HomeMy WebLinkAboutDocumentation_Environmental Advisory Committee_Tab 06_11/12/2025 Agenda Item #6.
Environmental Advisory Committee (EAC)
STAFF MEMO
R
Meeting: Environmental Advisory Committee (EAC) - Nov 12 2025
Staff Contact: Brad Freese Department: Community Development
Single Stream v. Dual Stream Recycling: History and Data
SUMMARY
Single-stream recycling, introduced in the 1990s, boosted convenience and participation but also
increased contamination. A 2018 University of Miami study of 15 Florida counties found average
contamination rates of 18.54% for single-stream systems versus 3.89% for dual-stream systems.
After processing, single-stream paper and cardboard had contamination rates of 7.13% and 5.12%,
respectively. Because paper mills typically allow only minimal contamination, fewer than 5% of
samples met saleable quality standards. Overall, single-stream systems collect more material but
yield significantly less marketable recyclables due to higher contamination.
This document and any attachments may be reproduced upon request in an alternative format by
completing our Accessibility Feedback Form, sending an e-mail to the Village Clerk or calling 561-
768-0443.
BUDGET INFORMATION:
BUDGET AMOUNT NA AMOUNT AVAILABLE NA EXPENDITURE AMOUNT: NA
FUNDING SOURCES: NA IS THIS A PIGGYBACK:
❑ Yes ❑ N/A
DID YOU OBTAIN 3 QUOTES?
❑ Yes ❑ N/A
COMMENTS/EXPLANATION ON SELECTIONNA
Staff Memo Recvclinq
Assessment and Evaluation of Contamination - EAC 11.12.25
Single-Stream-Recycling Back up - EAC 11.12.25
Page 9 of 67
Agenda Item #6.
V *111age of T
345 Tequesta Drive 561-768-0700
Tequesta, FL 33469 www.tequesta.org
TO: Environmental Advisory Committee
FROM: Brad Freese, Chair EAC
DATE: 11/12/2025
SUBJECT: Single Stream v. Dual Stream Recycling: History and Data
Single-stream recycling was introduced in the 1990s and significantly improved the recycling
process in the U.S. This system allows all recyclable materials, such as paper, plastic, and metal,
to be collected in a single bin, making it easier for consumers to participate in recycling. In
response the national recycling rate increased from 10.1% in 1985 to 25.7% in 1995 and nearly
32% in 2005. Since 2005 the national recycling rate has been stagnate at 32%.
Prior to 2018, China was a major importer of worldwide solid waste including recyclables. For
example, 70% of US plastic collected for recycling was sold and shipped to Chinese processors.
In 2018 China enacted a policy called The National Sword Ban that prohibited the import of 24
kinds of solid waste from foreign countries. Solid wastes including plastics, paper products, and
textiles, etc. The policy decreased the importation of recyclables through stricter
contamination rates of 10%to 0.5%. This new policy has seen a sharp decrease in solid waste
including recyclables shipped to China. Other major importers of recyclables have also enacted
similar policies. These new policies have changed the economic viability of recycling programs
and have put greater emphasis on clean, non-contaminated recyclable material.
Does single stream collection increase contamination rates? If so, by how much?
See attached a study from 2018 by Nurcin Celik PhD et al at the University of Miami
Department of Industrial Engineering, "Assessment and Evaluation of Contamination in Single
Stream Recycling Systems Due to Broken Glass and Other Non-recyclables". In summary, 15
Florida counties were evaluated for rates of contamination (6 single stream, 9 dual stream).
The average contamination rate for single stream was 18.54 and the average contamination
rate for dual stream was 3.89 for inbound material prior to processing. The study also reported
on the contamination rates of single stream after processing ie outbound material. The
investigators looked at paper products and corrugated cardboard products. Outbound
contamination rates were 7.13 percent and 5.12 percent respectively. How does this translate
This document may be reproduced upon request in an alternative format by contacting the Village
Clerk's Office at 561-768-0440 or by completing our accessibility form: https://bit.ly/3mnfeU4
Page 10 of 67
Agenda Item #6.
to viable recyclable material to be sold on the open market. Based on allowable contamination
rates by paper mills at the time of the study, 1 of the 266 paper samples and 12 of the 35
corrugated cardboard samples had allowable contamination rates.
Page 2 of 8
Page 11 of 67
Agenda Item #6.
1
A,
r
t �&
a r
h j
RA
r: 'Eke r k'�.
x �
a
{y r
i � F
F TSy
F Ar rj.di sy+° ti. ay�f
+-d
IN`
jr
PA.� x + ^1 4
•i g Y
�. .
1
t
Fjo
„
rrrryryrFjaj1 3 y $.r.f: *}ir�y,t`t*P'
fall Y�' l }'-�1'iTTIp{,�eP�'��"�,�� ..�`'��P, ;h •''�'�.L f r�''..
Page 12 of 67
Agenda Item #6.
gTT4ii■r LL*•`yy .�'qq■
.�. ,LIL
y l 4 4y 4
PrIL
S
yS2�` �� al� 6?'•. 7F h ,wk-.tr�„' '•.. .4- g
!►h� 4 'tij� �,�
A4 i
� L
u � .. sr �
E '
y �F
,Lgby:4 `t ,4$*'}' . •r y P'FR, � `'')ir.... �'7
,w.J °"
&VN
ih
is
� .. ..y�.�I���� s ter•,,,: �'Cy(:� A �1 �
yam- a
b#1� f ► �e
r �}R�..JY —Y�, L ll.: 4.. '��' -i �y{���•�� #,f7..yi.yj.�f
10
kr
41
a
Z r.
r
,r
It
_
}t. 4
n fi v
r
s . I
M1 ~ r
,
Y
,
y ~�°• tl
e
b� r
k
k
■ r � r '
f /
n
+y .
a _ ,
n
x °
,
Page 4 of 8
Page 13 of 67
Agenda Item #6.
ME 1*1
r.
000,
�4-0-1
A.
y
-
� 5 k K f,'.� h r 5 W1 5y Y w L-� Y ..• ! r ! I ;�><.:;,
IA * �''k+ /f� X '�IF�.+T�."yy.'�+•ram'
�� 4 4 ]F 'r' `��� �-41 oh � �1 i�l���. .r f 'C i.�..� �R�iy��•S may..-.
YYt5-� -�1�!'1'x' ii� �Fr S�ti f.'vlkA .`'_L �iF ��' �.r•4a. �°'x .
s
f
_ C> I LJ 11'3 ►I
I Lei 0
4 jfe
- i -
� •'p ram.. ;. .x,i � �� �. �, ��i.F 4'^iR'•
MIR
. - � ., p .ti � .g'T�''.a ti't'ry"•s Sa.r `'._h
k
•
i
Page 14 of 67
Agenda Item #6.
fa i:CR
r
{ P
- * ArOF
µ t
S%E
. _ 10
Or
j
11�11512R:Q'I
ram
Al
16,3
-020
V!-, -4
PK
AM
Image ID:2AT83AR
{a
'I� r
- F
r:
I,
1 I
my www.alamy.com
Surface Water Drain
IJ1
Rain & Storm Water only
No other liquids
to enter the drain
Page 6 of 8
Page 15 of 67
Agenda Item #6.
,
,
wcY' Y
Y ? . Y.
lire v
jy
e2:",iirr
`'�' � •Ye/�� a.� r �� _y art� �_ '/
r
C 40
yam'
.. , A.� q..,..�.'..-, ,+•�_,t- ^,;'.,...1 '.' (� :' A,: I �I � � -. ? IT�II�j. ILn.. I°.-. 9�ii� 'll'f'I �1 -. ,,II�y S'yi�l�,
I'i � *�i��":...15,t'�'4 11 I'� y Z'�;• I 1 / �� hd. I�; I � �'.
4 y
r
.. I
.I R ^
^
.. � ..,-I I i i.4.. •;i li°W Y 3e..:1', I 'r„ � i S
I
E
1A.
_ F 1G 'TRi�• �4;�R1.'i�On"
�..ly A
I fti
i'•' a '�"'''t,,,x^' � Y"'+ "+e� -«rot i'�+ ♦„w
y i
• d 9 w t
l � :
'.I
a
,►`
> • I
°.
x: •
s
y
Car
w4 �
I
I
zl:
Page 16 of 67
c 1
1 � ,
F
J,
• I � V
F'
n
. r 1
F ,
qp
o-'-
p
a r � r —4 i
�F�� ,. '�' I,y •� 1`.� �`�� 1 �t •J � � � T 1" 1"�IMI , � i , IL
J '`� y i• ��� a '4 �,Y'F '-e 1 '� • ,. .1 f � , �F
ow ,
: '
Agenda Item N.
Assessment and Evaluation of Contamination in Single Stream Recycling
Systems Due to Broken Glass and Other Non-Recyclables
by
Haluk Damgacioglu,Lea Perez,and Nurcin Celik(PI)
University of Miami, Coral Gables,FL,USA
Department of Industrial Engineering
Submitted to
Hinkley Center for Solid and Hazardous Waste Management
University of Florida
P. O. Box 116016
Gainesville, FL 32611
www.hinkleycenter.org
Report# 11667
I
Page 18 of 67
Agenda Item #6.
TABLE OF CONTENTS
LIST OF FIGURES III
LIST OF TABLES IV
LIST OF ACRONYMS V
EXECUTIVE SUMMARY XI
DISSEMINATION ACTIVITIES XV
1 INTRODUCTION 1
2 BACKGROUND AND LITERATURE REVIEW 2
3 MATERIAL RECOVERY FACILITIES 7
3.1 Material Definitions 8
3.2 Single Stream MRF Process Flow 10
4 DATA COLLECTION AND ANALYSIS 11
4.1 Inbound Contamination Analysis 11
4.2 Outbound Contamination Analysis 18
43 Material Recovery Facility(MRF)Audit Analysis 20
5 CONCLUSION 26
6 REFERENCES 28
7 APPENDICES 31
APPENDIX A: ANALYSIS OF VARIANCE 31
II
Page 19 of 67
Agenda Item #6.
LIST OF FIGURES
Figure 1. Distribution of paper/paperboard collection techniques in the United States (AF&PA,
2015) 1
Figure 2: End-market for collected glass (Kinsella& Gertman, 2007) 2
Figure 3. Residual characterization for each MRF type in California(Beck, 2006) 3
Figure 4. Effects of single stream on NORPAC Paper Mill (Morawski, 2010) 3
Figure 5. Chittenden County residue rates before and after their switch to single stream (Moreau,
2015) 4
Figure 6. End markets for glass collected from different systems (DSM Environmental Services,
2011) 5
Figure 7. Glass losses across the recovery process (Sustainability Victoria, 2014) 6
Figure 8. Glass losses in Ohio by collection system(CRI, 2009) 6
Figure 9. Number of Gullet processors in 35 states (GPI, 2018) 6
Figure 10. Average recycling rates (AR) and glass recycling rates (AGR) of counties that
switched to SSR in Florida (FDEP, 2017) 7
Figure 11. Single stream MRF process flow with equipment configuration to recover fiber, glass,
plastic and metal materials 11
Figure 12. Boxplot of SSR and DSR contamination data 12
Figure 13. Histograms of SSR and DSR contamination rates 13
Figure 14. Results of AD normality test for SSR contamination data 14
Figure 15. Results of AD normality test for DSR contamination data 14
Figure 16. Histograms of contamination rates of each county 15
Figure 17. Contamination rates in SSR counties 16
Figure 18. Contamination rates in SSR counties 17
Figure 19. County correlation matrix 18
Figure 20. OCC prohibitive materials analysis 19
Figure 21. ONP prohibitive materials analysis 19
Figure 22. Types of contamination in corrugated cardboard(a) and newspapers (b) 20
III
Page 20 of 67
Agenda Item #6.
LIST OF TABLES
Table 1. Construction application of glass Gullet(GPI,N.D.) 5
Table 2. Plastic material definitions 8
Table 3. Paper recyclables recovered in MRFs 9
Table 4. Glass recyclables recovered in MRFs 9
Table 5. Metal recyclables recovered in MRFs 9
Table 6. Residues found in MRFs during single stream recycling 10
Table 7. ANOVA results of SSR-DSR comparison 14
Table 8. Descriptive analysis of contamination rates for each county 15
Table 9. ANOVA results of contamination on the mean contamination rates of six selected SSR
counties 16
Table 10. SSR separation of efficiencies in percentage (Presley et al., 2015) 21
Table 11. MRF efficiency table based on the results of mass balance analysis 23
Table 12. Single stream MRF sort results 24
Table 13. Summary of ANOVA formulas 32
IV
Page 21 of 67
Agenda Item #6.
LIST OF ACRONYMS
AD : Anderson-Darling normality test
AF&PA : American Forest and Paper Association
ANOVA : Analysis of variance
CRI : Container Recycling Institute
DSR : Dual stream recycling
EPA : Environmental Protection Agency
FDEP : Florida Department of Environmental Protection
GPI : Glass Packaging Institute
HDPE : High density polyethylene
ISRI : Institute of Scrap Recycling Industries
MRF : Material recovery facility
OCC : Old corrugated containers
ONP : Old newspapers
PET : Polyethylene terephthalate
SSR : Single stream recycling
TAG : Technical awareness group
V
Page 22 of 67
Agenda Item #6.
FINAL REPORT
(April 2017 -March 2018)
PROJECT TITLE: Assessment and Evaluation of Contamination in Single
Stream Recycling Systems Due to Broken Glass and Other
Non-recyclables
PRINCIPAL INVESTIGATOR: Nurcin Celik, Ph.D.
AFFILIATION: Department of Industrial Engineering, University of Miami
EMAIL: celikgmiami.edu
PHONE NUMBER: 1-305-284-2391
PROJECT WEBSITE: http://www.coe.miami.edu/simlab/swm.html
COMPLETION DATE: March 31'% 2018
TAG MEMBERS: David Gregory, Jeremy O'Brien, Karen Moore, Paul
Valenti, Eddie McManus, Sean Williams, Michael
Heimbach, Michael J. Fernandez, Brenda S. Clark, Katie
Brown, Samuel B. Levin, Patrick Sullivan, Florin Gradinar,
Himanshu Mehta, Stephanie Watson, Sally Palmi, Ron
Hixson, Scott Harper, Teddy Lhoutellier, Greg Schaffer,
Johnny Gold, Leonard Marion.
