- Email: [email protected]
Separation of plastic mixtures using liquid-fluidized bed technology
Chemosphere 63 (2006) 893–902 www.elsevier.com/locate/chemosphere
Separation of plastic mixtures using liquid-fluidized bed technology Takehiko Kinoshita, Kazuaki Okamoto, Koichi Yamaguchi, Shigendo Akita
*
Nagoya Municipal Industrial Research Institute, 3-4-41 Rokuban, Atsuta-ku, Nagoya 456-0058, Japan Received 20 April 2005; received in revised form 14 September 2005; accepted 14 September 2005 Available online 22 November 2005
Abstract Separation of heavier-than-water plastic mixtures had been investigated via the fluidization of their packed beds induced by an upward flow of water. The samples examined were resin pellets and crushed plastic products including PVC, PET and PBT. On the onset of a flow, a mixed bed was swelled to the state of fluidization and separated into layers of respective resins depending on their density. The effects of the flow rate, an amount of the samples and their density difference were examined on the separation of resin pellets. Under an appropriate condition, the process was completed within a few minutes, and satisfactory separation was attained when the density difference of the samples exceeded 0.05 g cm3. By using a column equipped with several sample outlets sorted resins could be recovered with their purity intact by withdrawing each layer successively from above through a suitable outlet. Multi-stage separation was also found to be effective in treating close-density samples. Compared with uniform-sized resin pellets, size distribution of crushed plastic samples deteriorated the separation to some extent. 2005 Elsevier Ltd. All rights reserved. Keywords: Separation; Plastic mixture; Fluidized bed; Density; Water
1. Introduction Waste management of plastics has been prompted by increasing awareness for environmental protection and resource conservation. In Japan, the domestic production of plastics in 2002 reached 13.85 million tons and the waste generation came to as high as 9.90 million tons in the same year. Since 2000, a number of recyclingrelating laws have been enacted by the government for the purpose of achieving sustainable development. * Corresponding author. Tel.: +81 52 654 9899; fax: +81 52 654 6788. E-mail address: [email protected] (S. Akita).
Thanks to these initiatives, the effective utilization rate of plastic wastes has been steadily increasing to 55% in 2002. Among 45% of the non-utilized plastic wastes, 17% is simply incinerated and the rest, 28%, is treated in landfilling (Plastic Waste Management Institute, 2004). Recycling of plastics is categorized into three main groups of mechanical recycling, feedstock recycling and energy recovery. The first one is preferred, if available, to the other two from the viewpoint of energy efficiency, and increasing its rate is of great significance. At present, most sources of mechanical recycled plastics come from the loss in production and processing because of their stable quality and supply. On the other hand, mechanical recycling of used plastic products
0045-6535/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2005.09.068
894
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
from end users is still in the minority, since the wastes often consist of many kinds of plastics and need to be dismantled and sorted based on their raw material. For the separation of plastic wastes, wet process is relatively easy to apply and a multiple of techniques have been developed and some are in practical use. The followings are typical examples. Sink and float separation is the most fundamental and mainly used for separating heavier- and lighter-than-water plastics. Some modifications have been done on liquid media to meet a wide density range of plastics (Altland et al., 1995; Ferrara and Meloy, 1999; Pascoe and Hou, 1999). Jig separation, which has been widely used in coal separation, was reported to successfully separate small plastics particles of polyethylene and polyvinyl chloride (Tsunekawa et al., 2003). Recently, froth flotation is attracting attention as a recycling operation for plastic wastes. The technique was originally developed for mineral processing, and makes use of the adsorption of bubbles on solid particles to make them floatable in liquid media. By changing surface properties of plastics, selective flotation of different kinds of plastics has been attained (Shent et al., 1999; Fraunholcz, 2004). All these process have both good and bad points. Liquid-fluidized bed technology is another wet process worth searching for its availability in separating plastic wastes. It is known that mixed particles undergo segregation based on their density and size when fluidized in a liquid flow (Hu, 2002). However, little work has been conducted on its application to recycling plastics and there seems to be plenty of room for fundamental research. Most plastics, excluding polyethylene and polypropylene, have density of more than 1.0 g cm3 and their separation is crucial in many situations. The separation of polyvinylchloride and polyethylene terephthalate is the most prominent example (Paci and La Mantia, 1999; Pascoe and Hou, 1999). Unlike conventional sink and float separation using water, the technique is applicable to the mutual separation of these heavier-than-water plastics. In this study, we investigated the feasibility and limitations of this environmentally benign technique, with the emphasis being placed on the simplicity of equipment and procedure to make the process more realistic.
top. In some experiments a column with 46 mm ID was used. As sample, seven resin pellets with a variety of density were used; they were TPU (thermoplastic urethane), PVC (polyvinyl chloride), PBT (polybutylene terephthalate) and PET (polyethylene terephthalate). The density was determined by the immersion method using kerosene. A prescribed amount of resin pellets was inserted into the column with water, and a trace quantity of a surface active reagent was added to make the pellet surface hydrophilic. After degassing and mixing the resin bed, water was introduced upwardly from the inlet tube using a magnetic pump and discharged through the outlet tube. The flow rate was adjusted using a flow meter, and a height of the resin bed was recorded before and after the onset of water flow. The transition of the resin bed against the elapsed time was captured on a video and image-analyzed by color (Scion Image b 4.0.2) at an appropriate time interval. All the experiments were carried out at a room temperature.
2. Methods
3. Results and discussion
2.1. Binary separation of resin pellets in column
3.1. Binary separation of resin pellets in column
A glass column used in this study was 550 mm in height and 35 mm in inner diameter (ID), and a glass-filter was fixed at a position of 100 mm above the bottom. An inlet tube was vertically attached to the bottom of the column, while an outlet tube was horizontally connected to the side of the column at 50 mm below the
In Table 1 are summarized some physical properties of resin pellets used in this study. The density varies from 1.105 g cm3 for the lightest TPU to 1.396 g cm3 for the heaviest PVC. Each pellet has a rather uniform figure, and their sphere equivalent diameters based on volume fall within a small range of 3.10–4.17 mm. The
2.2. Resin withdrawal from column A series of operations consisting of charging, separation and withdrawal of samples was examined using a column equipped with six sample outlets. The sample outlets were attached to the side of the column at an interval of 50 mm with the undermost one situated at 22 mm above the filter. The inner diameter of the outlets was 17 mm, and the other dimensions were the same as those of the column without sample outlets. After separating plastics in the column in a similar manner as described above, each sorted portion, or layer, was withdrawn from an appropriate outlet. The withdrawal was carried out by removing a silicon cap from the sample outlets while keeping water flow to avoid a blockade of samples at the outlets. The resin content in the collections was determined by handselecting each resin by color and measuring its mass after drying. For the experiments, both resin pellets and crushed plastics were used as sample. The crushed samples were prepared by grinding bulk plastic products using a one-axis crushing machine and screening out fractions less than 1.4 mm.
