title :- weee plastics separation technologies

Final Report –

DEFRA –

Waste and Resources Evidence Programme

Title :-

WEEE Plastics Separation Technologies

Project code : WRO 202 (WRT 095)

Project Title : Developing bulk WEEE polymer separation and analysis techniques

Published : July 2007 Authors : Keith Freegard, Gayle Tan, Sebastien Frisch

Table of Contents Page 1. Summary 2

2. Introduction 4

3. Size Reduction 7

4. Metal Separation 29

5. Wood & Rubber Removal 59

6. Centrifugal Separation 67

7. Spectrographic Methods 104

8. Colour Sorting 122

9. Electrostatic Techniques 141

10. Analytical Methods 162

11. Sampling Methods 167

12. LCA and Environmental Impact 169

13. Conclusions, Recommendations & Implications 172

14. Appendix – Glossary of Polymer names 177

15. References 178

WEEE Plastic Separation Technologies 1

1 Executive Summary This is the Final report of a two year DEFRA funded project carried out by Axion Recycling Ltd which investigated sorting and separation technologies for the processing of WEEE plastics waste streams. The work commenced in June 2005 and ended in June 2007. Progress during the project has been reported in three interim reports. In these reports the development of a laboratory analysis technique for the rapid assessment of polymer types in mixture has been described. This main aim of this final report is to disseminate the acquired knowledge and findings of over twenty equipment trials carried out on separation and sorting technologies. It has therefore been assumed that the target audience for this document are process designers or plant operators involved in the WEEE treatment sector. It will also be of interest to Local Authorities and others engaged in collecting WEEE, in terms of segregation and care in handling. The plastics waste stream arising from the primary WEEE treatment processors in the UK and Europe is not a simple mixture of a few polymer types. Any process that attempts to extract a high-grade polymer product from this mixed stream of material must be able to handle a wide range of contaminants. This can include:-

• Metal – ferrous & non-ferrous – 2-7% • Dust / fines – e.g. PU foam – up to 15% • Rubber / elastomers • Fibres / fluff • Wood / Paper • Cable / electrical components • Dirt / grass • Glass / stones • AND some PLASTIC!

The report therefore includes assessment of pre-treatment processes that are necessary to clean-up the plastic material before any polymer sorting can take place. Several different plastic sorting technologies were trialled using large scale samples of real-life WEEE plastic material, where practicable. These are grouped into liquid phase, density separation methods and dry sorting systems using spectrographic, electrostatic and X-ray techniques. A brief description of the sampling methodology, the main analytical instruments employed and the environmental impact of plastics recycling is also included. The detailed conclusions from the trials of individual technologies are included in the body of the report, however the main conclusions are as follows:-


A complete process to treat the co-mingled materials found in WEEE plastics waste will be divided into two main stages:- Firstly, the plant will require technologies that carry out the initial clean-up of the material, by removal of the non-plastic contaminants as follows:-

• Size Reduction – primary • Metal removal – ferrous / non-ferrous / stainless steel • Fines and dust extraction • Wood and rubber removal • Stone and glass removal • Further Size Reduction - secondary

The above process steps should deliver a clean, accurately-sized mix of granules into the second polymer separation stage, at which point there is a choice of technologies to be used:-

• Increased-G Liquid Density Separation • Infra-Red Light sorting • X-Ray Transmission sorting • Electrostatics Separations • Colour Sorting

The benefits and limitations of these technologies are described in detail in the report. A list of recommendations is given to assist process designers and potential plant operators with the selection of suitable equipment for the task of recycling this complex, but valuable, waste resource. Axion Recycling gratefully acknowledge DEFRA funding support for this research project. Many examples of commercially available equipment and the names of machinery suppliers are quoted in reporting this work programme. Neither Axion Recycling nor DEFRA wish to offer any endorsement of particular brand names or manufacturers in doing so. K M Freegard July 2007


2 Introduction This report concludes a two year project carried out by Axion Recycling Limited on behalf of DEFRA Waste Evidence Branch to investigate separation and recycling technologies for WEEE plastic waste streams. The UK has implemented the EU WEEE Directive in Jan 2007 and the tonnages of waste electronics being collected separately for specialist treatment is now on the increase. It is expected that the total UK tonnage of waste plastic arising from this material will be between 200 – 300 thousand tonnes pa. A large proportion of this plastic will have to be recycled if the targets set for recycling of each WEEE catagory are going to be met (e.g. Target of 65% by re-use or recyling for components from IT, telecomms and consumer equipment). Extraction of the metal fraction alone will yield less than 50% recycling rate in most categories. The main aim of this final report is to disseminate the findings of the research project to those people involved in the challenge of recycling WEEE plastic in the UK and Europe. The work is therefore described by a series of trial reports on the different stages involved in a total process. In each case we have given an introduction to the technology used, together with an explanation of the theory behind each of the novel separation methods. The individual trials report are presented in summary only and comments are made where appropriate on the performance of each piece of equipment for the set task. The author is of the opinion that photographs and diagrams are the best way to describe the type of work carried out in the project, so these have been used to explain the technologies and to give visual representation of the results. At the end of each section we have included some web-links and contact information so that the reader can easily make contact with the manufacturers of the items of equipment used in the trials. We have also listed some alternative suppliers of similar process machinery for those who want to conduct further investigations of particular technologies. Abbreviations of the common polymer types have been used throughout the report, so a brief glossary of these terms is given as an appendix to aid the unfamiliar reader. We trust that this format of the report makes it a useful point of reference for those production engineers and process technologists who wish to learn about the available technologies to handle this complex, but valuable, waste resource. Keith Freegard MIChemE MBA


2.1 Processing Requirement for WEEE Plastics The first task in the project was to define the range of materials under investigation and to set limits on the processing steps needed to turn the input waste into a finished product. On looking at the type of waste plastic being generated by primary WEEE processing plants across Europe, it was clear that the task of ‘plastics separation’ was not going to be limited to a single-stage split of two mixed polymers to produce a high-purity finished product. The processing need is best described by Figure 2.1 below; which shows the flow of collected WEEE material through primary treatment and onto recycled end-products:-

Figure 2.1 WEEE Processing

WEEE Collection Schemes

The process limits are shown by the dotted rectangle on the diagram. Input material is the plastic fraction produced by the primary WEEE treatment plants, and it comes from the following main streams:-

• Treatment of Fridges under ODS Regulations • Treatment of CRTs under Hazardous Waste Regs. • Plastic from IT Recyclers • Mixed Small WEEE treatment

Each of these streams has particular requirements for processing and contains a characteristic mix of polymer types related to the ‘streaming’ that has occurred due to the differing regulations that apply. The delivery format of the input streams can also vary, including:-

• Whole, dismantled casings • Rough or fine shredded plastic • Baled plastic

Primary WEEE Treatment Plant

WEEE Plastics Plant

a ‘Re-Processor’

Metal Glass Waste

Plastic

Waste Metal

High-grade Polymer Chips


These materials can be delivered in big-bags, on pallets or in bulk-tipper

utput products from the process are:-

• Some residual metal , fines, wood, glass etc

or the purposes of this study, the final conversion of the sorted, clean chips

he input materials are still defined as a ‘Waste’ under UK legislation, so the

t

t s,

he processes required to convert the infeed waste into a clean, single-nput

• Metal – ferrous & non-ferrous – 2-7%

r mponents

es STIC!

he processes studied for the large-scale separation of polymers found in

he following sections describe the equipment types investigated and the

skips. O

• Waste – such as dust• Cleaned, sorted polymer chips or flakes. • OR Extruded polymer pellets

Finto an extruded pellet has not been included because this technology is well established in the marketplace. TRe-Processor site needs to be registered as a waste transfer station in order to be able to accept, store and handle the material. At present the main outpuplastic material, in the form of cleaned chips, is still classified as a waste material. It can only be delivered to sites that are also registered to accepwaste. If the plastic is converted into extruded pellet using a thermal procesthen it is no longer classified as ‘waste’ and it can be sold-on as a normal commercial ‘product’. Tpolymer type chip must therefore be able to handle the variations in the imaterial and also be able to remove the wide range of ‘non-plastics’ mixed in with the input waste. This can include:-

• Dust / fines – e.g. PU foam – up to 15%• Rubber / elastomers • Fibres / fluff • Wood / Pape• Cable / elec. co• Dirt / grass • Glass / ston• AND some PLA

TWEEE plastic waste, therefore needed to include a wider set of unit operations than simply plastic/plastic separation steps. Ttrials carried out using real-life WEEE waste material samples.


3.0 Size Reduction One of the first items to consider for any WEEE plastics recycling plant is the need for size reduction as part of an efficient overall process. This is because the waste plastic feed-stocks can arrive in a number of different formats into the recycling factory, depending upon the design of the upstream primary WEEE treatment plant. For example the following are typical of the types of plastic waste material seen in the European market:- Stream Waste Plastic Delivered Format CRT Waste HIPS ex TV casings Whole casings ABS ex VDU monitors Baled – 300 – 600 kilo Rough shredded 25 – 150mm Fridge Waste PS ex fridge liners Always shredded 5 – 50mm IT waste ABS, PC/ABS, HIPS Whole components Rough Shred 25 – 150mm From this table, it is clear that an understanding of the delivery format is needed to enable sensible decisions to be made about the ‘correct’ type of size reduction required. 3.1 Why size reduce? Size reduction of the input material is needed for several reasons:-

• To reduce large components to a manageable size for downstream processing

• To liberate metal and other parts from large pieces • To achieve regular particle size, often important for efficient separation • To reach the required size distribution for output product specifications

The task of size reduction can be broken down into three broad stages:- 3.1.1 Shredding of Whole Casings This often is an integral part of the upstream WEEE process, when the plastic components are broken away from the metal and other parts of the whole waste electrical item. However in manual dismantling operations, such as CRT processing sites, the outer casing is removed whole with no further processing. This can yield large back-of-TV casings of dimensions up to 1000 x 500 x 500mm overall, with weight up to 4 kilos per item.


Figure 3.1 Large whole TV casing

3.1.2 Oversize Shredding to Regular, Intermediate Particle Size The plastic fraction from upstream WEEE treatment plants often has a wide size distribution, from 10mm up to 200/300 mm for the largest pieces. It is therefore necessary to size reduce the largest of these particles down to a regular size in order to adequately feed the process with well-liberated individual plastic pieces. This can be achieved by isolating the largest size fractions (say > 50mm particles) and then submitting them to a size reduction and screening process. 3.1.3 Final Granulation Usually at a later stage in the overall process, it will be necessary to produce a small plastic ‘granule’ or ‘chip’ that is within a defined particle size range to meet the needs of the end-user or customer for the plant’s output product. In WEEE polymers the downstream process is usually some form of thermal treatment, such as extrusion compounding. This type of equipment can only be effectively operated if the input is within a tight and controlled size range. Typically the granulate feed for an industrial extruder needs to be in the range of 3 – 8mm.

Figure 3.2 Typical granule size of final product


3.2 Choice of Size Reduction Method There is a wide range of equipment available for the tasks described above. The choice of the right machine for each particular task is a difficult one, given the wide range of different equipment designs and the often confusing advice given by vendors of size reduction machinery. Furthermore, the terms used to describe the different equipment designs are not well defined and tend to be misused if operating across a language translation barrier. For the purposes of this report, and in relation to the task of rigid plastics size reduction, we have decided to apply the following definitions to the commonly used items of equipment:- 3.2.1 Shredder A shredder is used to carry out primary size-reduction on whole WEEE casings or baled material, or for reducing down large oversize pieces (i.e. > 200mm in one dimension). Typically the working parts are rotary and slow speed employing chopping, ripping and tearing actions to part the material. Output material size usually controlled by forcing through a fixed aperture screen. Shredders can handle a degree of light metal contamination such as clips, screws or thin sheet parts (but NOT tramp scrap items). Typical output size – 20 – 50mm in at least one dimension.

Figure 3.3 Tramp scrap

3.2.2 Granulator A granulator is used to produce fine granules of plastic. It employs scissor or guillotine cutting via close tolerance sharp blades, spinning at high rotary speeds. Mostly the infeed has been pre-shredded to a regular input size, below 50mm. Output particle size is controlled by a rigid mesh screen, often positioned close to the cutting zone. The output particles are usually below 10mm in size for plastic. Granulators can also be called ‘Grinders’ – hence the common term ‘re-grind’ for generic recycled plastic. These machines have a low tolerance for metal contamination in the infeed, which will quickly blunt or damage blade edges. 3.2.3 Hammer Mill A hammer mill uses high speed rotating hammer-heads to impact against the plastic and the stationary ‘anvils’ inside the machine casing. Size reduction is


caused by fragmenting of plastic on impact with machine or other particles in the chamber. A hammer mill can have a screen or grid fitted to control size, or use an adjustable exit clearance-gap as the size limiter. This machine can also be called a ‘fragmentiser’ or ‘impact mill’. It is able to handle lightweight metal parts in the mixed infeed material. There are many other designs of size reduction equipment used in industries such as minerals processing and pharmaceuticals. These include ball-mills, jaw-crushers, pin-mills and fluid-energy milling methods, but in the author’s experience these are not often applied to plastics processing. The three broad categories defined above cover 99% of plastics size reduction equipment.

3.3 Power Consumption and Throughput Size reduction equipment will be the largest user of power in most plastics recycling plants, with the required power being dependent upon:-

• Size reduction ratio – average size of input divided by average size of output.(e.g. 10:1 for 100mm feed down to 10mm granules)

• Mass throughput required – tonnes/hour

• Hardness or Toughness of the material.

• Shape of the particles, in terms of the amount of new free-surface created by the process. Solid lumps will require more energy than thin sheet plastic.

• Efficiency of the machine design in terms of noise, vibration and heat produced during operation.

• Percentage of the input plastic that ends up within the required size range – fines production will represent a high waste of power and loss of material / throughput.

Some attempts have been made to develop theoretical equations that allow the calculation of the power required to carry out a known amount of size reduction, based upon the cleaving forces needed to create new surface area per unit of volume. This work is material specific and no empirical data exists for plastics. Also, attempts to scale up this theoretical approach have shown that very large efficiency factors need to be built in to give meaningful results (e.g. calculated power = 25% of actual power required). On this basis, is it probably best to rely cautiously on the information supplied by equipment vendors as the first method of machine power selection for a given throughput. This should be checked with large-scale practical trials where possible and, even then, a generous allowance made for over-sizing of the selected machine. This should give some ‘head-room’ for normal operation to take place in the 50 – 75% range of full load power at the desired throughput, so that the machine is not always working at the upper limit of its design capacity.


3.4 Maintenance and Repairs Size reduction equipment will also be likely to represent the highest level of repair and maintenance cost in a plastics recycling process. Essentially the costs split into two types:-

• Normal wear and planned maintenance expenditure

• Breakdown and catastrophic failure. With correct operation, well within the design range of the machinery, the equipment should give long running time between failure. Regular maintenance routines will have to be set up to inspect and monitor the wear on the cutting blades (or hammers). The wear rate will be very dependent upon the infeed material, with clean, uncontaminated plastic giving much longer blade-life than dirty mixed waste plastics. In all cases some serious consideration has to given to protecting the machinery from unwanted metal pieces entering the unit. On large scale shredding machines, this can be brought about with a large over-band magnet, but this will not prevent all items of ‘tramp’ metal from entering the shredder (e.g. stainless steel ). A belt-and-braces approach is to have good magnetic removal, but also to provide a metal detector set to alarm if ‘large pieces’ get past the magnet removal stage. For granulators, the need to protect against unwanted metal entering the unit is even greater. Small screws and metal fixings will cause more rapid blunting of the blade tips, which will lead to an increase in machine power consumption for the required throughput and also create more fines with output particles having a poorer appearance of ‘cut quality’. It is strongly recommended to provide a high-standard of both ferrous and non-ferrous removal from the input plastic stream before feeding into a granulator. This system should be backed-up with further metal detection devices to alarm the operator and to removed unwanted metal particles, before they enter the cutting zone. Almost every plant visited by the author of this report has at least one horror-story to relate concerning unwanted ingress of large items of metal into expensive size reduction machinery. These tales are often followed by long periods of downtime and very expensive replacement costs.


3.5 Size Reduction Trials During this project, several size reduction trials were carried out on the following types of machinery:-

• Single Shaft Shredder – Wagner

• Impact Mill – BHS

• Hammer Mill – Alpine Hosokawa

• Granulator – Herbold A brief summary of the trials, main results and findings are given below. 3.5.1 Trial Summary – Single Shaft Shredder – Wagner Single shaft shredders are widely used in the recycling industry and most machines have a similar layout of working parts. Typically a single horizontal shaft is mounted close to a curved screen, with perforations to allow a certain size of particle past the screen. The shaft is fitted with a series of cutting knives which protrude from the surface. These knives rotate past a serrated row of stationary cutters, where the shredding action takes place. Material is fed into the shredder from an open top hopper, and is pushed into the rotary cutter shaft by means of hydraulically powered ram. Any pieces that have been cut-off at the point between the rotating knives and the stationary blade, are then forced against the curved screen as the shaft rotates. In this section further size reduction can occur until the pieces are below the hole size of the screen.

Figure 3.4 Single shaft shredder


Figure 3.5 Main shaft and pushing ram

A moving diagram of the internal workings can be found at this web-link:- http://www.wagner-shredder.at/DE/contentDD/cms/wygBilder/ews.swf

Technical Parameters – Wagner Model 40 Slow running single shaft rotor with double fly-wheel Motor power 2 x 30 kW, possibility of upgrading Hydraulic feeder (push bottom) Rotor speed approx. 80 rpm 183 chevron shaped rotor knives Stator knives in 4 lines top and bottom (each 2 lines) 2 screen baskets with cubic holes Trial Material Whole TV casings, black, post-consumer. Size: approx. 1500mm x 800mm x 500mm. Bulk density: 50 kg/m3 approx. Metal parts: screws, cables, metal inserts (Fe, Cu, Al)

Figure 3.6 Trial material


http://www.wagner-shredder.at/DE/contentDD/cms/wygBilder/ews.swf

Aim of Trial

• Check effective size reduction • Mode of operation (continuous or batch, because of feeding system) • Metal acceptance of unit

Summary Results Throughput estimated at 1.6 t/h. A size reduction from 1500mm to 20mm in one pass, with whole TV casings. There is a clean cut at the particles edge. The surface of the particles does not bear many traces of additional cuts There was no disturbance caused by metal particles. Photograph below shows typical material post 20mm screen.

Figure 3.7 Material post 20mm screen

Conclusion A single shaft shredder can provide a useful function to size reduce either whole casings or oversize particles as part of a polymer process. They are able to handle a low-level of thin metal parts, but should be protected by suitable metal removal system against ingress of ‘tramp scrap’.


3.5.2 Trial – Impact Mill – BHS, Germany The Impact Mill was included in the trial programme because it represents a novel application of technology from a different industry being applied to WEEE scrap processing. Impact mills have been developed in the stone and quarry industry to produce aggregates of regular milled size. The machine does not have any close-tolerance cutting blades, but relies upon impact between a series of rotating, horse-shoe shaped hammers and the stationary anvils around the walls. Size reduction occurs as a result of :-

• Impact with the machine parts • Friction between hammers and anvils • ‘Autogenous’ milling – meaning the particles are milling themselves

There is no screen basket, with output size being controlled by an adjustable gap between the hammer / wall. The scope of this trial was to investigate how effective the impact mill is with respect to metal pelletisation, disaggregation of composites (plastic-plastic and metal-plastic), harmonization of particle shape, flexibility of operation, wear and size reduction. Working Principle

Figure 3.8 BHS Impact Mill

The input to be milled is fed centrally from the top into the milling chamber. Upon contact with the rotor, the material is accelerated and centrifugally thrown against the external impact walls, into the ‘striking zone’ of the horse-shoe shaped impeller bars.


Friction and the impacts between individual particles, any metal pieces in the input and the impeller bars, results in a reduction and in size, smoothing of rough edges, balling up of any metal pieces and curling up of metal wires.

Figure 3.9 Internal working of mill

Input

Anvil Ring

Annular Gap Horse-shaped Impeller Bars

The shape and size of the final particles are determined by:

• Size of the annular gap • Rotor speed • Number of passes through the unit

The technical specification of the trial machine:- Type: RPMV 1113 Rotor diameter: 1150 mm Rotor height: 135 mm Number of impellers: 8 Circumferential speed: 71.2 m/s Gap width: 16 mm Drive power: 90-110 kW Total height of machine: 2704 mm


Trial Procedure The plastic to be milled was fed by a conveyor belt system into the mill. In contrast to a normal operation mode for an impact mill, the trial was done semi-continuously. After each pass, the material was filled into a bulk bag. Then the material was unloaded again to run a second and later a third pass through the mill. The results of the 3 passes are then compared for particle size distribution, and changes in shape for both the plastic and the metal parts. A photograph of the trial station is shown below.

Figure 3.10 Trial station

The photograph below shows the internals of the impact mill. Note the adjustable gap between hammer / anvil plates.

Figure 3.11 Internals of impact mill


Throughput for a single machine is dependent on the number of recycle passes needed to obtain the required size and shape distribution. The higher the number of passes required to achieve the desired result, the lower the throughput. Input Materials The input material for this trial was about 800 kg of IT WEEE material from a UK treatment plant. A photograph of this material is shown below.

Figure 3.12 IT WEEE material

The low metal content in this input material, (less than 1%) was insufficient to show the effect of forming metal balls from the metallic fraction, therefore it was decided to add 15 kg of metal in the second run. Summary Results There was nil loss of material in the trial, 815 kg was fed into the machine and the same mass was collected for recycling in a second and third pass. The processing parameters of the trial were as follows: Variable 1st Pass 2nd Pass 3rd Pass Motor Power (kw) 90 110 100 Throughput (t/h) 11.5 20.0 10.0 Rotor Speed (m/s) 71.2 71.2 71.2 The unit is fully enclosed with integral dust removal and generates a high noise during operation.