KEY WORDS: Single stream recycling, assessment of inbound
contamination rates, evaluation of outbound contamination
rates, material recovery facilities, prohibitive materials
VI
Page 23 of 67
Agenda Item #6.
ABSTRACT
In single stream recycling (SSR), all recyclables are separated from municipal solid waste into a
sole compartment in a collection truck. Amongst the curbside recycling programs, SSR has
grown rapidly in the past decade. This growth is likely due to SSR's ability to better attract the
participation of communities, reduce the cost of collection of curbside recyclables, and avoid a
cumbersome sorting process at the individual level. In Florida, 24 counties have recently
switched their recycling programs from dual stream recycling (DSR) to SSR. These changes
include the counties covering major cities such as Brevard, Broward, Hillsborough, and Miami-
Dade. In a DSR system, two collection bins are provided to consumers. The consumer then
separates paper products from the other glass, metal, and plastic recyclables. This separation
results in lower participation rates and smaller amounts of collected recyclables compared to
SSR. Despite an SSR system's inherent advantages on reducing collection costs and increasing
community recycling participation, it may not necessarily lead to a higher recycling rate. The use
of the mixed-material bin brings up issues of contamination, as materials sometimes cannot be
sorted properly with other recyclables and non-recyclables. This further leads to higher costs
during processes in material recovery facilities (MRFs) and recycled material discard. In order to
better understand the impact of SSR on the Florida solid waste management system, in this
study, we conducted an assessment of contamination in MRFs in Florida, aiming to determine to
which extent broken glass and other non-recyclables are responsible for the contamination of
recovered material.
Researchers collected various data regarding the inbound contamination rates from waste
composition studies provided by 15 counties around Florida, the outbound contamination rates
from four currently operating facilities in Florida, and the assessment of existing processing
technologies from one MRF audit. After assessing the waste composition studies, researchers
obtained 170 samples from SSR and 45 samples from DSR for inbound contamination rates.
Based on the results of the descriptive statistical analysis and one-way analysis of variance
(ANOVA) technique on the inbound contamination data, researchers concluded that there is a
statistical significance of the difference between the mean contamination rates of SSR and DSR
systems. Researchers further tested the statistical significance of the difference between the mean
contamination rates of the counties which currently implement SSR. A correlation analysis was
also carried out between the contamination rates and demographics of the considered counties.
Based on the results, contrary to popular belief, no statistical correlation between inbound
contamination and ethnicity was found in county level analysis. However, a positive correlation
between median age and contamination, as well as a negative correlation between income level
and contamination were identified in our analysis.
In the assessment of outbound contamination, researchers obtained 266 old newsprint (ONP)
samples and 35 old corrugated cardboard (OCC) samples from four currently operating facilities
in Florida. In this assessment, researchers focused on prohibitive materials in the sorted and
separated recyclable due to the scope of the project. Average rates (weight of the
contamination/weight of total sample) of prohibitive materials in the ONP stream and the OCC
stream were 7.13 percent and 5.12 percent, respectively. Considering the paper mill standards are
only 2 percent for ONP and 3 percent for OCC, 34 percent of OCC samples and only 0.3 percent
of ONP samples could meet the paper mill standards. The most common types of prohibitive
VII
Page 24 of 67
Agenda Item #6.
materials in the OCC streams apart from residues were film plastic, and high density
polyethylene (HDPE). On the other hand, residue, polyethylene terephthalate (PET), and film
plastic were the most common types of prohibitive materials found in the ONP streams.
Lastly, researchers investigated which of the current methods being applied to MRFs (i.e., for
sorting, etc.) lead to increased contamination rates in the outbound stream. Here, researchers
assessed MRF audit data using mass balance analysis and achieved two main tables: an
efficiency table and a sorting results table. According to the efficiency table, the efficiency rate
of Disc Screen 1, which separates OCC from other types of fiber, was found as approximately 45
percent. Despite the low efficiency rate, the contamination was approximately 5 percent
according to the sorting table. This means that even slight improvement on the efficiency of Disc
Screen I can provide much lower contamination since the screen eliminates OCC from other
recyclables and non-recyclables in the incoming stream. Based on the sorting results, the
acceptable rate of the mixed paper was 92 percent, which is considered a very low rate according
to mill standards. The efficiency rate of Optical Sorters that separate PET and HDPE plastics was
approximately 85 percent. This rate is very close to industry standards. In plastic recycling,
colored HDPE was the most contaminated recyclable with 65 percent of material weight being
acceptable. In glass recycling, the efficiency of the Glass Breaker Screen was found to be 95
percent; however, the sort results indicated that the screen failed to separate grit, fines and
sweepings from glass cullet. The acceptable rate of mixed glass was only 70 percent despite the
high level of efficiency or the processing techniques. Here, efforts could be shaped to decrease
the amount of grit, fines and sweepings in the inbound stream.
VIII
Page 25 of 67
Agenda Item #6.
METRICS:
1. List graduate student or postdoctoral researchers funded by THIS Hinkley Center project.
Last name, first Rank Department Professor Institution
name
Damgacioglu, Ph.D. Candidate Department of Prof. Nurcin University
Haluk Industrial Engineering Celik of Miami
2. List undergraduate researchers working on THIS Hinkley Center project.
Last name, first Rank Department Professor Institution
name
Perez, Lea Undergraduate Department of Prof. Nurcin University
Student Industrial Engineering Celik of Miami
3. List research publications resulting from THIS Hinkley Center project (use format for
publications as outlined in Section 1.13 of this Report Guide).
• Damgacioglu, H., Perez, L., Celik, N. "Assessment and Evaluation of Inbound
Contamination in Single Stream Recycling Systems — A Case Study of Florida" working
journalpaper.
• Damgacioglu, H., Hornilla, M., Celik, N. "Assessment and Evaluation of Outbound
Contamination in Single Stream Material Recovery Facilities —A Case Study of Florida"
working journal paper.
4. List research presentations (as outlined in 1.13.6 of this Report Guide) resulting from THIS
Hinkley Center project.
• Technical Awareness Group (TAG) I Meeting took place on July 14th, 2017 at the
McArthur Engineering Building of the University of Miami with 13 attendees (both on-
site and via conference call).
• Technical Awareness Group (TAG) II Meeting took place on March 9th, 2018 at the
McArthur Engineering Building of the University of Miami with 14 attendees (both on-
site and via conference call).
S. List of individuals who have referenced or cited publications from this project.
• No publications to date.
6. How have the research results from THIS Hinkley Center project been leveraged to secure
additional research funding?
• PI Celik has secured an additional funding through the DOE-funded REMADE Institute.
The project will start in the second half of 2018 once the Institute administration is fully
launched. University of Miami is also becoming a Tier 1 member of this Institute where
PI Celik is serving as the Technical Lead. The initial foundational funding will be at
110K from the Institute with a 1:1 cost share from University of Miami.
7. What new collaborations were initiated based on THIS Hinkley Center project?
IX
Page 26 of 67
Agenda Item #6.
• Eddie McManus, the director of material recovery facilities owned by Waste
Management in Florida has been very helpful for providing necessary information. He is
our direct point of contact if we need any further information from Waste Management.
• Jonathan Gold, a renowned expert in the paper recovery industry has been very helpful
for explaining how the single stream collection affected paper recycling industry. He has
a wealth of experience in paper recycling industry, public and private. He has been
interested in the finding of both the last year's project, "Assessment of the Impact of
Single Stream Recycling on Paper Contamination in Recovery Facilities and Paper
Mills," and the current project. He also serves in the TAG committee of the current
project.
8. How have the results from THIS Hinkley Center funded project been used (not will be used)
by FDEP or other stakeholders? (1 paragraph maximum).
• The results of this study have not yet been presented.
X
Page 27 of 67
Agenda Item #6.
EXECUTIVE SUMMARY
PROJECT TITLE: Assessment and Evaluation of Contamination in Single
Stream Recycling Systems Due to Broken Glass and Other
Non-recyclables
PRINCIPAL INVESTIGATOR: Nurcin Celik, Ph.D.
AFFILIATION: Department of Industrial Engineering, University of Miami
PROJECT WEBSITE: http://www.coe.miami.edu/simlab/swm.html
TAG MEMBERS: David Gregory, Jeremy O'Brien, Karen Moore, Paul
Valenti, Eddie McManus, Sean Williams, Michael
Heimbach, Michael J. Fernandez, Brenda S. Clark, Katie
Brown, Samuel B. Levin, Patrick Sullivan, Florin Gradinar,
Himanshu Mehta, Stephanie Watson, Sally Palmi, Ron
Hixson, Scott Harper, Teddy Lhoutellier, Greg Schaffer,
Johnny Gold, Leonard Marion.
COMPLETION DATE: March 31, 2018
OBJECTIVES:
In single stream recycling (SSR), all recyclables are separated from municipal solid waste into a
sole compartment in a collection truck. Amongst the curbside recycling programs, SSR has
grown rapidly in the past decade. This growth is likely due to S SR's ability to better attract the
participation of communities, reduce the cost of collection of curbside recyclables, and avoid a
cumbersome sorting process at the individual level. According to an American Forest and Paper
Association report, in 2005, 22 percent of the United States population had access to SSR, but by
2014, that had grown to 73 percent (AF&PA, 2015). This surge in popularity is also evident in
Florida where, based on our investigations, at least 24 of the 67 counties have currently operating
SSR programs in place. The list of counties include Brevard, Dade, Escambia, Marion, and
Orange counties. While SSR increases the number of recyclables collected and eases the
collection process, SSR systems may operate at the expense of higher volumes of materials
requiring pre-sorting at regional material recovery facilities (MRFs), and higher contamination
rates in the materials directed to the mills. In SSR, increasing amounts of non-recyclable
materials (e.g., plastic bags, food waste) in single collection bins and incoming comingled
recyclables (e.g., small glass pieces and shards may get stuck in fiber items, making them unable
to be recycled) fosters issues of contamination. This further leads to higher costs during
processes in MRFs and recycled material discard. To better understand the impact of SSR on the
Florida solid waste management system, in this study, we assessed the contamination issues
found in MRFs operating around the state of Florida. Our aim is to determine the extent which
broken glass and other non-recyclables are responsible for the contamination of recovered
material.
XI
Page 28 of 67
Agenda Item #6.
In this study, researchers first assessed and evaluated the impact of inbound contamination
resulting from SSR systems and compared the results with inbound contamination resulting from
the dual stream recycling (DSR) system. The use of DSR system necessitates the residents to
separate fiber (newspaper, magazines and catalogs, mixed paper, cardboard, etc.) and other
recyclables for collection. The inbound contamination rates in both SSR and DSR were obtained
by analyzing the collected waste composition studies from 15 Florida counties—Alachua,
Brevard, Broward, Citrus, Escambia, Hillsborough, Indian River, Lee, Leon, Marion, Okaloosa,
Pasco, Santa Rosa, Sarasota, and Seminole. The inbound contamination data was analyzed using
the analysis of variance (ANOVA) techniques, which measures the statistical significance, if
any, of the difference between the mean contamination rates of the two systems. Based on
ANOVA statistics and descriptive statistical analysis, the difference was found to be significant
and variance in contamination rates in SSR samples were much higher than the variance in DSR
samples. Further analysis was conducted to better understand the high variance in SSR samples.
Here, researchers compared the mean contamination rates of the counties which currently
implement SSR. This analysis showed that the inbound contamination rates were higher in
counties with higher median age and lower income level. In addition to inbound contamination,
outbound contamination rates of single stream MRFs in Florida were assessed to represent the
impact of SSR on the recovered material quality. Finally, one MRF audit was analyzed to
investigate which of the current methods being applied to MRFs (i.e., for sorting, etc.) lead to
increased contamination rates in the outbound stream.
METHODOLOGY:
The study consisted of the following phases:
1. Collection and Analysis of Inbound Contamination Rates in SSR and DSR: Researchers
identified and assessed the impact SSR has on inbound contamination rates by analyzing
collected waste composition studies from 15 Florida counties. Researchers obtained 170
samples from SSR and 45 samples from DSR to find contamination rates in the inbound
stream. Based on descriptive statistical analysis, the mean contamination rates were found to
be 18.54 percent for samples from SSR and 3.89 percent for samples from DSR. It was also
shown that SSR contamination rates were more spread out over a wide range of values than
were DSR contamination rates. The data was further analyzed using ANOVA to test if there
is statistically significant evidence that shows contamination rates in SSR systems are higher
than the rates in DSR systems. The analysis rejected the null hypothesis that the
contamination rates in SSR and DSR systems are the same and supported the alternative
hypothesis that the contamination rates in single stream recycling are higher than those in
dual-stream recycling at 0.999 significance level.
2. Assessment of Inbound Contamination Rates at County Level: The standard deviation (8.97)
of the SSR contamination samples was found to be higher than that of the DSR
contamination samples (3.08). To better understand this significant variance in inbound
contamination rates among SSR samples, researchers investigated the difference in
contamination rates of the SSR and DSR systems at a county level. Here, three counties
operationing with DSR demonstrate the lowest contamination rates at less than 5 percent,
while three other counties, after switching to SSR from DSR demonstrated the highest
contamination rates at more than 20 percent. It was also found that a wider variation of
contamination rates exists between counties operating with SSR. An ANOVA test was
XII
Page 29 of 67
Agenda Item #6.
designed to determine the statistical significance of the difference between the mean
contamination rates of the six selected counties. The results indicated that two counties, with
respective contamination rates of 28.2 and 22.3 percent on average, have means
significantly higher than the other counties based on a 95 percent confidence level.
Researchers then conducted a correlation analysis to further investigate the differences
between counties and find if any relationship exists between contamination rates and the
demographics of these six counties. The results found no correlation between inbound
contamination rates and ethnicity; however, the test showed a positive correlation between
median age and contamination, as well as a negative correlation between income level and
contamination.