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902 Table 1 Physical properties of resin pellets used in this study No.
Resin
Color
Density (g cm3)
Sphere equivalent diameter (mm)
1 2 3 4 5 6 7
TPU PVC PBT PVC PVC PET PVC
Black Red White Grayish-black Grayish-white White Green
1.105 1.216 1.288 1.308 1.358 1.372 1.396
3.10 3.45 3.47 3.32 3.88 3.36 4.17
following experiments were carried out on the assumption that the shape and the surface condition of pellets have negligible effects on their separation. Fig. 1 gives a typical time course for the separation of PVC (1.216 g cm3) and PET (1.372 g cm3). Each resin pellet (40 g) was packed in the column (ID 35 mm) and well mixed (a). On starting a flow of water at 1.15 l min1, the mixed bed expands and the pellets start to change mutual position rather freely in the upward flow. In other words, the resin bed reaches the state of fluidization. In Fig. 1, the photographs (b)–(e) show the situation after 0.5, 1.0, 2.0 and 5.0 min of the flow, respectively. Gradually, the lighter resin comes together at the upper part of the bed while the heavier one accumulates at the lower part; the two distinct layers are formed with time. The equilibrium is attained within 5 min, and the volume of the bed decreases to the initial value when the flow stops (f). In this manner, this simple technique easily realizes fine separation of heavier-than-water plastics having similar density.
895
The history of separation was picture-recorded and the abundance ratio of each resin at a particular position in the resin bed was determined by means of image-analysis based on color. A typical result for the separation of PVC (1.216 g cm3) and PET (1.372 g cm3) is shown in Fig. 2, where the percentage of the lighter resin, PVC, is plotted along the longitudinal line of the resin bed. At the beginning, both resins are well mixed and the percentage falls between 40% and 70% at any height of the bed. As the separation proceeds with time, the
Fig. 2. Distribution of two resins in packed bed as function of time.
Fig. 1. Time course of separation of PVC/PET in fluidized bed.
896
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
percentage becomes higher at the upper half of the bed and lower at the lower half. The separation seems to complete within 2 min, and the two resins are obtained as homogeneous layer with little contamination at both sides. A clear interface, however, was not observed between the two layers since they were moving actively, and the separation deteriorates at a middle portion of the resin bed, which we call as mixed layer hereinafter. This leads to a S-shaped curve along the vertical line, as seen in Fig. 2. In the liquid-fluidized bed method, a turbulent flow inevitably occurs to some extent in a column and this leads to the back mixing of samples at the interface. The width of the mixed layer varies depending on the experimental conditions. For the sake of convenience, however, the resin bed is divided into three equi-length portions along the longitudinal line to quantitatively estimate the separation. An average percentage of the lighter resin is calculated for the upper (Rupper) and the lower (Rlower) one-third portions, and the overall separation factor (S) is defined as their difference S ¼ Rupper Rlower
ð1Þ
In Fig. 3, the data in Fig. 2 are re-plotted as R and S against the elapsed time. As time proceeds, the percentage for the upper portion increases rapidly while that for the lower portion is decreasing to null. In consequence, the overall separation factor approaches 100% and complete separation is attained within 3 min. In the following experiments, the equilibrium time was set to be 5 min. Fig. 4 shows the effect of the flow rate on the separation of PVC (1.216 g cm3) and PET (1.372 g cm3) as well as on the expansion of the resin bed. The flow rate was varied from 1.00 to 2.50 l min1, which corresponds from 1.73 to 4.33 cm s1 in terms of the linear velocity.
Fig. 3. Time course of percentage of lighter resin and overall separation factor.
Fig. 4. Effect of flow rate on percentage of lighter resin and overall separation factor.
The bed expansion in a fluidized state is expressed as follows: E ¼ 100ðH =H ini Þ
ð2Þ
where H denotes the height of the resin bed and the subscript, ini, the initial state. The initial height was 88 mm. As expected, the bed expansion increases proportionally with an increase in the flow rate and reaches over 180% at 2.50 l min1. When the flow rate falls below 1.20 l min1, a lack of free motion of the resins in the bed results in incomplete separation. At a higher rate, however, the separation factor comes close to 100% and the two unmingled resins can be obtained as both upper and lower portions. The effect of the amount of each resin brought into the column was investigated on the separation of the two resin pellets, PVC (1.216 g cm3) and PET (1.372 g cm3), and typical results are summarized in Table 2. The bed expansion was invariably 113% at the flow rate of 1.20 l min1 independent of the amount. A change in the amount of each resin from 20 to 50 g gives no significant influence on the separation, and the separation factor more than 98% is obtained for all the runs. The initial ratio of the two resins was also found to be a small factor for the present system. In this case, 20% and 50% portions of the resin bed from top or bottom were used to calculate the average percentage, R, for the resins in smaller and larger amounts, respectively. When the flow rate was increased to 1.50 l min1, though the data are not shown here, similar results were obtained despite an increase in the bed expansion to 128%. In Table 3 are summarized the results for the separation of the above-mentioned resins using a column with a wider inner diameter of 46 mm. The amount of each resin was altered from 30 to 70 g and the bed expansion was 130% at the flow rate of 2.86 l min1. Compared
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
897
Table 2 Effect of resin amount on mutual separation of PVC/PET pellets PVC (g)
PET (g)
F (l min1)
H/Hini (%)
Rupper (%)
Rlower (%)
S (%)
20 30 40 50 20 60
20 30 40 50 60 20
1.20 1.20 1.20 1.20 1.20 1.20
113 113 113 114 112 118
100 100 100 100 100a 100c
1 2 0 0 0b 0d
99 98 100 100 100 100
a b c d
20% 50% 50% 20%
from from from from
top. bottom. top. bottom.
