RepTileä, manufactured by SelecTech Inc., is a resilient
floor tile made from a blend of PVC and mixed post-consumer carpet waste.
During early production of this product, problems arose with respect
to processing efficiency, and distortion (, or
“peaking)” , of
the tiles at
the four mating corners
Phase 1 of the project focused on
designing and installing a custom extruder system capable of feeding and
homogenizing a mixture of post-consumer carpet, post-industrial PVC, and
proprietary additives, to produce a suitable feedstock for use in finished
tiles. The production process was modified
to incorporate the revised feedstock material.
Phase 2 of the project focused on testing to determine the
impact of feedstock formulation and production process with respect to final
product quality. Tests included (1) the
effect of recycled post-consumer carpet concentration on the length and weight
of the tiles; (2) measurements of length and weight from a sample production
run to determine the repeatability of the modified injection molding process;
(3) measurement of tile length and weight before and after weathering exposure
to heat and humidity for a two month period; and (4) an observation test to
determine the relationship between peaking and blending efficiency (i.e.
recycled carpet content particle size and distribution).
Phase 1 resulted in the design and installation of a customized
extruder system that successfully combined three existing process steps
(Molding Into Prep Tile, Granulating Prep Tile, and Dry Blending) into one
revised process step (Extrusion into Nuggets).
The nuggets are then used as a pellet feedstock for the tile injection
molding process.
Phase 2 testing results indicated that the revised processing
and production system could repeatably produce injection-molded tiles that show
no evidence of peaking at recycled carpet concentrations as high as 20%. Weathering observations on a 30% recycled
carpet concentration batch indicated a trend for the length and weight of the
tiles to increase slightly, however, the data was insufficient to draw
conclusions with respect to the potential relationship of the increases with
respect to peaking. In an observation
test of the relationship between peaking and blending efficiency, it was noted
that tiles made with a concentration of 40% post-consumer carpet content did
not exhibit peaking in response to weathering if the material passed through
the blending/extrusion process more than once.
Note that all tiles made at 40% concentration (with a single pass
through the blending/extrusion process) had demonstrated peaking in previous
tests. This indicates that it is
possible to maintain product quality (i.e., resistance to peaking) at
post-consumer recycled content concentrations above 20% if the
grinding/blending process can be sufficiently improved.
The market for
resilient flooring is approximately 1.5 billion square feet per year. Resilient flooring is typically sold to
commercial facilities such as factories, warehouses, hotels, hospitals,
schools, municipalities, and office buildings.
The market is currently dominated by products manufactured from virgin
materials.
Some recyclers,
such as DuPont Flooring Systems, have the infrastructure to both collect used
carpeting from remodeling and demolition projects, and to recycle most of what
is collected. However, in many areas of
the country, used carpeting still goes into landfills.
The many different types of resins and materials used in
carpeting and carpet backing make this recovered material stream very
non-homogenous, and thus difficult to recycle without costly separation. One type of carpet in particular,
vinyl-backed carpet tile, is a problem stream for recycling. Vinyl-backed carpet tile consists of a vinyl
backing, a fiberglass reinforcing sheet, and nylon or polyester face fiber. These tiles are too costly to separate for
recycling, and when ground in an attempt to use as a typical re-grind
feedstock, the fibers clog conventional plastics manufacturing machinery and
can cause equipment damage.
SelecTech has
developed a patent-pending method for producing an industrial-grade resilient
floor tile, by re-using mixed carpet waste, including vinyl-backed carpet, in a
process that retains the fiberglass, nylon, and polyester intact as fibers to
reinforce the finished product. The
resilient floor tile, known as RepTileä, is manufactured from a
commingled mixture of shredded vinyl-backed carpet tiles, post-industrial PVC,
and proprietary additives. The blend
combines the flexibility of PVC with the strength of glass, nylon, and
polyester fibers.
Each RepTile weighs 6.25 lbs and covers an
area of approximately 4 square feet (1.6 pounds per square foot). The potential market for this material, in
pounds, is therefore 2.4 billion pounds per year. It is clear that even a small percentage of market share
represents the potential for a substantial diversion of carpet waste from
landfills.
The ability to
compete in the resilient flooring market is based on quality and price. Testing has shown the RepTile to be
comparable or superior in quality to competing resilient floor tiles made with
virgin materials. At present, the
critical stage of development in the product depends upon increasing efficiency
and reducing production costs, while maximizing recycled content.