Size Reduction and Distribution The sieve analysis shown in the figure below shows the size distribution of the particles in each pass and the input.

Figure 3.13 Size distribution of particles

Sieve Analysis BHS Vertical Impact Mill Trial

0%

20%

40%

60%

80%

100%

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00

Particle Size in mm

Mas

s Pe

rcen

tage

s in

%

Input 1st Pass 2nd Pass 3rd Pass

Figure 3.13 shows that there is a significant size reduction in the particles from the input to the 1st pass. The average (50% distribution) particle size fell from 16mm in the input to 11.2mm in the 1st and 2nd pass, and 8mm in the 3rd pass. Photographs show appearance of plastic and metal parts after third pass.

Figure 3.14 Material after third pass


Discussion of Results The picture above shows how the edges of the plastic chips become smoother and rounded in the impact mill. This aids downstream polymer separation processes. A unique feature of the Impact Mill is its ability to form the metal pieces into round balls or pellets, and also to cause thin copper wires to curl up. This emphasizes the characteristic difference between the metal pieces and the plastic chips, and thus aids the removal of metal downstream via a gravity separator or other metal removal step. Operators using this type of equipment are able to obtain a high value metal stream with low impurity contaminant. Disaggregation of Composites Another interesting feature of the impact mill, which differentiates it from the other standard size reduction equipment (shredders, granulators) is its ability to break up composites in the WEEE infeed. These composites include: screw thread-inserts that are moulded into plastic casings, PU foam adhered to the surface of fridge plastic and automotive dashboard waste which consists of 3 layers of different types of plastic / foam. Conclusion Bullet Points – Impact Mill

- High investment cost of £150,000-£180,000 for 3 to 5t/h - Needs a gravity separator in addition to separate metal from

polymer - does not work alone - Creates a significant amount of dust - Low operational costs (low wear, no blades necessary) - Typically used in metal recycling application / whole WEEE

application; less in plastic recycling application - Flexible in operation; can be used as 1-pass or multi-pass

machine - Size reduction through "auto-milling" - Enables the user to recycler multi-layer secondary raw materials


3.5.3 Alpine Hosokowa – Hammer Mill Trial This trial investigated the possibility of using a hammer mill for size reduction, and also as a direct comparison to the impact mill. In a hammer mill particles are subjected to similar conditions to those in the vertical mill, i.e. impact, friction, shearing and “autogenous” milling. Working Principle The photograph below shows the milling chamber of the Ominplex Hammer Mill.

Figure 3.15 Omniplex milling chamber

Bar

Milling Chamber Oscillating

Hammers

Rotor

The material to be milled is fed centrally from the top into the milling chamber. On contact with the oscillating hammers on the rotor the material is highly accelerated and thrown against the impact bars. After striking these bars the material falls into a gap between bar and milling chamber. There is no screen for determining the particle size. The gap, which is adjustable, and the rotor speed determine the particle size. In contrast to a vertical impact mill, the operation mode is not to run the material in a circle, but in one pass only. Equipment The trial station was spread over 3 levels. The infeed was fed by gravity from the top level (1st floor) into the hammer mill which was located on the 2nd level. The output from the hammer mill was also fed by gravity into a bag on the bottom level (3rd floor). Due to time constraints, there was only 1 pass in this trial.


The machine used for the trial is shown in the photograph below.

Alpine Omniplex Hammer Mill Type 80/63

Gap

Bar

The technical specification of the trial machine are: Type of hammer mill: Alpine Omniplex Hammer Mill Type 80/63 Type of hammers: Oscillating beaters Number of hammers: 18 Size of gap between bars: 20mm Mill speed: 1000 rpm Motor speed: 1500 rpm Drive power: 75kW Infeed material was the same as the Impact Mill trial (see above) and 800 kilos were processed, in a single pass only. Summary Results & Discussion The effect of the size reduction in the hammer mill can be seen below, with graphs of the particle size for before and after processing.


Sieve analysis for Alpine-Hosokawa input

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

25.0 - 11.2 11.2 - 8.0 8 - 5.6 5.6 - 4.0 4.0 - 2.0 < 2.0

Granule diameter, mm

% M

ass

Sieve analysis for Alpine-Hosokawa pass 1

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

25.0 - 11.2 11.2 - 8.0 8 - 5.6 5.6 - 4.0 4.0 - 2.0 < 2.0


% M

ass

It can be seen that there was over 60% of the input material in the 25 – 11mm size range, and that after milling 60% was in the 11 – 8 mm size range. The following photo shows the material before and after milling. Also shown is a view of the metal pieces, which have been rolled-up into spheroids in a similar manner to the BHS impact mill. The hammer mill gave an approximate 50% reduction in particle size on this single pass trials and also converted the metal into a format where it could be easily removed in downstream processing.


Figure 3.16 Before and after hammer mill

Figure 3.17 Metal ex-hammer


3.5.4 Herbold Granulation Trial The Herbold is a classic design of granulator for the size reduction of pre-shredded plastic materials. The main working part is a rotating cutter drum which is fitted with a series of axial knife blades with sharp edges. These blades rotate at high speed and pass very close to a set of matching stationary cutter blades. The two sets of blades are set at a slight and opposite angle to the horizontal, which creates a smoother scissor cut and less fines are claimed. The plastic is introduced through a feed chute and falls onto the rotating blades. Cutting takes place by fast scissor action between the blades. The cut pieces are swept around with the spinning drum and pass across a curved perforated sieve basket. The size of the holes in the sieve determines which particles can pass out of the unit to be conveyed away pneumatically.

Figure 3.18 View of Herbold granulator showing knives, rotor and screen mesh

Equipment Specification Type: Herbold SML 60/100 Rotor diameter: 600 mm Width: 1,000 mm Inlet: 750 x 980 mm Rotor knives: 7, in scissor cut operation Screen: 10mm round holes Driving power: 90 kW Suction system: MFT 55/500, 11kW

Trial Results 400 kg of mixed WEEE plastic was used for the trial. Throughput was estimated at 4.8 tonnes per hour with a current of 180Amps. At lower


current – 80 – 90 Amps throughput drops to the 2 – 2.5 tonnes per hour level. These photographs show the material before and after granulation. Note:- there was an aspirator on the outlet system to remove any fines generated. It is estimated that around 5% fines (below 2mm size) will be created in most granulation stages.

Figure 3.19 Material before and after granulation

Particle size distribution of the before and after plastics is shown below. This was based upon running the machine with a 10mm screen

Herbold granulator sieve analysis

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

25.0 - 11.2 11.2 - 8.0 8 - 5.6 5.6 - 4.0 4.0 - 2.0 < 2.0


% M

ass

Pre-granulatorPost-granulator


Conclusions The Herbold granulation trial demonstrated a good performance using a standard piece of equipment. The metal content of the infeed plastic was very low and the material had been pre-shredded to a suitable size. Average particle size has been reduced from around 8.7mm (D50) down to 6.2mm, but also the particle size distribution has been narrowed, as can been seen on the bar graph. Noise level can be high when granulating rigid plastics and there is noticeable vibration in the area of the machine. Infeed needs to be a constant flow and MUST be metal free to avoid blade damage. When operated properly, granulators will give several hundred tonnes of throughput between blade changes. Increased current draw can be used as a means to judge blunting of the blades. In summary the trial demonstrated that the machine does exactly what it supposed to do.

3.6 Main Findings Size Reduction The main outcome from the trials on size reduction technologies is that there is a broad choice in approach to the task. As follows:- 3.6.1 Traditional Two-Stage Approach The size reduction of the infeed plastic waste is carried out using a combination of shredder to do primary size reduction, followed by a knife-blade granulator to generate the final small particle size of plastic particles. In this instance, size reduction equipment must be associated with appropriate metal removal methods before each stage in order to protect the cutting machinery from damage during operation. 3.6.2 Milling Approach This would be suitable to a more integrated WEEE treatment process where the metal / plastic mixture arising from an upstream dismantling or fragmentiser unit could be processed together in an impact type mill. This is judged to be suitable when the metal content is above 15 - 20% of the total waste flow and the primary function of the plant is to maximize metal removal. For a secondary processor of plastic which has already been de-metalled, this type of equipment is not likely to prove economical as the metal yield would be too low. 3.6.3 Further Choice It was not possible within the scope of this report to conduct a full set of trials across all available designs of shredding machinery. It is important however to note that the Twin-Shaft shredder has found wide use within the WEEE sector as a means to size reduce whole items. A development of this design is the Four-shaft shredding unit. Both of these have intermeshing rotary cutter knives and produce strips of output material.


3.7 Equipment Suppliers Wagner

A-8385 Neuhaus/Klb. Panoramastrasse 13

Tel: +43 (0) 3329 / 2478-0 Fax: +43 (0) 3329 / 2478-15

Email: [email protected]

Herbold Granulators UK Ltd

27 Weaver Park Ind Est Mill Lane Frodsham Cheshire WA6 7JH

Tel: 0845 230 7464 Fax: 0845 230 7465


BHS-Sonthofen GmbH Hans-Böckler-Str. 7 D-87527 Sonthofen

Tel: +49 8321 802-200 Fax:+49 8321 802-220


Alpine-Hosokawa P.O. Box 10 11 51

DE – 86001 Augsburg Germany

Tel.: +49 821 5906-0 Fax: +49 821 5906-101


NOTE – These supplier names are given as an example only and should not be treated as any particular endorsement of the technology or

machinery on offer. Readers are encouraged to make their own search of potential suppliers.


mailto:[email protected]



4.0 Metal Separation

A key element of any polymer separation process, where the infeed material is derived from an upstream WEEE treatment plant, is the removal of unwanted metallic contamination from the mixed plastic waste. This metal needs to be removed in order to protect the downstream size reduction equipment from potential damage and also to ensure that the end plastic product is ‘metal-free’. Metal Separation can be applied at 3 different stages in a complete plastics treatment process. These stages are related to the scale of the metal particles being removed and occur in relation to the size reduction stages of the process. In broad terms these are:-

1. Large scale removal of ‘tramp’ metal items before entering any shredding equipment. (Items over ~50mm in any dimension)

2. Removal of smaller metal items post-shredding, but before fine granulation of the plastic. (e.g. screws, clips, connectors, wires etc)

3. Removal of metallic fines post-granulation as part of the refinement stage. (e.g. shards, splinters, wire filaments).

There is a wide range of equipment available for metal removal, some of which are listed below:-

• Permanent or over-band magnets – for large ferrous removal • Magnetic head rollers – ferrous and some stainless removal • Eddy-Current Separators – for removal of non-ferrous metals • Inductive Metal removal methods – all metal types • Vibrating Tables and Air-classifiers – for fine particle removal

More recently, advanced spectrographic methods are being developed which can also be configured as metal removal devices, and these are discussed in a separate section. Under the scope of this project, Axion Recycling conducted trials on several of the available designs in order to gain an understanding of the different methods, but this was by no means an exhaustive testing of all the choices available in the marketplace for metal removal. The results of those trials, together with some explanation of the operating principles for the methods used, are presented below.


4.1 Theory of Metal Removal Methods It is useful to have an understanding of the theory behind the various metal removal processes, as this will enable the user of such techniques to be aware of the capabilities and limitations of the different devices. 4.2 Magnetism Magnetic forces are widely used in metal removal systems. The magnets used can be either permanent magnets or electrically powered magnets. In either case the strength of magnetic force available to attract (or repel) metallic objects is important as this defines the ability of any magnet to remove metal from a mixed material stream. Magnetism can be thought of as a series of ‘lines of force’ emitting from the surface of a magnet. These lines naturally from continuous loops between the poles of a permanent magnet, as represented below:-

Figure 4.1 Iron filings used to visualise lines of magnetic force

The strength of the magnetic force varies in two ways; firstly the force reduces with distance away from the magnetic surface, and secondly with the density of the magnetic flux. In simple terms, the magnetic flux density can be thought of as the number of lines of magnetic force which pass through any fixed area. The more tightly packed the force lines, then the higher the flux density. Measurement of magnetic force is a complex subject, with units of ‘Gauss’ or ‘Tesla’ being used to quantify the density of magnetic flux through a given area. In simple terms the higher the ‘Gauss’ the stronger the magnetic force. In recent history, material scientists have developed increasingly more powerful permanent magnets using specialist alloys which can deliver very high levels of magnetic flux per unit volume. This has enabled the designers of magnetic separation equipment to pack more separating force into the working parts of metal removal equipment. This is shown in the following figure, where BHmax , a measure of the magnetic energy packed into the material, is plotted against the year of development for different compounds.


The Neodymium Iron Boron (Nd-Fe-B) magnets developed in the 1980’s represent a 5 fold increase in magnetic power when compared to those from the 1950’s.

Figure 4.2 Magnetic force

Metals react to a magnetic field in different ways, according to their own particular magnetic behaviour. There are three common types of magnetism displayed by metallic objects:- Ferromagnetism - Permanent ferromagnetic materials have a high retainment for magnetization, and a common example is a traditional refrigerator door magnet. In the periodic table of elements only iron, cobalt and nickel (Fe, Co, Ni) are ferromagnetic at and above room temperature. These elements are therefore always included in the compounds or alloys which have been developed to display very high levels of permanent magnetism. Normal iron and mild steel are also ‘ferromagnetic’ because they become self-magnetised when exposed to a magnetic field. Steel paperclips are a good example of this behaviour, as they will form a continuous clump of mutually attracting ‘magnetic’ pieces when touched by a magnet. When removed from the magnet they revert to individual pieces of steel. However extended working and thermal treatment of ferromagnetic materials can lead to them taking on some permanent magnetic behaviour.(e.g. compass needle) Paramagnetism. Virtually all materials have some detectable magnetic properties. The forces that act during paramagnetic behaviour are much smaller, however, and during this behaviour the material does not produce its own spontaneous magnetic field. Strictly, they can not be considered non-magnetic but as very weak ‘ferro-magnets’. Diamagnetism - Usually referred to as 'non-magnetic,' diamagnetic materials actually exhibit some magnetic behaviour - just to very small magnitudes and with opposing polarity to the applied field.


4.3 Electromagnets

In general, electromagnets consist of a core of magnetic or ferromagnetic material, for example soft iron around which a coil has been wound. As long as an electric current flows through the coil, the core remains magnetic.

A magnetic field is generated around a conductive wire through which an electric current flows. The magnetic flux generated can be expressed as follows:

Φ = L * I

In which Φ the magnetic flux expressed in Weber L the self-induction in Henry and I the current in Ampere

A strong magnetic field can be achieved either by using high currents or large self-induction effect. High currents are not always feasible, which is why a large self-induction is obtained by using multiple current-loops in the form of a coil or solenoid. The induced magnetic fields of all the individual current-loops in the coil are then added up, which results in a strong magnetic field being transmitted through the central metal core.

This type of electromagnet is used in simple solenoid switches and in large scale mechanical cranes for moving ferrous metal.

Figure 4.3 Electromagnet in action

A detailed scientific explanation of the theory of magnetism is beyond the scope of this report, but interested readers can find a thorough explanation of the subject at :-

http://www.aacg.bham.ac.uk/magnetic_materials/origin_of_magnetism.htm


http://nl.wikipedia.org/wiki/Elektrische_stroom

http://www.aacg.bham.ac.uk/magnetic_materials/origin_of_magnetism.htm

4.4 Eddy-Current Forces The term ‘eddy-current’ describes the internal electrical currents created inside a conducting body when it passes through a varying magnetic field. Thus if a piece of copper metal passes through a changing magnetic field, an internal flow of electrons will be created within the body of the metal item. This ‘eddy-current’ creates its own, smaller electromagnet field, which opposes the direction of the primary field. The effect of these opposing magnetic forces is seen as an acceleration directed away from the original magnetic field. Figure 4.4 Diagram of Eddy-currents being induced in a moving metal object

Thus a small piece of copper plate on a conveyor belt which passes across a varying magnetic field will be seen to ‘fly-off’ as a result of the strong, but invisible forces associated with the internal eddy-current. The size of the magnetic force generated to remove any particular item will be related to its electrical conductivity and the resulting acceleration away from the sorting zone will depend upon its mass ( i.e. F = ma ). Therefore lightweight, highly-conductive metals such aluminium will display the strongest effect as they accelerate away from the changing magnetic field. 4.5 Induction

Electromagnetic induction is the production of voltage across a conductor situated in a changing magnetic field or a conductor moving through a stationary magnetic field.

Figure 4.5 Electromagnet


In the figure above an ammeter is connected in the circuit of a conducting loop. When the bar magnet is moved closer to, or farther from, the loop, an electromotive force (emf) is induced in the loop. The ammeter indicates currents in different directions depending on the relative motion of the magnet and loop. In a sensing induction loop, conducting metal objects are detected in close proximity to the wire-loop, without touching them or having to pass through the loop. This is achieved by using a second powered wire carrying an alternating electric current to produce its own local magnetic field. Metal items passing through this field cause an induced current in the sensing loop, indicating the presence of metal. Thus induction loops form the basic technology used in many modern metal detection devices such as airport security scanners, car-park barrier sensors and hand-held metal detectors used for hobbies and mine sweeping. Increased sophistication of the excitation and sensing loops combined with microcomputer electronics provides the basis for complex detection tasks to be carried out at very high speed.


4.6 Operating Principle of Four Types of Metal Separator for Plastics Mixtures

4.6.1 Magnetic Head Rollers

A simple way to remove ferrous metal particles from a mixed material stream is to install a magnetic roller at the discharge end (or ‘head’) of a belt conveyor unit. The diagram below indicates a typical installation where the head roller emits a magnetic field all around its circumference and the flexible conveyor belt carries the mixed particle stream through the magnetic field. Any metallic parts which are attracted towards the magnetic head roller are diverted away from the normal ‘free-fall’ trajectory of non-magnetic material. The magnetised items remain held in the magnetic field of the roller until they either fall away by the influence of gravity or are ‘peeled’ away from the roller by the return path of the conveyor belt.

Figure 4.6 Magnetic head roller

Several factors contribute towards the effective operation of such head roller metal separators:-

• Belt speed, which must be fast enough to give a good trajectory for the non-metallic parts, but not too fast to avoid dense ferrous items being thrown forward with sufficient momentum to escape the magnetic field

• Depth of material on the belt – to ensure that all particles are close to the belt surface and enter the strongest part of the magnetic field

• Flux density of the magnet roller – to impart maximum separation forces. It is recommended to use a high-Guass permanent magnet, using the latest developments in high-energy magnetic materials.

• Particle size – to ensure that all particles are well separated and are free to react individually to the separating force, without being influenced by other large pieces in the mix (e.g. balls of fluff, or sheets of wood / plastic). Material that has been shredded down to below 50mm particle size will be suitable for this type of device.

• Position of the splitter plate – this needs to be adjusted accurately by the operator for each different infeed mixture.

In the case of metal removal from WEEE plastic streams, the ‘ferrous’ fraction will include a wide range of items that are attracted to the magnetic head roller, this includes electrical components, some circuit boards, screws and clips, magnetic storage media and steel sheet. It should also be noted that


items which appear to be made of stainless steel (which is normally assumed to be non-magnetic) can sometimes be influenced by very high flux magnetic forces. This behaviour has been attributed to changes in the metallic structure caused by upstream shredding processes and the low grades of stainless used in electrical products. 4.6.2 Eddy Current Separators (ECS) Eddy-Current Forces The term ‘eddy-current’ describes the internal electrical currents created inside a conducting body when it passes through a varying magnetic field. Thus if a piece of copper metal passes through a changing magnetic field, an internal flow of electrons will be created within the body of the metal item. This ‘eddy-current’ creates its own, smaller electromagnet field, which opposes the direction of the primary field. The effect of these opposing magnetic forces is seen as an acceleration away from the original magnetic field. Thus a small piece of copper plate on a conveyor belt which passes across a varying magnetic field will be seen to ‘fly-off’ as a result of the strong, but invisible forces associated with the internal eddy-current. The size of the magnetic force generated to remove any particular item will be related to its electrical conductivity and the resulting acceleration away from the sorting zone will depend upon its mass. Therefore lightweight, highly-conductive metals such aluminium will display the strongest effect as they accelerate away from the changing magnetic field. Eddy Current Separators (ECS) In metal recycling operations the application of eddy current separators is well proven as the primary method to sort non-ferrous metals from mixed shredder residue waste, post magnetic removal of steel and iron. The ECS removes copper, aluminium, brass and other metals from the waste stream. In most primary WEEE treatment plants, the ECS is used to recover value from the plastic stream by taking out copper, aluminium and brass items. This non-ferrous stream can form a significant proportion of the recovered value from certain types of WEEE waste being 5 - 10% of the material mass. The resulting waste plastic streams from the primary WEEE treatment process does still contain a variable level of non-ferrous, either due to the inefficient operation of the upstream ECS unit, or because the residual non-ferrous is of small dimensions and very difficult to remove by normal eddy-current techniques. The diagram below gives a schematic representation of an eddy current device. The most important part of the unit is the driven head roller, which incorporates a high-speed rotating central core rotating within an outer roller shell. The inner roller incorporates a set of magnets which are arranged to give an alternating North-South field polarity around the circumference of the unit (see diagram). This inner roller can rotate as fast as 3000 rpm within the


outer roll which will be rotating at much lower speed to drive the conveyor belt. (~50 – 200 rpm)

Figure 4.7 Diagram of Eddy Current separator

Non Ferrous

Non Metallic Ferrous

Important Note :- This diagram shows a ferrous stream also being separated due to its attraction to the magnetic roller. However most suppliers do not recommend this approach due to the risk of ferrous metal getting stuck to the roller and becoming hot. Steel and iron should be removed upstream of the ECS.

Figure 4.8 Magnetic polarity arrangement in head-roller

Metal objects entering the rapidly alternating magnetic field created by the fast-spinning head roller, will develop their own internal eddy-currents as described earlier. Different metal will react to the alternating field to a different degree in relation to their ‘conductivity/density’ which is a measure of the ease with which internal eddy-currents can flow in the metal and the density of the material. High conductivity/density material such as aluminium will be the most reactive to the eddy-current sorting zone. The table below ranks metals according to this measurement, and for metals with a conductivity/density below ~1.0 (shaded area) it is very difficult for the eddy current device to effect a separation.