Table 3 Separation of PVC/PET pellets using column with 45.7 mm ID PVC (g)
PET (g)
F (l min1)
H/Hini (%)
Rupper (%)
Rlower (%)
S (%)
30 40 50 60 70 25 75
30 40 50 60 70 75 25
2.86 2.86 2.86 2.86 2.86 2.86 2.86
131 128 134 128 128 126 133
97 100 99 98 99 98a 100c
2 1 2 2 1 0b 5d
95 99 97 96 98 98 95
a b c d
10% 50% 50% 10%
from from from from
top. bottom. top. bottom.
with the results in Table 2 for the narrower column, a small increase in Rlower was observed. The flow of water is easily disturbed by the column configuration, and this might be a primary cause for a small decline in the separation. On the ground of the present simple technique, however, the separation factor as high as 95% is thought to be satisfactory. 3.2. Effect of difference in density of resin pellets on separation For mutual separation of plastics, a difference in their density (Dq) is thought to be a crucial factor in determining the efficiency of process. In this section, we selected several combinations of the resin pellets listed in Table 1 and investigated their separation as a function of the flow rate. In this series of experiments, the column with 35 mm inner diameter was used. Fig. 5 shows the relationship between the flow rate and the bed expansion for seven resin pellets with their density varying from 1.105 to 1.396 g cm3. The charged amount was fixed to be 40 g. As the flow rate increases, the bed expansion increases proportionally for all the resins examined. By increasing the resin density, the bed expansion is much suppressed and a higher flow rate
Fig. 5. Effects of flow rate and resin density on bed expansion.
is required to attain apparent fluidization. For instance, the expansion for the heaviest PVC pellet (1.396 g cm3) is no more than 130% even at the flow rate of 2.00 l min1. In Fig. 6 is plotted the bed expansion against the resin density at a constant flow rate of 1.00 l min1 in
898
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
both single and binary resin systems. For the binary system, an arithmetic average of the density of two resins was adopted and is shown as closed symbols. As previously described in Fig. 5, the expansion decreases with increasing the density, and the bed of the heaviest PVC hardly expands at this flow rate. The data for the binary system are also correlated well with the same single line, implying additivity of the bed expansion in this system. For several combinations of two resins, their mutual separation was examined and the results are shown in Fig. 7 in terms of the separation factor. As the flow rate increases, the separation improves and the separation factor approaches 100% in all the runs with the exception of the two combinations. When the density difference of
two resins is reduced to 0.072 g cm3, the separation factor hits its ceiling of 70% even at a higher flow rate. A similar trend is also observed in the binary system with the difference of 0.050 g cm3. In these cases, the sorted resin layers are inclined to be disturbed at their interface by the upward flow and the mixed layer becomes wider than those observed in the other successful cases. This leads to lower separation. Thus, the density difference of target samples is required to be 0.050 g cm3 at a minimum for achieving satisfactory separation in the present system. In Fig. 7, a clear relationship was not observed between the separation factor and the density difference of sample resins. Therefore, in Fig. 8 the data are replotted against the bed expansion, which is thought to represent both effects of the flow rate and the density difference. All the data show a similar tendency: the separation factor increases with an increase in the bed expansion and reaches plateau at 120% expansion. Even the combinations, which fail to complete separation, trace a similar line up to a point. In this manner, the separation is well correlated with the bed expansion irrespective of the density difference, and this finding might give useful information in the scale up of the technique. Separation factors were read out from Fig. 8 for three values of the bed expansion and plotted against the density difference of the corresponding resins in Fig. 9. As previously stated, the separation improves by increasing the flow rate, and reaches the maximum at the bed expansion of 120% or more. On the other hand, the expansion of 105% results in little or negligible separation due to a lack of free motion of the resin pellets in the flow. Moderate separation with the factor of 60% is given when the flow rate is adjusted to give
Fig. 7. Effects of flow rate and density difference on overall separation factor.
Fig. 8. Relationship between bed expansion and overall separation factor at various density difference.
Fig. 6. Relationship between resin density and bed expansion at constant flow rate.
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
899
maximum performance is available in separating plastics having the density difference between 0.1 and 0.2 g cm3 in this technology. 3.3. Resin withdrawal from column
Fig. 9. Relationship between density difference of resins and overall separation factor.
110% expansion. Though the data are rather scattered, a convex trend in the separation factor is seen against the density difference for each expansion. Unexpectedly lower efficiency is observed for separating resins having larger density difference. A plausible explanation is as follows. As the density difference increases, a higher flow rate is required to attain the prescribed bed expansion since the combination inevitably contains much heavier resin. This leads to an excess movement of lighter resin at the interface, resulting in a decline in the separation. A dull movement of heavier resin also helps to deteriorate the separation. Therefore, it is thought that the
In this section, a simple setup is proposed for attaining both separation and subsequent recovery of sorted plastics. A typical procedure is shown in Fig. 10, where the separation of two resin pellets, PVC (1.216 g cm3) and PET (1.372 g cm3), is demonstrated. Each 60 g of the resin is put into the column with six sample outlets, though only five are shown in the figure, and well mixed (Fig. 10a). After fluidizing the resin bed for 5 min under the water flow of 1.60 l min1, the mixed bed is separated into two distinct layers (Fig. 10b). The upper layer is withdrawn from the outlet just above the interface of the two layers (Fig. 10c). From the outlet just beneath the interface obtained a mixed collection of the two resins (Fig. 10d). Finally, the lower layer is obtained from the lowest outlet (Fig. 10e). The first and third collections exclusively contain the PVC and PET pellets, respectively. A quantitative description of mutual separation is shown in Table 4, where the weight and the purity of each resin are given for the respective three collections. The flow rate was arbitrarily adjusted to place the interface of the resin layers suitable for withdrawal from an appropriate sample outlet. In the case of the separation shown in Fig. 10, the PVC content in the upper collection reaches 95% while that of PET is as high as 98% in the lower one. PVC of 55.2 g is recovered from the
Fig. 10. Separation of PVC/PET and their withdrawal from column.