1
SelecTech’s
original process equipment was specifically designed to handle the fiberglass,
nylon, and polyester fibers and utilize their beneficial properties in a
value-added final product. The tiles
are molded in an innovative injection molding technology that is ideally suited
for processing commingled waste plastics.
The original process is shown in Figure 1. Whole carpet tiles were shredded, then mixed
with post-industrial PVC and proprietary additives. The clumpy mixture was then dry-blended and hand-fed into the
injection molding machine to be molded into a preparatory (or intermediate)
tile. The preparatory tile was
granulated, and then dry-blended. The
blended mixture was then fed into the injection molding unit, where the blend
was melted and then injected under high pressure into a mold that forms the RepTile. The tile was then cured by cooling.
Figure 1: Original
Process Flowchart
|
Whole Carpet Tiles
|
|
Shred
|
|
Shredded Carpet
+
Post-Industrial PVC
+
Proprietary Additives
|
|
Dry Blend
|
|
Mold into Prep
Tile
|
|
Granulate Prep Tile
|
|
Dry-Blend
|
|
Mold
Finished RepTile
|
1
Phase 1 resulted in the design and installation of a
customized extruder system that successfully combined three existing process
steps (Molding Into Prep Tile, Granulating Prep Tile, and Dry Blending) into
one revised process step (Extrusion into Nuggets), thus eliminating the
inefficiencies of the old process.
Figure 2:
Revised Process Flowchart
|
Whole Carpet Tiles
|
|
Modified Particle
Size Reduction
|
|
Diced Carpet
+
Post-Industrial PVC
+
Proprietary Additives
|
|
Dry Blend
|
|
Extrusion into
Nuggets
(Proposed New Equipment)
|
|
Mold
Finished RepTile
|
Whole carpet tiles (roughly 18” x 18” square x Ľ”
thick) are sorted from other carpet waste at DuPont’s Carpet Reclamation
Facility in Calhoun, GA. These sorted
tiles are then sent to Conigliaro Industries in Framingham, MA where they are
shredded and granulated in a modified, low-speed granulator that reduces the
carpet into a more uniform size that is easier to handle and mix. The blended material is then fed into an
extruder that further homogenizes the blended material and forms the mixture
into relatively uniform nuggets. These
nuggets can then be automatically fed into the injection molding machine to
create the finished RepTile.
2
The goal of the
Phase 1 of this project was to develop an extruder system that could homogenize
the carpet/PVC mixture sufficiently for the desired product. The critical requirements for the system
were:
·
the ability to feed a non-uniform, fluffy feedstock;
and
·
the ability to break-up and disperse carpet clumps in
the mixture.
The final system is shown in Figure 3-1. As shown, this system consists of the
following components:
·
Feed Hopper Bin
·
6” Screw Feed Conveyor
·
4 ˝” Cram Feeder
·
4 ˝” Extruder
·
4 ˝” Custom Extruder Die
·
Stainless Steel Water Bath with Air Knife
·
8” Pelletizer with Pull Rollers
·
Custom Hopper Catcher
·
Vacuum Loader Mounted on Gaylord Stand
Because the waste carpeting/PVC
blend is unconventional, many of these components are not standard auxiliaries
used for conventional plastic compounding.
This system has been specifically designed to handle the hard-to-convey
mixture of PVC and carpeting, and individual components were modified to ensure
that the material feeds positively into the extruder. The die on the end of the extruder and the cutting system was
designed to ensure maximum throughput of this particular feedstock.
Figure 3-1
Extruder System Design
3
Phase
2 of the project focused on testing to determine the impact of feedstock
formulation and production process with respect to final product quality.
3
Typically, RepTile has a size of
24L ´
24W ´
0.22T inches and a weight of six pounds.
In this project, size and weight of RepTile were chosen as the control
factors and measured precisely. Size
was represented by three dimensions and measured by a stainless steel ruler
with one 64th inch precision. Two of
the dimensions are the lengths of a pair of parallel edges of a tile. The third length was measured parallel to
the two edges at the center of the tile.
Tile weight was measured using an electronic scale with 0.01 pound precision.
This testing was divided into four
parts as follows:
·
Part 1 – Effect of recycled carpet concentration on the
length and weight of the tile.
·
Part 2 – Repeatability of the injection molding
process.
·
Part 3 – Effect of weathering on length, weight, and
shape of the tile.
·
Part 4 – Effect of blending efficiency on tile distortion or peaking.