3. Assessment of Outbound Contamination Rates: Researchers assessed the contamination
levels in the recovered fiber samples baled by MRFs. We obtained 35 old corrugated
cardboard (OCC) and 266 old newspaper (ONP) samples from audit reports conducted in
2016 for four currently operating facilities in Florida. In our assessment, we focused on
prohibitive materials in the outbound stream, due to the scope of the study. Average rates
(weight of the contamination/weight of total sample) of prohibitive materials in the OCC
stream and the ONP stream were 5.12 and 7.13 percent, respectively. The assessment of the
contamination rates of outbound materials produced alarming results. Considering the paper
mill standards are only 3 percent for OCC and 2 percent for ONP, these results are
disconcerting. With these standards in place, only 34 percent of OCC samples and 0.3
percent of ONP samples meet paper mill standards. Our analysis also found that both OCC
and ONP samples were most frequently contaminated with residue and plastics. In the OCC
stream, these contaminants mostly included film plastics and high density polyethylene
(HDPE). Comparatively, ONP samples were mostly contaminated by polyethylene
terephthalate (PET) and MRF film plastics.
4. Assessment of Material Recovery Processes: Researchers obtained MRF audit data
conducted by one of the largest counties in Florida, in terms of population and waste
generation. Based on the process flow of the single stream MRF, efficiency of each process
and sort results of each recyclable stream were analyzed using mass balance analysis. Based
on the results, researchers found the most common types of contaminants in each recyclable
stream and the most inefficient processes in the separation and sorting of recyclables. Here,
the efficiency rates of Disc Screen 1 and Eddy Current Separator were remarkably low. Disc
Screen I, which separates OCC from other types of fiber, had an efficiency rate of 45
percent, and Eddy Current Separator, used to separate aluminum cans from the stream, had
an efficiency rate of 75 percent. With respect to the sorting results, the lowest recovery rates
among colored HDPE materials and mixed glass were found to be 65 and 70 percent,
respectively.
XIII
Page 30 of 67
Agenda Item #6.
RESEARCH TEAM
Nurcin Celik (P.I.) is an Associate Professor in the Department of Industrial
Engineering at the University of Miami. She received her M.S. and Ph.D.
degrees in Systems and Industrial Engineering from the University of Arizona.
Her research interests lie in the areas of integrated modeling and decision
making for large-scale, complex, and dynamic systems such as solid waste
management systems and electric power networks. She received the
Presidential Early Career Award for Scientists and Engineers in 2017 from the
White House. This is the highest honor bestowed by the U.S. government on
outstanding scientists and engineers beginning their independent careers. She has also received
many other awards including the IAMOT Young Investigator Award (2016), Eliahu I. Jury
Career Research Award (2014), AFOSR Young Investigator Research Award (2013), University
of Miami Provost Award (2011, 2016), IAMOT Outstanding Research Project Award (2011),
IERC Best Ph.D. Scientific Poster Award (2009), and Diversity in Science and Engineering
Award from Women in Science and Engineering Program (2007). She can be reached at
celikkmiami.edu.
Haluk Damgacioglu is a Ph.D. candidate in the Department of Industrial
Engineering at the University of Miami. He received a B.S. and an M.S. in
Industrial Engineering from the Middle East Technical University in Turkey.
He has received several awards including the 2017 IISE Gilbreth Memorial
Fellowship Award, 2016 University of Miami GSA Academic Excellence,
Leadership and Service Award and 2015 ICCS Best Workshop Paper Award.
His research interests lie broadly in application of dynamic data driven
adaptive simulations and simulation optimization under uncertainty. His email
address is haluk.damgacioglukmiami.edu.
Lea Perez is an undergraduate student studying Industrial Engineering and
., minoring in Architecture at the University of Miami. She graduated from
�. g y
Design & Architecture Senior High in the Design District of Miami, where she
studied fashion design. She is presently Internal Affairs Chair and member of
Tufaan, UM's premier South Asian aca ella group. Her interests include
R p pp g p
designing systems in areas like fashion and architecture that give priority to
i' sustainability and the environment. She can be reached at 1mp72kmiami.edu.
XIV
Page 31 of 67
Agenda Item #6.
DISSEMINATION ACTIVITIES
JOURNAL PAPERS AND BOOK CHAPTERS:
1. Damgacioglu, H., Perez, L., Celik, N. "Assessment and Evaluation of Inbound
Contamination in Single Stream Recycling Systems — A Case Study of Florida" working
journalpaper.
2. Damgacioglu, H., Hornilla, M., Celik, N. "Assessment and Evaluation of Outbound
Contamination in Single Stream Material Recovery Facilities — A Case Study of Florida"
working journal paper.
PRESENTATIONS AND SITE VISITS:
1. TAG I Meeting: The first TAG meeting took place on July 14, 2017 at the McArthur
Engineering Building of the University of Miami with 13 attendees. A conference call for
those who wanted to attend the meeting remotely was established. Several comments made
during the first TAG meeting were recorded.
2. TAG II Meeting: The second TAG meeting took place on March 9, 2018 at the McArthur
Engineering Building at the University of Miami with 14 attendees. A conference call for
those who wanted to attend the meeting remotely was established. Multiple comments were
offered during the second TAG meeting.
TAG MEETINGS:
The project team hosted two TAG meetings on July 14, 2017 and on March 9, 2018.
WEBSITE: The team created and posted an enhanced website describing the project, accessible
at http://www.coe.miami.edu/simlab/swm.html.
Xv
Page 32 of 67
Agenda Item #6.
1 INTRODUCTION
Over the past two decades, the quantity of municipal solid waste produced in the United States
and particularly in Florida has increased significantly (by over 73 percent), outpacing all other
nations (Environmental Protection Agency, 2017). In order to deal with this burgeoning waste
stream, as well as a multitude of other environmental, economic, and social factors, different
curbside recycling programs, including single stream recycling (SSR), have been introduced to
practice over time. SSR has become increasingly popular due to its capabilities to better attract
the participation of communities in recycling while reducing the cost of collection of curbside
recyclables compared to multi or dual stream recycling. In SSR, all recyclables are separated
from municipal solid waste into a sole compartment in a collection truck, where payloads and
costs are significantly reduced. According to an American Forest and Paper Association report,
in 2005, 22 percent of the United States population had access to SSR, but by 2014, that had
grown to 73 percent (AF&PA, 2015) as shown in Figure 1. Similarly, 77 percent of material
recovery facilities (MRFs) built in the United States after 2007 are equipped to process single
stream (Berenyi, 2015). This surge in popularity is also evident in Florida where, based on our
investigations, at least 24 out of a total of 67 counties have currently operating SSR programs in
place including Brevard, Dade, Escambia, Marion, and Orange counties.
100
0
80 70 73
ct
60 49
53
o >, 42
40 34
° 22
18
20 8..... .... ....
12 ....
7 7
1 1
�fffffa ■
0
2005 2007 2010 2014
Two or More Streams Single Stream Including Galsss
Single Stream Excluding Glass Combination
Figure 1. Distribution of paper/paperboard collection techniques in the United States (AF&PA,
2015)
The advantages of SSR systems are comprised of lower worker injuries on collection routes,
reduced collection costs due to single-driver trucks, increased participation rates by residents,
and higher volumes of collected recyclables. However, SSR systems may operate at the expense
of higher volumes of materials requiring pre-sorting at regional MRFs, as well as higher
contamination rates in the materials directed to the mills (Lakhan, 2015). To this end, glass,
plastic bags, and the increasing amount of food-waste in the incoming comingled stream pose
significant challenges for MRFs (Cascadia Consulting Group, Inc., 2006). Small glass pieces and
shards may get stuck in paper, cardboard, and plastic items, making them unrecyclable. Glass
also reduces the life of the equipment at the MRFs intermediary processors and increases the
operational costs for increased facility downtime. The impact of the damage leads to a loss in
labor for repairs, new parts and equipment, and broken pieces of glass shards, which present
safety hazards for MRF employees. For instance, Shoreline Recycling & Transfer Station in DC
estimated that the damage caused by glass shards costs their facility about $500,000 per year,
solely for replacement of their screen basket valves (Cascadia Consulting Group, Inc., 2006).
1
Page 33 of 67
Agenda Item #6.
On average, 40 percent of glass from single stream collection winds up in landfills and another
20 percent constitutes small broken glass ("glass fines"), used for low-end applications. Only 40
percent remains to be recycled into containers and fiberglass (Kinsella & Gertman, 2007). In
contrast, 90 percent of the mixed glass from dual stream recycling (DSR) systems is recycled
into containers and fiberglass, and the remaining 10 percent comprises glass fines used for low-
end applications, with nearly nothing being sent to landfills (see Figure 2). The broken glass
adversely affects MRF operations as it leads to higher rates of residuals and degraded product
quality. In the recycling supply chain, receiving poorly sorted materials from a processor
discourages manufacturers from expanding upon or investing in new recycled product
manufacturing capacity. Additionally, the manufacturer may return to using raw virgin resources
(Kinsella & Gertman, 2007). The low quality of recovered materials may further lead to
imperfections in the finished products and low customer satisfaction, resulting in lower demand.
Percentage
100
80
60
40
20
0
Single Stream Dual Stream
Landfill Glass Fines Recycled(containers and fiberglass)
Figure 2: End-market for collected glass (Kinsella& Gertman, 2007)
To this end, we investigate the extent to which recovered materials are contaminated and
evaluate if SSR programs have a significant role on the contamination of recycled materials. We
also explore new and emerging advanced material recovery processes that may help with the
issue of contamination in recovered materials. The remainder of this report is organized as
follows:
• Section 2 provides background information, along with a review of works in the
literature.
• Section 3 details the considered material recovery facilities.
• Section 4 explains the specifics of the collected data, details of the statistical analysis
methods conducted, and the results obtained in this study.
• Finally, Section 5 discusses the conclusions drawn in this work.
2 BACKGROUND AND LITERATURE REVIEW
In SSR systems, all recyclables are separated from municipal solid waste into a sole
compartment in a collection truck. Amongst the curbside recycling programs, SSR has been a
growing trend in the last decade due to its ability to reduce collection costs, increase the amount
of collected material and increase participation. SSR accomplishes this by avoiding a
cumbersome sorting process at the individual level (i.e., customers use a single all-purpose bin
2
Page 34 of 67
Agenda Item #6.
for recycling) (Wang, 2006), making the process more consolidated and user-friendly. However,
this system does not necessarily lead to a higher recycling rate since improperly-sorted materials
may be mixed with other recyclables and go on to contaminate the recycling stream.
Contamination in SSR systems has been found to result in down-cycling of the products
(Chameidas, 2013). Contaminated recyclables pose serious problems for processors and end-
users such as paper mills and glass manufacturers. These problems include but are not limited to
equipment damage, lower quality in recyclables, and frequent downtime due to equipment
cleaning and repair.
Composition and quantity of contaminants vary among communities and waste collection types.
Beck (2006) analyzed this in different types of MRFs in California, including single stream,
multi stream, and mixed waste facilities. Results obtained from 83 MRF lines have shown that
multi stream processing facilities have the least inbound contamination rate (6 percent) with less
moisture and food contamination within the fiber material than expected. Comparatively, the
average contamination rate in the incoming material for single stream was found to be 14
percent, with a range of 2 to 50 percent. Beck (2006) also provided a contaminant
characterization of different MRF types in California in terms of the average proportion of each
material type by weight in the total contaminant weight, as shown in Figure 3.
35%
30%
t 2 5%
20%
15%
10%
5%
C&D Paper Glass Metal Electronics Plastic Organic
Single Stream ■Multi Stream ■Mixed Waste
Figure 3. Residual characterization for each MRF type in California (Beck, 2006)
While overall profiling of these recycling systems provides valuable insights for the
improvements and investments on MRF,
the Beck (2006) study is limited to the 68%Commingled 0.33% 0 15%
incoming stream of MRFs. In a related
Stream 3.40/o
and complementary study, Department
42%Commingled 0.10%
of Ecology of State of Washington 6.30j0%2010 studied the effectiveness ofStream 5.70%
sorting each material in the commingled 100%Curbside
collection systems. As a result of this Sort 0.25/
study, on average, approximately 93 to
96 percent of recyclable paper and 84 Glass Rate Prohibitives Outthrows
percent of plastic containers were found Figure 4. Effects of single stream on NORPAC Paper
to be properly sorted and sent to correct Mill (Morawski, 2010)
3
Page 35 of 67
Agenda Item #6.
markets, while 14 percent of plastic containers were found to cause contamination in the paper
that was being sent to paper mills, and the remaining 2 percent were disposed of immediately.
On average, one-third of aluminum cans and two-thirds of aluminum foil were found to be
improperly sorted and mixed with other materials. Despite these occurrences, the processing
facilities were shown to sort the materials quite effectively (i.e., on average, 92-94 percent of the
materials were transferred to the proper markets). However, significant loss may occur at end-
user facilities such as paper mills and recycled glass processors as well as glass bottle
manufacturers even if the level of contamination is small. For instance, only 0.46 percent of glass
contaminants in the total material stream caused NORPAC Paper Mill an additional $360,000
per year (Morawski, 2010). Figure 4 further shows that this facility experienced drastic increases
in the rates of out-throws, prohibitive materials, and glass contamination once its suppliers
switched to single stream collection.
Some studies in the literature focus on the economic impacts of high rates of contamination on
the sorting facilities. For example, even though commodity prices were declining (Robinson,
2014), Waste Management Inc.'s processing costs increased by 20 percent between 2012 and
2014, due to contamination in single stream collection and low-quality recyclables. A similar
study conducted by McCormick (2015) concluded that in Florida, the cost of inbound
contamination to the MRFs was estimated to be around $125 per ton between 2013 and 2015. In
another study, Moreau (2015) investigated the Chittenden County MRF in Vermont. The facility
was built as a dual stream facility in 1993 and converted into single stream in 2003, and its glass
processing system was also updated in 2016. Residue rates (including the non-recyclable
materials in the incoming stream and the recyclables lost during the sorting process) were
measured before and after the conversion to single stream (see Figure 5). Following the
conversion to single stream, the amount of materials collected as well as the residue rates
increased simultaneously. However, the residue rates diminished immediately after an equipment
upgrade in 2014, suggesting that the way sorting is handled in MRFs also has an impact in these
rates.