Non-ferrous Metal Conductivity/density (s/103* M2/V*Kg)

Aluminium 14.0 Magnesium 12.9 Copper 6.7 Silver 6.0 Zinc 2.4 Gold 2.1 Brass (Messing) 1.8 Tin 1.2 Lead 0.45 Stainless Steel 0.18 Glass 0.0 Plastic 0.0

(Ack – Info ex Allcontrols)

Another issue to consider is the size of the ECS unit in relation to the particle size of the metal that needs to be removed. The items typically found in WEEE plastic waste are not large pieces of copper or aluminium, (i.e. >50mm) as these have mostly been removed by large scale eddy-current separators at the upstream primary processor. In the residual plastic scrap stream, it is more typical to find small pieces of wire, copper tubing, small push-fit connectors etc. These are in the 10 – 40mm size range in at least one dimension, and were not removed by exposure to a large-scale ECS unit. Thus a very high speed magnetic roller, fitted with high flux-density magnets is needed to produce a sufficiently intense and rapidly fluctuating magnetic field in the eddy-current inducing zone of the device. This gives a greater chance of creating the necessary forces to remove thin, narrow items from the waste stream. Such a system is called a ‘fine-pole’ eddy-current device.


4.6.3 Induction Sorting Systems The induction sorting system is essentially a form of detect-eject machine, which has a similar layout to optical sorting devices (e.g. a bottle sorting machine). The material is fed at a controlled rate by a vibratory feeder and is then transported on a wide conveyor with a speed selected to give an even spread of particles across the belt. For best results, each particle should be individually located, with minimal overlap or coverage by adjacent pieces of metal / plastic. The core of the unit is an array of induction sensor loops located underneath the surface of the belt. As described earlier, metal items passing through the electromagnetic field of the sensor loop, create a small induced current which is used to record the position of the item across the belt’s width. This information is then combined with the exact speed of transport to determine when the sensed item reaches the end of the belt. Here a row of accurate air ejectors is used to deliver a timed jet of air, coinciding with the ‘reject’ item falling from the end of the conveyor. In this way rejected metal items are pushed over a deflector plate into the collecting bin. Material that did not create any induction current (non-metallic) is allowed to fall freely under its normal trajectory into the ‘accept’ stream of mainly plastic product.

Figure 4.9 Schematic of induction sorting system


4.6.4 Gravity Separators These devices can be used for the removal of fine metal particles from a stream of plastic waste. In parallel to separation of ‘heavies,’ it is also possible to create a separation of lightweight fluff or fines on this type of device. A gravity separator or air shaking-table is thus a flexible method of sorting particulate mixtures which can be adjusted to give both heavies and lights removal from the main product stream.

Figure 4.10 Gravity separator

Material Infeed High

Side

Main Discharge

Low Side

Vibrating Deck

The sorting process occurs on a vibrating mesh table (or deck), through which a volume of air is blown using an integral fan. During operation, the deck can be tilted in both directions to control the movement of material across the machine surface. The photograph below shows a demonstration version of the unit, loaded with plastic particles (note tilted mesh shaking deck).

Figure 4.11 Vibrating mesh table

Air is used as the separating medium through the process of ‘stratification’. Stratification occurs by forcing air through the particle mixture so that the particles rise or fall by their weight relative to the average bulk density of the aerated ‘shaking bed’. However it is important to ensure that full ‘fluidisation’ does not occur on the deck (caused by excess air flow), as this would induce


a well-mixed bed of materials and destroy the potential for separation to take place by stratification.

Figure 4.12 Stratifying and sorting zones

Top Flaps

Discharge Flaps to split outlet flows

The figure above represents a plan view of the ideal situation to be achieved in operation of a gravity table. The particle mixture falls from the feeder onto the deck in the bottom right hand corner. The area immediately around the feeder is called the ‘Stratifying Zone’. In this area, the vibration of the deck and the lifting action of the air combine to stratify the material into layers. This leaves heavier layers on the bottom and lighter layers on the top. Separation cannot occur until the material becomes stratified. The size of the stratification area will depend on the difficulty of separation and on the capacity at which the machine is processing. The shaking action of the table, combined with the lifting effect of the stratifying air-flow, creates a gradual bulk movement of material towards the upper left hand corner of the machine. During this progressive shaking motion the heavy particles are pushed towards the top edge of the deck and the lighter fraction flows from the upper layer of stratified material down towards the bottom edge of the machine. A middle fraction of cleaned plastic particles can be collected at the discharge chute, with the lights being separated using an integral diverter plate. The heavier (metal containing) fraction is removed along the top edge of the deck by opening simple flap-gates to release the right amount of material for adequate cleaning. It can be deduced from this description, that efficient operation of gravity separators relies to a large extent upon the skill and observations of the process operator. There are several variables to control and the split of heavies and lights needs to be gauged against the need for a high yield of cleaned product material. Getting this balance right, requires constant attention and adjustment by a diligent process operator.


4.7 Summary of Trials Gravity Separators – as Metal Removal Units

Figure 4.13 Test unit

Dust

WEEE Plastic

Claimed Features: • Separates products with very slight differences in size • Separates products with very slight differences in weight • Dust free (with extraction fitted) • Divider blades provide control of the discharge product • Optional Recycle Elevator which allows the lights or middling to

be recycled back to the gravity's deck

Working Principle Material is fed to the fluidised bed via a vibratory feeder. After a few seconds, the bed vibrates, causing the material to distribute over the surface of the bed. The airflow is adjusted, along with the throughput andvibrations to achieve the required separation. Flap valves and diverter chutes are used to adjust the split between, heavies, lights and product fraction – some of which can be recycled for further separation.

Mechanical tilting motor

Extraction

Vibrating Table

Heavy, Light and Middle discharge fractions

Separation Technologies 42

.

W

Machine Specifications – for Prime Range of machines – ‘Crown’ brand.Capacity: 1 - 18te/hr Deck drive: 0.5 - 1.5kW Hydraulics: 1.1 - 1.5kW Fan: 2.0 - 15kW Length: 1,520 – 4,015mm Width: 1,120 – 2,115mm Height: 1,270 – 1,950mm Weight: 600 – 2,000kg Deck area: 0.4 - 4.6m2 Airflow: 7,000 - 40,000m3/h Recycling Exhaust: 1,750 - 7,000m3/h Material Types: seeds, cereals, plastic particulates

Figure 4.14 Gravity Separator

This figure indicates the working parts of a typical gravity separator (i.e. Trenn-so unit). It is often necessary to use a air collection system above the unit to remove the fines and dust which become entrained in the air-stream.

EEE Plastic Separation Technologies 43

Figure 4.15 Block diagram of input / output flows

Dust Extraction

Metal Removal Trials with Gravity Separators In this section we report on the two trials carried out to investigate metal removal using the Gravity Separator. Trials were also conducted with different input streams to investigate removal of lightweight fines (e.g. wood particles) and for the removal of rubber particles. These are reported elsewhere in the report. Two designs of Gravity Separator were tested during the Project – Prime and Trenn-so. Metal Removal Trial using Prime – ‘Crown’ Gravity Separator

Figure 4.16 In feed material

The photograph above shows the in-feed material to the Prime gravity separator. It is apparent that there is a significant quantity of metal present.

Input material

Heavy Gravity Separator Middle Fraction

Inlet Air

Light Fraction


The input fraction contained 1.7% by mass of metals, which was well mixed

across the total sample of 400 kg of plastic mixed material.

4

TmT

W

Summary of Results from Trial – Prime Gravity Separator from the

fter performing this test, it was found that 1.7% metal was found in the

Material

Total Weight

Tested (g)

Non-Metals Metals

The gravity separator was adjusted to extract the metal fractionbulk material containing plastic and metal mixture. Atarget fraction (light end) sample.

Present (g)

Present (g)

Post Consumer 117 114.9 2.1

his indicates that for this particular material there was little separation

the

unit.

Teffect seen on the machine to remove the metals. This was put down to particle size of the large flakes used – these are well above the recommended particle size of 2 – 10mm specified for the Prime

.7.1 Metal Separation with Trenn-So Gravity Separator

Figure 4.17 Gravity Separator

his design of gravity separator was us d to separate metals from a plastic

eixture that had previously been passed through a hammer mill process. hese metal particles and wires had been balled-up by the hammer milling


process and thus represented a different separation challenge to that trialledon the Prime unit. Tref GT34R2 was passed through a sieving machine with a sieve of screen size 3mm and 8mm. This resulted in 3 size-differentiated fractions which arpassed through the gravity separator separately. Size separating the infeed first allows for more efficient separation by the gravity separator. The details the trial machines and operating conditions for each size fraction is as listed below.

his was performed on the output of the hammer mill. About 60 kg of sample

e

of

etails of Trial Machines 2 3

DRun. No. 1 Size of particles mm mm mm >8 3-8 <3Machine Model TTS-600 TTS-300 TTS-300Screen perforation size

0.6 mm 0.6 mm 0.4 mm

Angle of screen 10 degrees 10 degrees 10 degrees Air flow 58% 52% 54% Throughput

33.8 kg 19.2 kg 5 kg Total Input Amount Time taken 3 min 4 min 2 min Throughput 660 kg/h /h h 270 kg 120kg/

Figure 4.18 Input plastic (post hammer mill)

ue to the different formats of the infeed materials between the Prime and e

DTrenn-So trial, it must be stressed that the results of the two trials can not bdirectly compared.


Trial Results

he table below shows the mass balance for the 3 different screen

Run No. 1 2 3

Tfractions after passing through the Trenn-So unit:-

Size Distribution mm mm mm >8 3-8 <3Input (Total) 33.8 kg 19.2 kg 5 kg Plastic Fraction (light fraction) 29.5 kg 15.9 kg 3.5 kgMetal Fraction (heavy fraction) 2.5 kg 1.5 kg 0.4 kg Dust (Filter fraction) 0.05 kg 0.06 kg 0.1 kg Total Output 31.05 kg 17.46 kg 4 kg Loss 2.75 kg 1.74 kg 1 kg

On the Trenn-So trial the output product from the trial was lost in transit. So

e

he photographs below show the plastic and metal fractions for the three

Figure 4.19

results for the degree of separation can only be made visually based upon thphotos below. Tdifferent sized outputs after gravity separation.

<3mm metal fraction

<3mm plastic fraction

3-8mm metal fraction 3-8mm plastic fraction


>8mm metal fraction >8mm plastic fraction

These visual results indicate the clean quality of metal that can be achieved on a gravity separator, and also how the spherical shape of the metal from a hammer mill process makes it more suitable for further recovery. In summary our conclusions from the trials for metal removal from WEEE plastics using gravity separators were:-

• Gravity separators are not suitable for removal of metal from rough shredded plastic particles with size range above 15mm, where the metal is in the form of clips, sheets and long wires.

• Good quality metal removal can be achieved from plastic feedstock when the particle size has been controlled and the metal particles have been balled-up by processing in a hammer mill or similar.


4.7.2 Induction Sorting Trials Two sets of induction sorting machine trials were carried out. A detailed high volume trial at Steinert and a shorter trial at Scan & Sort. 4.7.2.1 Steinert Electromagnetbau GmbH – ISS

DTNmarti

W

Features: • Automated recovery of metals • Eliminates the need for hand picking • Recovery rate greater than 90% • Speed of belts and sensitivity of sensors can be adjusted to optimize

results

escription of Operation he material is fed by a vibrating feeder onto a fast-moving conveyor belt. ear the end roller of the belt is an array of induction sensors that analyse the aterials across the width of the belt. Detection of any metallic particles sends n electronic signal to the control computer which triggers one or more of a ow of compressed air jets, mounted just beyond the end of the belt. An air-jet hen diverts the detected metallic particle from the main flow of material and nto a separate collecting bin.

Figure 4.20 Overall view of trial unit


W

Machine Specifications – full range of equipment Working Width: 600 – 2,000mm Weight: 3,000 – 7,300kg Length: 1,230 – 3,600mm Height: 1,180 – 2,500mm Material Types: Municipal waste, WTE – ash, electronic waste, woodchips, glass and mould sand

Figure 4.21 View of conveyor showing location of induction sensors

Metal detector

Figure 4.22 View of sorting section – shows bank of air-jets

Figure 4.23 Block diagram of input/output flows Input material

Metal Fraction Induction Sorting

System Non-metal Fraction


Large Scale Trial with Steinert ISS – Summary of Results Two types of WEEE material were processed through the ISS. Mixed IT waste and shredded CRT casings containing metal were separated by the ISS unit. The table below gives the mass delivered to each output stream. The IT plastic material had an average particle size of 15mm, the CRT shred was in the range 30 – 50mm, but with some larger pieces.

Material

Output Plastic Weight

(kg)

Reject ‘Metal’ Weight

(kg)

Clean Plastic stream

(%)

Metal Rich

Stream (%)

Mixed IT

plastic 614 17.5 97.22 2.78 CRT

Casings 429 24.9 94.51 5.49 The trial was to assess the efficiency of the metal separation. Throughputs of 2.5 – 3.4 tonnes/hour were achieved in the trial. Results of the first pass separation indicated that the residual (cleaned) plastic fraction had very low level of metal remaining in the material:- Cleaned IT Plastic - Metal Content = 8g in 55kg (<0.02%) Cleaned CRT Plastic - Metal Content = 1g in 45kg (<0.01%) This indicates a very efficient removal of metal particles from the bulk WEEE plastic infeed. Analysis of the metal rich ‘reject’ stream indicated that it was between 20 – 50% metal content, indicating some loss of plastic yield to the reject material. It was possible to further recovery plastic (and thus improve metal purity) on a second pass through the machine, but this was judged to be near the limit of economic processing. Photographs showing output of induction separation trial:-

Figure 4.24 Output from induction separation trial

Cleaned Plastic Fraction – CRT shred Metal Rich Fraction ex CRT


4.7.2.2 Induction Sorting Trial – Scan & Sort, Hamburg As a comparison exercise, a similar trial was carried out on an Induction sorter at Scan & Sort, in Hamburg.

Figure 4.25 Scan & Sort induction sorter

A similar feed material was used, from shredded IT plastic waste with 15.5 kilos being used in the trial. The degree of separation can be seen in the photograph below (note the very clean plastic fraction on right).

Figure 4.26 Example of separation


The induction sorting machine ejected 6.8% material. It failed to eject only a single screw and four pieces of wire. An additional 0.23kg of plastic was ejected with the metal fraction. This indicates that the original metal content of the infeed material was 5.7%.


4.7.3 Goudsmit Ltd – Eddy-Current Separator & High-Flux Head Roller – Summary of Trial This trial combined tests on a high-flux magnetic head-roller with further separation on a fine-pole eddy current unit.

W

Claimed Features: • Automated recovery of metals • Recovery rate greater than 99% • Advanced software to suit any application • Ferrous and Non-Ferrous removal can be assembled in one

machine in series, giving metal free product

Working Principle The mixed WEEE stream passes over the ferrous metal separator where the magnetic material is separated. The non-magnetic material is further processed to the ECS unit. A rotor which is part of the ECS unit rotates at extremely high speed. This generates an alternating magnetic field. The metal comes under the influence of the alternating magnetic field and an eddy current is created. This creates its own magnetic field. This field is diametric to the original field. As opposite magnetic poles repel, the magnetis “shot away”.

Figure 4.27 Goudsmit ECS unit (stand alone)

Machine Specifications - Goudsmit Belt width: 2 x 600mm Consists of: vibrator feeder, 2 Neoflux magnetic rollers diameter 300 x 600mm built upon U-profiles Type: Neoflux with flux control for an extra deep and strong field Magnets: GSN 44, Br value 12.4 Gauss Magnetic width: 300 x 600mm Weight: ca 1400kgs Type: 2 drum motors van de Graaf 2.2kW Dimension: ca 600 x 200 x 280mm Capacity: 3T/hr specific weight ca 0.5 kg/m3

Material Types: Al from glass recycling, brass from wood recycling andcopper from WEEE scrap processing.


Summary Results of Trial - Goudsmit The metal removal trial included two different types of metal removal system:-

• High-flux magnetic head roller • Fine-pole eddy current separation

In the trial unit these two devices are set up as individual pieces of equipment, but it is possible to combine the ferrous and non-ferrous removal into one machine, which only needs a single vibratory feeder to process the mixed plastic waste.

Figure 4.28 Drawing of combined Goudsmit ferrous and non ferrous separator

In the trials IT plastic and fridge p

Figure Fridge plastic waste

Magnetic Head Roller Tests The IT plastic was passed over thto remove ferrous and other magamount of ‘magnetic dust’).

WEEE Plastic Separation Technologies

Vibratory feeder

s

lastic was us

4.29 Infeed ma

e Neoflux hignetic material

2 x Neoflux head roller

.

ed as the infe

terial IT waste p

h density he (this include

Eddy current Sep

ed material.

lastic infeed

ad-roller unit twice d a significant

55

First Pass 9% of material removed Second Pass 0.6% of material removed This indicates the high level of residual ferrous material that can be recovered from some WEEE input streams. The effect of the second pass can be seen by the further 0.6% of ferrous material recovered. However not all of the reject ‘metal’ fraction is clean steel, as often pieces of circuit board, electrical connector and other part-metal items are diverted into this stream.

Figure 4.30 Metal items and other materials held by strong Neoflux magnetic head roller

Figure 4.31 Magnetic dust recovered by head roller on fridge plastic sample

The fridge material in this sample was passed twice over the magnetic head-roller system and the following material was rejected:- First Pass 1.7% ferrous metal removed + dust. Second Pass 0.3% ferrous metal removed The photo above highlights an interesting result of the trial which was the level of dust and fines that were diverted into the ‘ferrous reject’ – it was proven that


there is some magnetic attraction by introducing a hand-held magnet into the bin. ECS Tests After processing the fridge plastic through 2 stage ferrous removal rollers, it was fed into the fine-pole ECS unit. A high level of non-ferrous was liberated from the plastic mixture, over two batches 9.1% and 9.8% metal was removed from the bulk polymer.

Figure 4.32 Final, clean, de-metalled fridge plastic

Figure 4.33 Copper and aluminium removed by ECS

The IT plastic sample taken from the head-roller test was also processed on the ECS unit but it only yielded 0.25% non-ferrous metal fraction. As a stand-alone test result this level of non-ferrous would not justify the investment in an eddy-current unit. The large difference between the content of copper/aluminium residual scrap metal in the infeed WEEE plastic samples (i.e. 9.4% for fridge and 0.25% for IT), demonstrates the variable nature of this secondary raw material. There is a wide variation between the different input streams of WEEE waste and also between different upstream plant designs.


4.8 Equipment Suppliers

S+S Separation and Sorting Technology GmbH

Regener Strasse 130 D-94513 Schönberg - Germany

Tel.: +49 (0) 8554 / 308-0

CommoDaS GmbH

Rosengarten 10 D-22880 Wedel

Tel. +49 4103 - 1888-0 Fax +49 4103 - 1888-

Master Magnets Burnt Meadow Road North Moons Moat

Worcestershire B98 9PA

Tel: +44 (0)1527 65858 Fax: +44 (0)1527 65868


Magnapower 11 North Street Industrial Estate,

Droitwich, Worcestershire,

WR9 8JB Tel: +44 (0) 1905 779 157 Fax: +44 (0) 1905 779 867

E mail: [email protected]





5.0 Wood & Rubber Removal This section of the report follows on from the work reported under ‘Metal Removal’ using gravity separators. In both of the trials carried out, the opportunity was also taken to trial some plastics material that had been contaminated with wood particles and, in one case, a sample of plastic with rubber contamination. The wood found was largely chipboard. A level of wood and rubber contamination is inevitable when handling a waste stream derived form co-mingled, post-consumer WEEE. In some instances the level of wood contamination can be very high (up to circa 10%), particularly in mixed waste from households. This is largely attributed to the use of wood in stereo speakers and Hi-Fi cabinets throughout the 1970’s & ‘80’s. Wood composites are also found in TV cases, bases for video and tape recorders etc. During the shredding process, the wood tends to shatter into chips and releases a lot of smaller splinters / fines that stay with the plastic fraction for much of the process. Rubber has been used in a wide range of electrical applications, such as:-

• Door seals on fridges • Hoses and drive belts in washing machines • Drive belts & bumping strips on vacuum cleaners • Foot pads on desk-top equipment

It should be noted that the generic term ‘rubber’ is used in this instance to describe a wide range of elastomers found in electrical equipment. It will include natural rubber compounds, synthetic rubbers, foamed polymers, silicone seals and anything which has an ‘elastic feel’. Generally if the particle in question is not rigid plastic, is not metallic or wood, and has a squashy feel – then it will be classified as a ‘rubber’ piece. Like wood, these rubber parts tend to follow the plastic during some separation stages, and therefore need to be removed in a dedicated process step. The operating principle of gravity separators has been described in the previous section, so here we will concentrate upon a brief report of the trials and the results seen.


5.1 Trial on Prime – ‘Crown’ Gravity Separator – Wood Removal The technical parameters of the gravity separator used for this trial are:- Type of Gravity Separator: GS2 Capacity of Gravity Separator: 2 tonne/hr nominal throughput Deck Area : 1.0 m2 for trial Airflow: up to 11,000 m3/h Separation of plastic and wood – trial description The infeed material used for this trial to separate plastic from wood was similar to that used in the Trenn-So gravity separator trial (see below). It was mainly polypropylene (PP) and wood from post-consumer WEEE plastics, that had been recovered from a previous float-sink test on water. The majority of the infeed had been previously dried, but was mixed with some slightly damp material in order to have sufficient total infeed to cover the deck of the gravity separator. The mass balance for the trial is shown below:

Input Mass (kg) Dry (wood + PP) 11 Wet (wood + PP) 5 Output Light wood 3 Middling (PP with some wood) 7 PP - clean 4.5

(Note – some loss of mass as dust & fines to exhaust air flow.) Throughput of trial: ~1.5 tonne/hr approx. Results Analysis - Separation of plastic and wood Figure 5.1 below shows the input material-


The pictures below show the deck of the gravity separator during the trial. The 1st photograph shows the plastic fraction on the left/upper side of the deck, while the 2nd photograph shows the wood fraction on the right/lower side of the deck.