3.2
Pa
In
Part 1, the effect of recycled carpet concentration on the weight and length of
a tile was investigated. For black and
gray RepTiles, the carpet regrind was composed of irregularly shaped particles
with projected areas of from 0.25 to 5 square inches. This chopped vinyl-backed carpet material was collected by DuPont
Flooring Systems in whole form and then shipped to Conigliaro Industries in
Framingham, MA where it was chopped using a special, low-speed granulator to
produce the particles as described above.
For colored (yellow in this test) RepTiles, powdered carpet backing was
used as the post-consumer carpet content.
This powdered carpet backing is a byproduct of DuPont Flooring Systems
nylon carpet recycling operation located in Chatanooga, TN.
The formulations are showed below.
|
Experimental trials
|
Color
|
Carpet Concentration
(Weight %)
|
Flexible PVC
|
|
1
|
Black
|
10%
|
90%
|
|
2
|
20%
|
80%
|
|
3
|
30%
|
70%
|
|
4
|
40%
|
60%
|
|
1
|
Gray
|
10%
|
90%
|
|
2
|
20%
|
80%
|
|
3
|
30%
|
70%
|
|
4
|
40%
|
60%
|
|
1
|
Yellow
|
5%
|
95%
|
|
2
|
7.5%
|
92.5%
|
|
3
|
10%
|
90%
|
All trials were conducted under the
same processing conditions that consisted of:
·
Melt Temperature 170
°C
·
Mold Cooling Temperature 34 °C
·
Injection Pressure 150
bar
·
Mold Clamp Pressure 200
Metric Tons
·
Cooling Time 60
seconds
Fifty tiles were made in
each trial and 20 tiles were randomly chosen from the fifty to represent the
trial. After measuring the sizes and weights of the twenty tiles,,
12 tiles were interlocked with each other to form a 3´ 4 unit and placed
outdoors to test
for weathering.
Past problems with distortion
or peaking at customer locations were known to be exacerbated with rises in
temperature and humidity. When interlocked
tiles distort due to material inconsistencies, they bump up at the corner
where the four tiles come together. This
"bump" resembles a little peak. The primary reason for placing the 3 x 4 unit outdoors was ""
under high heat and humidity.
3.3 Part
2 – Repeatablity of the injection molding process
In
Part 2, the repeatability of the injection molding process was
investigated. A test manufacturing run
was conducted to produce tiles with 20% post-consumer carpet content. During this production run of 500 tiles, 41
RepTiles were chosen randomly. The lengths and weights of these tiles were
measured.
3.4 Part
3 – Effect of weathering on length, weight, and shape of the tile.
In Part
3, the effect of weathering on the weight and length of the tiles was investigated.
It is known that increases in heat and humidity increase the likelihood
that the tiles will distort or “peak” after installation.
To determine if exposure to heat and humidity actually caused changes
in the weight and length of the tiles, tiles were exposed for two months of
direct sunlight on a black surface (roof of the Selectech facility). In this environment, tile surface temperatures
were measured to be as high 150 °F. During the trials, actual weather conditions were
not recorded. The weather varied over
the time period, from warm and rainy to extremely hot and humid, thus exposing
the tiles to a wide variation in temperature and humidity.
Ten filled, gray tiles were
then selected at
random, from the exposed batch and measured for size and weight.
In
Part 4, tiles were manufactured using materials that had been processed through
the extruder (homogenized) from 0 to 4 times to observe the “peaking” behavior
of the tiles in a field environment as a function of the degree of compounding
or blending.
4.0
4.1 Part
1 Results – Effect of recycled carpet
concentration on the length and weight
of the tile
In
Part 1, the effect of recycled carpet content on the length and weight of
RepTile was investigated. Data are
provided in the Appendices, and summarized in Table 1 below.
Increasing
recycled carpet content led to an increase in both the length and weight of the
tiles. When the carpet content
increased from 10% to 40%, the tile length increased approximately 0.07 to 0.1
inches, and the weight increased approximately 0.1 to 0.3 pounds. More significantly, variations in the sizes
and weights (as measured by standard deviation), from tile to tile, also
increased. This indicates that the
higher carpet content contributes to reduced homogeneity of the final
product. Finally, it should be noted
that higher carpet content decreased the processability of the material. When the carpet content was increased to
40%, the material would not completely fill the mold. As carpet content increased, melt viscosity increased
accordingly, making it more difficult to push the blend into the cavities of
the mold under the same processing conditions.