900
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
Table 4 Stepwise withdrawal of resin pellets from sample outlets of column Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
PVC PET
1.216 1.372
60.0 60.0
1.60
137
55.2 (95) 3.2 (5)
4.5 (14) 26.9 (86)
0.5 (2) 30.2 (98)
PVC PET
1.216 1.372
80.0 80.0
2.00
167
76.5 (96) 3.2 (4)
3.5 (13) 22.9 (87)
0.1 (0) 53.9 (100)
TPU PBT
1.105 1.288
60.0 60.0
1.15
139
58.2 (98) 1.1 (2)
1.6 (6) 26.5 (94)
0.2 (1) 33.0 (99)
initial 60.0 g and the yield is calculated to be 92%. That for PET is 50%. Although the middle collection contains much more PET than PVC, this proportion can vary depending on the flow rate, i.e. a relative position of the interface against the sample outlet. In a practical use, this mixed collection would be added to the next feed. A change in the amount of each resin to 80.0 g has no significant effects on the separation. Moreover, when the density difference is increased to 0.183 g cm3, both lighter resin, TPU, and heavier resin, PBT, are obtained with negligible contamination. In this case, the yield for TPU reaches as high as 97%. The results for the separation of two PVC pellets are shown in Table 5. A small density difference, 0.050 g cm3, between the two pellets originates from a difference in their filler content. The heavier PVC can be satisfactorily separated from the mixture with its purity of 96%, while the purity of the lighter one decreases to 79% in the upper collection. The middle collection also contains both PVCs. The experiment was repeated three times and each collection was added together to obtain sufficient samples for the next step. To improve the separation, the collections from both upper and middle portions were treated again using the same technique, and the results are also shown as second step in Table 5. For the upper collection, the purity of the lighter PVC is improved from 79% to 91%. The amount recovered is
134 g from the initial 161 g, corresponding to the yield of 83%. In the same manner, the heavier PVC is recovered exclusively from the middle collection in the first step to the lower collection in the second step. The middle collection in the second step also contains as high as 94% heavier PVC, and the yield amounts to 72% by adding them up. Thus, multi-stage fluidized bed separation is a useful method for separating plastics having close density. In Table 6 are summarized the results for the separation of three PVC pellets. The density difference of each resin is about 0.09 g cm3. After the separation at 1.70 l min1 for 5 min, the withdrawal of each resin layer was carried out by adjusting the flow rate to place the interfaces just at an appropriate sample outlet. The purity of the lighter PVC in the upper collection is 84%, while that of the heavier PVC in the lower collection is 94%. On the other hand, the middle collection contains all the three PVCs and the purity of the PVC having an intermediate density is reduced to 79%. Even without the withdrawal of mixed layers, reasonable separation is attained in the technique. Additionally, compared with the previous experiments a marked increase in the yield is observed due to a lack of sampling loss in mixed layers; the yield is found to be 92%, 77% and 88% for the PVCs in order of increasing density. Further improvement would be expected by precisely controlling the flow rate.
Table 5 Multi-stage separation of resin pellets having close density Resin
Density (g cm3)
First step (sum of three runs) PVC 1.308 PVC 1.358
W (g)
60.0 · 3 60.0 · 3
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
1.90
152
160.9 (79) 41.6 (21)
16.5 (20) 68.1 (80)
2.6 (4) 69.9 (96)
Second step of upper part PVC 1.308 PVC 1.358
160.9 41.6
1.53
136
134.2 (91) 12.6 (9)
19.7 (68) 9.1 (32)
6.9 (26) 19.9 (74)
Second step of middle part PVC 1.308 PVC 1.358
16.5 68.1
1.90
141
14.6 (44) 19.0 (56)
1.6 (6) 24.8 (94)
0.2 (1) 24.3 (99)
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
901
Table 6 Separation of three PVC pellets Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
PVC PVC PVC
1.216 1.308 1.396
50.0 50.0 50.0
1.70
142
45.8 (84) 9.1 (16) 0.0 (0)
4.0 (8) 38.3 (79) 6.2 (13)
0.2 (0) 2.6 (6) 43.8 (94)
Table 7 Application of present technique to crushed plastics Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each part, g (%) Upper
Middle
Lower
PC PVC
1.194 1.413
40.0 40.0
2.00
154
30.4 (83) 6.4 (17)
7.5 (40) 11.2 (60)
2.2 (9) 22.6 (91)
PC PVC
1.194 1.383
40.0 40.0
2.00
169
31.9 (86) 5.2 (14)
7.5 (37) 12.5 (63)
0.6 (3) 22.5 (97)
PA PC
1.138 1.194
40.0 40.0
1.40
152
28.7 (76) 9.3 (24)
8.9 (44) 11.2 (56)
2.4 (11) 19.7 (89)
Finally, the technique was applied to the crushed samples prepared from bulk plastic products with the intention of examining the effects of size distribution on the separation. The results are summarized in Table 7. Compared with the results for the uniform-shaped pellets described above, the separation deteriorates to some extent for all the combinations examined. Albeit a large difference in the density, 0.219 g cm3, in one combination, the purity is suppressed to 83% and 91% for the respective lighter and heavier resins. During the separation, classification based on the size was observed in each resin layer. Consequently, larger fragments of the lighter resin tended to blend with smaller fragments of the heavier one around the interface, resulting in the widening of the mixed layer. Although the separation might be improved by applying multi-stage treatment, pretreatment of samples such as sieving is desirable to apply. Given that the study had been carried out within the limited combinations of samples due to a difficulty in providing varied plastics with relevant density, reasonable performance was attained in terms of rapidity and accuracy with this simple method. Further study is expected. Now in our laboratory, the separation of lighter-than-water plastics using a similar technique is in the pipeline.
4. Conclusion Fluidized bed technology has been applied to the separation of heavier-than-water plastics. In an upward flow of water, a mixed resin bed was swelled to fluidize
and separated into respective resin layers dependent on their difference in density. The following is knowledge obtained in this study: (1) For the binary system of resin pellets, the separation was completed within a few minutes and little affected by the initial amount of the samples and the column diameter used. (2) The maximum efficiency was obtained when the flow rate was adjusted to swell the resin bed more than 120% of its initial height. (3) The density difference of 0.05 g cm3 at a minimum was required to attain reasonable separation. (4) Resin withdrawal after the separation was easily attained by using the column with several sample outlets vertically aligned along the column side. (5) Multi-stage operation was useful for separating plastics having very close density. (6) Size distribution of sample plastics showed a negative effect on the separation because of the classification effect in the flow.
References Altland, B.L., Cox, D., Enick, R.M., Beckman, E.J., 1995. Optimization of the high-pressure, near-critical liquid-based microsortation of recyclable post-consumer plastics. Resour. Conserv. Recy. 15, 203–217. Ferrara, G., Meloy, T.P., 1999. Low dense media process: a new process for low-density solid separation. Power Technol. 103, 151–155. Fraunholcz, N., 2004. Separation of waste plastics by froth flotation—a review, part I. Miner. Eng. 17, 261–268.
902
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
Hu, X., 2002. Prediction of the inversion velocity in the binary-solid liquid fluidized bed. Chem. Eng. Sci. 57, 3149– 3153. Paci, M., La Mantia, F.P., 1999. Influence of small amounts of polyvinylchloride on the recycling of polyethylene terephthalate. Polym. Degrad. Stabil. 63, 11–14. Pascoe, R.D., Hou, Y.Y., 1999. Investigation of the importance of particle shape and surface wettability on the separation of plastics in a larcodems separator. Miner. Eng. 12, 423–431.
Plastic Waste Management Institute (Japan). An Introduction to Plastic Recycling, May 2004. Shent, H., Pugh, R.J., Forssberg, E., 1999. A review of plastics waste recycling and the flotation of plastics. Resour. Conserv. Recy. 25, 85–109. Tsunekawa, M., Hori, K., Hirajima, T., Hiroyoshi, N., 2003. Jig separation of plastics particles. In: Proceedings of The Japan/Korea International Symposium on Resources Recycling and Material Sciences, Osaka, Japan, pp. 95– 100.