Figure 5.2 Plastic fraction

Figure 5.3 Wood fraction

These photographs demonstrate the capability of the gravity separator to concentrate the wood fraction in a mix of plastics. There is not a complete separation taking place, as a significant loss of plastic occurs down the lights discharge from the vibrating deck. The plastic used in this trial was polypropylene, which has a relatively low density that is similar to that of wood. It was noted during the trial that the ‘dry’ wood particles tend to be easier to separate than the ‘wet’ particles, which may reflect a change in density as the material dries out. Therefore drying of the infeed is probably a necessary step in order to deliver an acceptable separation effect.


The table below shows the composition of the primary streams in the mass balance:- Input Mass (kg)

Plastic Wood Fraction

Rubber Fraction

Metal/ Stone/ Other

Dry (wood + PP) 11 10.3 0.5 0.2 0.0 Wet (wood + PP) 5 4.7 0.2 0.1 0.0 Total Input 16 15.0 0.7 0.3 0.0 Output Light wood-rich 3.5 2.7 0.6 0.2 0.0 Middling (PP + trace of wood) 7.5 7.2 0.1 0.2 0.0

PP 5 4.6 0.0 0.0 0.3 Total Output 16 14.6 0.7 0.4 0.3 For the particular set of conditions used in this trial, it was possible to recover 30% of the infeed as a clean PP plastic fraction and a further 47% as PP middling fraction which had only a 1% wood contamination and approx 2% rubber. However it must be noted that some ‘heavy’ contaminant was found at a low level in the PP stream, which is a function of how the gravity separator works. It may be possible to ‘bleed-off’ this trace level of heavy contaminant (metal / stone) by using one of the top-edge output gates, but this was not tested during the trial. Around 85% of the wood contamination was separated from the main plastic stream. If a recycle stream was set up to recycle the middling fraction, then a greater percentage of wood would be removed to yield a higher purity end-product, but at a throughput trade-off. From this set of results, the gravity separator appears to be more efficient at removing wood than the rubber fraction, because the rubber is distributed evenly between the lights and middlings streams. A visual inspection of the output streams is given below:-

Figure 5.4 Output plastic fraction cleaned of wood particles


Figure 5.5 Output ‘lights’ fraction – wood rich waste

Separation of plastic and rubber – Prime Gravity Separator The separation of plastic and rubber was carried out using an infeed that consisted of polyvinylchloride (PVC) plastic and rubber. This material was chosen because the rubber particles were mostly black/grey and could easily be seen against the white coloured plastic, allowing a direct visual assessment of the separation effect. The mass balance for this trial is shown below:

Input Mass (kg) PVC + rubber mix 42.5 Output Rubber 6 Middling (PVC + trace rubber) 7.5 PVC 29

Figure 5.6 below shows the input for this trial.

The photographs below show the deck of the gravity separator during the trial. Figure 5.7 shows the plastic fraction on the left side of the deck, while figure 5.8 shows the rubber fraction on the right side of the deck.


Figure 5.7 Plastic fraction

Figure 5.8 Rubber fraction

Rubber-rich fraction

These photographs give a visual representation of the separation that was achieved. It can be seen that the PVC fraction in the upper photo still contains some particles of dense rubber material. The lighter foamed rubber parts have been successfully concentrated to the right hand / lower side of the shaking deck, but there is still a significant loss of PVC plastic with the ‘reject stream’. It should be noted that this trial was carried out with PVC plastic due to a lack of available WEEE material at the time. Also the PVC is all white and the contamination is primarily ‘black/grey’ in colour, thus enabling a rapid visual assessment of the separation effect during the tests. PVC is much heavier than most of the common WEEE polymers, so this should only be considered as a ‘model’ of what might be achieved with ‘real’ WEEE mixtures. It is clear that only a concentration effect is happening here; there is not a complete separation of rubber from the bulk PVC granules.


5.2 Trial on Trenn-So TTS Gravity Separator – Wood Removal The trial was to use the gravity separator to remove wood from plastic granules. The input was 40 kg of post-consumer WEEE polypropylene granules with wood contaminant, which had been recovered as the ‘floaters’ fraction from an earlier density separation trial in water. The average size of the particles was 5mm. The input material had been stored in a sealed container with a residual amount of water, so the plastic granules and the wood were slightly damp and it was judged that this would affect the results. Thus, the input was dried in the sun for a few days before being passed through the gravity separator

Figure 5.9 Input plastic material with wood contamination

This was separated on the gravity separator to yield the following sample of reject wood material:-

Figure 5.10 Reject wood material

For the removal of wood from plastic, the photograph above shows that the gravity separator has been partly successful, although there is still some


plastic visible in the wood fraction. Due to a lack of samples from the trial, (they were lost in transit) an assumption is made that all the light fraction is wood and all the heavy fraction is plastic. The table below shows the efficiency of the Trenn-So gravity separator on wood removal. Plastic Wood Input Composition 94% 6% Input Amount 37.2 kg 2.4 kg Output Amount 33 kg 5.7 kg As can be seen from the table, there is a ~9% loss of plastic into the wood fraction (i.e. in the 5.7kg of wood + plastic). Although all the wood has been removed, there is a loss of yield of the plastic, and this loss of yield can be reduced by recycling the wood fraction again, or re-tuning the gravity separator. Unfortunately, due to the loss of samples, it has not been possible to assess the purity of the ‘cleaned’ plastic fraction from this trial. 5.3 Equipment Suppliers

Prime Process Systems Ltd Unit N4 Inchbrook Trading Estate

Woodchester Stroud GL5 5EY

Tel: 01452 227025

Trenso Technik Trenn- und Sortiertechnik GmbH

Siemensstrasse 3 D-89264

Weissenhorn Tel: 073 09 96 20 0

Fax: 073 09 96 20 30 Email: info:trennso-trechnik.de




6.0 Centrifugal Separators

A wide range of devices exist that use the forces created by a spinning motion to bring about separation between particles of different densities suspended in a fluid. This section explains some of the theory behind the operation of such ‘centrifuges’ and then describes the practical experience of trials using examples of this equipment type.

Figure 6.1 Examples of two different designs of separating centrifuge

Hydrocylone Decanter

The separation effect in a centrifuge is created by the forces involved in moving a body through a circular motion. The rate at which the desired separation takes place is then controlled by the speed each particle can move through the separating fluid. A brief explanation of how these forces work is given below.

6.1 The Separating Force

A centripetal (or centre-seeking) force F is required to sustain a body of mass M moving along a curve trajectory. The force acts perpendicular to the direction of motion and is directed radially inward. The centripetal acceleration, ‘a’, which follows the same direction as the force, is given by the kinematic relationship:-

rV

a2θ=

Where is the tangential velocity at a given point on the trajectory and θV r is the radius of curvature at that point. Thus a high spin-speed and a small radius of curvature will deliver the stongest centripetal force. It can be seen in the photographs above that the


physical design of centrifuge equipment embodies this principle in the tubular shape of the separating chambers. This analysis holds for the motion of a body in a fixed reference frame,for example,a stationary laboratory. However in a mechanical centrifuge the frame is often rotating at the same angular speed as the fluid. Here, additional forces and accelerations arise, some of which are absent in the fixed frame. A particle in the rotating frame experiences a centrifugal acceleration directed radially outward from the axis of rotation with magnitude:

ra 2Ω= Where Ω is the angular velocity, r is the radius of the frame. Thus a solid particle suspended in a spinning liquid will be subjected to two opposing forces:-

1. The centripetal force transmitted through the fluid to pull it into the circular motion, and

2. The centrifugal force pulling it towards the outer walls of the circular vessel.

The larger of these two forces acting on any individual particle will determine in which direction it moves within the centrifuge. Factors which determine the result of this balance of forces include relative solid and liquid density, particle shape and size, fluid viscosity and concentration of solids in the liquid mixture. In addition to these theoretical factors affecting the movement of each particle as it moves through the centrifuge, there are also several factors about the mechanical design of each system which have significant impact upon the separation flow-route. The degree of turbulence created in the fluid and the way each particle interacts with the physical boundaries of the machine can also have large effect on how efficient a particle separation occurs within the process. 6.2 G-Level

Centrifugal acceleration G is measured in multiples of earth gravity ’ ’: g

DngG 20000142.0=

Where D is the centrifuge diameter in inches, n is the rotational speed in r.p.m.. G can be as low as 100g for slow speed, large basket units and up to 10,000 – 15,000g for high speed, decanter or disk centrifuges. In analytical ultra-centrifuges used to process small samples, G can be as much as 50,000g to effectively separate two phases with very small density difference. Because G is usually very much greater than g, the effect due to the earth’s gravity is negligible. So the spin axis of centrifuges can be either horizontal or vertical with little difference in performance.


The greater the G that can be attained, the greater the separating effect achieved. This is the main advantage that spinning centrifuges hold over stationary fluid separation. Much sharper separations can be achieved by using spinning fluids as opposed to keeping the fluid stationary. Greater spin-speed increases the separation force, as the resulting acceleration is directly proportional to the angular velocity squared. Therefore doubling the spin speed increases the separating force four fold. (see figure below).

Figure 6.2 Graph of separating force and revolutions per minute

0

1020

3040

50

6070

8090

100

0 2 4 6 8

Separating Force

Revo

lutio

ns p

er m

inut

e

10

6.3 Settling Forces Acting On a Particle in a Fluid

The speed at which any solid particle settles through a fluid is governed by Stoke’s Law. This equation predicts the theoretical speed at which a particle of know density and size will fall in a liquid of fixed density. Variations in particle shape, surface roughness and fluid viscosity are included in this empirical equation by means of a drag coefficient.

6.4 Stokes Law

ρρρ

PD

pps AC

gVV

)(2 −=

In the equation above, Vs is the settling velocity, ρ p is the density of the particle, ρ is the density of the fluid, g is the gravitational constant, Vp is the volume of the particle, CD is the drag coefficient and Ap is the area of the particle.


It can be seen that, for a particle of fixed density and size in a liquid of constant density, the speed at which the particle moves through the fluid will be determined by the square-root of the available G-force. Thus in centrifuge separations, the ability to create a much increased G-factor will bring about a more rapid movement of particles through the separating fluid. The movement being in the direction of that G-force. The table below indicates how increased rotational speed, and hence increased G-force, can create wide differences in settling velocity for given designs of separation unit. The sink/float velocity is typical of that measured for plastic particles of around 10mm size with a density of 1.05 g/cc in stationary water.

Stationary Sink/Float Hydrocyclone Decanter Centrifuge

RPM 0 500 5,000 10,000

Multiple g-forces attained

1g 50g 2,000g 5,000g

Settling Velocity

metre/min 1 7 44 70

In the case where the density of the solid particle is less than the liquid density, Stokes Law will predict the upwards floating velocity of the particle. This force will be in the opposite direction to the applied G-force and in centrifuges this means towards the central axis of rotation of the separating fluid. So for a mixture of solid particles in water, in which some particles are of a density below 1.00 g/cc and some are heavier than 1.00 g/cc, there will be a split created within the spinning liquid as some ‘float’ towards the central axis and others ‘sink’ to the outer circumference of the spin-chamber. As a result of this increase in settling velocity created by the high-G separation force, it is possible to use a relatively small volume of liquid in the physical separating zone of each type of centrifuge. Solid particles pass rapidly through the separating zone and exit the equipment in seconds rather that minutes, allowing the liquid medium to be re-circulated and re-used for successive separation steps. This has the advantage of making the high-G separation units much more compact for a given solids throughput, because a large volume of material can be separated in very short residence times inside the equipment. An example of this can be seen by comparing a sink-float tank with a single hydrocyclone unit, designed for the same solids volume throughput.


However as the process designer seeks ever increasing G-force by selecting higher rotational speed machinery, there is a payback in terms of the mechanical engineering design needed to support the high-speed rotating parts of the equipment. As the very high spin-speed equipment requires more substantial bearings and drive mechanisms, it become increasingly difficult to design adequate infeed and outfeed flow-paths to ensure a high-volume supply of solid/liquid mixture into the separating zone.

As a result of this increasingly complex mechanical design, the capital cost of high-speed centrifuges and their associated infeed and outfeed systems becomes very much higher than for simple low-G separation methods. As a final comment on Stoke’s Law the reader should note the impact of particle Area, ‘A’. Many plastic particles are naturally ‘plate-like’ in shape and therefore can present a much larger surface area to the separating fluid than a sphere or cuboid shape. Thus a flat particle can often take a different route through the centrifuge than that predicted for a more regular shaped item. Turbulence within the fluid will also have a greater impact upon flat platelets than other shape of particles.


6.5 Main types of centrifugal separators:-

6.5.1 Hydrocyclones –

• Typical g forces reached – 50g • Typical rotational speeds – 500rpm • Power source – provided by the pumping energy

A cyclone is a commonly used apparatus that makes use of gravity and centrifugal force to separate solid particles from a gas stream. A typical cyclone is a cylindrical vessel with a tangential inlet and conical shaped top and bottom outlets. Cyclones are widely used in various industries because they are easy to build, inspect and maintain.

Figure 6.3 Simple air cyclone for separating dust/particles from air stream (from www.kongskilde.com)

Hydrocyclones are similar devices to cyclones, but the operating fluid is a liquid rather than a gas. Hydrocyclones operate under pressure. The feed, a mixture of solids suspended in the operating fluid, enters the hydrocyclone tangentially through the inlet. This forces the mixture to spin inside the hydrocyclone, accelerating the separating fluid and solids mixture into a spiral flow-path.

This spinning motion generates centrifugal forces which cause the heavier particles to move outwards relative to the carrier fluid because of their greater density. The lighter particles move in the opposite direction towards the vortex finder, and exit via the central top outlet, termed ‘overflow’. The heavier solids that are pushed against the wall then travel down the length of the conical section of the hydrocyclone in a spiral pattern, towards the bottom outlet, termed the ‘underflow’.


Figure 6.4 Typical Configuration of a hydrocyclone and schematic representation of the flow-paths

6.5.2 Mechanical Centrifuges or Decanters

• Typical g forces reached – 5000g • Typical rotational speeds – 2,000 – 10,000rpm • Power source – external motor drive.

In mechanical centrifuges the rotational motion of the fluid is created by an external motor/drive system. This direct drive of the internal separating chamber imparts a spinning motion to the body of liquid (and solids) held within it, which produces the high-G zone needed for the separation. The liquid is usually introduced into the rotational chamber by pumping through hollow drive shafts, which necessitates well-engineered sealing and bearing systems at both the inlet and outlet points of the units. Very high forces are produced in the rotating parts such as the mechanically driven bowls or baskets, usually made of metal, turning inside a stationary casing. Rotating a cylinder at high speed induces a considerable tensile stress in the cylinder wall, particularly at larger diameters. This wall-stress limits the centrifugal force which can be generated in a unit of a given size and material of construction. Very high forces, therefore, can only be developed in very small centrifuges.


Figure 6.5 Tubular bowl centrifuge

The figure above is a schematic of a tubular bowl centrifuge which has been used for longer than most other designs. It is based on a simple geometry: formed by a tube, of length several times its diameter, rotating between bearings at each end. The process stream enters at the bottom of the centrifuge and high centrifugal forces act to separate out the solids. The bulk of the solids will adhere on the walls of the bowl, while the liquid phase exits at the top of the centrifuge.

As this type of system lacks a provision for solids rejection, the solids can only be removed by stopping the machine, dismantling it and scraping or flushing the solids out manually. Tubular bowl centrifuges have dewatering capacity, but limited solids capacity. Foaming can be a problem unless the system includes special skimming or centripetal pumps.

The figure below shows a disk stack separator. The simplest type of design is a ‘closed bowl’, containing the disk stack, with any solids present collecting at the outer part of the bowl, from which they have to be removed manually after stopping rotation. In a continuous disk-stack, the solids are discharged from the bowl by a number of methods, including the use of nozzles, which are open continuously, allowing thick slurry to discharge. In the more complicated designs, valve nozzles open automatically when the solid depth in the bowl reaches a certain value, and then close again when most of the solids have been discharged.


Figure 6.6 Disk stack separator

Whilst this design of equipment has primarily been developed for the removal of fine solid sludge or powders from liquid streams, there are some examples of the design where larger particles can be separated and discharged from the mechanical centrifuge body. Two types of mechanically driven decanters were tested under this project.


6.5.3 High Speed Mechanical centrifuges

• Typical g forces reached – 10,000g • Typical rotational speeds – 8,000– 15,000rpm • Power source – external motor drive

The angular velocity of this type of centrifuge is much greater than a typical mechanically driven decanter leading to an even greater separation force. This type of centrifuge is more commonly used for separating small particles, such as fine powders, as opposed to larger particles such as plastic chips. The operating principle is identical to other mechanically driven centrifuges. However the higher spin speeds demand a much higher standard of precision engineering and stronger components have to be used.

Figure 6.7 High speed mechanical centrifuges

More difficult separation can be achieved using a high speed centrifuge. With greater G forces being attained in this type of a separator, power consumption also increases. Capital cost increases per unit of throughput capacity for this type of machinery – making it suitable for small batch separations only. This type of equipment is unlikely to be applied to plastic separation, and it is more likely to be found in laboratory or pharmaceutical applications.


6.6 Description of Practical Trials Plastic separation trials were carried out on two different designs of mechanical centrifuge (Decanters) and on a hydrocyclone separating unit:-

• Mechanical Centrifuges:- o Flottweg ‘Sorticanter’ o Foma ‘Centrec’

• Hydrocyclone Rig – Berlin University A description and results from each of these trials is given below:- 6.6.1 Flottweg GmbH, Vilsbiburg, Germany – ‘Sorticanter’

Claimed Features:

• Control of density cuts to within 0.01g/cm3 relative density • High speed centrifuge • Excellent sharpness of cuts • Can operate on mixed plastics with more than 40% heavy material and

mixed plastics with less than 10% heavy metal • Dry output fractions prevent the need to install a dryer downstream • Can reach capacities up to 1000kg/hr

Working Principle The figure below shows the internals of the Sorticanter. The moving parts consist of a conical cylindrical bowl and a tapered cone section fitted with a variable pitch scroll, both rotating on a common horizontal axis. The drive mechanism is designed so that the bowl and the scroll section are rotated at different speeds, thus imparting a drive to the solids particles through the length of the separating chamber.

Figure 6.8 Internals of Sorticanter

The liquid feed mixture containing the suspended solids particles, is pumped into the Sorticanter via the hollow main shaft. The liquid mixture is then thrown out to the walls of the chamber by the centrifugal force, where a spinning liquid layer is created around the tips of the scroll flights. The centrifugal force


causes solids with a higher specific density than the liquid phase to settle against the outer wall. The scroll then conveys the separated solids up a conical ‘beach’ where they are discharged through ports in the main housing. Solids lighter than the liquid phase float and are carried with the liquid flow to the other end of the centrifuge, where they enter the conical bowl to be skimmed away from the liquid surface and discharged from the unit.

Thewhesepwas

WEEE

Machine Specification Max Bowl Speed: 3500min-1

Max Sediment Density: 1.3kg/dm3

Min/Max Product Temp: 0/1000C Rotor Spin Speed: 2500rpm Bowl Diameter: 400mm Material Types: post-consumer material, fibres and films, ground bottles,recovery of PET flakes, cable insulations, carpet fibres and general WEEE

Figure 6.9 Experimental Plant Layout:-

liquid used as the separating medium is recycled back to the feed tank re it is mixed with further solids material and fed to the Sorticanter for aration once again. Therefore there is no accumulation of wastewater and tewater treatment is minimized.

Plastic Separation Technologies 78

Figure 6.10 Block diagram of input/output flows

Large Scale Trial with Flottweg Sorticanter

Material A variety of WEEE material was processed through the Sorticanter. The post consumer WEEE contained polymers of various densities, and a level of metal contamination. The products tested were:-

1. CRT plastic – recovered from end-of-life TV and VDU recyclers. 2. Fridge plastic – recovered from end-of-life Fridge recyclers 3. IT plastic – recovered from end-of-life IT equipment recyclers 4. Post Consumer WEEE – recovered from collection stations for WEEE

from household waste Separations were carried out with salt solutions of 1.00 and 1.09 SG. These are typical density ‘cut-points’ that might be used to extract a useful polymer fraction from a mixture of WEEE plastics. In theory, polymers in the range of 1.0 and 1.09g/cm3 should be positively identified and polymers below 1.0 and above 1.09g/cm3 should be removed from the target fraction. Two separate runs were performed on the Sorticanter. The first was to remove the floaters using a salt solution of 1.09g/cm3. Then the floaters were used as the input material in the second run but this time using 1.0g/cm3 as the separating fluid. The sinkers were separated from the second run to leave polymers of density between 1.0 and 1.09g/cm3 – the ‘target product’ fraction. The aim of trial was to assess the efficiency of the density separation. If material greater than 1.09 g/cm3 or less than 1.0g/cm3 was identified in the target fraction, it would be counted against the performance of the machine.


A more detailed explanation of the off-line density fraction analysis that took place on the feed and products is explained in the ‘Analytical Methods’ section of this report.

Figure 6.11 TV/VDU WEEE material processed using the Sorticanter

Figure 6.12 Fridge WEEE material processed using the Sorticanter

Density Fraction Analysis of the In-feed Materials Used

This graph shows the mass percentage of solids particles falling within a range of density ‘fractions’ across the full range of the material. The dotted lines indicate the proportion of the infeed plastics particles that fall within the defined ‘target fraction’ density range of 1.00 to 1.09 SG.