TABLE 1 - Effect of recycled
carpet concentration on the length and weight of the tile
Color
|
Carpet content
(%)
|
Right Edge
(in)
|
Middle
(in)
|
Left Edge
(in)
|
Weight
(lbs)
|
|
|
10
|
24.00
(0.01)
|
24.00
(0.01)
|
24.00
(0.01)
|
5.93
(0.01)
|
|
Black
|
20
|
24.04
(0.01)
|
24.02
(0.01)
|
24.04
(0.01)
|
5.97
(0.02)
|
|
|
30
|
24.08
(0.02)
|
24.05
(0.01)
|
24.08
(0.01)
|
6.11
(0.07)
|
|
|
40
|
24.10
(0.02)
|
24.07
(0.01)
|
24.09
(0.01)
|
6.30
(0.03)
|
|
|
|
|
|
|
|
|
|
10
|
24.03
(0.01)
|
24.02
(0.01)
|
24.03
(0.01)
|
5.99
(0.02)
|
|
Gray
|
20
|
24.09
(0.01)
|
24.07
(0.01)
|
24.09
(0.01)
|
6.06
(0.03)
|
|
|
30
|
24.09
(0.02)
|
24.07
(0.01)
|
24.09
(0.01)
|
6.06
(0.01)
|
|
|
40
|
24.12
(0.02)
|
24.10
(0.01)
|
24.12
(0.01)
|
6.09
(0.03)
|
|
|
|
|
|
|
|
|
|
5
|
23.99
(0.01)
|
23.98
(0.01)
|
23.98
(0.01)
|
5.69
(0.03)
|
|
Yellow
|
7.5
|
23.99
(0.01)
|
23.98
(0.01)
|
23.98
(0.01)
|
5.74
(0.03)
|
|
|
10
|
24.00
(0.02)
|
23.99
(0.01)
|
23.99
(0.02)
|
5.80
(0.02)
|
Values given are
averages. Standard deviations are
presented in parenthesis.
The two charts that follow provide
a graphic representation of the trends observed with respect to the effect of
the percentage of recycled carpet content on the average tile dimensions and
weight. It is clear that the length and
weight of the tiles increased as the percentage of recycled carpet content
increased, and these increases in length and weight were relatively consistent
throughout the sample.
When
tiles were placed outdoors in the direct sun on a black surface, the 40%
carpet-content tiles showed a great distortion at the interlocking joints,
which caused the tiles to rise off the floor at this intersection (referred to
as “peaking”). Moreover, gaps in the
interlocks were observed due to this deformation. Almost every joint showed peaking on the 3 ´ 4
tiles mat (6 joints). The height of
peaking joints was a quarter of inch to half an inch above the flat
surface. When carpet content decreased
to 30%, the amount and height of peaking joints decreased accordingly. When carpet content decreased to 20%, 10%,
7.5%, and 5%, no peaking was observed.
This indicates that carpet content affects the tile quality with respect
to the peaking problem. It was
generally observed that when carpet content is below 20% tiles perform well;
and above 30%, tiles show distortion.
Finally, tiles were sliced to
determine if there was any visual difference within the tiles. When tiles were cut, the cross-sections
revealed agglomerates of post-consumer carpet within the tile. Tiles made from increased carpet
concentrations showed increases in the presence of these agglomerates. Because the thermal expansion rates of these
agglomerates and the surrounding PVC resins are different, when the temperature
increases, stresses are generated within the tile that result in peaking.
4
In
Part 2, the repeatability of the injection molding process was
investigated. The length and weight of
41 randomly chosen tiles, from a 500 tile production run with a 20% recycled
carpet concentration, is provided in the Appendices. Table 2 provides a summary of this data.
Table 2 – Repeatability of Injection Molding
|
|
Size (in)
|
Weight
(lbs)
|
|
Right edge
|
Middle
|
Left edge
|
|
Maximum
Value
|
24.11
|
24.11
|
24.13
|
6.21
|
|
Minimum
Value
|
24.08
|
24.06
|
24.06
|
6.01
|
|
Average
Value
|
24.09
|
24.08
|
24.09
|
6.10
|
|
Standard
Derivation
|
0.01
|
0.01
|
0.01
|
0.05
|
The
length variation was found to be approximately 0.07 inches with a standard
deviation of 0.01. The weight variation
was found to be approximately 0.20 pounds with a standard deviation of
0.05. Referring to the scatter plots of
length and weight, with linear trend lines shown below, it can be seen that the
injection molding process to make RepTiles is repeatable within acceptable
production limits.
