, H.F., & et al., eds., 1985, Encyclopedia of Polmer Science & Engineering,
Wiley.

Miller, K.J., M.S. Klima, and RP. Killlneyer, 1991, “Selection and Production of
Dense-Medium Solids for the Micro-Mag Process”, Topical Report
DOE/PETC/TR-91/5 (DE91011943).

Moder, J. and DA. Dahlstrom, 1952, “Fine Size, Close Specific Gravity Solids
Separation with the Liquid-Solid Cyclone”, Chemical Engineering Progress,
48(2), 75. '

Nugent, DC, 1991, Process for the Separation and Recovery of Plastics, U.S.
Patent (5,022,985).

Perry, RH., D.W. Green, and J .0. Maloney, eds., 1984, Fey’s Chemical
Engineers’ Handbook, 6th ed., McGraw-I-Iill Book Company.

Petty, C.A‘., S.K. Ali, E.A. Grulke, and SE. Selke, 1993, “Hydrocyclone
Classifiers for Microsorting Mixed Thermoplastics from Consumer Waste”,
Proceedings of Waste Minimization and Utilization of Innovative Concepts - An
Experimental Technology Exchange, organized by the U.S. Department of Energy,
Austin, TX.

Svarovsky, L., 1984, Hydrogclones, Holt, Rinehart and Winston Ltd.

"I111111111111111“