Separation of plastic mixtures using liquid-fluidized bed technology Takehiko Kinoshita, Kazuaki Okamoto, Koichi Yamaguchi, Shigendo Akita
*
Nagoya Municipal Industrial Research Institute, 3-4-41 Rokuban, Atsuta-ku, Nagoya 456-0058, Japan Received 20 April 2005; received in revised form 14 September 2005; accepted 14 September 2005 Available online 22 November 2005
Abstract Separation of heavier-than-water plastic mixtures had been investigated via the fluidization of their packed beds induced by an upward flow of water. The samples examined were resin pellets and crushed plastic products including PVC, PET and PBT. On the onset of a flow, a mixed bed was swelled to the state of fluidization and separated into layers of respective resins depending on their density. The effects of the flow rate, an amount of the samples and their density difference were examined on the separation of resin pellets. Under an appropriate condition, the process was completed within a few minutes, and satisfactory separation was attained when the density difference of the samples exceeded 0.05 g cm3. By using a column equipped with several sample outlets sorted resins could be recovered with their purity intact by withdrawing each layer successively from above through a suitable outlet. Multi-stage separation was also found to be effective in treating close-density samples. Compared with uniform-sized resin pellets, size distribution of crushed plastic samples deteriorated the separation to some extent. 2005 Elsevier Ltd. All rights reserved. Keywords: Separation; Plastic mixture; Fluidized bed; Density; Water
1. Introduction Waste management of plastics has been prompted by increasing awareness for environmental protection and resource conservation. In Japan, the domestic production of plastics in 2002 reached 13.85 million tons and the waste generation came to as high as 9.90 million tons in the same year. Since 2000, a number of recyclingrelating laws have been enacted by the government for the purpose of achieving sustainable development. * Corresponding author. Tel.: +81 52 654 9899; fax: +81 52 654 6788. E-mail address: [email protected] (S. Akita).
Thanks to these initiatives, the effective utilization rate of plastic wastes has been steadily increasing to 55% in 2002. Among 45% of the non-utilized plastic wastes, 17% is simply incinerated and the rest, 28%, is treated in landfilling (Plastic Waste Management Institute, 2004). Recycling of plastics is categorized into three main groups of mechanical recycling, feedstock recycling and energy recovery. The first one is preferred, if available, to the other two from the viewpoint of energy efficiency, and increasing its rate is of great significance. At present, most sources of mechanical recycled plastics come from the loss in production and processing because of their stable quality and supply. On the other hand, mechanical recycling of used plastic products
0045-6535/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2005.09.068
894
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
from end users is still in the minority, since the wastes often consist of many kinds of plastics and need to be dismantled and sorted based on their raw material. For the separation of plastic wastes, wet process is relatively easy to apply and a multiple of techniques have been developed and some are in practical use. The followings are typical examples. Sink and float separation is the most fundamental and mainly used for separating heavier- and lighter-than-water plastics. Some modifications have been done on liquid media to meet a wide density range of plastics (Altland et al., 1995; Ferrara and Meloy, 1999; Pascoe and Hou, 1999). Jig separation, which has been widely used in coal separation, was reported to successfully separate small plastics particles of polyethylene and polyvinyl chloride (Tsunekawa et al., 2003). Recently, froth flotation is attracting attention as a recycling operation for plastic wastes. The technique was originally developed for mineral processing, and makes use of the adsorption of bubbles on solid particles to make them floatable in liquid media. By changing surface properties of plastics, selective flotation of different kinds of plastics has been attained (Shent et al., 1999; Fraunholcz, 2004). All these process have both good and bad points. Liquid-fluidized bed technology is another wet process worth searching for its availability in separating plastic wastes. It is known that mixed particles undergo segregation based on their density and size when fluidized in a liquid flow (Hu, 2002). However, little work has been conducted on its application to recycling plastics and there seems to be plenty of room for fundamental research. Most plastics, excluding polyethylene and polypropylene, have density of more than 1.0 g cm3 and their separation is crucial in many situations. The separation of polyvinylchloride and polyethylene terephthalate is the most prominent example (Paci and La Mantia, 1999; Pascoe and Hou, 1999). Unlike conventional sink and float separation using water, the technique is applicable to the mutual separation of these heavier-than-water plastics. In this study, we investigated the feasibility and limitations of this environmentally benign technique, with the emphasis being placed on the simplicity of equipment and procedure to make the process more realistic.
top. In some experiments a column with 46 mm ID was used. As sample, seven resin pellets with a variety of density were used; they were TPU (thermoplastic urethane), PVC (polyvinyl chloride), PBT (polybutylene terephthalate) and PET (polyethylene terephthalate). The density was determined by the immersion method using kerosene. A prescribed amount of resin pellets was inserted into the column with water, and a trace quantity of a surface active reagent was added to make the pellet surface hydrophilic. After degassing and mixing the resin bed, water was introduced upwardly from the inlet tube using a magnetic pump and discharged through the outlet tube. The flow rate was adjusted using a flow meter, and a height of the resin bed was recorded before and after the onset of water flow. The transition of the resin bed against the elapsed time was captured on a video and image-analyzed by color (Scion Image b 4.0.2) at an appropriate time interval. All the experiments were carried out at a room temperature.
2. Methods
3. Results and discussion
2.1. Binary separation of resin pellets in column
3.1. Binary separation of resin pellets in column
A glass column used in this study was 550 mm in height and 35 mm in inner diameter (ID), and a glass-filter was fixed at a position of 100 mm above the bottom. An inlet tube was vertically attached to the bottom of the column, while an outlet tube was horizontally connected to the side of the column at 50 mm below the
In Table 1 are summarized some physical properties of resin pellets used in this study. The density varies from 1.105 g cm3 for the lightest TPU to 1.396 g cm3 for the heaviest PVC. Each pellet has a rather uniform figure, and their sphere equivalent diameters based on volume fall within a small range of 3.10–4.17 mm. The
2.2. Resin withdrawal from column A series of operations consisting of charging, separation and withdrawal of samples was examined using a column equipped with six sample outlets. The sample outlets were attached to the side of the column at an interval of 50 mm with the undermost one situated at 22 mm above the filter. The inner diameter of the outlets was 17 mm, and the other dimensions were the same as those of the column without sample outlets. After separating plastics in the column in a similar manner as described above, each sorted portion, or layer, was withdrawn from an appropriate outlet. The withdrawal was carried out by removing a silicon cap from the sample outlets while keeping water flow to avoid a blockade of samples at the outlets. The resin content in the collections was determined by handselecting each resin by color and measuring its mass after drying. For the experiments, both resin pellets and crushed plastics were used as sample. The crushed samples were prepared by grinding bulk plastic products using a one-axis crushing machine and screening out fractions less than 1.4 mm.