5000
°
; Single Stream °
45,000 Dual Stream ' � i Upgrade 9%
40,000 8%
35,000 ; ; 7%
30 000
6%
° 25 000 ' 5%
20,000 ; 4%
1500 ; 3%
►� 1000 2%
5,000 1%
0 0%
i
O\ O� d1 41 d1 O O O O O O O O O O O O O O O O O
N N N N N N N N N N N N N N N N N
Fiscal Year
Figure 5. Chittenden County residue rates before and after their switch to single stream (Moreau,
2015)
4
Page 36 of 67
Agenda Item #6.
Several studies (AF&PA, 2013; Conservatree, 2016; Yasar et. al., 2017) have concluded that the
SSR-driven paper recycling industry faces contamination issues from prohibited materials like
glass, plastics and other polymer materials. The glass recycling industry suffers from similar
problems. Even though glass is 100 percent recyclable and does not lose its quality in the
recycling process (GPI, N.D.), much of the glass recycling process is hindered by contaminants
in the commingled stream.
Cullet and glass fines are the two main types of recovered glass. Cullet is a higher-grade product
(worth $100-$149 per ton) that is produces new glass when it is mixed with virgin material in
glass manufacturing furnaces (Sustainability Victoria, 2014). Glass cullet is defined as
contaminant-free container glass, which is sorted by color (clear amber, and green) and is
expected to meet particle size specifications before it is used in glass manufacturing (Cattaneo,
2010). The other main type of recovered glass is glass fines, which are very small pieces of glass
that are used for non-container glass products. The collection and processing of glass using SSR
poses the following challenges to the recycling industry:
• SSR causes glass to break into very small pieces that cannot be sorted with currently
available technologies in MRFs.
• To be reused, glass cullet needs to be sorted by color, and the optical color sorting
technologies require high capital investment.
• Contaminants such as non-recyclable material, metal, paper, plastic, etc. in glass Gullet
decrease the quality of recovered glass, cause equipment damage, and increase the cost
for glass manufacturers and MRFs.
Table 1: Construction application of glass cullet (GPI, N.D.)
Application Cullet Debris Recycled Glass Fines
Content Level 100%
Retaining wall(structural 0%-30% 5%
ackfll) 50%
Retaining wall 100% 10%
(Non-structural backfill) 0%
Utility Bedding 100% 5% Single Dual Deposit
Embankments 0%-30% 5% Stream Stream Return
Foundation Drainage 100% 5% Figure 6. End markets for lass collected from
Pavement Base 0% 5% g g
different systems (DSM Environmental Services,
Trench and Foundation 2011
Back�ll 100% 5%-10% )
Contaminated glass cullet obtained from SSR is down-cycled and cannot be used for new glass
production. Luckily, the glass container industry has seen an increase in average glass cullet use
from 25 percent to approximately 34 percent since December 2008 (Bragg, 2015). Table 1
represents the glass cullet purity and usage in construction.
Some studies in the literature analyze the impacts of different collection systems on glass
recycling. For example, DSM Environmental Services (2011) observed that glass from source
separated drop-offs are often found to be less contaminated than the glass collected from dual or
5
Page 37 of 67
Agenda Item #6.
single stream. At Strategic Materials, Inc., which is the largest glass processor in North America,
98 percent of color sorted glass from deposit return collection and 90 percent of mixed glass
from dual stream collection are used for containers and fiberglass, while only 40 percent of
commingled glass from single stream collection can be recycled into containers and fiberglass
(see Figure 6) (Cattaneo, 2010). In a related study conducted for the State of Victoria in
Australia, researchers analyzed glass loss across the recovery process (Sustainability Victoria,
2014). They found the highest loss occurred due to splintering during collection and compaction
of recyclables (see Figures 7 and 8). According to a Container Recycling Institute (CRI) report,
breakage during collection and processing resulted in lower-grade glass fines (CRI, 2009). Then,
DSM Environmental Services (2011) analyzed the glass losses for different types of glass
collection methods in Ohio. Eighteen percent of glass collected is lost during the Optical Sorting
process after commingled collection. Source separation by color resulted in the lowest glass loss
rate with 1 percent (Figure 8).
0 1 30
3 0A�o % 3 25
24 0. 20 17
r
% 15
26 18 10 7 6 4 4
5 3 3 2 2 2 2 2 2 2
N % 0
Source Separated `a, 1`a ° `a5�1�`a�1� ° `a���� o�� °����,��5
■Loss to Landfill ,�.��1�° O�� � �,�o��o ono boo fie,� .155�c�
I Three Mix �����`� e �� �'
■Breakage During Collection
Commingled
Figure 7. Glass losses Figure 8. Glass Figure 9. Number of cullet processors in 35
across the recovery losses in Ohio by states (GPI, 2018)
process (Sustainability collection system
Victoria, 2014) (CRI, 2009)
The most important users of recycled glass are glass container manufacturers. In one study,
Goodman (2006) interviewed recycled glass end users in Minnesota about contaminant types in
their feedstock, operational problems caused by contamination, and the change in the quantity of
recycled cullet that they receive. Ceramics, pottery, Pyrex, and mixed glass are listed as the most
problematic contaminants for their operation. According to a GPI report (GPI, 2018), Florida is
amongst the top 10 states in U.S. with respect to number of glass processing facilities (see Figure
9). However, Florida's average glass recycling rates have decreased since the number of SSR
systems in Florida has increased over the past decade (FDEP, 2017). Collier (2005), Miami Dade
(2008), Broward (2009), Brevard (2009), and Charlotte (2010) are examples of counties in
Florida that have made the switch to SSR. The average recycling (AR) rates and the average
glass recycling (AGR) rates of these counties before and after switching to SSR are shown in
Figure 10, respectively. Here, all five counties have experienced an increase in AR with a
decrease in their AGR after switching to SSR. Investigating this issue, in this work, we evaluate
if the SSR programs have a significant role on the contamination of material from broken glass
and other non-recyclables.
6
Page 38 of 67
Agenda Item #6.
■ AGR % before SSR ■ AGR % after SSR
■ AR % before SSR Ali °o after SSR
■
2005 2008 2009 2009 2410
Collier ?vlimni Broward Brevard Charlotte
Figure 10. Average recycling rates (AR) and glass recycling rates (AGR) of counties that
switched to SSR in Florida(FDEP, 2017)
3 MATERIAL RECOVERY FACILITIES
Material recovery facilities (MRFs) are one of the critical components in the integrated solid
waste management system. Understanding the way MRFs operate and handle challenges given
their recent technological and organizational developments is critical for establishing an effective
solid waste management system. MRFs are the processing facilities where received materials are
separated, sorted, and prepared for marketing either to end users (manufacturers) or to other
facilities for additional processing. The design of the MRF processing depends heavily upon the
type of the waste collection system they use. Below are the main types of waste collection
systems and their descriptions.
• Source separated: Residents sort the recyclables by type for collection. The primary
purpose of the facility is to remove contaminants often by baling, flattening, or crushing
prior to sending them to their respective markets.
• Dual stream: Residents sort the materials in two separate streams, generally fiber
(newspaper, magazines and catalogs, mixed paper, cardboard, etc.) and commingled
containers (plastic, glass, metal, and sometimes aseptic containers). Dual stream systems
utilize a combination of automated equipment and manual sorting to separate and sort the
received materials.
• Single stream: Residents put fiber and commingled containers in the same bin for
collection. The separation of materials into two separate streams (fiber and containers)
occurs during the first stages of processing. The rest of the processes are similar to those
of dual stream.
• Mixed waste: Unsegregated mixed waste is separated using various technologies.
Collected waste is first dumped on a tipping floor where recyclable materials are then
processed using equipment similar to that of single stream MRFs.
7
Page 39 of 67
Agenda Item #6.
While each type of MRF employs common design principles and sequencing in the configuration
of equipment and labor, each one has different additional steps and processes for sorting and
recovering waste streams. Seventy-seven percent of all MRFs built after 2007 in the US are
single stream MRFs due to increasing popularity of the single stream collection system (Berenyi,
2015). These new single stream plants process, on average, 225 tons of recycled materials per
day with multiple mechanical and optical technologies to achieve high levels of sorting
efficiency(Berenyi, 2015).
3.1 Material Definitions
The understanding and categorization of the materials (both recovered and discarded) is essential
to the proper comprehension of the functioning of any MRF. In this report, we adopt the same
material definitions used in waste composition studies provided by Kessler Consulting Inc
(Kessler Consulting Inc, 2009, 2010, 2013, 2014a, 2014b, 2016). The entirety of the materials
processed in a recycling facility are organized into five categories. These categories are paper,
plastic, glass, metal, and residue. The recoveries of plastics occur throughout the entire MRF
processing and can be seen in Table 2.
Table 2. Plastic material definitions
D
Plastic Sub-Category escription/Exampies
PET Bottles(SPI#1) Clear and colored plastic bottles coded PET#1 such as soda bottles,
Polyethylene terephthalate water bottles label with SPI #1. Does not include loose caps.
Natural HDPE Bottles (SPI #2) High Clear/natural plastic bottles coded HDPE #2 such as milk jugs,
density polyethylene vinegar bottles and gallon water bottles. Does not include loose
caps and lids.
Colored HDPE Bottles (SPI #2) High Pigmented plastic bottles coded HDPE #2 such as detergent,
density polyethylene shampoo,and orange juice bottles. Does not include loose caps and
lids.
Non-Bottle PET Clear and colored plastic items labeled PET #1 such as clamshell
containers, frozen food trays, disposable cups and other items
labeled PET#1.
Non-Bottle HDPE Wide-mouthed tubs and containers labeled HDPE #2 including lids.
Examples include yogurt cups, margarine tubs, Cool Whip°tubs and
other non-bottle HDPE items.
Expanded Polystyrene(Styrofoam) Styrofoam° containers such as egg cartons and clamshell food
containers.
Mixed Plastic Containers All plastic containers coded #3-#7, such as containers, pill bottles,
Arizona Iced Tea TM gallon jugs, etc.
Bulky Rigid Plastics Consists of non-container rigid plastic items such as plastic drums,
crates, buckets, baskets, toys, refuse totes, and lawn furniture,
flower pots, laundry baskets, and other large plastic items. Does not
include electronic toys.
Plastic Film(Residue) Loose and bagged plastic bags, garbage bags, shrink wrap,
resealable bags, etc.
Non-Container Expanded Polystyrene Non-container Styrofoam° such as packaging peanuts and other
(Styrofoam)(Residue) packaging.
8
Page 40 of 67
Agenda Item #6.
The next tables are organized in the order in which each material is recovered in a typical MRF.
The first type of material to be recovered is paper (see Table 3), which is immediately followed
by glass (see Table 4). Because glass is crushed in the early stages of the process, and then
optically sorted by color during the SSR process.it does not necessarily have multiple sub-
categories. Still, we defined a table for it in order to provide details of the collected materials.
Lastly, there are metals, or ferrous materials (see Table 5).
Table 3. Paper recyclables recovered in MRFs
Paper Sub-Category I Description/Examples
Newspaper(loose or tied) including other paper normally distributed inside
Newspaper newspaper such as ads,flyers, etc. Newspaper found inside plastic sleeve will be
removed from plastic and sorted accordingly.
Corrugated Cardboard Brown "cardboard" boxes with a wavy core(no plastic liners or packaging
OCC Styrofoam'). Does not include small pieces of OCC within shrink-wrap plastic such
(OCC) as those from a case of bottled water.
Waxy Cardboard All wax coated OCC will be sorted and weighed separately from non-wax OCC.
Printed or unprinted paper including white,colored,coated and uncoated papers,
Mixed Paper
manila and pastel colored file folders magazines,telephone books,catalogs,
paperboard,chipboard, brown paper bags, mail, bagged shredded paper and other
printed material on glossy and non-glossy paper.
Loose Shredded Paper Loose shredded residential mixed paper or newspaper.
Aseptic Containers Gable top milk cartons,juice boxes,and other similar containers.
Table 4. Glass recyclables recovered in MRFs
Glass Sub-Category Description/Examples
Mixed Glass Containers
and Jars(Glass Clear,green,and amber glass bottles and jars as well as broken glass pieces larger
than half a square inch.
Containers)
Table 5. Metal recyclables recovered in MRFs
Metal Sub-Category Description/Examples
Aluminum Cans Aluminum soft drink, beer, and some food cans.
Aluminum Foil and Pie
Plates Aluminum foil, pie plates,and clean catering trays.
Tin/Steel Cans
Tin-plated steel cans, usually food containers,and aerosol cans, including labels
and steel caps.
Scrap Metals Rejects Non-container ferrous scrap metals such as pipes,coat hangers,and miscellaneous
scrap metal.
The previous tables contain the materials recovered during the SSR process. The residues that
are discarded during the recycling process are described in Table 6. These are non-recyclables
that are sorted out by any of the different recycling processes.
9
Page 41 of 67
Agenda Item #6.
Table 6. Residues found in MRFs during single stream recycling
Residue Sub-Category Description/Examples
Materials not included in the other categories,such as bagged garbage,fast food
Rejects
lids and straws, CDs and VHS tapes,composite materials,Christmas lights, hoses,
electronics, recyclable items full of food (non-liquid), loose plastic caps and lids,
or plastic cutlery and plates.
Grit All items that fall through a half inch mesh.
Liquids All liquids found within recyclable containers.
3.2 Single Stream MRF Process Flow
Single stream MRFs are capable of recovering fiber, glass, metal and plastic from a commingled
recyclables stream with the appropriate equipment and design. Based on our literature review
(Kessler Consulting Inc., 2009; Combs, 2011; Pressley et al., 2015), synthesis of our discussions
with the counties in Florida and our experience from previous studies, we demonstrated a typical
single stream MRF design that is capable of paper, glass, plastic, and metal recycling, as shown
in Figure 11.
• Paper recycling process: The collection truck unloads the collected recyclables on the
tipping floor and a claw pushes the material into the Drum Feeder. The Drum Feeder
distributes the collected material to the Conveyor, where large items and prohibitive
materials that can damage downstream equipment are removed by manual sorting.