Limits of the target fraction


IT

WEEE TV/VDU WEEE

Fridge WEEE

Post Consumer WEEE

Density Range:- % Below Target

Fraction 0.4 5 4 5

% Above Target Fraction 59 33 33 62

% Within Target

Fraction 40.6 62 63 33 The table above shows the percentage of in-feed material falling above, below and within the target density range. Summary of Results from Trial The Sorticanter extracted a light and heavy fraction at each density cut, so that the required target fraction between 1.0 g/cm3 and 1.09 g/cm3 was isolated as the ‘target product’. In order to measure the efficiency of the trial separation, a laboratory density analysis was performed on the collected target fraction. Any material remaining in the target fraction which was outside of the desired range was deemed to be ‘contamination’ of the required product. After performing this off-line analysis test, the results are summarised in the table below. Density Fraction Analysis of the Individual Products Separated

IT WEEE TV/VDU WEEE

Fridge WEEE

Post Consumer WEEE

Weight of material tested (g) 77.5 77.5 62.4 50.9

% below target fraction 0 2.71 0 0 weight (g) 0 3.5 0 0

% above target fraction 0 0 0 0

weight (g) 0 0 0 0

% within target fraction 100 96.5 100 100 weight (g) 77.5 74.8 62.4 50.9

The table above summarises the 4 streams processed through the Sorticanter which were subjected to laboratory density analysis. It can be seen from the analysis that for IT , Fridge and Post-Consumer WEEE 100% of the sample was within the target fraction. The machine


successfully separated the desired product from the bulk material very accurately achieving zero contamination in the target fraction. For the TV/VDU plastic, 2.7% of material was found to be contaminant in the target fraction. This contaminant should have been ‘floater’ material in the 1.0 g/cc water solution. These results indicate a very successful performance of the Sorticanter in terms of the efficiency of separation achieved with the type of plastic used.


6.6.2 FOMA Engineering, Leeuwarden Holland – CenTrec Claimed Features:

• Handles particle sizes between 3-20mm • Separation accuracy >99.5% • Low power consumption of 35kW • Maximum sediment density approximately 1,300kg/m3 • Density Difference between Heavy and Light Fraction: >0.02kg/dm3

Working Principle The separating principle of the Centrec unit works by creating a spinning cone of liquid in an open-topped bowl which has a separate external drive. A spinning disc is positioned near the base of this bowl to aid removal of the sinking particles, with floaters being skimmed-off by a funnel-tube near the top of the liquid cone. The following photograph and diagram shows the conical shaped rotating body of the unit, which has a flat disc near the base. The disc (3) rotates at a lower speed than the outer drum (1) and causes the liquid in contact with it to be drawn down the central outlet pipe (10). The liquid that rises up the walls of the spinning cone exits though a curved pipe (2) under its own momentum. A view of the machine and its overall size can be seen in the photograph below. To the right are two outlet liquid sieves to remove the solid particle from the re-circulating water flow. Particle size has an important effect on the operation of the machine. Larger pieces tend to cause the machine to block which then requires shutting down and unblocking. This problem can be mitigated by using flakes sizes below 10mm.

Figure 6.13 Centrec unit


Detail of Operation

Initially the outside rotor (1) is accelerated to the required speed, after which the separation liquid is fed to the centrifuge through connection (11) As soon as a stable liquid column (7) is built up and a continuous flow of liquid is achieved through the light fraction discharge tube (2) and the heavy fraction discharge port(10), the separation process is started.

Figure 6.14 Diagram of internal workings

Under the influence of centrifugal force which is acting on the particles as well as the liquid, the light particles flow to the surface of the liquid and are removed on the upper side (2) of the centrifuge. The heavy particles sink through the liquid column to the outside (9) and are removed from the centrifuge with a special construction (3 and 6) to the bottom centre outlet. The two fractions, the light- and heavy fraction, are now separated and are transported to downstream equipment, such as de-watering sieves and mechanical dryers.

The photographbelow shows a view down into spinning drum, where the floaters outlet tube can be seen supported by clamps on the left. The inlet liquid tube feeds into the central column below the solids hopper-cone.


Figure 6.15 Block diagram of input/output flows

-

W

Machine Specification:- Capacity: 1,000 - 1,200 kg/hr Max bowl speed - 2500 min-1

Separation Costs: 0.01 - 0.04 Euro/kg Separation accuracy: >99.5% Particle Size: 3 - 20mm Density Difference between Heavy and Light Fraction: >0.02kg/dm3 Particle to Liquid Ratio: 1:20 (solid material : liquid) Liquid usage: 3-8% remaining mixture of separated fractions Power: 35kW Air Consumption: 10nl/min – 6Bar Water: 2 bar, 1m3/hour Dimensions: Length 4250mm, Width 2550mm and Height 3500mm Material Types: PET bottles from PP caps, PMMA and ABS parts of rear lights of cars, PP pipes and rubber gaskets from PVC pipes, Aluminium foillids from PS cups and Poly-olefin fractions from house-hold waste


Large Scale Trial with FOMA CenTrec

Material The products tested were:-

1. CRT plastic – recovered from end-of-life TV and VDU recyclers. 2. Fridge plastic – recovered from end-of-life Fridge recyclers 3. IT plastic – recovered from end-of-life IT equipment recyclers 4. Post Consumer WEEE – recovered from collection stations of WEEE

from household waste

These materials were drawn from exactly the same sample batch as those used in the Sorticanter trials, in order to allow a direct comparison of the two equipment types.

Separations were carried out with salt solutions of 1.00 and 1.09 SG. Again to give an exact repeat test for comparison. Two separate runs were performed on the CenTrec. The first was to remove the floaters using a salt solution of 1.09g/cm3. Then the floaters were used as the input material in the second run but this time using 1.0g/cm3 as the separating fluid. The sinkers were separated from the second run to leave polymers of density between 1.0 and 1.09g/cm3 – the ‘target product’ fraction. The trial was to assess the sharpness of the density separation. If material greater than 1.09 g/cm3 or less than 1.0g/cm3 was identified in the target fraction, it would be counted against the performance of the machine. A more detailed explanation of the off-line density fraction analysis that took place on the feed and products is explained in the ‘Analytical Methods’ section of this report.

Figure 6.16 TV/VDU WEEE material processed using the CenTrec

Figure 6.17 Fridge WEEE material processed using the CenTrec


Density Fraction Analysis of the Individual In-feed Material Used

Limits of the target fraction

IT

WEEE TV/VDU WEEE

Fridge WEEE

Post Consumer WEEE

% Below Target

Fraction 0.4 5 4 5

% Above Target Fraction 59 33 33 62

% Within Target

Fraction 40.6 62 63 33

The table above shows how the mix of plastic particles in the feed material are distributed according to their densities. The first and second rows are the percentages of WEEE that are sinkers or floaters when placed in solutions of relative density 1.0g/cm3 or 1.09g/cm3. The third row tells us the percentages of material that are actually within the target density range. The aim of the trial was to successfully remove the polymers less than 1.0g/cm3 and greater than 1.09g/cm3 and to leave polymers only in the density range of 1.0 to 1.09g/cm3.


Summary of Results from Trial The Centrec extracted a light and heavy fraction at each density cut, so that the required target fraction between 1.0 g/cm3 and 1.09 g/cm3 was isolated as the ‘target product’. In order to measure the efficiency of the trial separation, a laboratory density analysis was performed on the collected ‘target’ fraction. Any material remaining in the target fraction which was outside of the desired range was deemed to be ‘contamination’ of the required product. After performing this off-line analysis test, the results are summarised in the table below.

IT WEEE TV/VDU WEEE

Fridge WEEE

Post Consumer WEEE

Weight of material tested (g) 77.4 72.9 97.0 94.8

% below target fraction 2.3 0 5.6 15.0 weight (g) 2.0 0 5.6 14.1

% above target fraction 1.7 15.0 5.4 7.0

weight (g) 1.3 10.6 5.2 6.8

% within target fraction 96.0 85.0 89.0 78.0 weight (g) 74.1 62.3 86.2 73.9

Discussion of Results For IT WEEE, 96% of the separated bulk material was found to be in the correct density range. 2.3% of the target fraction should actually have been separated into the light fraction and 1.7% should have been in the heavy fraction. Therefore a total of 4% contamination was found in the target fraction. The TV/VDU WEEE was analysed and it was found that 85% of the target fraction had a density between 1.0 and 1.09g/cm3. 15% of the target fraction was contaminated with material greater than 1.09g/cm3 density but material less than 1.0g/cm3 was not found in the target fraction. The Fridge WEEE was analysed and it was found that 89% of the target fraction was in the correct density range. However, 5.4% of the target fraction should have been separated as heavies and 5.6% as lights as this percentage of the target polymer had a density greater than 1.09g/cm3 or less than 1.0g/cm3. 22% contamination was found in the target fraction for the Post Consumer WEEE material. 78% of the separated fraction was in the correct density range, but 15% was less than 1.0g/cm3 and 7% was greater than 1.09g/cm3 density.


Density Fraction Analysis for Output Materials A visual comparison of the separation performance of each mechanical centrifuge can be seen in the following bar charts. This shows the density fraction analysis of the output product fractions for the Sorticanter and Centrec trials.

AxBond of Defra Polymer Type Separation Material

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

Float 1.00 1.00 - 1.025 1.025 - 1.05 1.05 - 1.075 1.075 - 1.9 Sink 1.09

Floa

ter %

Post Consumer WEEE 1- 19ITPost Consumer WEEE 20- 38TV/VDUFridge

Comparison of Sorticanter (top) and Centrec (bottom) output product density range. Target range shown by dotted lines.


6.7 Comparison of Mechanical Centrifuges It is clear for this comparison that the Sorticanter delivers a much sharper separation efficiency than the Centrec unit, with only one of the test samples having a small percentage of contaminant material from outside of the desired target range. The Centrec’s performance as measured in these tests indicates that an acceptable purity level could only be achieved by running a second pass through the machine – which is expensive both in terms of loss of capacity and operating cost per tonne of material. This difference in performance is reflected in the capital and operating costs of the two units. The Sorticanter costs in the order of €500,000 Euro compared with a cost estimate of €270,000 for the Centrec unit. (These capital costs include the infeed conveyor equipment and the de-watering sieves needed on the Centrec). It should be noted that the Sorticanter claims a ‘dry’ output plastic granule, removing the need for further mechanical drying equipment after the process. The higher capital cost of the more complex Sorticanter design also comes with the negative of potentially high maintenance and wear costs. Users should factor-in an element of rotor wear cost of €40 – 50k every 12 months and consider the need to hold an expensive spare rotor for the avoidance of long downtime when awaiting rotor refurbishment by the manufacturer. Summary Table - Sorticanter Vs CenTrec

Capital Cost

€000’s

Facilitate post

consumer WEEE

Maximum Throughput

Kg/hr Maintenance

Costs Separation Efficiency

Centrifuge Spin

Speed rpm

Flottweg Sorticanter 500 Yes 1000 High Excellent 3500

FOMA CenTrec 270 Yes 1200 Medium Average 2500


6.8 Hydrocyclone Trials – conducted at University of Berlin Material A similar set of polymer materials to those used on the mechanical centrifuge trials were sent to the University of Berlin. Three infeed materials were separated in the hydrocyclone:

1. IT plastic fraction 2. WEEE plastic fraction 3. Fridge plastic fraction

The fractions were separated in a conical hydrocyclone into density fractions as follows:

- below and above RD 1.00 polymers using water as separation liquid - below and above RD 1.09 polymers using a salt solution as separation

liquid Different vortex and apex diameters were used to find an optimised vortex/apex ratio to achieve the best split (i.e. a separation that was as close as possible to the ‘target fraction’, with 100% of the solids particles falling between 1.00 and 1.09 SG). All fractions were analysed before and after the separation to evaluate the efficiency of the separation.

Figure 6.18 Test rig used

This test rig re-circulates the pumped liquid and plastic particle mixture through the hydrocyclone and both the overflow and underflow streams fall back into the feed mix tank. In this manner it is possible to maintain a good control of the solids concentration and to run the hydrocyclone in steady-state conditions at a high flow-rate, with the minimum of material handling requirements.


A detailed multi-variable test programme was conducted over a period of several weeks. From this wealth of data, we have extracted the most important results for this report, concentrating only upon those separation tests which gave close to the desired target fraction in the output product stream. Pump Speed, Inlet Pressure and Flow-rate As part of the test programme the liquids flow-rate was varied across the available range of the feed pump speed, using a fixed hydrocyclone configuration. This yields the following graphs:-

Figure 6.19 Pump Speed versus Inlet Pressure HC seperation with 1090kg/m3 fridge polymer: speed vs. pressure for cone =75cm(29.5") votex finder

=64mm(2.52") Apex orifice =40mm(1.57")

0

2

4

6

8

10

12

14

16

51.5 61 69

Pump speed in %

Pres

sure

in p

si

Inlet pressure to the hydrocyclone is essentially linear in relation to pump speed over the range tested (50 – 70% of full range).

Figure 6.20 Pump Speed vs. Total Flowrate HC seperation with 1090kg/m3 Fridge Polymer: speed vs flow rate for cone =75cm(29.5") votex finder

=64mm(2.52") Apex orifice =40mm(1.57")

0

0.02

0.04

0.06

0.08

0.1

0.12

51.5 61 69

pump speed in %

flow

rate

in m

³/10s

This shows that liquid flow is also directly related to pump speed.


Figure 6.21 Inlet Pressure vs Overflow and Underflow Volume Flow HC seperation with 1090kg/m3 fridge polymer: Flow rate UF Vs OF for cone =75cm(29.5") votex

finder =64mm(2.52") Apex orifice =40mm(1.57")

0

5

10

15

20

25

30

35

7.975 11.313 14.504

pressure in psi

flow

rate

m3/

h

flow rate UFflow rate OF

This graph demonstrates an important issue for hydrocyclones, which is that the flow ‘split’ between the overflow (OF - top vortex outlet) and the underflow (UF - base apex outlet) does not vary in direct proportion to the input flow-rate. For the particular configuration in this test, the vortex finder orifice is set at 64mm and the apex outlet cone orifice is set at 40mm. The graph indicates that, as the pump speed is increased, the available increase in pressure head at the inlet to the hydrocyclone is causing a greater increase in flow out through the vortex than the percentage increase in flow at the base outlet, where flow rate is virtually stable across the tested range. The liquid is following the route of least resistance; out through the larger orifice of the vortex finder rather than via the base apex outlet, which appears to be running at near full capacity, regardless of the pressure increase. In terms of the ability to perform accurate separations, it can be seen that more of the entrained solids material will be driven with the flow towards the vortex as pump speed (and pressure) are increased. To the operator of the equipment, this can be seen as a decrease in separation efficiency when running at higher volume throughputs. A greater proportion of ‘sinker’ particles (i.e. with solids density heavier than the liquid medium), will report to the overflow stream as erroneous ‘floaters’ when pumping volume is increased. As mentioned earlier in this report, the major flow regime inside the hydrocyclone is the rotational flow as the liquid and solids mixture corkscrews down the conical shape of the cyclone body and back up the central liquid vortex. However the other bulk movements of the liquid from inlet down to the base outlet point and, more importantly, radially inwards to the central vortex finder, also have a large impact upon the accuracy with which each solid particle ‘reports’ to the correct side of the desired density split. Higher flows create greater shear rates in the fluid and with this comes increased turbulence and bulk flow towards the vortex finder; both having large effects on the quality of the separation taking place.


Particle Size of Sample Plastics For each of the supplied materials the particle size was investigated by sieve analysis, as follows:-

Material Particle Size (mean) Size Range (d10 – d90) IT plastic 2.4 mm 1.7 - 3.6 mm WEEE plastic 3.6 2.2 - 5.9 mm Fridge plastic 3.7 1.8 6.2 mm It can be seen that the materials have broadly similar particle size, although it should be noted that the fridge material had more flat shaped particles with thinner walled pieces than the other two materials. Having investigated the flow and pressure response of the test unit and measured the particle size of the sample materials, the following separations were then studied:- 6.8.1 IT Plastic Separation The input material sample was analysed by a laboratory float-sink test in water across the range of the sample bags sent for testing, this shows the consistency of the infeed material through the sample set. IT- input separated in RD 1.00 (tested on 18 samples to show the consistency of the test method):

IT Seperation with wet density 1000kg/m^3 Evaluation of sink & float in bags

-

10

20

30

40

50

60

70

80

90

100

IT 3 IT 6 IT 9 IT 12 IT 15 IT18 IT 21 IT 24 IT 27 IT 31 IT 33 IT 36 IT 39 IT 42 IT 45 IT 48 IT 51 IT 54

IT Material

Bag

Con

tent

s Floater

Sink

This material was then loaded into the test rig and separated using a range of different process conditions and hydrocyclone configurations. Samples of the


overflow and underflow streams were collected and then tested in off-line liquid solutions to find out the accuracy of the separation. This was assessed by measuring the percentage of solids that had reported erroneously to the OF / UF fractions, as shown in the graph below:- IT- output separated in RD 1.00 (tested on 3 samples to analyse the influence of the pump pressure):

HC seperation w ith 1000kg/m3: Overflow output vs Underflow output for cone =75cm(29.53") votex finder =40mm(1.675") Apex orifice =40mm(1.675") by Input Polymer Consentration = 121.238kg

0

10

20

30

40

50

60

70

80

90

100

7.975 12.6 14.5

Pressure in psi

Mat

eria

l in

% % OF

% UF

% sink in OF

% float in UF

It can be seen that there is primarily sinker material in the IT sample at 1.00 liquid density, with >95% of the solids reporting to the UF stream. Only a small percentage of sinkers were ‘lost’ to the overflow stream, and nil floaters were found in the underflow material. This graph also indicates that inlet pressure had little effect upon the separation across the range tested. In this example, it is important to note that the quality of the split obtained depends upon the desired separation required by the operator. If the aim of the separation was to deliver a high purity of overflow plastic material (i.e the polyolefin, floaters fraction), then even a 3 – 5% contamination level of sinkers in the OF may be considered unacceptable. However the sinkers fraction is essentially free of any floating material – indicating a very good separation on that basis. The lesson to learn from this example, is that hydrocyclones rarely produce a 100% perfect split of material between the overflow and underflow fractions. It is necessary for the operator to decide in advance which of the streams are required to be produced at near 100% purity and in making this decision, it must be accepted that there is some lost of useful yield to the other output flow. IT Separation at higher liquid density of 1.09 SG The input material was also tested in a salt solution of 1.09 density using an off-line laboratory test to determine the split of material in the sample set.


IT- input separated in RD 1.09 (tested on 18 samples to analyse the reproducibility of the test method):

IT seperation w ith wet density 1090kg/m3 evaluation of sink & float in bags

-

10

20

30

40

50

60

70

80

90

100

IT 3 IT 6 IT 9 IT 12 IT 15 IT18 IT 21 IT 24 IT 27 IT 31 IT 33 IT 36 IT 39 IT 42 IT 45 IT 48 IT 51 IT 54

IT

Bag

con

tent

in %

Floater

Sink

This test indicates that between 50 – 65% of the IT plastic had a particle density greater than 1.09 SG, and therefore is shown as ‘sink’ on the above graph. The IT material which had been collected from the sinkers fraction at 1.0 SG was then processed on the hydrocyclone test rig using a salt solution density of 1.09SG. The following material splits were measured at 3 different densities:- IT- output separated in RD 1.09 (tested on 3 samples to analyse the influence of the pump pressure). Input material taken from the sinkers of the water split (PS-rich):

HC seperation with 1090kg/m3 IT Polymer: Overflow output vs Underflow output for cone =131cm(51.57") votex finder =53mm(2.09") Apex orifice =30mm(1.18")

0

10

20

30

40

50

60

70

80

90

100

7,975 11,31 14,5

Pressure in p si

% OF% UF% sink in OF% f loat in UF


This test gives the following results:-

• The pump pressure seems to have no influence on the results. • The underflow fraction has got 3 - 5% of floaters which is a loss of

desired product yield. • The overflow fraction has got impurities of 3 - 5% which are impurities

in the final product. Purity of 95 - 97% required product. • The final product yield is approx. 40 - 42%, based upon the total input

mass. In this particular example the target product fraction has been set to recover polymers with density between 1.00 – 1.09 SG, this being the fraction usually associated with a high yield of styrenic polymer types (polystyrene and ABS plastic types). It can be seen that, for this material, it was not possible to find a hydrocyclone configuration that gave the perfect split of 100% pure output plastic in the target range. In order to reach a high purity fraction of output product, while still operating at high throughputs, it is often necessary to have more than one hydrocyclone operating in series on full-scale production plants. This incurs additional complexity and cost for the total separation plant design. Similar separation trials were then carried out on the WEEE plastic samples and Fridge material as described briefly below:- 6.8.2 – WEEE Plastic Separation WEEE- input separated in RD 1.00 off-line lab. test (tested on 4 samples to analyse the reproducibility of the test method):

WEEE Seperation with wet density 1000kg/m^3 Evaluation of sink & float in bags

-

10

20

30

40

50

60

70

80

90

100

WEEE1 WEEE 2 WEEE 3 WEEE 4

Material WEEE

Tota

l bag

s co

nten

tin %

Floater

Sink


This shows between 5 – 15% floaters (poly-olefins) in the WEEE plastic. WEEE- output separated in RD 1.00 (tested on 3 samples to analyse the influence of the pump pressure):

HC seperation w ith 1000kg/m3: Overflow output vs Underflow output for cone =121cm(47.6") votex finder =64mm(2.52") Apex orifice =38mm(1.496") by Input Polymer Consentration = 98kg

0

10

20

30

40

50

60

70

80

90

100

7.975 10.44 13.78

ressure in psi

Mat

eria

l in

% % OF

% UF

% sink in OF

% float in UF

• The pump pressure seems to have no influence on the results. • The underflow fraction is very pure and does not contain any poly-

olefins. • The overflow fraction has got approx. 15 - 20% sinkers which is a loss

of product yield. WEEE- input separated in RD 1.09 (tested on 4 samples to analyse the reproducibility of the test method):