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902 Table 1 Physical properties of resin pellets used in this study No.
Resin
Color
Density (g cm3)
Sphere equivalent diameter (mm)
1 2 3 4 5 6 7
TPU PVC PBT PVC PVC PET PVC
Black Red White Grayish-black Grayish-white White Green
1.105 1.216 1.288 1.308 1.358 1.372 1.396
3.10 3.45 3.47 3.32 3.88 3.36 4.17
following experiments were carried out on the assumption that the shape and the surface condition of pellets have negligible effects on their separation. Fig. 1 gives a typical time course for the separation of PVC (1.216 g cm3) and PET (1.372 g cm3). Each resin pellet (40 g) was packed in the column (ID 35 mm) and well mixed (a). On starting a flow of water at 1.15 l min1, the mixed bed expands and the pellets start to change mutual position rather freely in the upward flow. In other words, the resin bed reaches the state of fluidization. In Fig. 1, the photographs (b)–(e) show the situation after 0.5, 1.0, 2.0 and 5.0 min of the flow, respectively. Gradually, the lighter resin comes together at the upper part of the bed while the heavier one accumulates at the lower part; the two distinct layers are formed with time. The equilibrium is attained within 5 min, and the volume of the bed decreases to the initial value when the flow stops (f). In this manner, this simple technique easily realizes fine separation of heavier-than-water plastics having similar density.
895
The history of separation was picture-recorded and the abundance ratio of each resin at a particular position in the resin bed was determined by means of image-analysis based on color. A typical result for the separation of PVC (1.216 g cm3) and PET (1.372 g cm3) is shown in Fig. 2, where the percentage of the lighter resin, PVC, is plotted along the longitudinal line of the resin bed. At the beginning, both resins are well mixed and the percentage falls between 40% and 70% at any height of the bed. As the separation proceeds with time, the
Fig. 2. Distribution of two resins in packed bed as function of time.
Fig. 1. Time course of separation of PVC/PET in fluidized bed.
896
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
percentage becomes higher at the upper half of the bed and lower at the lower half. The separation seems to complete within 2 min, and the two resins are obtained as homogeneous layer with little contamination at both sides. A clear interface, however, was not observed between the two layers since they were moving actively, and the separation deteriorates at a middle portion of the resin bed, which we call as mixed layer hereinafter. This leads to a S-shaped curve along the vertical line, as seen in Fig. 2. In the liquid-fluidized bed method, a turbulent flow inevitably occurs to some extent in a column and this leads to the back mixing of samples at the interface. The width of the mixed layer varies depending on the experimental conditions. For the sake of convenience, however, the resin bed is divided into three equi-length portions along the longitudinal line to quantitatively estimate the separation. An average percentage of the lighter resin is calculated for the upper (Rupper) and the lower (Rlower) one-third portions, and the overall separation factor (S) is defined as their difference S ¼ Rupper Rlower
ð1Þ
In Fig. 3, the data in Fig. 2 are re-plotted as R and S against the elapsed time. As time proceeds, the percentage for the upper portion increases rapidly while that for the lower portion is decreasing to null. In consequence, the overall separation factor approaches 100% and complete separation is attained within 3 min. In the following experiments, the equilibrium time was set to be 5 min. Fig. 4 shows the effect of the flow rate on the separation of PVC (1.216 g cm3) and PET (1.372 g cm3) as well as on the expansion of the resin bed. The flow rate was varied from 1.00 to 2.50 l min1, which corresponds from 1.73 to 4.33 cm s1 in terms of the linear velocity.
Fig. 3. Time course of percentage of lighter resin and overall separation factor.
Fig. 4. Effect of flow rate on percentage of lighter resin and overall separation factor.
The bed expansion in a fluidized state is expressed as follows: E ¼ 100ðH =H ini Þ
ð2Þ
where H denotes the height of the resin bed and the subscript, ini, the initial state. The initial height was 88 mm. As expected, the bed expansion increases proportionally with an increase in the flow rate and reaches over 180% at 2.50 l min1. When the flow rate falls below 1.20 l min1, a lack of free motion of the resins in the bed results in incomplete separation. At a higher rate, however, the separation factor comes close to 100% and the two unmingled resins can be obtained as both upper and lower portions. The effect of the amount of each resin brought into the column was investigated on the separation of the two resin pellets, PVC (1.216 g cm3) and PET (1.372 g cm3), and typical results are summarized in Table 2. The bed expansion was invariably 113% at the flow rate of 1.20 l min1 independent of the amount. A change in the amount of each resin from 20 to 50 g gives no significant influence on the separation, and the separation factor more than 98% is obtained for all the runs. The initial ratio of the two resins was also found to be a small factor for the present system. In this case, 20% and 50% portions of the resin bed from top or bottom were used to calculate the average percentage, R, for the resins in smaller and larger amounts, respectively. When the flow rate was increased to 1.50 l min1, though the data are not shown here, similar results were obtained despite an increase in the bed expansion to 128%. In Table 3 are summarized the results for the separation of the above-mentioned resins using a column with a wider inner diameter of 46 mm. The amount of each resin was altered from 30 to 70 g and the bed expansion was 130% at the flow rate of 2.86 l min1. Compared
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
897
Table 2 Effect of resin amount on mutual separation of PVC/PET pellets PVC (g)
PET (g)
F (l min1)
H/Hini (%)
Rupper (%)
Rlower (%)
S (%)
20 30 40 50 20 60
20 30 40 50 60 20
1.20 1.20 1.20 1.20 1.20 1.20
113 113 113 114 112 118
100 100 100 100 100a 100c
1 2 0 0 0b 0d
99 98 100 100 100 100
a b c d
20% 50% 50% 20%
from from from from
top. bottom. top. bottom.