During manual sorting, plastic films (e.g., plastic bags) are also removed. Then, the
remaining recyclables continue to Disc Screen 1, which separates the OCC. Remaining
mass travels to Disc Screen 2, which removes ONP. Residual material from Screen 2
continue to the Scalping Screen, which then separates fiber from containers and other
materials. The fiber streams from Disc Screen 2 and the Scalping Screen proceed to a
manual sort to remove contaminants before the baling process. After the baling process,
the paper bales are sent to paper mills and processors.
• Glass recycling process: After glass and other containers are separated from fiber
materials by Disc Screens, all containers continue to the Glass Breaker Screen. The Glass
Breaker Screen is designed to separate glass from other materials like plastic bottles, milk
cartons, and metal cans. It allows crushed glass to fall between discs, while fiber and
remaining containers (i.e., plastic, aluminum, and metal) pass over the top. It also
provides more the uniform sizing of glass cullet required for optical separation and
subsequent remanufacturing. Then Air Knife eliminates light contaminants (i.e., fiber,
light debris, etc.) to maximize the end product quality. In the last step, Optical Sorter
identifies pre-determined material(s) using optical technology (e.g., cameras, lasers,
sensors) and sorts glass cullet by color.
• Plastic and metal recycling process: The materials passing over the Glass Breaker
Screen are conveyed to an Optical Sorter that recovers PET. The remaining stream is
conveyed to a second Optical Sorter that removes all colors of HDPE. The remaining
stream proceeds to a magnet for ferrous recovery. The material remaining after the
magnet proceeds to an Eddy Current Separator for aluminum recovery. The remaining
residual stream goes to a manual sort, where any recyclable materials missed by the
separation equipment are recovered by pickers and sent to the 2-way Baler. The
aluminum, ferrous, HDPE, and PET streams are separated and stored in cages prior to
baling. Each stream is inspected for contaminants prior to baling.
10
Page 42 of 67
Agenda Item #6.
I Plastic&Metal Glass recycling Process I
2-Way
I Recycling � I
I Baler Process Glass 1
Flint
Optical
l<11 u'Ll _ _ Glass �
I ------------------------- ,.------------------------, ,r Knife 'lass
Sort Green I
I Al` Fe HDPE. :PET Sorter
c;ll, Glass
Eddy Magnetic Glass Amber �
Current Meal Optical Optical Breaker
Sorter Sorter
Se Sorter Screen I
Glass and
'
Other
Film
Containers
Conveyor 1 Paper 1
� Dnum .1nual Disc Disc Scalping recycling I
Input Feeder Sorg 1 Screen I Screen Screen Process
1 I
I . i
Manual
Sort
Residual
Mixed,
Lead I LCaer Mixed Paper 1
I 1
Mass`Re naini ng' 1-Way
1
I
Mass`Ret ox,ed" I Baler 1
i-
Figure 11. Single stream MRF process flow with equipment configuration to recover fiber, glass,
plastic and metal materials
4 DATA COLLECTION AND ANALYSIS
Researchers collected various data regarding the inbound contamination rates from waste
composition studies provided by 15 counties around Florida, the outbound contamination rates
from four currently operating facilities in Florida, and the assessment of existing processing
technologies from one MRF audit.
4.1 Inbound Contamination Analysis
One of the major tasks in this work is to investigate if the contamination rates of the inbound
stream have been impacted by switching to S SR programs. In order to identify and evaluate this
impact, the contamination rates in different recycling programs should be compared. Researchers
first collected the data regarding the inbound contamination rates by bringing together the
recyclable composition studies conducted for the inbound waste stream in the following counties
and major cities in Florida: Alachua, Brevard, Broward, Citrus, Escambia, Hillsborough, Indian
River, Lee, Leon, Marion, Okaloosa, Pasco, Santa Rosa, Sarasota, Seminole, Fort Lauderdale,
Margate, and Lauderdale by the Sea. The waste composition studies provided individual sample
results showing the breakdown of recyclable materials, as well as the contamination rates. DSR
data consisted of the contamination in the paper stream. For confidentiality purposes, county
names are removed in the rest of the report.
After assessing the weights of non-recyclables in the collected waste composition studies, we
obtained 170 samples from SSR and 45 samples from DSR for inbound contamination rates. We
11
Page 43 of 67
Agenda Item #6.
conducted descriptive statistical analysis and analysis of variance (ANOVA) to investigate if
there is a statistical significance of the difference between the mean contamination rates of the
inbound SSR and DSR systems. In order to visualize the descriptive statistics of the data
collected for SSR and DSR systems, we built a boxplot combining the samples of both systems
along with their outliers (see Figure 12). Here, the standard deviation (8.97) and mean (18.54) of
the SSR contamination dataset were found to be higher than that of the DSR contamination data
set (3.08 and 3.89, respectively). Comparison of the two boxplots reveals the difference in rates
between the two systems in terms of mean, median, and overall distribution of outliers. It is also
shown that SSR contamination rates were more spread out over a wide range of values than were
DSR contamination rates. A higher standard deviation in the SSR data denotes the differences in
implementation and success rates of variations of SSR programs and people's contributions
towards it. For example, some may assume that even non-recyclable materials can be thrown in
SSR bins (i.e., researchers found diapers, food waste in the SSR bins). Histograms for the SSR
and DSR contamination rates were plotted to better represent the wide range of SSR
contamination rates (see Figure 13). The contamination rate of one SSR sample was 48.2 percent
which means that almost half of the collected materials were actually non-recyclables. However,
approximately 3 percent of samples promised that even in SSR systems, low levels of
contamination rates could be achieved. Based on the significant difference observed through
these boxplots and histograms, we have conducted further detailed analysis on the data obtained
as the following.
ILL
4
i
i
DS SSR
Figure 12. Boxplot of SSR and DSR contamination data
12
Page 44 of 67
Agenda Item #6.
40 20
3
05
Q O9
1
Q 1 5
- r r,-10
V
SSR DSR
Figure 13. Histograms of SSR and DSR contamination rates
We use ANOVA to test if there is statistically significant evidence that shows contamination
rates in SSR systems are higher than the rates in DSR systems. ANOVA primarily assumes that
the data of each group are normally distributed. Since the histogram plot of DSR samples did not
follow a bell-shaped curve, we first conduct a normality test on the collected data for each group.
Here, we conducted the Anderson-Darling (AD) normality test, a non-parametric test that comes
in handy when there are a limited number of samples available. AD takes both the means of
samples and differences in shape and variability of the distributions into account. Normality test
plots and statistics for SSR and DSR samples are shown in Figures 14 and 15, respectively. In
these figures, blue points represent each sample and red lines shows the normality test statistic
curves. It is expected that all the samples lie on the red line. Based on AD normality test
statistics, p-values of both groups were smaller than 0.005 which means that both SSR and DSR
samples comply with the normality assumption of ANOVA.
Once the available data passes the normality test as described, an ANOVA test is conducted to
differentiate between the sources of variances amongst these available data. ANOVA is a
widely-used technique to compare the ratio of between-group variance to within-group variance
using the F-critical ratio test statistic. The detailed information about ANOVA is provided in
Appendix A. Here, our null hypothesis was "SSR and DSR systems have equal contamination
rates" while the alternative hypothesis was "the contamination rates in these systems are
different". Our ANOVA resulted in very high F-ratio (112.954) which corresponds to very low
p-values (almost equals to 0) as shown in Table 7. The ANOVA results rejected the null
hypothesis at more than 0.999 (a = 1 — p — value) significance level. In other words, the
ANOVA analysis below reveals significant statistical evidence in rej ection of the hypothesis that
the contamination rates in SSR and DSR systems are the same and supports the alternative
hypothesis that the contamination rates in single stream recycling are higher than those in dual-
stream recycling.
13
Page 45 of 67
Agenda Item #6.
Probability Plot of Single Stream Contamination Probability Plot of Dual Stream Contamination
Non-nal Normal
99
Mean 1954 Mean 3.898
99 StDev 8.973 StDev 3.084
N 170 all 95 N 45
•
AD 1.505 AD 1941
P-Value<0.005
95- P-Value<0.005 �
90 8d
80 70
0 60 60
� 50 � 50
� 30 � AO 20 L
10 L 20-
5 —1— F-— ID
1 — —
5-
0-1-
-ID 0 M 20 30 40 50 1-5 0 5 10 15 I
Single Stream Contamination Dual Stream Contamination
Figure 14. Results of AD normality test for Figure 15. Results of AD normality test for
SSR contamination data DSR contamination data
Table 7. ANOVA results of SSR-DSR comparison
Source of variation
Sum of Degrees of Mean F (test P-value
squares freedom square statistic
q q )
Between Groups 5093.447 1 5093.447 112.954 0.000
Within Groups 3968.194 88 45.093
Total 9061.641 89
Once we analyzed the data in its totality as mentioned above, we further investigated the
difference in contamination rates of the SSR and DSR systems at a county level. Here, we first
plotted a combined histogram of contamination rates of each county as shown in Figure 16,
followed by some descriptive parameters as in Table 8. The names of the counties here are noted
as County 1 through 16 for confidentiality purposes. With respect to descriptive analysis, the
samples of Counties 11-16 are removed from some of our analysis due to their sample size (<5).
The combined histogram in Figure 16 visibly shows that DSR systems in general have lower
contamination rates than those of SSR systems. Of these, Counties 5, 8, and 9 (DSR)
demonstrate the lowest contamination rates at less than 5 percent, while Counties 3, 4, and 14
(SSR) demonstrate the highest contamination rates at more than 20 percent. From this histogram,
we can conclude a wider variation of contamination rates between counties operating with SSR,
with the largest variation occurring between Counties 1-4. On the contrary, County 10
demonstrates the lowest and most promising contamination rate of all counties operating with
SSR at just 5.98 percent.
14
Page 46 of 67
Agenda Item #6.
0.00
a.) 215.00
20.00
15.00
10.00
U
5.00
CU 0_00
Figure 16. Histograms of contamination rates of each county
Table 8. Descriptive analysis of contamination rates for each county
County Recycling Number of Mean Standard
Maximum Minimum
Index System Samples Deviation
n
County 1 SSR 45 16.85 7.02 38.65 6.83
County 2 SSR 33 19.80 8.74 48.20 6.70
County 3 SSR 24 28.22 8.11 42.50 13.80
County 4 SSR 23 22.33 7.75 34.80 7.50
County 5 DSR 21 4.42 3.17 12.50 1.20
County 6 SSR 16 11.98 5.24 23.20 4.20
County 7 SSR 16 13.46 6.11 25.20 3.70
County 8 DSR 14 2.82 2.13 7.05 0.57
County 9 DSR 8 3.51 2.31 7.80 1.00
County 10 SSR 5 5.98 4.95 14.22 1.76
County 11 SSR I 1 S n S 7.71 1111n n 7.59
County 12 DSR SSR 8.41 DSR 1.60
County 13 SSR 2 14.77 1.00 15.48 14.06
County 14 SSR 1 22.86 NA 22.86 22.86
County 15 SSR 1 7.70 NA 7.70 7.70
County 16 SSR 1 7.56 NA 7.56 7.56
Supplementary analysis on inbound contamination data was conducted to better understand the
variance in contamination rates among counties. Here, six SSR counties (Counties 1-4 and
Counties 6-7), which had more than 10 samples, were selected. A boxplot of the contamination
rates in the six SSR counties is shown below in Figure 17. Then, another ANOVA test was
designed to determine the statistical significance of the difference between the mean
15
Page 47 of 67
Agenda Item #6.
contamination rates of these counties (see Table 9). Here, the difference in the counties' inbound
contamination rates was found to be statistically significant at a 99 percent significance level. A
more detailed comparison was conducted using a multi-comparison test on these six S SR
counties. Our results (see Figure 18) indicated that Counties 3 and 4, with respective
contamination rates of 28.2 and 22.3 percent on average, have means significantly higher than
the other counties based on a 95 percent confidence level.
50
45
T
40
+ I
I I
� 35 I T
� I I
� I
30 I
o T I
I I
. 25 I
20 I I
c I I
V I I
15
� I
I I
10 I I
I I
5
County 1 County 2 County 3 County 4 County 6 County 7
Figure 17. Contamination rates in S SR counties
Table 9. ANOVA results of contamination on the mean contamination rates of six selected S SR
counties
Source of variation Sum of squares Degrees of Mean F (test P-value
freedom square statistic)
Between Groups 3799.5 5 759.907 13.63 0.000
Within Groups 8416.8 151 55.541
Total 12216.4 156
16
Page 48 of 67
Agenda Item #6.
County 1
County 2
County 3
County 4
County 6
County 7
5 10 15 20 25 30 35
Contamination Rates
Figure 18. Contamination rates in S SR counties
To further understand the differences between counties, our data was analyzed to find if any
correlation exists between contamination rates and the demographics of these six counties (see
Figure 19). These factors included population, population density, median age, average income
level, poverty level, White population percentage, Hispanic population percentage, and Black
population percentage. Contrary to popular belief(Kessler Consulting Inc. 2013; 2014a; 2014b;
EPA, 2015), our research showed no correlation between inbound contamination rates and
ethnicity, but a positive correlation between median age and contamination, and a negative
correlation between income level and contamination. Here, the counties with lower income
levels and higher ages can develop specific education programs and/or incentives to reduce the
contamination levels in the collected recyclables.