WEEE Seperation with wet density 1090kg/m 3 Evaluation of sink & float in bags

-

10

20

30

40

50

60

70

80

90

100

WEEE1 WEEE 2 WEEE 3 WEEE 4

WEEE

bag

cont

ent i

n %

Floater

Sink


WEEE- output separated in RD 1.09 (tested on 3 samples to analyse the influence of the pump pressure). Input material taken from the sinkers of the water split (PS-rich):

HC seperation with 1090kg/m3 WEEE Polymer: Overflow output vs Underflow output for cone =75cm(29.5") votex finder =40mm(1.57") Apex orifice =35mm(1.38")

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

7.975 11.313 14.504

pressure in psi

Mat

eria

l in

% % OF

% UF

% sink in OF

% float in UF

• The pump pressure seems to have no influence on the results. • The underflow fraction has got 3 - 5% of floaters which is a lost of

sinker yield. • The overflow fraction has got impurities of 3 - 5% which are impurities

in the final product. Purity of 95 - 97% PS. • The final product yield is approximately 38% (overflow material on 1.09

density). 6.8.2 Fridge Plastic Separations

Fridge - input separated in RD 1.00 (tested on 2 samples):

Fridge Seperation with wet density 1000kg/m3 Evaluation of sink & float in bags

0

10

20

30

40

50

60

70

80

90

100

F1 F 2

Fridge materials

Bag

con

tent

in %

FS1 Floater

FS1 Sink


Fridge - output separated in RD 1.00 (tested on 1 sample in a tank because particle size blocked the pump):

Friedge Tank Flow-Sink Seperation by Wet Density = 1000kg/m³

6.9%

93.1%

Sink Floater

In the fridge fraction are 93% sinker in water and 7% floater in water (mainly poly-olefins – PP/PE plastic). Fridge- input separated in RD 1.09 (tested on 2 samples):

Fridge Seperation w ith wet density 1090kg/m 3 Evaluation of sink & float in bags

-

10

20

30

40

50

60

70

80

90

100

F1 F 2

Fridge materials

bag

cont

ent i

n %

f loater

sink


Fridge- output separated in RD 1.09 (tested on 3 samples to analyse the influence of the pump pressure). Input material taken from the sinkers of the water split (PS-rich):

HC seperation w ith 1090kg/m3 Fridge Polymer: Overflow output vs Underflow output for cone =75cm(29.5") votex finder =64mm(2.52") Apex orifice =40mm(1.57")

0

10

20

30

40

50

60

70

80

90

100

7.975 11.313 14.504

pressure in psi

Mat

eria

l in

% % OF

% UF

% sink in OF

% float in UF

Main findings:- • This particular experiment delivered a ‘perfect split’ on the infeed fridge

plastic material. • The underflow fraction has got no loss of product yield – nil% float in

UF (green bar on chart – nil height) • The overflow fraction has got no sinkers in the overflow and the purity

is 100%. (cream bar on chart – nil height) • The final product yield is approx. 70 - 72%.(floaters on 1.09; sink in

1.00SG) 6.8.3 Discussion – Hydrocyclone Trials The results presented in this section represent only a fraction of the many different variations used during the trial to find an ‘optimum’ setting for each material being separated. The physical configuration of the test hydrocyclone can be altered using the change parts for :-

• Vortex finder outlet diameter – at overflow • Apex spigot diameter – at underflow • Cone angle and length

In addition it is possible to vary the pumping speed to alter inlet pressure and flow rate, which is then split between vortex and spigot outlets in varying ratio. The liquid density can be altered and the solids concentration within the liquid is another variable. Particle shape and size can also have a significant impact upon the level and accuracy of separation.


Therefore, while hydrocyclones appear to be an essentially simple design of centrifugal separator, it is clear that the number of variables which influence the complex flow regimes inside the unit, can create a technical challenge to the operator. For each input stream of mixed plastic, it is necessary to conduct a series of iterative experiments to find the optimum settings that can deliver the required product density split. If the make-up, size or shape of the solids feed changes then it may be necessary to make further changes in configuration to return to an acceptable separation efficiency. In nearly all cases the ‘perfect split’ is very difficult to achieve –( i.e. 100% purity in both the overflow floaters and underflow sinkers). This means that in a practical application there will nearly always be a loss of useful product yield to one of the output flows. In order to recover this yield of material (and give acceptable purity to the secondary stream) it may be necessary to make a second pass through the process or to design a system with a series of hydrocyclone units, each operating at a different configuration.



Werner-Habig-Straße 1 59302 Oelde (Germany) Phone: +49 2522 77-0 Fax: +49 2522 77-2488

KREBS 21-22 Carrick Business Centre

Commercial Road, Penryn Nr. Falmouth

TR10 8AR Tel: +44 1326 379393






7.0 Spectrographic Methods This section of the report looks at rapid on-line sorting of polymer waste streams using two spectral analysis methods:-

• Near Infra Red light; • X-Ray Transmission

A brief explanation of the technology is given, followed by summary findings of the practical trials. 7.1 What is Spectrography?

Spectrography is the quantitative study of electromagnetic spectra. The term electromagnetic spectroscopy is also used to describe the field of spectral analysis across a wide range of wavelengths. More specifically the term ‘spectrophotometry’ deals only with the wavelengths of visible light, near ultraviolet, and near infrared. Hence the use of NIR light as a tool in polymer analysis falls within this definition. The use of X-rays as the source of emission energy, falls within a different range of the electromagnetic spectrum, as can be seen in the following figure.

Figure 7.1 Electromagnetic spectrum

In this diagram the visible light section of the spectrum has been expanded out into the range of colours from violet to red. In most spectrometers a prism or similar device is used to spread the range of light reflected back from (or passing through) an object so that quantitative analysis of the wavelengths can take place. The measurement of this reflected light spectrum across the chosen range of wavelengths is then used to produce a characteristic ‘fingerprint’ for each type of material.


Figure 7.2 MIR scan of HIPS plastic sample

The different peaks or troughs on the IR curve are representative of particular chemical groups within the specimen. Infrared spectroscopy works because chemical bonds have specific frequencies at which they vibrate or resonate in response to the incident wavelength of light. Thus a particular bond, such as C=O bond in a hydrocarbon compound, will display one or more identifying peaks as the bond vibrates at its characteristic frequencies across the range. As an example, the atoms in a CH2 group can vibrate in six different ways, symmetrical and asymmetrical stretching, scissoring, rocking, wagging and twisting.

Figure 7.3 Vibrations of CH2 group

Identification of more complex molecular groups, such as those found in a polymer like ABS (Acrylonitrile, Butadiene, Styrene co-polymer), will rely upon comparison of the complex range of peaks in the IR spectrum with a library of stored data of ‘know’ polymer types. The correct identification thus depends upon very rapid mapping and comparison of the collected spectrum for the ‘unknown’ sample with the large number of stored spectra in the memory of the measuring device. Often this matching process is accomplished using statistical and mathematical methods such a Fourier Transform analysis; to speed up the process and to give a ‘degree of fit’ measurement (e.g. 75% fit with ABS known sample in library).


7.2 Industrial IR Sorting Machines The above principles of using IR spectrography to identify polymers and other materials have been embodied into industrial scale equipment to provide rapid on-line sorting capability. This type of machinery has been most commonly employed in the sorting of whole plastic bottles into PET and HDPE fractions. The machinery typically consists of an IR light source and a set of optical detectors mounted over a conveyor belt that carries the material to be sorted. Shortly after the detectors the material passes over an array of air jet ejectors mounted at the end of the conveyor belt.

Figure 7.4 TiTech diagram NIR – Sorter layout

Compressed air impulse

TTiiTTeecchh PPoollyySSoorrtt®®

Scanner

Conveyor belt

Reject

Light source

Product

The detectors scan each item as it passes beneath them to measure the spectrum of light reflected across the near-infra red range. A computer built into the machine very rapidly analyses each spectrum and compares this to the reference spectrum for the target polymer to be sorted. If the spectra match, then an accurately timed and positioned pulse of compressed air is used to shoot the detected item over a deflector plate and into the collection bin.

Figure 7.5 3-D Sketch of bottle sorting unit.

7.3 X-Ray Transmission Sorting


The ability to distinguish different materials by means of XRT is explained by considering basic X-ray transmission theory. Lambert’s law gives the transmission damping of a sample of thickness d at X-ray source intensity Io

In this equation:- Idet is the detected intensity of the X-ray after passing through the sample. µ is the linear damping coefficient, which is a function of the wavelength, λ of the X-ray radiation being used. This indicates that in single wavelength X-ray detectors, the intensity of the detected signal is largely affected by the thickness of the sample. In this manner a thick paperback book passing through an X-ray scanner would give a much higher level of damping to the transmitted energy, than a thin magazine or newspaper. In the sorting of waste plastics, it is clear that the damping effects of different thicknesses of material ‘d’, will cause large variation in the transmitted X-ray signal. These changes in thickness will swamp the differences in transmitted signal that result from the atomic structure of the materials themselves. Hence in order to negate this problem, dual-energy X-ray transmission methods have been developed :– DE-XRT. Two wavelengths of X-ray are transmitted onto the sample items simultaneously, and the level of detected intensity is recorded at each wavelength. By using this information at two different energy levels to mathematically solve the equation of Lambert’s law, it is possible to rule out the effect of sample thickness. The results can then be used to approximate the average atomic number of the material in the sample, independent of thickness of material. In an example of this equipment seen in a laboratory situation, it was possible to carry out the above processing of the dual-energy transmission data across a fine grid of the viewed sample, down to a ‘pixel’ size of 1.5mm. This enabled a detailed picture to built up across randomly shaped plastic pieces, representing the average atomic density in each pixel. In this manner inclusions such as metal screw inserts set into plastic components could be rapidly identified. The following photo shows a set of plastic parts in normal light and as they are seen after processing on the DE-XRT machine. Colours have been used to indicate different atomic densities within the sample.

Figure 7.6 Scan of sample set of plastic pieces


(Ack – T de Jong, Delft University)

The level of X-Ray transmission is affected by the average atomic number of the material within each ‘pixel’ of the sample. Thus most polyolefin products (Carbon-Hydrogen rich) will have significantly different transmission characteristic than PVC, for example. A further level of analysis can be added by comparing the measured transmission data (after correcting for ‘d’ variations), with a transmission curve for a known material. The processed data can then be used to plot a graph showing the grouping of the total data set in relation to the known regression curve. Figure 7. 7 Graph of processed DE-XRT data for IT plastic samples – note groupings

of similar material types (yellow & green ellipse)


By using this sophisticated processing of the X-ray transmission data, it is possible to develop a sorting methodology based upon the specified ‘target’ zones. In the above example graph, it could be determined that plastic items falling into the ‘yellow ellipse’ zone are characteristic of a particular polymer type, and this could be set as the accept/reject criteria. Additives and fillers within the polymer will create significant differences in the transmission. This provides a basis for detection of heavy metal additives and for halogenated compounds. It may also be possible to remove polymers with high levels of mineral fillers from more ‘pure’ plastic pieces. This technique has been developed into a full-scale sorting process that utilizes the same basic layout as other camera-based sorting units. The level of transmitted X-rays are measured using a detector unit mounted below the transfer belts, and ejection takes place using a similar bank of air-jets. Reference :- Sorting of Plastics Using X-Ray Transmission, Identiplast paper 2005 Tako P.R. de Jong Delft University of Technology Faculty of Civil Engineering and Geosciences Department of Geotechnology Mijnbouwstraat 120, NL-2628 RX Delft E-mail: [email protected]


7.4 TiTech GmbH Germany- Near Infra-Red Sorting Trial Claimed Features:

• High capacity – 2 to 10 tonne/hour • High reliability – 7000 hours MTBF • Easy to operate and maintain • Determination of chemical constituents possible • Sorting of different shapes based on spectral features • Removal of foreign materials, like polymers from natural materials

Working Principle The mixed plastic material in transferred via an inclined conveyor onto a vibratory feeder. This feeder ensures that the mixed polymer is distributed evenly onto the surface of the sensing conveyor. The layer of well-spaced plastic particles passes underneath the IR detection head, where two sensors analyse the material against operator set criteria. When a reject is detected by the sensor, the air-jets are activated and the reject is blown over the deflection plate. There are 96 nozzles supplying the air-jets and each nozzle is individually controlled for accuracy.

Figure 7.8 Test unit

Titech Polysort Sensor

Inclined Conveyor

This equipment is normally used for plastic items above 40mm in particle size, such as whole plastic bottles. So during this trial it was important to find out if adequate identification and sorting could be carried out on particles below 40mm in size. An important factor in the sorting process is that the unit can be set to give a ‘positive sort’. This means that only the desired material type is ejected from the mixed stream of plastics, which should deliver very high purity of output.


Figure 7.9 Titech Polysort

Titech Polysort

Air Jets

Deflection Plate

Products Reject

LT

BPpt

W

Machine Specification Speed of Conveyor (fast running conveyor): 2.5 – 3.0m/s Speed of Conveyor (slow running conveyor): 0.5m/sDistance Between Sensor and Belt: 425mm Width of Belt: 1000mm Number and Position of Sensors: 1000mm wide, Two scanning systems Type of Light Source: 12V, 20W. Combined visual and NIR halogen lamp Material Types: post-consumer material, fibres and films, ground bottles, recovery of PET flakes, cable insulations, carpet fibres and general WEEE

arge Scale Trial with Titech NIR Sorter he materials used in the trial are as below.

Material Type Type Colour Size Estimation Post Consumer

WEEE VDU monitor

case light coloured 40 mm

Post Consumer WEEE

WEEE mainly black 40 mm

oth materials contained a significant quantity of ABS, PS, PC and olyolefins. The aim of the trial was to separate them into their individual olymer types, by positively sorting each polymer on successive runs through he Titech NIR unit.


Figure 7.10 Material used in trial

VDU monitor case plastic Mixed WEEE plastic

Trial Targets in Detail The targets for the trial were defined as follows:

1. Positive detection of ABS 2. Positive detection of PS 3. Positive detection of Polyolefins – PP/PE 4. Positive detection of PC

The trial sequence was as follows:- Trial 1 Runs 1-3 The input sample (VDU monitor cases) was passed through positive ABS detection, the reject from this run was then passed through positive PS detection, subsequently, and the reject from this run was passed through positive PP/PE detection. Trial 2 Runs 4-5 As for the sample above, this input (mixed WEEE) is passed through positive PS detection and positive PP/PE (polyolefins) detection. Trial 3 Run 6 To check the accuracy of the ABS detection in the first run, the target output, from the first run was passed through positive PS detection, to check the amount of PS impurity in the target output. Trial 4 Runs 1-6 Similarly for the input sample above, the second sample (post consumer WEEE) was passed through positive detection for ABS, and the reject of this run was tested for positive detection of PS, followed by PP/PE and PC.


Summary of Results from Trial The material was separated into its apparent individual polymer types. The results from each of the runs are presented below. Trial/Run No.

Target Polymer Type

Input Mass (kg)

Sorted Target Output (kg)

Reject Output Amount (kg)

Throughput (tonne / hr) Est.

Loss (kg)

T1R1 ABS 25.24 14.17 11.07 1.2 0T1R2 PS 11.07 2.07 7.58 1.2 1.42T1R3 PP/PE 7.58 0.07 6.27 1.2 1.24 T2R4 PS 27.47 4.95 22.52 1.4 0T2R5 PP/PE 30.5 0.12 30.38 1.4 0 T3R6 PS 14.17 0.98 12.05 - 1.14 T4R1 ABS 21.49 3.51 17.98 - -T4R2 PS 17.98 1.42 16.56 - -T4R3 PP/PE 16.56 0.55 16.01 - -T4R4 PC 16.01 0.26 15.75 - -

Note :– T1 – T3 – beige shading all from VDU sample. Run T4 – green shading, mixed WEEE It can be seen from the above data, that the reject fraction from one run is used as the input material for the next run. So that the successive processing of the input to remove the desired polymer type should leave a residue of ‘other’ (or unidentified materials) as the final reject. The material used in run T1 to T3 was mainly light coloured ‘beige’ plastic from shredded VDU casings. There was a sort of 56% ABS on run 1, 8% PS on run 2 and the remainder was either ‘reject’ or losses. On the re-run of the positively sorted ABS from sample T1R1, the 14.17 kg of ‘ABS’ yielded 0.98 kg of ‘PS’ on the second pass through the unit. This indicates either a erroneous identification on the first pass or a miss-sort on the second pass. The trial with the mixed WEEE plastic gave disappointing results, only a low percentage of plastic was positively identified on each successive pass through the machine (i.e. 16% as ABS, 7% as PS, 2.5% as PP/PE). After 4 sorting runs, there was still 73% of the original feed material left as ‘reject’ or unidentified polymer type. This was attributed to the large proportion of black/grey plastics in the mixed WEEE sample. Dark colours tend to absorb a lot of the incident light (including infra-red wavelengths) and thus the level of reflected light with which to make the signal analysis is very low. Often this makes the rapid identification of black or dark colours near impossible with this technique.


Analysis of Sorted Plastic Fractions In order to assess the efficiency of each plastic sort carried out on the TiTech unit, a sample of 15 or 20 plastic chips was taken from the input and output of each trial run. These chips were then individually identified using a laboratory Mid-Infra Red analyser (Thermo Nicolet model – 380). This instrument compares the MIR spectrum for each particle with a library of known materials, and gives a percentage-fit for the ‘unknown’ material spectrum to the one selected as the best fit. For this analysis a positive identification was taken when the percent-fit was above 80%. Mid-Infra Red has been seen to give better results on darker plastic colours. Table of Laboratory Plastic Identification

Trial/Run no

Target Polymer

Number of targeted

polymer in INPUT

Number of targeted polymer in OUTPUT

Number of targeted

polymer in REJECT

Separation efficiency

T1R1 ABS 8/15 13/20; 15/20 5/15 Low T1R2 PS 3/15 12/15 3/15 Good T1R3 PP/PE 1/15 14/15 --- Good T2R4 PS 3/15 ------- 1/10 ------ T2R5 PP/PE 1/15 11/15, 13/20 0/15 Fair T3R6 PS 4/15 11/15 0/15 Fair

The figure shown as a fraction, such as ‘8/15’, indicates the number of individual chips which were positively identified correctly as the ‘target’ material for each run. So for run number T1R1, where ABS was the target –pick, 8 out 15 chips in the input were analysed as ‘ABS’ and in the output fraction 13 and 15 out of 20 were positively identified in two sample sets. If the Titech unit had given 100% accurate sorting of ABS from the mixed input material, then one could expect to see 20/20 results for the analysis of the output material. The sorting of PS and then PP/PE on subsequent trial runs R2 and R3 gave better sorting performance with 12 and 14 out of 15 correctly picked. It must be noted that the off-line sampling test used did identify plastic types which were similar to the desired target material. For example, the two tests used to check the ABS target output, show that SAN, PPE/PS and PS were the plastics that had been erroneously picked on the Titech trial, as shown below. Each of the non-ABS plastics do contain a ‘styrene’ element in them:- Example result of off-line plastic ID tests for ABS

T1R1 Output Colour Test 1 Test 2ABS White 13 15

PPE/PS White 5 0 SAN White 2 2 PS White 0 3

TOTALS 20 20


Discussion IR Sorting Trial - Titech The two trial materials gave very different results, with poor sorting on the dark coloured plastic in the ‘mixed WEEE’ samples. The sorting of the VDU monitor casing material was encouraging, because there was a positive identification and sort of the required plastic types. In the analysis carried out, it is important to note that when the ‘best fit’ polymer was not ABS, the identified material was of a very similar type (e.g. SAN – styrene Acrylonitrile). Also ABS was often seen as the ‘next best fit’ in the library of polymer spectra, indicating similarity of some of the stored spectra. It appears that the Titech IS positively picking out ‘styrenic’ characteristics, in that it selects polymers of a similar ‘family’ of chemical types. It should also be noted that the samples used for this trial were described as nominal 40mm particle size, but on inspection it is clear that there is a range of smaller particle sizes in the sample and some significant variation in shape. This makes the task of identification and accurate ejection more difficult at the very high processing speeds required on the machine. In conclusion, it appears that the Titech unit would be best suited to doing a sort of similar polymers from a mixture. For example the task of separating ‘styrenics’ from ‘polyolefins’ might be the best approach to take with this equipment, before attempting more fine sorting based upon the exact polymer compound. It must also be emphasised that ONLY the trial with mostly light coloured plastics gave acceptable results on identification of polymer types. The IR sorting method did not perform well on dark colours, as seen in the mixed WEEE plastic trial. This means that the application of this technology to the wide mix of colours mostly found in E&E plastics will be difficult. IR sorting can not be used to sort on additive content, such as bromine level to indicate presence of flame retardants.


7.5 Scan and Sort, Hamburg Germany – X-Ray Transmission Sorting Claimed Features:

• Handles particle sizes between 10-40mm • Separation accuracy >99.5% • Lower power consumption than other similar technology

Operating Principle Materials are fed by a vibrating feeder onto a 60cm wide belt moving at 2m/s to give even spacing of individual pieces, with minimal particle overlap. The length of the belt (approx 5 metres) ensures that particles are stationary relative to the belt when they reach the detector. An X-Ray generator is mounted over the belt near the end roller and is focussed across the width of the belt. The X-Rays are generated electronically so there are no radioactive sources present and the radiation ceases when the power is switched off. The detector is mounted under the belt next to the end roller. Particles that trigger the detector send a signal to the control computer which activates one or more of a row of compressed air jets mounted just beyond the end of the belt. Firing of the air jet at the correct moment diverts the particle from the main stream of particles into a second collection bin.