Table 3 Separation of PVC/PET pellets using column with 45.7 mm ID PVC (g)
PET (g)
F (l min1)
H/Hini (%)
Rupper (%)
Rlower (%)
S (%)
30 40 50 60 70 25 75
30 40 50 60 70 75 25
2.86 2.86 2.86 2.86 2.86 2.86 2.86
131 128 134 128 128 126 133
97 100 99 98 99 98a 100c
2 1 2 2 1 0b 5d
95 99 97 96 98 98 95
a b c d
10% 50% 50% 10%
from from from from
top. bottom. top. bottom.
with the results in Table 2 for the narrower column, a small increase in Rlower was observed. The flow of water is easily disturbed by the column configuration, and this might be a primary cause for a small decline in the separation. On the ground of the present simple technique, however, the separation factor as high as 95% is thought to be satisfactory. 3.2. Effect of difference in density of resin pellets on separation For mutual separation of plastics, a difference in their density (Dq) is thought to be a crucial factor in determining the efficiency of process. In this section, we selected several combinations of the resin pellets listed in Table 1 and investigated their separation as a function of the flow rate. In this series of experiments, the column with 35 mm inner diameter was used. Fig. 5 shows the relationship between the flow rate and the bed expansion for seven resin pellets with their density varying from 1.105 to 1.396 g cm3. The charged amount was fixed to be 40 g. As the flow rate increases, the bed expansion increases proportionally for all the resins examined. By increasing the resin density, the bed expansion is much suppressed and a higher flow rate
Fig. 5. Effects of flow rate and resin density on bed expansion.
is required to attain apparent fluidization. For instance, the expansion for the heaviest PVC pellet (1.396 g cm3) is no more than 130% even at the flow rate of 2.00 l min1. In Fig. 6 is plotted the bed expansion against the resin density at a constant flow rate of 1.00 l min1 in
898
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
both single and binary resin systems. For the binary system, an arithmetic average of the density of two resins was adopted and is shown as closed symbols. As previously described in Fig. 5, the expansion decreases with increasing the density, and the bed of the heaviest PVC hardly expands at this flow rate. The data for the binary system are also correlated well with the same single line, implying additivity of the bed expansion in this system. For several combinations of two resins, their mutual separation was examined and the results are shown in Fig. 7 in terms of the separation factor. As the flow rate increases, the separation improves and the separation factor approaches 100% in all the runs with the exception of the two combinations. When the density difference of
two resins is reduced to 0.072 g cm3, the separation factor hits its ceiling of 70% even at a higher flow rate. A similar trend is also observed in the binary system with the difference of 0.050 g cm3. In these cases, the sorted resin layers are inclined to be disturbed at their interface by the upward flow and the mixed layer becomes wider than those observed in the other successful cases. This leads to lower separation. Thus, the density difference of target samples is required to be 0.050 g cm3 at a minimum for achieving satisfactory separation in the present system. In Fig. 7, a clear relationship was not observed between the separation factor and the density difference of sample resins. Therefore, in Fig. 8 the data are replotted against the bed expansion, which is thought to represent both effects of the flow rate and the density difference. All the data show a similar tendency: the separation factor increases with an increase in the bed expansion and reaches plateau at 120% expansion. Even the combinations, which fail to complete separation, trace a similar line up to a point. In this manner, the separation is well correlated with the bed expansion irrespective of the density difference, and this finding might give useful information in the scale up of the technique. Separation factors were read out from Fig. 8 for three values of the bed expansion and plotted against the density difference of the corresponding resins in Fig. 9. As previously stated, the separation improves by increasing the flow rate, and reaches the maximum at the bed expansion of 120% or more. On the other hand, the expansion of 105% results in little or negligible separation due to a lack of free motion of the resin pellets in the flow. Moderate separation with the factor of 60% is given when the flow rate is adjusted to give
Fig. 7. Effects of flow rate and density difference on overall separation factor.
Fig. 8. Relationship between bed expansion and overall separation factor at various density difference.
Fig. 6. Relationship between resin density and bed expansion at constant flow rate.
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
899
maximum performance is available in separating plastics having the density difference between 0.1 and 0.2 g cm3 in this technology. 3.3. Resin withdrawal from column
Fig. 9. Relationship between density difference of resins and overall separation factor.
110% expansion. Though the data are rather scattered, a convex trend in the separation factor is seen against the density difference for each expansion. Unexpectedly lower efficiency is observed for separating resins having larger density difference. A plausible explanation is as follows. As the density difference increases, a higher flow rate is required to attain the prescribed bed expansion since the combination inevitably contains much heavier resin. This leads to an excess movement of lighter resin at the interface, resulting in a decline in the separation. A dull movement of heavier resin also helps to deteriorate the separation. Therefore, it is thought that the
In this section, a simple setup is proposed for attaining both separation and subsequent recovery of sorted plastics. A typical procedure is shown in Fig. 10, where the separation of two resin pellets, PVC (1.216 g cm3) and PET (1.372 g cm3), is demonstrated. Each 60 g of the resin is put into the column with six sample outlets, though only five are shown in the figure, and well mixed (Fig. 10a). After fluidizing the resin bed for 5 min under the water flow of 1.60 l min1, the mixed bed is separated into two distinct layers (Fig. 10b). The upper layer is withdrawn from the outlet just above the interface of the two layers (Fig. 10c). From the outlet just beneath the interface obtained a mixed collection of the two resins (Fig. 10d). Finally, the lower layer is obtained from the lowest outlet (Fig. 10e). The first and third collections exclusively contain the PVC and PET pellets, respectively. A quantitative description of mutual separation is shown in Table 4, where the weight and the purity of each resin are given for the respective three collections. The flow rate was arbitrarily adjusted to place the interface of the resin layers suitable for withdrawal from an appropriate sample outlet. In the case of the separation shown in Fig. 10, the PVC content in the upper collection reaches 95% while that of PET is as high as 98% in the lower one. PVC of 55.2 g is recovered from the
Fig. 10. Separation of PVC/PET and their withdrawal from column.
900
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
Table 4 Stepwise withdrawal of resin pellets from sample outlets of column Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
PVC PET
1.216 1.372
60.0 60.0
1.60
137
55.2 (95) 3.2 (5)
4.5 (14) 26.9 (86)
0.5 (2) 30.2 (98)
PVC PET
1.216 1.372
80.0 80.0
2.00
167
76.5 (96) 3.2 (4)
3.5 (13) 22.9 (87)
0.1 (0) 53.9 (100)
TPU PBT
1.105 1.288
60.0 60.0
1.15
139
58.2 (98) 1.1 (2)
1.6 (6) 26.5 (94)
0.2 (1) 33.0 (99)
initial 60.0 g and the yield is calculated to be 92%. That for PET is 50%. Although the middle collection contains much more PET than PVC, this proportion can vary depending on the flow rate, i.e. a relative position of the interface against the sample outlet. In a practical use, this mixed collection would be added to the next feed. A change in the amount of each resin to 80.0 g has no significant effects on the separation. Moreover, when the density difference is increased to 0.183 g cm3, both lighter resin, TPU, and heavier resin, PBT, are obtained with negligible contamination. In this case, the yield for TPU reaches as high as 97%. The results for the separation of two PVC pellets are shown in Table 5. A small density difference, 0.050 g cm3, between the two pellets originates from a difference in their filler content. The heavier PVC can be satisfactorily separated from the mixture with its purity of 96%, while the purity of the lighter one decreases to 79% in the upper collection. The middle collection also contains both PVCs. The experiment was repeated three times and each collection was added together to obtain sufficient samples for the next step. To improve the separation, the collections from both upper and middle portions were treated again using the same technique, and the results are also shown as second step in Table 5. For the upper collection, the purity of the lighter PVC is improved from 79% to 91%. The amount recovered is
134 g from the initial 161 g, corresponding to the yield of 83%. In the same manner, the heavier PVC is recovered exclusively from the middle collection in the first step to the lower collection in the second step. The middle collection in the second step also contains as high as 94% heavier PVC, and the yield amounts to 72% by adding them up. Thus, multi-stage fluidized bed separation is a useful method for separating plastics having close density. In Table 6 are summarized the results for the separation of three PVC pellets. The density difference of each resin is about 0.09 g cm3. After the separation at 1.70 l min1 for 5 min, the withdrawal of each resin layer was carried out by adjusting the flow rate to place the interfaces just at an appropriate sample outlet. The purity of the lighter PVC in the upper collection is 84%, while that of the heavier PVC in the lower collection is 94%. On the other hand, the middle collection contains all the three PVCs and the purity of the PVC having an intermediate density is reduced to 79%. Even without the withdrawal of mixed layers, reasonable separation is attained in the technique. Additionally, compared with the previous experiments a marked increase in the yield is observed due to a lack of sampling loss in mixed layers; the yield is found to be 92%, 77% and 88% for the PVCs in order of increasing density. Further improvement would be expected by precisely controlling the flow rate.