17
Page 49 of 67
Agenda Item #6.
iLI a.12 0.11• 0.70 -0.60' -0.010 -0.13 0.113 0.110
+r„ o • a --v v a a s
L
4 a
!'.!h#AA3J.f 1J-x.S.S} 9 v 4 4 q q ■ O
v ixa`aaa • 4 P q.' • a q
0.12 x -024 -0.20 0.07 0-91 -0.30 0.50 0-07
13
i III aaaa+a+i++
4 q • ■ O S q a +xa�ax-xa� _
0.11 -0 24 038 -0.71 0.06 -0.79 0.53 0.94
41_11_11_I
S q ° - # � q a ° -__O_ 7 , v-'` P -lam■ P BQ
0.70 4 -020 ° 0.30! -0 33 a -0.30 r• -0.01 ° -020 0.35 q
4 5
41J - _
. CPO 4 P O y O • 9 • P e p O O
15
-0,10 a.0 -02.1 -0.33, .... �14 0.76 p
T 1 -0.64
1 "' q .. _- , vxa.'.ta*a•.x o+jxxr4 , - . q`v e e•. q
L
4 4-
-0.01 C.91 0.06 a -0.30 -0.14 -a.59 0.72 _= 0 36
MR
14 _-
s
------------
12
-0,13 -0.3D -0 7.9 -0.01 0.76 459 _ .92 _-0,87
� 41_I
as
1.1 0.18 MC 0.53 -0 0 -0.71 0.72 -0..92 � 0J63
° . . c
sir 1 9 e' O 9 `9` -- -_ SJ g .+'� .:•R l
a,11 0.07 0.94 ' 035 -0.64 036° °-0.07 a.63 I"-10--
- r g
II
III '�_� U U 600 1 LI[ILI 0 4000 36 40 46 4.5 6 5,6 6 12 14
Cant Pop Don Age Inc X104 Pov 'Alice }Bsp Black
histograrnplot - county data- fitted line arcurve
Figure 19. County correlation matrix
4.2 Outbound Contamination Analysis
S SR affects not only the collection systems, but also the MRFs, as adaptation is needed to
process the commingled recyclables. As shown in the previous section, S SR systems had
significantly higher contamination rates compared to DSR. In this section, we assessed the
contamination levels in end products baled by MRFs. To conduct this assessment, we obtained
35 old corrugated cardboard (OCC) samples and 266 old newspaper (ONP) samples from audit
reports for four currently operating facilities in Florida. In our assessment, we only focused on
prohibitive materials in the outbound stream due to the scope of the study. Based on the collected
audit reports, glass, plastic, metals and residue were considered prohibitive materials for both
ONP and OCC streams. In our assessment, we compared mill standards of accepted
contamination rates and the number of samples in S SR that meet these standards. The assessment
of the contamination rates of outbound materials produced distressing results. Average rates
(weight of the contamination/weight of total sample) of prohibitive materials in the OCC stream
and the ONP stream were 5.12 and 7.13 percent, respectively. Figures 20 and 21 show how these
samples compared to allowable paper mill limits for each of the OCC and ONP samples.
Considering the paper mill standards are only 3 percent for OCC and 2 percent for ONP, these
results are disconcerting. With these standards in place, only 34 percent of OCC samples and 0.3
percent of ONP samples meet paper mill standards.
18
Page 50 of 67
Agenda Item #6.
14
l
l
C Sample Mean
Mill Standard
sampk!
Figure 20. OCC prohibitive materials analysis
16
14
amp[`Mean
�-%
4 d ill standard
1ti 1 '?G 25 411 4- ii
Sample's
Figure 21. ONP prohibitive materials analysis
Figure 22(a) shows that OCC samples were most frequently contaminated with residue and
plastics, which includes film plastics and high density polyethylene (HDPE). Similarly, Figure
22(b) shows ONP samples were mostly contaminated by residue and plastics, specifically
polyethylene terephthalate (PET) and MRF film plastics. The prohibitive material categories
with respect to SSR are defined as the following.
• Residues: Residues make up the largest portion of prohibitive materials in the outbound
stream. Residues in the end products of MRFs arise from the following: 1) collected non-
recyclables in the inbound stream (inbound contamination) and 2) collected recyclables
that are not separated out in the MRFs. As shown in Section 4.1, inbound contamination
rates were much higher in SSR systems compared to DSR. Also, since recyclables are
commingled in the SSR bins, residue rates in single stream MRFs are expected to be
higher than residue rates in dual stream MRFs.
• Plastics: Plastics are collected in separate bins in DSR, which naturally decreases the
amount of plastics in the outbound stream, compared to SSR. In single stream MRFs,
19
Page 51 of 67
Agenda Item #6.
sticky plastics such as film plastic and plastic bags were problematic in the fiber stream
separation process.
• Metals: Tin is the most problematic contaminant metal in fiber recycling. It makes up an
average of 80 percent of metal contaminants in fiber stream.
• Glass: The average rates of glass (weight of the glass/weight of total sample) in the OCC
and ONP streams were 0.04 percent and 0.2 percent, respectively. While these numbers
are perceivably very low, even a small amount of glass can cause significant equipment
damage in MRFs. For example, in the NORPAC Paper Mill case, 0.33 percent of glass in
the fiber stream resulted in $400,000 equipment damage (Morawski, 2010).
Moving forward, MRFs should improve process efficiency where the HDPE and PET are
separated out from the OCC and ONP streams. Since film plastics are generally considered non-
recyclable, the amount of collected film plastics can be reduced in the inbound stream by
educating the residents and through advertisement.
r
a) b)
Plastics
Glass
Metal
own
Residue
*%Mom
o
Figure 22. Types of contamination in corrugated cardboard (a) and newspapers (b)
4.3 Material Recovery Facility (MRF) Audit Analysis
In this study, MRF audit data (obtained from one of the largest counties in Florida in terms of
population and waste generation) was analyzed using a mass balance analysis. Mass balance
analysis represents the mass of each proportion of recycled materials (i.e., paper, glass, metal,
etc.) that passes through a process or equipment in the MRF, recorded in terms of mass per hour.
To the best of our knowledge, this study is the first to analyze the separation of efficiencies for
each part of the MRF process. The information that exists is based on shared information from
experienced MRF operators and examination of operating MRF equipment (Pressley et al.,
2015).
More specifically, the mass of the waste proportion, i, recovered by each process or equipment;
j, is calculated as follows.
removed separation
M)j = f.i X m P (1)
In Equation (1), m,,moved shows the mass of waste proportion i removed b e ui ment
ij p p Y q p I
separation represents the separation efficiency of equipment for waste proportion i, and mT P
fj,i p p Y1 p p i,l
demonstrates the incoming mass to equipment j for waste proportion i. The units used for all
analysis of mass throughput are megagrams (Mg). The mass of waste proportion i remaining
20
Page 52 of 67
Agenda Item #6.
after equipment j is calculated as in Equation (2) where mremaining is the mass of waste
proportion i remaining after equipment j.
mremaining = mTj,iP _ mremoved (2)
j,i
Understanding the effectiveness of SSR requires knowledge of process operation. The amount of
material recovered is what determines effectiveness in MRFs. Contaminants in the recovered
material may tamper with the quality of recyclables to the extent that the entire bale, and thus
each part of the process leading to the creation of the bale, is wasted. Details of the recovery
processes in an SSR system are shown in Figure 11, where the recovery process is divided into
three major categories boxed in by red-dashed lines. The first important category to note is on
the bottom right of the figure. This part of the process is paper and recovery, and it includes OCC
and mixed paper material. In the top right box is glass recovery and sorting. The last box is
plastic and metal recovery, and this includes PET and HDPE for plastic materials, and aluminum
and other ferrous materials for metals. Previous studies have been conducted in the literature on
the efficiency of SSR. A mass balance analysis from a past study (Presley et al., 2015), was used
to better understand the "separation efficiencies" within this process. Separation efficiencies
consist of the proportion of the material that is separated by each process of the SSR system. The
separation efficiencies presented on the aforementioned study can be seen in Table 10.
The main problem with the current understanding of MRF separation efficiencies derives from
the fact that it completely focuses on the amount of material recovered, without placing much
attention on the contamination found in such recovered material. As it has been expressed before,
contamination on baled materials may damage both the consumer of the recycled material and
the trust companies place on the viability of using recycled materials. Hence, a more detailed
analysis on contamination assessment is needed with respect to not only the quantity of
recyclables, but also the quality of them.
Table 10. SSR separation of efficiencies in percentage (Presley et al., 2015)
Waste Manual Disc Disc Disc Glass Optical Optical Optical Magnet Eddy
fraction sort/ Screen Screen Screen Breaker glass PET HDPE Current
Vacuum 1 2 3 Screen Separator
OCC 70 85 91
Non-OCC 85 91
fiber
Plastic film 90
HDPE 98
PET 98
Fe 98
Al 97
Glass 97 98
Using a mass balance analysis, the percentage of material removed by each process was
calculated based on MRF audit data. The results are shown in Table 11. In this table, green cells
21
Page 53 of 67
Agenda Item #6.
show the true separation rates of processes while the blank and orange cells represent the false
separation percentages. For example, the efficiency of Glass Breaker Screen was 95.11 percent,
and the Eddy Current Separator incorrectly separated out 25.59 percent of aluminum foil as
aluminum can. Here, orange cells indicate the most problematic materials in terms of false
separation rates of MRF processes. Additionally, the sort results of each recyclable stream in the
single stream MRF were obtained using equations (1) and (2) (see Table 12). In Table 12, green
cells represent the accepted material rate for the corresponding recyclable stream. For example,
the greatest acceptable materials percentage occurred with OCC and natural HDPE streams,
which were found to be 96.48 and 97.59 percent respectively. The lowest recovery rates were
found among colored HDPE materials and Mixed Glass, with acceptable material rates of 65.04
and 70.16 percent, respectively. The blank and orange cells indicate the contamination rates of
specific material in the corresponding streams. For instance, Other Mixed Plastics were the
major contaminant in the colored HDPE stream with a rate of 14.75 percent. Here, similar to
Table 11, orange cells highlight the important contaminants in the corresponding stream.
22
Page 54 of 67
Agenda Item #6.
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
WOO
cl o 0 0 0 0 0 0 0 00000000000000 0 0
0 00 0 0 0 0 0 0 00000000000000 0 0
� o0 0 0 0 0 0 0 00000000000000 0 0
O O O O O O O O O O O O O O O O O O O O O O O O
� o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 \ o 0 OF"
l� l� 00 l�
OF"
cM O*
WOO
�+ C� .
R*
o 0 0 0 0 0 0 0 0
0 0 0 0 � o 0 0 l� l� 00 0 0 � o o M o o � � o 0 0
M `O l� `O M .-� N r--a O N N r--a M 0o O
.� M M 00 l� a1 N oo ,� O
00 r--a O 00 `O
l� N O l� r--a
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
•� O O O M O o0 O `O 00
v
o\
y �, o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
y y O O O O O O oo O N O 0 0 0 O� O N IC
CD O
O O O O O O O O O O O O O O O O O 0c
INC N N O O r- N Vn o0 O O
MOM A O O O O `O `O r- N 0 0 0 M M l� 00 O O O
0000
� � w y � w o 0 0 0 0 0 \ o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 O O O N M O V) 0 0 `O O 00 M r--+ 0 0 O O
A A O O O O O O N N O l— M 0 O O O Ocl
O
6 6 oc o o �0000000000000 0 0
ct
OC)
w O O O O O (n O� N M O Ql O W N `O N 00 O `O O O
Q O O O O O O O N a1 00 O O O N 6 0 0 0 O O
C�
ct cd
� � � � •� � � O O � O ,-� O O O O � � � � '� � O N oo N O N O O
0 rA W yORO y � � O r- o 0 0 N M kn O `O o 0 o CIAoc o 0 0 0 0 o 0 0 0
•� N M O O� O M N O� N O O O A M Ok oc N O ,�_, �--+ O� q N O l� 00 M N O O O
0 0 0 0 0 0 0 0 0 0 0 \ o \ o 0 0 0 0
CJ o o 0 0 0 0 0 o 0 0 o 0 o N o N o 0 0 0 0
� N �-- `O `O N `o r--+ oc `C Oo r--+ oo O INCr-- V) oo O
.� •� .� Outoc
`O O o0 M M O `O r O
p y \ \ \ \ \ \ \ \ \ \ \ \ \
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00,� M O CA O O � O O M O 00
00
CIA � Q-) O N 0 0 0 0 0 O l—
ct
�
o 0 0 0 0 0 0 0 0 0 0 0
O ♦, ♦, o \ o 0 0 0 0 0 \ \ \ \ \ \ o o \ o \ \ o o
G� G� t N V) N kn � OM O O N M CA' O
Out CIA l� l� l� CIAO � O
p oo M O N O l� �
4 IC �
� P4 M M M M N N N M O N M M O 01
� M l M O M O O '--+ O O O
O O O O O `O O oc
O O� O O O O O O
•a� a A4a o0 0 0 0 0 0 0 o �rio � 0000000000 0 0
ct
cn
Ucr
4-j ct
ct
cn
v
ct
ct
ct O O N ct N ct U ct
ICI U U A-{ U � A-i �/1 �/1 U � M P4 M
Page 55 of 67
Agenda Item #6.
�r W�
•�•� o o c
0 0 0 0 0 0 0 0 0 0 0 0 0 0 � o 0 0
M '-- N M M 01 O `C 00 M .-� N C., 0\ O O O C
O O O O O -_ O O TOGO
0 c
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o O o 0 0 0 o c
. . . . . . . . . . . . . . . . . . . . . . .
O
v
M o 0 0 0
o 0 0 � o 0 0 0 0 0 o 0 0 0 0 0 0 0 o 0 0 0 0
y 00 00 � O
00 N M r- 00 �c M M � O
M M 00 V-)
r� r--� r--{
•� � o 0
01 00 M M O O O
O O O O O N O O O O M O O O O O N 0c M O 0 0 O
. . . . . . . . . . . . . . . . . . . . . .
\ \ \ \ \
� o 0 0 0 0 0 01 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o O
00
0 0 0 0 0 0 0 0 0 0 0 0 0
w 00 O
M O O N O 00 r- O M O O N ^� r- oO O N O O N
M 00 Vn C1 N O N C O
. . . . . . . . . . . . . . . . . . . . . . .