Figure 7.11 X Ray transmission sorting

WEEE Pl

Conveyor

Sorting Chamber

astic Separation Technologies 116

T MCw Cw Pcr

W

Machine Specification Type: dual energy X-ray Transmission sorter Working Width – 600-1200mmResolution: 0.8mm Sensors: single or dual X-ray sensor/metal detection Sensitivity: high dynamic range High Speed evaluation: 15,000 particle/sec Size Range: 5-150mm Throughput: Up to 40 tonne/hr Sorting Criteria: Atomic Density Material Types: CRT Glass, Ceramics, PVC halogenated polymers frommixed plastic waste, etc

Figure 7.12 Side view of X-Tract Sorter showing vibratory feeder, conveyor and sorting chamber (Ack-Commodas GMbH)

rial with XRT Sorting Machine

aterials Used RT Casings shred with nominal particle size of approx. 10mm-30mm, but ith numerous shorter pieces. (Sample 1)

RT Casings shred with nominal particle size of approx 30mm-100mm, but ith numerous longer pieces. (Sample 2)

rior to the trial and before shredding, the samples of whole TV & VDU asings were separated according to their colour and brominated flame-etardant content, as follows:-

TV casings – Black – only NON-brominated FR pieces selected - NFR VDU casings – Beige – only brominated FR pieces selected - BFR


This was so that a rapid visual assessment could be made during the trial, as theoretically all the light colour pieces should have been separated as the reject fraction and the black in the target fraction, based upon bromine content. The whole casings were manually sorted using a sliding spark gun. The sliding spark gun is a quick tool to analyse the flame retardant content within a polymer sample. Once the initial sorting was complete, the TV casings were shredded and then mixed prior to the trial at Scan and Sort.

Figure 7.13 Sample 1 TV Casings between the sizes 10mm to 30mm

Figure 7.14 Sample 2 between 30-100mm


Trial Results The following table summarises the mass balance for the two trials:-

MaterialSort Type

Input Weight

(kg)

Weight Passed

(kg)

Weight Ejected

(kg)

% Passed - NFR

% Ejected +BFR

Sample 1 X-Ray 63.86 51.3 12.56 80% 20%

Sample 1 X-Ray 51.3 45.5 5.8 89% 11%

Sample 2 X-Ray 13.78 10.76 3.02 78% 22%

Sample 2 X-Ray 10.76 10.48 0.28 97% 3%

Sample 1 was used to separate polymers which did not contain any halogens using the size range 3 – 30mm particles. This was separated into a target (non-FR) and reject fraction with the reject being polymers containing BFR. The photos below show the colour split of materials after two passes through the X-Ray sorting unit.

Figure 7.15 Colour split in Sample 1

Target - non-FR fraction (dark) Reject – BFR fraction (light)

The colour split did not indicate a very clear separation based upon the detected bromine content, even after two passes through the sorting machine. It was thought that this may have been due to poor analysis of the whole TV cases based upon erroneous results from the sliding spark detector. It was therefore decided to check a sample of the sorted chips using a hand-held XRF device, to give a more accurate analysis. From a sample of 15 chips from the ‘target’ non-FR fraction tested with the XRF unit, 14 were below 0.1% Br level but one was found with 3.5% bromine content. For the two samples of ‘rejected’ material (i.e. detected as +BFR content), on the first and second passes, it was found that half of the chips had been erroneously rejected in both runs (i.e. 8 out of 14 chips with below 0.02%


bromine content). This would represent a high loss of yield of the desired non-bromine plastic reporting to the ‘reject’ +BFR fraction. Trial 2 Sample 2 in the size range 30mm-100mm was used to separate halogen containing polymers. The initial separation gave an 80:20 split again in the material flow.

Figure 7.16 Split in sample 2 Target – NFR fraction Reject – BFR fraction

Similar results were seen with the larger particle size material, in that mixed colours were found in the target bin. The NFR target fraction was analysed using the XRF unit to show zero particles in the 14 chip sample with any BFR content. All the target fraction test pieces had below 0.1% bromine content, which was a good result.

There was the same high level of non-BFR material passed to the reject fraction with around 50% of the ‘reject’ fraction being found to have negligible bromine content. Discussion X-Ray Results The machine gave a good result in terms of producing a ‘target fraction’ of non-bromine plastic particles, with only one piece found that had a significant bromine level at the smaller particle size. In the larger particles, all pieces sampled in the target fraction were bromine free. However, if the aim of the process is to produce a ‘bromine-free’ fraction, then it only needs a low level of wrongly sorted +BFR chips, each with a few percent bromine, to create an average bromine level across a homogenised sample that would be above required levels (e.g. 0.1% level under RoHS regulations). For example if one chip in 100 was found to have 10% bromine level, this would give an overall average of 0.1% bromine when homogenised. There was a fairly high level of loss of target material to the reject stream (circa 50% non-bromine in reject), which can be considered a high loss of yield of valuable bromine-free polymer.


In defence of these poor results, it is clear form the photographs of the samples that the quality of shredding process had been very poor, resulting in a wide spread of particle size in both samples. This type of equipment works much better when the particles to be sorted are of a regular shape with a tight size distribution. In conclusion, this is an effective method for sorting on the basis of additive content where there is a detectable element to act as a ‘marker’ for the desired additive type. In this case, ‘bromine’ is being used as a general indicator of any type of brominated flame retardant compound, of which there are many different types. This approach could also be used for other additive types, such as cadmium stabilisers used in PVC compounds. For effective sorting a double pass of the machine is recommended and also better control of particle size than that seen in this trial. 7.6 Equipment Suppliers

RTT Systemtechnik GmbH

Hirschfelder Ring 9 D-02763 Zittau

Phone +49 3583 - 540 84-0 Fax +49 3583 - 540 84 44

E-Mail [email protected]

www.unisort.com

Binder+Co AG, CRITERION

Grazer Straße 19-25, A-8200 Gleisdorf, Phone: +43 (0) 3112/800-0, Fax: +43 (0) 3112/800-300, E-mail: [email protected]

www.binder-co.com

MSS, Inc. 3738 Keystone Av.

Nashville, TN 37211 USA

Phone: 615.781.2669 Fax: 615.781.2923 www.magsep.com





http://www.unisort.com/


http://www.binder-co.com/

http://www.magsep.com/

8 Colour Sorting It is increasingly important in the plastic recycling industry to colour sort materials. This may be to:-

1. Enhance the product’s uniform appearance 2. Remove defects and foreign material 3. Add value to the finished product 4. Sort to a light colour, allowing addition of colour dyes in subsequent

extrusion compounding This section of the report covers some of the theory of colour sorting, explains how the technology works and describes the trials carried out on two different sorting machines.

Figure 8.1 Benefit of Colour Sorting on Mixed Plastic Input


8.1 Electromagnetic Spectrum The electromagnetic spectrum covers a wide range of wavelengths and photon energies. Light used to ‘see’ an object must have a wavelength about the same size as or smaller than an object. The visible range of light is only a small section of the full electromagnetic spectrum, as this diagram shows.

Figure 8.2 Electromagnetic spectrum

Electromagnetic energy of a particular wavelength λ has an associated frequency f and photon energy E. Thus, the electromagnetic spectrum may be expressed equally well in terms of any of these three quantities. They are related according to the equations:

Wave speed = frequency x wavelength or λ×= fc

Also or hfE =λhcE = where c is the speed of light and h is Planck’s

constant High frequency electromagnetic waves have a short wavelength and high energy whereas low frequency electromagnetic waves have a long wavelength and low energy. Colour Sorters are defined as using the visible light range as the basis for the sorting technology, however some manufacturers are also adding non-visible light cameras into sorting equipment as a means to provide more functionality within one machine. The use of infra-red light for sorting plastics has been described in a separate section of the report.


8.2 Visible Light

Only a narrow part of the electromagnetic spectrum corresponds to the visible wavelengths of light. Wavelengths between 400 – 750nm cover the range of colours humans recognise as the ‘rainbow’. When mixed together these colours form ‘white light’ as seen in sunlight or from man-made light sources, such as electric lamps.

Figure 8.3 Visible spectrum

White l 8.3 Di

Generause ca Monocwaveleto the econtainsine wa

WEEE Pla

ight can be separated into its spectral colours by dispersion in a prism.

fferent Types of Light

lly speaking, optical particle sorters have light sensing cameras which n use either Monochromatic or Bichromatic visible wavelengths.

hromatic light is said to be one dimensional and therefore has one ngth. It is described by only one frequency. Monochromatic light looks ye as a pure colour, and can never be white or magenta. Since it s only one frequency, monochromatic light can be represented as a ve:-

Figure 8.4 Monochramatic light

stic Separation Technologies 124

The height of the sine is the amplitude or how bright the light is. The width of one period is the wavelength of the light, and is inversely related to the frequency: since the light travels at 3 x 108 m/s. This type of light is used in a black and white camera. Bichromatic light is said to have more than one wavelength. At different wavelengths, different frequencies will be generated. The actual term means ‘two colours’. The wavelength of Bichromatic light can be represented by the diagram below which shows the changes in wavelength observed. The two wavelengths represent two different colours of light.

Figure 8.5 Bichromatic light

8.4 CCD Cameras for Sensing Light The key component of all colour sorting devices is the light sensing camera (or cameras). These devices typically contain one or more light sensing devices known as CCDs :- A charge-coupled device (CCD) is an image sensor, consisting of an integrated circuit containing an array of linked or coupled, light sensitive capacitors. This device is also known as a Color – Capture Device

Figure 8.6 Charge coupled device


The photoactive region of the CCD is generally a micro-layer of silicon. It is impregnated with Boron and covered with a layer of phosphorus to produce a microscopic array of light sensitive devices. Further surface treatments are used to produce an array of Red/Green/Blue filters at the pixel level. The diagram below gives a representation of how the light active surface of the CCD is arranged and covered with the filter layer, showing two different patterns of the filter array – ‘Bayer’ and RGBE.

Figure 8.8 Filters on a CCD

A Bayer filter on a CCD An RGBE filter on a CCD

Digital color cameras generally use a Bayer ‘mask’ over the CCD. Each square of four pixels has one filtered red, one blue and two green. The result is that luminance information is collected at every pixel, but the color resolution is lower than the luminance resolution. Better color separation can be reached by using three CCD devices and a beam splitter prism that splits the image into red, green and blue components. Each of the three CCD’s is then tuned to respond to a particular color.

8.5 Available Options in Colour Sensing Devices Given a basic understanding of the different types of light, an appreciation of the operations of CCDs in sorting cameras and the ability to split the light using prisms; one can see that the designer of a colour-sorting machine has a range of options at his disposal in terms of the level of sophistication that the sorting device will use to bring about the desired colour separation. In addition to visible light, some machines have also been designed to include infra-red light sensors within the CCD array, adding another level of sorting capability in the one machine. The following figure summarises the different design options that can be used in modern colour sorting devices:-


Figure 8.9 Design options

The degree of ‘sorting power’ offered by any device is then directly related to the choice of light sensing method used. The information collected from the camera can then be analysed with greater sophistication in up to three dimensions, as represented by the following figure:-

Figure 8.10 Analysis of information collected by camera

From the above, it can be seen that colour sorting using ‘Monochromatic’ light sensors limits the capability of the sorting to the darkness of particles in relation to each other or to the background being used. This can be thought of as analogous to watching a game of snooker on a black-and-white TV set, where the different ball-colours appear with similar darkness of shade and can even appear to fade into the background ‘green’ felt.

When ‘Bichromatic’ or combined-power cameras are employed in the sorting device, a whole new dimension of analysis opens up, and tasks can be carried out with much greater selectivity. In this instance it would be possible to set the


colour sorter to remove (say) blue particles from a mass of red particles, even if both colours appear to have the same darkness of shade when viewed in B&W. 8.6 How Colour Sorting Machines Work

The main elements of a typical colour sorting machine are shown in the next figure. This shows how the light information collected from the CCD device is used by a microprocessor to drive an air-ejection system.

Figure 8.11 Diagram of the main elements in a typical colour sorting

machine

It can be seen above how the cameras analyse the passing flakes and detect which ones produce a colour-signal which deviates from the pre-set threshold value. If a signal greater than the upper or lower limit is ‘seen’, a pulse will be sent to the air ejector from the computer software blowing the unwanted flake from its normal trajectory.

The computer analysing software is one of the critical elements in the colour sorting system. It operates at a very high frequency, enabling the cameras to scan the falling cascade of particle to be sorted at a rate which essentially ‘freezes’ the movement of each particle against the background colour of the falling chute. In this snap-shot of time, each particle is assessed against the pre-set threshold values, which are used to control the level of sort, sometimes on several dimensions at once (e.g. darkness, RGB, and IR level). The software then controls the eject mechanism to remove unwanted material. Microprocessor speed is therefore vitally important in terms of enabling the vast number of calculations and signals which need to be generated at the processing speed demanded by evaluation of every single particle in a falling cascade of material.


It is often possible to store different settings to fine-tune the operation of the colour sorter to each particular mix of particles being separated and the particular type of sorting task required (e.g. to remove blue particles from clear flakes).


8.7 Colour Sorter Equipment Trials 8.7.1 Sortex, London UK – Optical Flake Sorter Features of Equipment:

• Adding value to all types of grains, seeds and particles • Quality assurance by removal of foreign materials such as colour

variations, spot defects, mud stained products and glass with plastic • Maximum ejection accuracy for higher yields • 0.5 to 35 tonnes per hour processing capacity • Reliable and easy to use with low running costs • Clean and safe operation

Figure 8.12 Typical example of Sortex equipment

Machine Specifications Capacity: 0.5 - 35te/hr Weight: 600 – 1,000kg Power: 1.5 – 3.0kW Air Requirements: 8.6 to 35l/sec at 4.6 bar Width: 800 – 2000mm Depth: 2,200mm Height: 2,000mm Material Types: rice, grain, seeds, coffee, nuts, plastics and other foods


Working Principle The mixed polymer is fed via the input hopper. A vibratory feeder is used to transport an even layer of particles onto the top of a steeply angled slide-chute. The material then falls down the surface of the chute where the cameras differentiate between the reject and acceptable material. The ejectors are activated when the signal for an unwanted flake is received. The ejector’s job is to deflect the bad material to the reject stream allowing the good material to pass into the acceptable stream.

Figure 8.13 Diagram of Working Principle and 3-D View of Sortex Machine

Operating Settings The machine can be arranged to give one high purity stream (e.g. the white particles), with minimal black ‘contamination’. In order to do this, it is set to eject several falling particles around the point where a single ‘dark’ item is detected. This means that some white particles will also be diverted into the reject stream on the first pass, in order to make sure that no black particles are missed by the ejector air-pulse. This first-pass reject stream can then be re-sorted in a second pass over one section of the detection chute, where a higher sensitivity will be used at a lower flow-rate to recover lost yield of white material In this manner the unit can be fine-tuned to give a high throughput on the first pass, using two-thirds of the available width of slide-chutes and detectors. The second pass function can then be used to increase the yield of desired product, utilising one-third of the feed-chute width. The photograph below shows a demonstration unit used in the trials. It has only a single pass slide-chute, so in effect, it is one third of a full-scale sorting unit.


Figure 8.14 Demonstration unit

Inlet User Interface Target

Reject

Camera Positions

Trial with Sortex Optical Flake Sorter Material Three types of WEEE material were processed through the Sortex machine.

• Post consumer mixed WEEE plastic • Plastic from coloured telephones • IT mixed plastic

Photographs below show the IT equipment plastic and mixed WEEE:-

Figure 8.15 Particle size range of 3-6mm was used for the trial

Summary of Results from Trial With each of the different materials, the optical flake sorter settings have to be adjusted and tuned to gain optimal results. Several runs were made at various settings, to determine the best conditions for each type of material.


Mixed Small-WEEE Plastic With the post-consumer small-WEEE plastic, the aim was to achieve a black fraction from the mixed colours. It took two passes through the machine to obtain an acceptable result, due to the wide range of shades and colours in the original mix.

Figure 8.16 Result of first pass

Figure 8.17 Result of second pass – small-mixed WEEE plastic

The results were assessed visually and indicate a purity level of >95% black particles being sorted from the mixed input stream.


Telephone Casings The plastic material from telephone casings had a high percentage of white / cream coloured chips mixed with reds, greys and other colours. For this trial the aim was to extract a light colour of white/cream particles with nil dark contaminants. The photograph below shows the results of a single pass through the single-chute unit. This was carried out at 1.9 tonnes/hour throughput rate (which would enable a 3 chute machine to deliver nearly 6 tonnes/hour).

Figure 8.18 Result of a single pass through the single chute unit

In numerical terms the input material was measured to have 30% of contamination. The ‘Accept’ stream was 52% of the input volume flow, with 48% being sent to ‘Reject’ on the first pass. The low level of dark particles in the ‘Accept’ product was easily removed in a second pass through the sorter unit. This second pass gave only a 2% level of reject flow to remove the estimate 1% level of dark contamination. In a production environment, this second pass would require additional machine capacity to be added as it involves re-processing the major flow of the ‘accept’ material from first-pass.


I.T. Plastic Sorting For the IT plastic material, it was estimated that the contamination level of dark particles in the light material was 26%. The machine made a very clean stream of ‘accept’ – light product on a single pass, with a 50/50 split between accept and reject streams. Throughput was again 1.9 tonnes/hr.

Figure 8.19 Result of sort of IT plastic

This trial indicates the compromise that has to found in terms of setting the colour sorter machine. If the operator requires 100% purity of the light fraction, then the machine will have to be set up to high-sensitivity which causes a larger proportion of the light product to be reject with each black/dark particle. However, if total purity is not required, then the sensitivity can be dropped to generate a higher yield of the accept material stream.


8.7.2 ASM, Bologna Italy – Optical Flake Sorter Features of equipment:

• Added value from pure substances obtained • Maximum ejection accuracy for higher yields • Reilable and easy to use with low running costs • Clean and safe operation • 1024 pixel cameras adding to the sensitivity of the system • Scan and separate defective products with differences up to 0.1mm in

size • Flake sorter is controlled by 32 microprocessors which are controlled

by an additional super microprocessor monitoring the environmental conditions of the machines working area

Figure 8.20 Example unit

Machine Specification Capacity: 1.0 - 35te/hr Weight: 800 – 1,200kg Power: 2.5 kW Air Requirements: 8.6 to 35l/sec at 3.0 bar Width: 800 – 2000mm Depth: 2,200mm Height: 2,000mm Material Types: rice, grain, seeds, Post Consumer WEEE


Working Principle The electronic colour sorting machine separates materials by their colour differences. The product is fed by a dosing vibrator to an inclined slide that conveys it under two vision systems (two digital CCD cameras per channel, front and rear). The vision system monitors the cascade of material on a particle by particle basis to sense any colours which fall outside of the pre-set threshold limits. The defect is removed from the falling curtain of particles by an means of a compressed air jet. This mode of operation is very similar to the Sortex equipment

Figure 8.21 Internals of colour sorter

Inlet to the colour sorter

Target Fraction

Reject Fraction

User Interface

Trial with ASMs’ colour sorter

Material A variety of WEEE material was processed through the ASM machine. The post consumer WEEE contained polymers of differing colours. The material sent to the trial was a mixture of IT and post consumer WEEE and two sizes were used for the colour sorting trial. For the purposes of the trial, it was assumed that the white material was the fraction which needed to be purified, with the black/dark material being treated as the ‘reject’ product.


Figure 8.22 Typical example of the infeed particles

It can be seen above how the white and black flakes were distributed evenly through the sample. Summary of Results from Trial The colour sorter processed both sizes of material. The objective was to liberate the white flakes from the dark coloured ones. The pictures below indicate what the target and reject fractions look like. The target fraction from one of the optimal runs is shown below. It can be seen that there is negligible amounts of black reject flakes present in the target fraction.

Figure 8.23Target fraction

The reject fraction from one of the optimal runs is shown below. It can be seen that this stream is predominantly black with some white flakes present.


Figure 8.24 Reject fraction

Analysis of A hand sort pfraction consfrom a single As with all cochoice betwedepend uponthroughput. 8.8 Compa Axion performand an ASM

Similar typesnot taken frodefinitive com Both machinconfigured ineach trial is ddifferent sen Both units deand mix of pmachines wilost-yield of tthat in order operation, thmaterial han

WEEE Plastic Sep

the Target Fraction

erformed on the target fraction showed that 98% of the target isted of white flakes whereas 2% were black flakes. This was pass through the machine.

lour sorting machinery, the operator is constantly faced with a en maximum product yield and the purity of that product. This will the individual settings for sensitivity used at the chosen volume

rison of Trial Equipment

ed separation trials on both the Sortex colour sorter in the UK colour sorter in Italy.

of WEEE plastic were used for each set of trials, but these were m exactly the same sample set. It is therefore difficult to make a parison between the two colour sorters based upon these trials.

es make use of similar technology for the vision cameras and are near identical layouts. The quality of the separation achieved in etermined by the skill of the machine operator and the range of

sitivity settings that can be programmed into the electronics.

monstrated very good colour sorting performance using the size lastics sent for the trials. There is an option to configure the th a partial re-sort of the rejected fraction as a way to recovery arget product from the reject stream. However it should be noted to make practical use of this ‘re-sort’ capability in a continuous ere is a need to design and build a more complex arrangement of dling systems around the machine.

aration Technologies 139

Colour sorting equipment operates at very high processing speeds and relies upon accurate sensing of defect by the camera arrays. Most particulate materials involve a level of dust when handled at high volume, it is therefore important to consider the regular cleaning of the machines to give reliable operation in industrial applications. Equally important, is the provision of a constant pressure supply for the compressed air, which must be cleaned and filtered to ensure reliable operation of the ejector systems. Equipment selection should be determined by demonstration of a large scale trial on the material in question, with a well-defined criteria for the required purity of output product at the specified throughput. Assuming that near equivalent colour sorting performance can be achieved between different supplier’s machinery, then selection of the right machine will probably be dependent upon capital cost and after sales service capability. 8.9 Equipment Suppliers

S+S Separation and Sorting Technology GmbH Regener Strasse 130

D-94513 Schönberg - Germany Tel.: +49 (0) 8554 / 308-0

SEA srl Colour / IR / UV sorters

Via Ercolani, 30 - 40026 Imola BO Italy – Tel: +39 0542 361423

Fax: +39 0542 643567 – Email: [email protected]

Satake Corporation UK Division Horsfield Way,Bredbury Industrial Park, Bredbury, Stockport, SK6 2FG

England Tel: +44(0) 161 406 3800 Fax: +44(0) 161 406 3801

E-mail: [email protected]




http://maps.google.it/maps?f=q&hl=it&q=via+ercolani+30+imola&om=1



9.0 Electrostatic Separation Techniques In this section of the report, we focus upon the use of electrostatic charging of plastic particles as a means to bring about separation effects. This technology can be used in two ways as part of an overall process for WEEE plastics:-

1. As a method to separate different polymer types based upon their differential charging properties

2. To remove contaminant materials such as wood or metals based upon differences in conductive charging of the particles.