4
Part
3 consisted of a weathering test to observe distortion, and measure changes in
the tile’s length and weight due to exposure to heat and humidity. The size and weight of 10 tiles before and
after two months exposure outdoors are provided in the Appendices. Table 3 below summarizes the weathering
data.
Table 3 – Effect of Weathering on Length and Weight
|
|
Size (in)
|
Weight
(lbs)
|
|
Right edge
|
Middle
|
Left edge
|
|
Before
|
After
|
Before
|
After
|
Before
|
After
|
Before
|
After
|
|
Maxim value
|
24.14
|
24.13
|
24.11
|
24.08
|
24.14
|
24.11
|
6.09
|
6.18
|
|
Minimum value
|
24.08
|
24.06
|
24.05
|
24.05
|
24.08
|
24.06
|
6.04
|
6.09
|
|
Average value
|
24.09
|
24.09
|
24.07
|
24.06
|
24.09
|
24.09
|
6.06
|
6.13
|
|
Standard derivation
|
0.02
|
0.02
|
0.01
|
0.01
|
0.01
|
0.02
|
0.01
|
0.03
|
It
was found that the weight of the tile increased only slightly after two months
weathering. Before the weathering test,
the weight ranged from between 6.04 to 6.09 pounds. After that, the weight increased slightly to a range of 6.09 to
6.18 pounds. This weight gain is considered to be small and possibly the result
of dirt sticking to the surface and moisture absorbed into the tile. Weight measurements were not done after
cleaning the tiles, or after drying the tiles, and thus it cannot be concluded
whether dirt or absorbed moisture contributed more to weight gain. With regard to size, average tile length did
not change significantly due to the weathering. Even in the tiles that distorted visibly at corners, overall
dimensions did not change substantially.
It
was also noticed that the amount and height of peaking joints were greater on a
hot day than that on a cool day, indicating that the peaking is a result of
non-uniform expansion of the non-homogeneous tile components due to temperature
changes.
4 Part 4 Results – Effect of blending efficiency on tile peaking.
Part 4 – Effect of compounding on tile performance
In
the fourth part of the program, tiles were made using materials that had been
compounded through the extruder process 0, 1, 2, and 3 times. In all cases, tiles were made from a
consistent blend of 40% post-consumer carpet content and 60% post-industrial
vinyl. The tiles were then exposed to
direct sunlight on a black surface and monitored for peaking. Tiles made from uncompounded material and
from material that had been compounded once all showed peaking. Tiles made from material that was compounded
once peaked less than tiles made from uncompounded material. Tiles that were made from material that had
been compounded two and three times did not peak.
This
test indicates that the level of compounding (i.e., the homogeneity of the
material in the tile) affects the performance of the tile and that peaking is
most likely a result of material that is not fully blended.
5
The following
conclusions resulted from this project:
·
The occurrence of distortion or “peaking” in the field
applications of finished RepTiles is directly related to the degree of
blending, or homogeneity, of the recycled carpet concentration in the
tiles. The higher the concentration,
the more likely the presence of agglomerates that will cause the tiles to
distort when subjected to temperature changes.
·
Increasing the recycled carpet concentration leads to a
corresponding increase in the weight of the finished tile.
·
Increasing the carpet content, without thorough
blending, reduces the processability of the material. When carpet content is increased beyond 30% using standard
in-process blending, the material does not consistently fill the mold resulting
in reject product.
·
The injection molding process for the tile production
is repeatable within typical manufacturing limits. The peaking was not caused by size variations inherent in the
molding process.
·
The peaking is related to variations in thermal
expansion of the feedstock materials.
Peaking is exacerbated when insufficient blending results in the
presence of agglomerates.
·
Increased compounding efficiency increases the
homogeneity of the final product and allows for higher post-consumer carpet
content without end-product distortion.
This
project resulted in the design and implementation of a revised system that is
capable of economically producing consistent field-quality tiles using up to
20% post-consumer vinyl-backed carpet waste.
The
concentration of post-consumer carpet content is currently limited by the level
of mixing and homogeneity that is achieved by the system. If this can be further improved, then the
concentration of post-consumer carpet could be increased to both lower costs of
the final product and, potentially, improve performance.