O O O O O 00 O O N O N 0 0 0 0 0 0 0 0 0 0 0 O
•^� O O M O N O O O O N M Q\ O l� N 00 O kn 0 O
6 0 O O O O 0 0 0 0 0 0 0 � 0 ^; O O O
. . . . . . . . . . . . . . . . . . . . . .
o \ o \ \ \ \ \ \ \ \ \ \ \
01 o N 01 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o O
M V) 00 ' `O N oo V 01 00 Vn o0 oo M a1 O r- N O
P `O
y \ Ln O
o0 V-) `O O r- M '-- Vn C1 �lc O M r- 01 o0 M I- VI) O
00 N '-- o0 O V) r--+ M
71- Ln Ln a1 O O O O '-- O M N O '--+ O O O 00 N O r--+ N O --+
o 0 0 0
4-40 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
00 O
N O M tt `O N kn O O ,C O cM 0 O
. . . . . . . . . . . . . . . . . . . . . . .
N
O v o 0 0
0
Outo 0 0 0 0 0 0 0 0 0 0 0 o 0 0 0 0 o N o 0 0
t o O
�n �n kn kn 01 `O C1 M � 7 N N O `O � kn
kn 00 \O N `O V') N O \O O N
M a1 O O O O O IC O O O O M O N 00 O --+
�'
CGO
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
\ \ O \ \ \ 0 \ \ \ \ \ \ \ \ O
� r- a1
r-- \O O `O 00 O 00 a1 01 r--+ 71- O 01 O 00 O O
r--+
cn
73
4-4 ct
GPM ct
ct
ct
4-4
ct ct
O Q N ct O w ct O O O .--� O ,—, ,—,
ICI �!1
Page 56 of 67
Agenda Item #6.
OCC: The efficiency rate of Disc Screen 1, which separates OCC from other types of fiber, was
found to be approximately 45 percent. The most problematic contaminants for Disc Screen 1
were polystyrene foam (1.40 percent) and plastic bags (2.78 percent), with respect to the
efficiency rate. Despite the low efficiency rate, the contamination was approximately 5 percent
according to the sorting table, due to light weights of polystyrene foam and plastic bags in the
inbound stream compared to OCC weight. This means that even slight improvement on the
efficiency of Disc Screen 1 can provide much lower contamination since the screen eliminates
OCC from other recyclables and non-recyclables in the incoming stream.
Mixed Paper: The efficiencies of Disc Screen 2 and the manual sorting that occurs before any
paper is baled are reflected by the efficiency of recovery for mixed paper (this includes
remaining flat OCC from Disc Screen 1 and newspaper of other paper grades). Disc Screen 2 is
mainly intended to recover newspaper, and for this reason, newspaper has the highest recovery
rate in Disc Screen 2 (94.91 percent). It is followed by residential mixed paper (92.00 percent),
which includes paper bags, printed paper, etc. Remaining efficiencies are carton food and
beverage containers (81.97 percent), and OCC left from Disc Screen 1 (50.56 percent). Disc
Screen 2 has several major contaminants. Some examples include polystyrene foam and residue,
which produce respective contaminations of 28.48 percent and 20.59 percent. Remaining major
contaminants are plastic film (12.92 percent), non-bottle PET (11.35 percent), and aluminum foil
(8.19 percent). The sort results for mixed paper revealed that of the material recovered in the
MRF process, about 91.90 percent was acceptable. By looking at the main contaminants, it is
possible to conclude that these objects might fall through Disc Screen 2 because of their flat
shape (e.g. aluminum cans are flattened before being thrown into the recycling bin). Increasing
public awareness of the correct way to dispose of recyclables could solve this problem.
Mixed glass: Glass recovery occurs at the Glass Breaker Screen. Since glass is not categorized
as much as other materials, it is relatively simpler to know the glass recovery efficiency over
other efficiencies. The efficiency rate of the Glass Breaker Screen was found to be 95.11 percent.
However, this efficiency was not emulated by the contamination rate incurred during the glass
sorting process. Of all the grit/fines/sweepings gathered during the entire SSR process, about
73.29 percent were mixed with the recovered mixed glass. Additionally, residue contaminants
were present in a significant level (16.26 percent). This amount of contamination greatly affected
the sorting result. The sorting results for mixed glass show that it was only possible to recover
70.16 percent of the material.
PET: Plastic sorting begins at the PET Optical Sorter, where PET material is separated from
HDPE and ferrous materials. PET recovery rates are among the highest in the entire process,
amounting to 84.84 percent of total PET bottles recovered in the full SSR stream, and 28.71
percent of the non-bottle PET recovered. Despite being categorized as plastic, part of the non-
bottle PET recovered is contaminated and cannot be recycled. The sorting results give PET
recovery in the Optical Sorter an 88.88 percent effectiveness.
HDPE: The plastic category of HDPE is further divided into two sub-categories in the HDPE
hopper: natural HDPE and colored HDPE. The efficiency rate of the natural HDPE hopper was
84.23 percent. Natural HDPE recovery includes very small amounts of non-bottle HDPE (1.29
25
Page 57 of 67
Agenda Item #6.
percent), yielding very positive results for recycled natural HDPE (97.59 percent). This result is
contrasted by that of colored HDPE. While the efficiency rate of the colored HDPE hopper was
88.25 percent, colored HDPE was found to be the lowest and most problematic result out of all
the recovered materials (65.04 percent). Here, the separation rate of the hopper was 34.21
percent for non-bottle HDPE and 24.56 percent for other mixed plastics. Other unusable plastic
materials caused most of the contamination incurred by colored HDPE.
Steel Can: The Magnetic Metal Sorter separated steel cans and other metals from the material
stream with an efficiency rate of 87.00 percent. The main shortcoming of this process was the
relatively high contamination rate, mostly caused by non-recyclable metals. The majority of this
contamination resulted from scrap metals (23.78 percent) and liquids (12.67 percent). The
sorting results reveal that 72.93 percent of the sorted steel cans could be recycled and baled.
Contamination, in the case of steel cans, is mostly a result of public ignorance or irresponsibility.
Recycling steel cans that still contain liquid hinders the entire recycling process and makes the
cans useless. It is the responsibility of the recycling companies and local governments to make
the public aware of the dos and don'ts of recycling, and it is the responsibility of the general
population to be receptive and responsible when recycling.
Aluminum Can: The Eddy Current Separator, used to separate aluminum cans from the stream,
operated with an efficiency of 75.20 percent. Additionally, it recovered 25.59 percent of the
aluminum foil in the single stream. Contamination was minimal. Sorting results put the recovery
rate of usable aluminum cans at 91.39 percent—the third highest after natural HDPE and OCC.
5 CONCLUSION
SSR systems are facing significant technical and economic barriers in terms of higher
contamination in collected materials, higher volumes of materials requiring pre-sorting at
regional MRFs, highly contaminated recyclable materials directed to the mills, and reliance on
export markets. To this end, this study was conducted to analyze the contamination rates of
MRFs in Florida, with the aim of determining the extent which broken glass and other non-
recyclables are responsible for the contamination in both inbound and outbound streams.
Researchers obtained several waste composition studies from 15 counties, one MRF audit and
266 old newsprint (ONP) samples and 35 old corrugated cardboard (OCC) samples from four
currently operating facilities in Florida. Regarding the inbound contamination rates, waste
composition studies were investigated with respect to the material definitions (see Section 3.1).
The standard deviation (8.96) and mean (18.54) contamination rates of the samples from SSR
were higher than the samples from DSR (3.08 and 3.89, respectively). Then, ANOVA was
conducted to test the difference in contamination rates. The inbound contamination rates of SSR
were found to be statistically higher than the those of DSR.
Further analysis on inbound contamination data was conducted to better understand the variance
in contamination rates among counties. Based on the descriptive analysis, County 3 (refer to
Section 4.1) had the largest recorded contamination rate with 28.2 percent on average, while
County 10 had the smallest contamination rate with 5.98 percent on average. Then, researchers
returned once more to ANOVA to test the statistical significance of the difference between the
mean contamination rates of the six counties that currently implement SSR. Based on the
ANOVA statistics, the difference in the counties' inbound contamination rates was found to be
26
Page 58 of 67
Agenda Item #6.
statistically significant. A more detailed comparison was conducted using a multi-comparison
test on these six SSR counties. Our results indicated that Counties 3 and 4 (contamination rates
were 28.2 percent and 22.3 percent on average, respectively) have means significantly higher
than Counties 6 and 7, based on a 95 percent confidence level.
To specify the difference in contamination rates among six counties, researchers developed a
correlation analysis that took into account population, population density, median age, average
income level, poverty level, White population percentage, Hispanic population percentage and
Black population percentage. Interestingly, no correlation was found between inbound
contamination rates and ethnicity data in the analysis. However, our correlation analysis
identified a positive correlation between median age and contamination, and a negative
correlation between income level and contamination. Moving forward, solid waste officials
should design specific educational programs and incentives for the counties with low income and
older residents to decrease inbound contamination.
As for the outbound contamination aspect, the acceptable OCC and ONP rates were 91.12
percent and 67.41 percent of the total weight on average for all samples. Average rates (weight
of the contamination/total sample weight) of prohibitive materials in 266 ONP samples and 35
OCC samples were 7.13 percent and 5.12 percent respectively. With such high rates of
prohibitive materials, only one of the 266 samples from the ONP stream could pass paper mill
standards. Among 35 samples from the OCC stream, only 34.3 percent of samples had
acceptable contamination rates according to paper mill standards. The most common types of
prohibitive materials in the OCC and ONP streams were residue,plastic film, HDPE and PET.
Additionally, researchers investigated which of the current methods being applied to MRFs (i.e.,
for sorting, etc.) lead to increased contamination rates in the outbound stream. Here, mass
balance equations were used to analyze the MRF's audit results. Our analysis provided two main
tables: an efficiency table and a sorting results table. Based on these tables, the existing
processing technologies for OCC, mixed paper, mixed glass, PET, HDPE, steel cans and
aluminum cans streams were assessed separately. Of these streams, mixed paper, mixed glass,
colored HDPE and steel cans were the most highly contaminated. As such, the acceptable mixed
paper, mixed glass, colored HDPE and steel cans rates were 92, 70, 65 and 73 percent,
respectively. To decrease contamination levels for these streams, single stream MRFs should
upgrade their current processing technologies and educate the people contributing to the waste
stream. These changes should decrease the presence of the most problematic contaminants (see
Section 4.3) found in the inbound stream.
We conducted this study to increase our knowledge of the impact of SSR contamination for both
inbound and outbound recovery operations in Florida. We achieved this by (i) evaluating if SSR
has a significant role on the inbound contamination, (ii) identifying to what extent the recovered
ONP and OCC samples are contaminated, and (iii) assessing the currently implemented recovery
processes for minimal outbound contamination. Further venues of this research include the
extension of this preliminary regional study to SSR programs implemented in and throughout the
US and the world. Additionally, we investigated cost-effective processing techniques designed
for sorting and separating commingled recyclables collected as a single stream and through
mixed waste recycling.
27
Page 59 of 67
Agenda Item #6.
6 REFERENCES
American Forest & Paper Association (AF&PA) (2015). Community survey. Retrieved April
2017 from http://www.paperrecycles.org/docs/default-source/default-document-library/2014-
afpa communit -survey exec-summary final.pdf
._
American Forest & Paper Association. (2013) Recovery & Use of Old Corrugated Containers.
Retrieved May 2017 from http://www.paperrecycles.org/statistics/recovery-use-of-old-
corrugated-containers-(occ)
Beck, R.W. (2006) Characterization and Quantification of Residuals from Materials Recovery
Facilities. Retrieved April 20171 from
http://www.calrec. cle.ca.gov/wastechar/PubExtracts/34106005/ExecSummarY df
Bragg, L. (2015). Glass Recycling Markets & Trends. Retrieved March 2016 from
http://www.indianarecyclin�org/wp-content/uploads/2014/10/Lynn-Bragg-Glass-Packaging
Institute.pdf
Cascadia Consulting Group, Inc. (2006). Waste Monitoring Program: 2006 Material Recovery
Facility (MRF) Assessment. Retrieved May 2017 from
htTyour.kin countygov/solidwaste/about/documents/MRF_assessment.pdf
Cattaneo, J. (2010). Glass Bottles: Reaching 50% Recycled Content. Retrieved March 2016 from
http://www.vrarecycles.org/LinkClick.aspx?fileticket jP3bJOxHPuo%3D&tabid=58
Chameidas, B. (2013). Single Stream Recycling. Retrieved April 2017 from
http://blogs.nicholas.duke.edu/the_rg een_rok/single-stream-rec. clin_/
Commingled Recycling Systems (2008). Preventing Contamination at the Curb, MRF, and Mill.
Retrieved April 2017 from
htlp:Hyosemite.epa.gov/rIO/HOMEPAGE.NSF/Topics/ccrs/$F ILE/11.+Marketing+White+P
a er. df
Conservatree (2016). Environmentally sound paper overview: Essential issues. Retrieved May
2017 from http://www.conservatree.org/learn/Essential%20Issues/EIPaperContent.shtml
Container Recycling Institute (2009). Understanding Economic and Environmental Impacts of
Single-stream Collection Systems. Retrieved March 2016 from hgp://www.container-
recyclin_�org/assets/pdfs/reports/2009-SingleStream.pdf
Department of Ecology of State of Washington. (2010). Beyond the Curb — Tracking the
Commingled Residential Recyclables from Southwest WA. Publication no. 10-07-009.
Retrieved May 2017, from http://www.container-recycling.org/assets/pdfs/2010-
BeyondCurb.pdf
DSM Environmental Services (2011) .Ohio Glass Recycling Study Final Report. Retrieved
March 2016 from http://epa.ohio.,,aov/Portals/41/rec.T�,g/OhioGlassRecyclin S. dypd
Dyer, T. D., & Dhir, R. K. (200 1). Use of glass cullet as a cement component in
concrete. Recycling and Reuse of Glass Cullet (Dhir RK, Limbachiya MC and Dyer TD
(eds)). Thomas Telford, London, UK, 157-166
Eileen Brettler Berenyi (2015) "What Comes after Single-Stream?" Retrieved April 2017 from:
https://nerc.org/documents/conferences/Spring%202015%20Conference/Berenyi%20Resour
ce%20Recycling%20Article_What%20Comes%20After%20Single%20Stream.pdf
Environmental Protection Agency (2017). Advancing Sustainable Materials Management: Facts
and Figures. Retrieved from https://www.epa.gov/smm/advancing-sustainable-
materialsmanagement-facts-and-fi G,.i1
28
Page 60 of 67
Agenda Item #6.