Table 5 Multi-stage separation of resin pellets having close density Resin
Density (g cm3)
First step (sum of three runs) PVC 1.308 PVC 1.358
W (g)
60.0 · 3 60.0 · 3
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
1.90
152
160.9 (79) 41.6 (21)
16.5 (20) 68.1 (80)
2.6 (4) 69.9 (96)
Second step of upper part PVC 1.308 PVC 1.358
160.9 41.6
1.53
136
134.2 (91) 12.6 (9)
19.7 (68) 9.1 (32)
6.9 (26) 19.9 (74)
Second step of middle part PVC 1.308 PVC 1.358
16.5 68.1
1.90
141
14.6 (44) 19.0 (56)
1.6 (6) 24.8 (94)
0.2 (1) 24.3 (99)
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
901
Table 6 Separation of three PVC pellets Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each collection, g (%) Upper
Middle
Lower
PVC PVC PVC
1.216 1.308 1.396
50.0 50.0 50.0
1.70
142
45.8 (84) 9.1 (16) 0.0 (0)
4.0 (8) 38.3 (79) 6.2 (13)
0.2 (0) 2.6 (6) 43.8 (94)
Table 7 Application of present technique to crushed plastics Resin
Density (g cm3)
W (g)
F (l min1)
H/Hini (%)
Weight in each part, g (%) Upper
Middle
Lower
PC PVC
1.194 1.413
40.0 40.0
2.00
154
30.4 (83) 6.4 (17)
7.5 (40) 11.2 (60)
2.2 (9) 22.6 (91)
PC PVC
1.194 1.383
40.0 40.0
2.00
169
31.9 (86) 5.2 (14)
7.5 (37) 12.5 (63)
0.6 (3) 22.5 (97)
PA PC
1.138 1.194
40.0 40.0
1.40
152
28.7 (76) 9.3 (24)
8.9 (44) 11.2 (56)
2.4 (11) 19.7 (89)
Finally, the technique was applied to the crushed samples prepared from bulk plastic products with the intention of examining the effects of size distribution on the separation. The results are summarized in Table 7. Compared with the results for the uniform-shaped pellets described above, the separation deteriorates to some extent for all the combinations examined. Albeit a large difference in the density, 0.219 g cm3, in one combination, the purity is suppressed to 83% and 91% for the respective lighter and heavier resins. During the separation, classification based on the size was observed in each resin layer. Consequently, larger fragments of the lighter resin tended to blend with smaller fragments of the heavier one around the interface, resulting in the widening of the mixed layer. Although the separation might be improved by applying multi-stage treatment, pretreatment of samples such as sieving is desirable to apply. Given that the study had been carried out within the limited combinations of samples due to a difficulty in providing varied plastics with relevant density, reasonable performance was attained in terms of rapidity and accuracy with this simple method. Further study is expected. Now in our laboratory, the separation of lighter-than-water plastics using a similar technique is in the pipeline.
4. Conclusion Fluidized bed technology has been applied to the separation of heavier-than-water plastics. In an upward flow of water, a mixed resin bed was swelled to fluidize
and separated into respective resin layers dependent on their difference in density. The following is knowledge obtained in this study: (1) For the binary system of resin pellets, the separation was completed within a few minutes and little affected by the initial amount of the samples and the column diameter used. (2) The maximum efficiency was obtained when the flow rate was adjusted to swell the resin bed more than 120% of its initial height. (3) The density difference of 0.05 g cm3 at a minimum was required to attain reasonable separation. (4) Resin withdrawal after the separation was easily attained by using the column with several sample outlets vertically aligned along the column side. (5) Multi-stage operation was useful for separating plastics having very close density. (6) Size distribution of sample plastics showed a negative effect on the separation because of the classification effect in the flow.
References Altland, B.L., Cox, D., Enick, R.M., Beckman, E.J., 1995. Optimization of the high-pressure, near-critical liquid-based microsortation of recyclable post-consumer plastics. Resour. Conserv. Recy. 15, 203–217. Ferrara, G., Meloy, T.P., 1999. Low dense media process: a new process for low-density solid separation. Power Technol. 103, 151–155. Fraunholcz, N., 2004. Separation of waste plastics by froth flotation—a review, part I. Miner. Eng. 17, 261–268.
902
T. Kinoshita et al. / Chemosphere 63 (2006) 893–902
Hu, X., 2002. Prediction of the inversion velocity in the binary-solid liquid fluidized bed. Chem. Eng. Sci. 57, 3149– 3153. Paci, M., La Mantia, F.P., 1999. Influence of small amounts of polyvinylchloride on the recycling of polyethylene terephthalate. Polym. Degrad. Stabil. 63, 11–14. Pascoe, R.D., Hou, Y.Y., 1999. Investigation of the importance of particle shape and surface wettability on the separation of plastics in a larcodems separator. Miner. Eng. 12, 423–431.
Plastic Waste Management Institute (Japan). An Introduction to Plastic Recycling, May 2004. Shent, H., Pugh, R.J., Forssberg, E., 1999. A review of plastics waste recycling and the flotation of plastics. Resour. Conserv. Recy. 25, 85–109. Tsunekawa, M., Hori, K., Hirajima, T., Hiroyoshi, N., 2003. Jig separation of plastics particles. In: Proceedings of The Japan/Korea International Symposium on Resources Recycling and Material Sciences, Osaka, Japan, pp. 95– 100.