Florida Department of Environmental Protection (FDEP) (2017). Solid Waste Management Data.
Retrieved March 2016 from
http://www.dep.state.fl.us/waste/categories/recycling/SWreportdata/09_data.htm
Glass Packaging Institute (2018), North American Glass Recycling (Processor) List, as of 2-23-
2018. Retrieved March 2018 from http://www.gpi.org/glass-resource-locator.
Glass Packaging Institute (N.D), Learn About Glass. Retrieved May 2017 from
http://www. p. i.org/learn-about-glass/benefits-glass 1packa .
Glass Packaging Institute. Glass Recycling
ss Facts. Retrieved March 2016 from
http://www. pg= i.or /,g recycling//,glass-rec.T� -facts
Goodman, T. (2006). Single Stream and Dual Stream Recycling: Comparative Impacts of
Commingled Recyclables Processing. Retrieved March 2016 from
https://www.pca.state.mn.us/sites/default/files/commingled.pdf
Hastie, T., Tibshirani, R., Friedman, J., & Franklin, J. (2005). The elements of statistical
learning: data mining, inference and prediction. The Mathematical Intelligences, 27(2), 83-
85.
Kessler Consulting Inc. (2009). Materials recovery facility technology review. Prepared for
Pinellas County. Retrieved January 20161 from
https://www.dep.state.fl.us/waste/quick topics/publications/shw/recycling/InnovativeGrants/
IGYear9/finalreport/Pinellas_IG8-06_Technology_Review.pdf
Kessler Consulting Inc. (2010). Beach recycling program report. Prepared for City of Fort
Lauderdale. Retrieved May 20161 from
https://www.dep.state.fl.us/waste/quick topics/publications/shw/rec clingg/InnovativeGrants/
IGYear9/fullprop/Recycling-Program-Report.pdf
Kessler Consulting Inc. (2013). Seminole County's 2013 Recyclables Composition Study.
(L.Rubino, personal communication, March 29, 2016).
Kessler Consulting Inc. (2014a). Town of Lauderdale-by-the-Sea Recyclable Materials
Composition Study. (V. Eckels, personal communication, April 11, 2016)
Kessler Consulting Inc. (2014b). Pasco County Recyclable Materials Composition Study. (J.
L.Seney, personal communication, May 6, 2016).
Kessler Consulting Inc. (2016). Sarasota County Recyclables Composition Study. (B. Usher,
personal communication, May 12, 2016).Combs, A. R. (2011). Life Cycle Analysis of
Recycling Facilities in a Carbon Constrained World.
Kinsella, S. & Gertman, R. (2007) "Single Stream Best Practices Manual and Implementation
Guide". Retrieved April 2017 from
http://conservatree.org/learn/SolidWaste/bestpractices.shtml.
Lakhan, C. (2015). A comparison of single and multi-stream recycling systems in Ontario,
Canada. Resources, 4(2), 384-397.
McCormick D (2015) Recycling industry report: The challenges of contamination. FDEP
Webinar, 16 December 2015. Retrieved September 2017 from:
http://www.dep.state.fl.us/waste/quick_topics/publications/shw/recycling/Single-Stream-
Recycling-Contam-Webinar 21 Dec 15.pdf
Morawski, C. (2010). Single stream uncovered. Reprinted from Resource Recycling. Retrieved
March 20169 from http://www.cmconsultinginc.com/wp-
content/uploads/2011/03/featured Single Stream l.pdf
29
Page 61 of 67
Agenda Item #6.
Moreau, T. (2015). Managing Recycling Contamination. EPA Sustainable Materials
Management Webinar. Retrieved March 20161 from
http://www.epa.gov/sites/production/files/2015-12/documents/moreau .pdf
Pressley, P. N., Levis, J. W., Damgaard, A., Barlaz, M. A., & DeCarolis, J. F. (2015). Analysis
of material recovery facilities for use in life-cycle assessment. Waste Management, 35, 307-
317.
Robinson, S. (2014) The Changing Waste Stream. EPA Webinar Series. Retrieved March 2016,
from https://www.epa.,gov/sites/production/files/2015-
09/documents/changng_wste_stream.pdf
Sustainability Victoria (2014). Market Summary-Recycled Glass. Retrieved May 2017 from
http://www.sustainability.vic.gov.au/-/media/resources/documents/publications-and-
research/research/market-analysis/market-analysis-glass-sept-2014-pdf.pdf
Wang, J. (2006). All in one: Do Single-Stream Curbside Recycling Programs Increase Recycling
Rates? Retrieved April 20171 from:
http://nature.berkeley.edu/classes/esl96/projects/2006final/wand df
Yasar, D., Damgacioglu, H., Bastani, M., & Celik, N. (2017). Assessment of the Impact of
Single Stream Recycling on Paper Contamination in Recovery Facilities and Paper Mills.
30
Page 62 of 67
Agenda Item #6.
7 APPENDICES
APPENDIX A: ANALYSIS OF VARIANCE
ANOVA (analysis of variance) is commonly used to compare differences of means between
more than two groups in the literature. It can be used in both experimental and observational
data. ANOVA analyzes this by looking at variation in the data. ANOVA considers both the
amount of variation between groups and within groups.
In the studies, when taking a sample rather than measuring the whole population, sampling error
needs to be considered in comparing differences in means among different groups. ANOVA
basically answers this question. Is the difference between groups greater than sampling error? In
other words, is there a real difference in the population means in different groups?
ANOVA can be represented mathematically as:
Xij = µi + Ei.1 (1)
In the equation (1), xij denotes the individual data point for group i and observation j, E shows
the unexplained variation and pi shows the population mean of group i. Based on Equation (1),
each individual data point can be represented as its group mean and error term.
Hypothesis testing
Similar to other statistical tests, ANOVA calculates a test statistic, the F-ratio. Using the F-ratio,
one can obtain the probability, called the p-value, of obtaining the data assuming the null
hypothesis. Based on the p-value, it is determined if at least one group's mean is statistically
different from the others. Usually if the p-value is smaller than 0.05, it is concluded that there is
statistical evidence for the alternative hypothesis.
Ho: µl = P2 = ... = µk
H1: Means are not all equal
Here Ho is the null hypothesis, which means all population means are equal, and H1 is the
alternative hypothesis.
F-ratio calculation
As discussed above, ANOVA separates the variation in the data into two parts: between-group
and within-group. As shown in Table 7 in the report, these variations are called the sum of
squares. Particularly, SSB shows the between-group variation and SSw shows the within-group
variation. The calculations of SSB and SSw in the case of K groups with total N observations are
given in the following equations.
K
SSB — ni(XL —X)2 (2)
i=1
31
Page 63 of 67
Agenda Item #6.
K
SSW = 6i(ni — 1) (3)
i=1
In Equations (2) and (3), ni is the number of observations in group i, Xl is the mean of group i,
6i is the variance of the group i and X is the population mean. In the calculation of F-ratio, the
mean squares are used. It is computed by dividing sums of squares by degrees of freedom.
Degrees of freedom is K — 1 for SSB and N — K for SSW. The calculation of mean squares is
given in the following.
SSB SB= (4)
K — 1
SSWMSW _ (5)
N — K
After the calculation of mean squares, the F-ratio is computed as:
F — ratio = MSB (6)
MSW
Then, to obtain the p-value, F-ratio is tested against the F-distribution of a random variable with
degrees of freedom associated with MSB and MSw. The summary of ANOVA is shown in Table
13.
Table 13. Summary of ANOVA formulas.
Source of Sum of squares Degrees of Mean square F ratio (test
variation freedom statistic)
Between SSB = EK 1 ni(XI )2—X K-1 MSB =SSB /(K-1) MSB /MSW
Within SSW = EK 1 6i(ni — 1) N-K MSS,=SSA, /(N-K)
Total SST = SSB + SSW N-1
32
Page 64 of 67
Agenda Item #6.
Depart Rick Scott
Florida Department f Governor
Environmental Protection
Carlos Lopez- antera
. Lt. Governor
` - Bob Martinez Center
600 Blair Stone Toad
Tallahassee Florida 3 399- 400 Jonathan P. Steverson
4 e n t a l 01V Secretary
Memorandum
TO: Recycling and Solid waste Directors
FROM: F. Joseph Ullo, Jr. P.E., Director
Division of waste Management
SUBJECT: Single Stream recycling Contamination
DATE: December 2, 2015
Florida counties have bccn given the responsibility of implementing a recycling program within
their boundaries under s. 40 .706, Florida Statutes (F.S.). Large counties, defined as counties
having more than 100,000 1n population, are responsible for implementing recycling programs to
meet the statewide 75% recycling goal by 2020. The state, counties, municipalities, and the Solid
waste and recycling industry have been working hard to make significant strides toward this
statewide goal.
In an effort to reach the 75% recycling goal, many counties and municipalities have instituted
single stream recycling programs. Single stream recycling programs have been successful by
Providing increased curbside collection efficiency while also increasing residential participation
and the amount of recyclables collected. while there are many advantages to incorporating single
stream recycling, it has not consistently yielded positive results for the recycling industry.
Coupled with other trends in the industry, single stream recycling has also resulted in the
collection of unwanted materials and poorly sorted recy lables, resulting in increased
contamination of recycled materials.
Contamination hinders processing at Material Processing Facilities MRFs when unwanted
items are placed into recycling bins. For example, many residential customers may not
understand that while retail plastic bags are recyclable, they are harmful to the automated
equipment typically used to process and separate recyclable materials from single stream
collections. while MRFs are equipped to handle some non-recyclable materials, excessive
contamination can undermine the recycling model resulting in additional sorting, processing,
energy consumption, and other increased costs for recycling due to equipment downtime., repair,
or replacement needs. In addition to increased recycling processing costs, contamination also
results in poorer quality rcyclabl , and increased rejection and landfilling of unusable
materials.
Page 65 of 67
Agenda Item #6.
Single Stream Recycling Contamination
Page 2
December 2, 2015
Single stream contamination can result from cross-contamination of recycicd materials. For
example., recent trends with decreases in the use of newsprint and other fibrous content have
contributed to the reduction in recyclable paper that would otherwise cushion glass in single
stream recycling. As a result, there 'is an increased likelihood that recycled glass will become
broken both in single stream containers, and during collection and transportation. Glass pieces
smaller than 3/8" resulting from breakage during collection and transport can be comingled with
paper, reducing the value of the paper or causing the entire load to be sent to the landfill.
The recycling industry has provided information showing that residential single stream
contamination has increasingly become an obstacle to implementation and expanding recycling
programs to meet the 75% recycling goal. While some local governments have implemented
successful single stream recycling programs with low contamination rates, contamination rates
for other programs have continued to rise, in some cases reaching contamination rates of more
than 30-40% by weight. This disparity in program Successes suggests that with proper education,
combined with adequate controls and oversight, the contamination problem is one that can be
better managed. Industry norms appear to suggest that contamination levels of 5% -10% by
weight are manageable and acceptable under many existing recycling, programs; however, MRFs
cannot manage higher contamination levels effectively and efficiently. Local governments are in
the best position to provide the leadership, vision and targeted incentives to ensure their
implementation of an effective recycling program. It is critical for local government leadership to
optimize long term recycling development by getting the best services that produce the highest
quality materials. Indeed it appears that reducing contamination rates will be an essential
component for many counties to better implement or expand their recycling programs as part of
their efforts to achieve the recycling goals outlined in s. 403.706(2), F.S.
For the State of Florida to reach our target recycling goal of 75%, it will take firm commitments
from all local government units. The Department strongly encourages counties and cities to take
steps today to address the increasing problem of contamination in your single stream recycling
programs. Below are some ideas to consider when reviewing the implementation of yoLir
recyclable materials recycling program.:
Review the design of your recycling process as a whole and consider what controls are in
place during collection to ensure proper identification,, screening, and handling of
incompatible materials that are being introduced in the single stream process.
Consider meeting with and discussing these issues with the MRFs and haulers that
service your county. What is the quality of the recyclable material products produced
from YOUr single stream program and the efficacy of your process.?
• What improvements may be required for successful processing to ensure the material can
be sorted to meet end use requirements or recycled product specifications?
Review your existing contracts so they may help support a sustainable recycling program.
Are they focused solely on diversion rates with little or no enforceable standards or
language addressing contamination?
What is known about the SOUrce(s) of contamination in your county?
Page 66 of 67
Agenda Item #6.
Single Stream Recycling Contamination
Page 3
December 2, 2015
• Consider options, including the use of people, to help increase understanding and
compliance by customers with directions on what items are suitable for recycling in
residential recycling containers.
• Consider performing waste audits to determine the composition of the single stream
materials sent to the MRF. (Your MRF may already have some infonilation and may be
able to help with this process.)
• Evaluate your Current community outreach and educational programs. Are they
effective? Do residents really know what can and should go in their cart? Are you making
it easy for them to comply? Is there a way to screen and inform residents when they have
included items in the single stream recycling bins that are incompatible?
The Department of Environmental Protection is partnering with the Florida Recycling
Partnership and Recycle Florida Today to host a webinar concerning the growing contamination
issuO, associated with single stream recycling. The purpose of the webinar is to discuss ways to
reduce cross-contamination of recycled materials, increase the quality and quantity of materials
recycled, and yield the highest percentage of materials that are accepted in your recycling
program. This webinar is scheduled to occur on December 16"' at 10:00 am. A "Save the Date"
announcement was sent last week, and additional registration information will be scat to you
within the next week.
The Department will also be organizing a stakeholders committee to examine the single stream
containination issue, as well as soliciting ideas on how best to help local governments work
towards resolving this and other emerging recycling challenges. If you would like to participate
or would benefit from technical assistance, or have ideas that you would like to share on
reaching the 75% recycling goal, please be sure to contact the Department's Waste Reduction &
Recycling Program at Recycling@dep.state.fl.us. If you have any questions or would like to
further discuss these topics in the meantimc., please contact Karen Moore,, Environmental
Administrator, at (850) 245-88641, or Karen.S.Moore Oa.,dep.state.fl.us.
Page 67 of 67