An overview of the theory of electrostatics is given below followed by a description of the machine operating principles for equipment that uses this method of separation. 9.1 Theory of Electrostatic Separation Electrostatic separation, also known as ‘high-tension’ separation, is a method of separating based on the differential attraction or repulsion of charged particles under the influence of an electrical field. Applying an electrostatic charge to the particles is a necessary step before particle separation can be accomplished. Various techniques can be used for charging. These include contact electrification, conductive induction and ion bombardment.

Figure 9.1 Typical electrostatic separators

9.2 Particle Charging Limitations Regardless of the method of charging, the amount of charge that can be accumulated on a particle is limited by the maximum achievable charge density and the surface area of the particle. The forces that bring about particle separation by electrostatics are thus limited by how much charge can be ‘packed’ onto the surface of each particle. These forces must be able to overcome the forces of gravity and inertia which usually govern the movement of particles in, for example, a falling stream of plastic granules. Surface area increases in relation to the square of linear dimensions, whereas volume (and hence mass) increases in a cube function to linear size. Therefore, as particle size in increased the forces that can be


created by electrostatics are limited to the increase in surface area, whereas the forces of gravity and inertia increase much more rapidly in proportion to the mass of the particle. In practical terms this means that electrostatic separation is effectively limited to small particle dimensions in the order of 5 – 8 mm and below. Above this size, the forces created due to surface charging will have to compete with stronger forces of gravity and inertia related to the large mass of the particles. This rule-of-thumb may vary when dealing with thin flakes of material or longer filament shaped material, in this case up to 25mm size may be acceptable. For plastic separation to occur efficiently, it is therefore best to ensure that particle size has been accurately controlled in any upstream size reduction process. Variation in surface charging of the particles will be easier to differentiate when the mass and shape of each particle is held within a tight distribution. 9.3 Charging Mechanisms There are three different charging mechanisms that can be used to create electrostatic forces on the surfaces of particulate materials:-

1. Contact Electrification or Tribo-electrification 2. Conductive Induction 3. Ion Bombardment

9.3.1 Contact Electrification When two dissimilar materials touch each other, there is an opportunity for the transfer of electric charges. The extent of charge transfer can create a significant surface charge of opposite sign when the materials are then separated. High temperature and low humidity favour the development of high surface charges through the mechanism of contact electrification. Rubbing the materials together to increase the area of effective contact can also lead to higher surface charges. This effect can be demonstrated by rubbing a rubber balloon onto a woollen pullover, when the build up of charge on the balloon can be used to make it stick to a wall. Particles carrying charges of opposite polarity due to contact electrification will be attracted to opposite electrodes when passing through an electric field and thus can be separated from each other.

The degree to which different materials will become oppositely charged is governed by their position in the triboelectric series. This ranks each material based upon its tendency to take on positive, neutral or negative static charge when brought into contact with another material in the series. The series can be used to quickly evaluate which materials will display a strong charging effect when brought into contact with each other, and thus to make a judgement about their propensity to separate using electrostatic methods. Often the behaviour of materials such as rabbit fur, leather and amber rods


are used as examples of different triboelectric charging behaviour, but a more useful table for the behaviour of plastics is given below:-

Triboelectric series

Positive charge (+) PA PC

PMMA ABS

METAL (Neutral) PBT PP

PVC PVDF

PE Negative charge (-)

Example triboelectric series – (Ack. G Hearn, Wolfson, Southampton University)

Applying this table to the use of contact charging as a method of plastic separation, it could be assumed that separation of Nylon (PA) from polyethylene (PE) is readily achievable, as one displays a tendency to become strongly positively charged and the other to become negatively charged. Separation of PMMA (‘Perspex’) from polycarbonate (PC) would appear to be more difficult on this basis. A further effect of triboelectric charging that needs to be considered in separation equipment is the effect of ‘wall charging’ that occurs in the chamber used to generate the particle charging. Often a vibrating chamber is used to shake the particles against each other and the walls of the vessel, in order to create many opportunities for the build-up of static charge by contact electrification. In this case the interaction between the wall material and the different materials of a mixed particle stream also affects the degree and polarity of individual particle charges.

Figure 9.2 Wall charging effect


9.3.2 Conductive Induction This describes the process by which an initially uncharged particle that comes into contact with a charged surface assumes the polarity and, eventually, the potential of the surface. A particle that is an electrical conductor will assume the polarity and potential of the charged surface very rapidly. However, a non-conducting particle will become polarized so that the side of the particle away from the charged surface develops the same polarity as the surface. If a conductor particle and a non-conductor particle are then separated from contact with the charged plate, the conductor particle will be repelled by the charged plate and the non-conducting particle will be neither repelled nor attracted by it. The particles which have been charged by this conductive method can therefore be separated by passing them through a strong electric field immediately on leaving the charging surface.

Figure 9.3 Mechanism of conductive induction

9.3.3 Ion Bombardment The most positive and strongest method of charging particles for electrostatic separation is ion bombardment. The use of ion bombardment may be visualized by considering conductor and non conductor particles touching a grounded conducting surface (per diagram below). Both particles are bombarded by ions of atmospheric gases generated by an electrical corona discharge from a high-voltage electrode (usually a fine tungsten-alloy wire at 20 to 30kV with respect to ground and several centimetres away from the particles). When ion bombardment ceases, the conductor particle loses its acquired charge to ground very rapidly and experiences an opposite electrostatic force tending to repel it from the conducting surface. The non-conducting particle, however, being coated on its side away from the conducting surface with ions of charge opposite in electrical polarity to that of the surface, experiences an electrostatic force tending to hold it to the surface. If the electrostatic force is larger than the force of gravity or other forces trying to separate the non-conducting particle from the surface, then the particle is held in contact with the surface and is said to be ‘pinned’.


Figure 9.4 Representation of ion bombardment

Ions from corona discharge

9.4 Separating Forces In all electrostatic separators the charging zone of the equipment is rapidly followed by exposure to a strong electric field, where the accumulated charge on each particle surface will influence it motion as it passes through the ‘sorting zone’. It is important that there is minimal delay in transport of the charged material into the sorting field, because the static charges will begin to dissipate immediately on leaving the charging area. The rate of this discharge depends upon the conductivity of the material itself, the moisture content of the atmosphere and the surrounding levels of dust. In the separating electric field the force on each particle is governed by the equation:- F = Q x E Where F is the vector sum of the forces; Q is the total electric charge and E is the intensity of the electric field. The magnitude of Q on each particle will have been determined by the pre-charging process, and the size of ‘E’ the electric field intensity, is largely dependent upon the voltage of the applied electrical potential across the electrode plates. This can be increased by using a small gap between electrodes at very high voltage, but is limited by the effect of arcing across an air-gap if the electrodes are held too close. Exposure to the electric field usually takes place in an area of ‘free-fall’, where the effect of the small differential forces on each particle can be maximized as they fall towards a splitting or diverter plate.


9.5 Types of electrostatic separators for plastic mixtures:- With this brief explanation of the operating mechanisms employed in electrostatic separators, it is easier to appreciate the two main types of machine used for commercial separation of particulates, including plastic/plastic mixtures and plastic/metal or other material mixtures. 9.5.1 Triboelectric Separators – This type of machine uses contact electrification in a separate tribo-electric charging chamber to generate the maximum possible charge on each particle in the mixture. The charged particle mix is then fed directly into separating zone of high intensity electric field where the differences in charging on the material types in the mix, is used to create opposing forces on the particles. This usually leads to a three way split of the infeed material into:-

• Positively charged • Negatively charged • Neutral or equal +/- charging.

An example of this type of separation method is shown in the figure below, where a tubular charging chamber is used to create multiple contacts between the walls and the particles and also inter-particle contacts.

Figure 9.5 Triboelectric separator

Tribo-charged particles

The charged particles then enter the sorting zone, between two electrode plates and fall towards the outlet where the three zones of charged material can be separately collected into bins via adjustable diverter plates. 9.5.2 Corona Discharge Electrostatic Separators


In this design of separator the principle of ion bombardment is employed to create the charging effect on the mixture of particles. Typically a steady, mono-layer feed of the plastic or metal/plastic mixture is transported onto the rotating surface of an electrically earthed rotating drum.

Figure 9.6 Corona discharge electrostatic separators

At the top of the drum the particles pass under a strong electrical discharge field (corona discharge) where the effect of ion bombardment occurs. This leads to a difference in the behaviour of the particles on the basis of their conductivity. As they progress around the drum, the conductors become oppositely charged to the drum surface and are repelled away past a splitter plate. The non-conductor particles form a dipole and are attracted to the drum surface with sufficient force to be held there until a brush or scraper is used to remove them.


9.6 Summary Description of Practical Trials Trials were conducted on different designs of electrostatic separator as follows:-

1. Plas-Sep, Canada – Triboelectric Charger 2. Hamos, Germany –

a. KWS – Corona Discharge Separator b. EKS – Triboelectric Unit

The materials used in the trial were:-

• Mixture of post consumer WEEE polyolefins with wood / rubber contamination

• Mixture of CRT plastics – ABS / PS mixture. • PS from fridges with low level of rubber contamination

A brief description of the trials and results is given below.

9.6.1 Plas-Sep, London, Ontario, Canada – Electrostatic Separation Trial Claimed Features:

• Provides 99%+ separation of impurities in a single pass • Can meet any required product purity using multiple passes • Is most effective on binary mixtures, is also effective on multi-resin

mixtures • 2000 lb per hour input - usable output up to 85% of input depending on

purity requirements • Low cost operation enables profitable separation of commodity plastics

such as polypropylene (PP) and polyvinyl chloride (PVC) • Robust, repeatable, controllable process tailored to high volume

automated or semi-automated operation

Operating Principle and Description of Trial The Plas-Sep machine works on the tribo-electric charging method as described above. A simple flow diagram is shown below which indicates the key process variables. The sample sent for processing was wet, so it was dried and then separated in a laboratory scale electrostatic unit. In the separator, the material was charged while passing through a slow rotating drum. It was then allowed to fall through a very strong electric field in the separation tower. The material was collected at the bottom of the tower in 6 equally sized bins. The bins were numbered 1 to 6, with bin 1 closest to the positive electrode, therefore collecting the most negatively charged material. This produced 6 output fractions obtained from the separator.


Figure 9.7 Input/output flow

Figure 9.8 Full scale Plas-Sep unit

In-Feed drum for charging

Separation Unit with electrode plates

Outlet Product splitter plates

Figure 9.9 Block diagram of input/output flow


TTvdpo

W

Machine Specification Input Feed Rate: Up to 1000 kilo per hour Granular Size: 5 - 8mm max.Input Material Appearance: Must be Dry, free from dust. Air Requirement: 140l/min @ 100 PSI Power Req. – 1 KW separation unit, plus est. 4KW material conveying. Unit Footprint: 2m by 4m Ceiling Height: 5m Material Types: post-consumer material, fibres and films, ground bottles,recovery of PET flakes, cable insulations, carpet fibres and general WEEE

rial Procedure with Plas-sep Electrostatic Separator he material received by Plas-Sep was sampled and separated by hand into isually distinct portions. The polymer composition of each portion was etermined and the total composition determined by calculation from these ortions. The particles were sized between 2 and 10mm and a total of 1,230g f material was processed. The compositions are detailed in the table below;

Polymer Weight

(%) Weight

(g) PP 83.7 1030

PC/ABS 2.6 32 ABS 2.7 33 PS 2.1 26 PE 0.1 1

PVC 0.1 1 Metal 0.08 1 Wood 6.8 84

Rubber/foam 1.8 22

Distribution of elements after hand sort

PPPC/ABSABSPSPEPVCMetalWoodRubber/foam

Distribution of materials after hand sort


It was noted that the material had to be dried BEFORE the electrostatic charging could take place, which is a constraint for some plants. Summary of Results from Trial Visually, the wood and other contaminants were seen to be concentrated in bins 4-6 close to the negative electrode. Percentages of the total material deposited in each bin are summarised below;

Bin Weight

(%) Weight

(g) 1, positive

side 10.6 130 2 26.8 330 3 33.7 415 4 13.4 165 5 9.8 120

6, negative side 5.7 70 Total 100 1,230

By analysis, the PP concentrations in bins 1-3 were found to be 99.7%, 99.9% and 95.5% respectively. If the three bins are combined, the concentration of PP would be 97.8% with the main impurity being PS. Initially 1,030g of PP was present in the feed and 855g of it was recovered leading to a recovery rate of 83%, if pure PP is treated as the ‘target fraction’. In comparison to the total weight of material processed, 70% of the infeed was recovered as cleaned PP material. In summary the machine recovered 83% of the desired ‘target product’ of PP at a purity of 97.8% on a single pass through the unit, which is very good. The pie chart below shows the distribution of elements when the first three bins were combined.

Hand Sort of Target Fraction

PPPS


Trial with Rubber removal from PS Fridge plastic Plas-Sep also conducted a test to remove trace levels of rubber from fridge plastic – polystyrene material. A sample of 4.2 kilos of PS plastic with a measured rubber content of 1.6% was processed through the laboratory scale unit. Most of the rubber particles were found in the middle 3 bins (Nos 3,4& 5) which indicates that there was neutral charging of the rubber and therefore little deflection from the free-fall path in the electric field. However it was noted that if the PS collected in bins 1 + 2 was taken as the cleaned product (1.72 kilos of material) then this would only have a rubber content of 210ppm. However this was only at a yield of 40% recovered material, and the remaining 60% of the mass would have to be reprocessed to extract further yield. Conclusion – Based upon this trial the Plas-Sep technology offers an interesting technology to deliver a plastic sorting performance that is difficult to achieve with many other types of separation equipment. The separation of wood particles to deliver a cleaned fraction of PP at a high yield was particularly impressive. This method should be considered by anybody with a need to remove wood contamination from mixed WEEE plastics. The process, being based upon electrostatics, is susceptible to changes in atmospheric humidity and is also sensitive to high level of dust in the feed material. Both of these conditions often apply to shredded WEEE plastics.


9.6.2 HAMOS GmbH GERMANY – Electrostatic Separation Trial

Cl

Hamos offerwas carried technology dpolyolefin anunit which usseparation.

9.6.2.1 Ham Operating PThe materialfeeder and thelectrostaticacharge very conductors ldrum and arpossible.

WEEE Plastic Sep

aimed Features:

• Dry separation process • High product purities and increased metal yield • High separating efficiency • Easy to Operate • Low energy consumption • Nil emissions

TWO designs of electrostatic separating machine. The first trial out on a KWS machine that uses the corona discharge escribed above. In this trial the infeed material was a mixture of d wood material, plus other plastics. A further trial used the EKS es the principle of tribo-electric charging for plastic/plastic

os KWS machine

rinciple to be separated is fed to a rotating metal drum by a vibration en transported to the area of a corona electrode. There it is lly charged in a high voltage field. Conductive products lose their

quickly and are thrown off the rotary drum. In contrast, the non-ose their charge very slowly. They stick to the surface of the metal e brushed off. Thus the dry separation of both fractions is

Figure 9.10 Diagram of Operating Principle - KWS

Feed Hopper

Corona discharge electrode

Earthed Drum

Brush

Non-conducting material

Conducting material

aration Technologies 153

TT

W Fmwp

W

Machine Specification – Hamos KWS Input Feed Rate: 200kg - 2 tonne per hour Drum Length: 1000 - 2500mmInput Material Appearance: Dry Material Types: post-consumer material, fibres and films, ground bottles,recovery of PET flakes, cable insulations, carpet fibres and general WEEE

rial with Hamos’ Electrostatic Separator he materials used in the trial are as below;

Figure 9.11 Material for trial

EEE stream containing PP contaminated with wood and other materials

rom analysis of the WEEE stream by hand sort, there was 30,605g of total aterial of which 25,620g was PP. This accounted for 83.7% of the feed hereas the remaining 14.3% of the material was a range of different olymers. These are summarised in the table below;

Polymer Weight

(%) Weight

(g) PP 83.7 25,620

PC/ABS 2.6 796 ABS 2.7 826 PS 2.1 643 PE 0.1 31

PVC 0.1 31 Metal 0.08 24 Wood 6.8 2,081

Rubber/foam 1.8 551


Hand Sort of Input Material

PPPC/ABSABSPSPEPVCMetalWoodRubber/foam

Summary of Results from Trial Referring to the table below, 30,605g of feed material was processed through the first stage of the KWS machine. 1,875g of wood rich ‘reject’ was removed to the conductive fraction accounting for 6% of the total input. 2,240g of conductive material was separated leaving 26,490g of mixed material which equates to 87% of the total feed to the 1st stage. The middle fraction was processed again through the KWS to separate a further 1,285g of wood (5%). A further 11% of the total throughput was removed as non-conductive plastic, leaving 22,360g of middling material from the 2nd stage separation run. From 30,605g of total material, 3,160g of wood was separated leaving 90% of the remaining material as plastic. This means that 10% of the total feed was separated as wood-rich, leaving the remaining 90% as a mixed plastic fraction, which still then required further separation. On the basis of these results, the KWS unit appears to offer a potential method to reduce wood particle contamination in plastic mixtures.

Feed

Material 1st

step2nd

step

Total (g) 30,605 30,605 26,490 conductive fraction (%) – wood rich ‘reject’ 10 6 5 non conductive fraction (%) – plastic 17 7 11 middling fraction (%) – mixed - plastic 73 87 84

It must be pointed out that the KWS machine is primarily designed for the removal of fine metal particles from plastic mixtures, for example in the


removal of copper wire filaments from PVC / PP cable reclaim. The application of this technology to wood removal was an attempt at a novel separation outside of the normal design conditions. 9.6.2.2 Trials Using Hamos EKS – Triboelectric sorting machine Two different sets of trials were conducted to investigate plastic/plastic separation using the EKS unit. In both cases the trial focussed upon a ‘difficult’ polymer separation often found in the WEEE waste stream; that is the sorting of ABS from PS plastic. These are probably the two most common plastics found in the WEEE plastic waste stream, and they are not easily separated by other methods. Operating Principle – Hamos EKS The plastic mixture is transferred from a storage silo via a vibrating feeder to the tribo-electric charging chamber. Contact electrostatic charging is caused by rigorous shaking and inter-particle impact inside the metal chamber. After charging, the particles are spread onto a moving belt and then fall through a high voltage electrical field, which provides the separation force. Two adjustable splitter plates enable some operator control of how the material splits into three fractions – ‘belt’, ‘middling’ and ‘electrode’, these are differently charged particles depending upon the electrical polarity selected for the high voltage charge used to generate the electrical field.

Figure 9.12 Diagram of operating principle

To achieve optimal separation results, it is recommended that the plastic mixtures to be separated must be constituted as follows:

• Mixture of non-conductors • Ideally two-component mixture • Different "electrostatic" charging behaviour (refer to triboelectric scale) • Dry material • Completely liberated material (no connection materials) • Dust-free material • Particle size - between 2 and 10 mm


Figure 9.13 Layout of the EKS internals

Vibrating Feeder

Tribo-electric Charging Chamber

High Voltage Separation Unit

Splitter – 3 different fractions

Summary of Trials and Results 1 – PS / ABS from WEEE The first trial was carried out with a mixed WEEE polymer stream which was primarily PS mixed with ABS and other plastics. This was measured as 90% polystyrene material (see photograph below).

Figure 9.14 90% Polystyrene material

The EKS machine produces 3 streams, which in this case should yield material as follows:- Electrode Fraction PS Middlings Fraction still mixed Belt Fraction ABS. Due to the large volume of the infeed that reported to the middle fraction (47% of the input mass), it was necessary to make a second pass through the machine with this material. The results of the two passes through the machine are summarised in this table, below:-


Stream Weight (g) PS Purity

(%) PS content

(g) ABS content

(g) Input 1st Stage 23100 91 21021 2079

Fraction A electrode fraction (PS) 7700 99 7623 77

Fraction B middle fraction 10900 76 8284 2616 Fraction C belt fraction

(ABS) 4500 2 90 4410

Input 2nd Stage(from B). 10625 76 8075 2550 Fraction D electrode

fraction 5860 98 5743 117 Fraction E middle fraction 2555 64 1635 920

Fraction F belt fraction 2210 15 332 1879 It can be seen that on both passes through the EKS machine the electrode fraction was clean PS material at 98 – 99% purity. In total a yield of 58% of the input mass was collected from the two passes. The ‘belt fraction’ of the first run was 98% pure ABS, at a 19% yield of input mass, and from the second pass a further 8% was collected but at a lower purity of 85% ABS. These products are of a purity that may make viable polymers of a saleable quality when further processed into recycled resin. However there is still a large proportion of mixed middle fraction, which does not yield a single polymer type. 2 – TV case HIPS + ABS ex IT casings Further separation work was done using a mixture of black HIPS collected form TV casings that was mixed with beige / light coloured ABS from VDU monitors in a 40/60 ratio. This was to enable a visual estimate to be made of the degree of separation achieved. (see photo below).

Figure 9.15 Infeed material HIPS/ABS mix


Figure 9.16 Visual split:– electrode- dark / middle-light

This material gave the following mass split on a single pass:- Mass Split % Mass % Dark % Light Electrode – Dark (HIPS)

3343 28 86 14

Middling Light (ABS)

8064 69 20 80

Belt - mixed 325 3 This simple analysis confirms that the EKS was able to generate an enrichment of the mixture for both polymers to give one stream that was 86% HIPS dark coloured and one stream 80% ABS light colour. It was not possible to improve the purity significantly with a second pass through the machine. The tribo-electrical method did not produce two streams with large, opposite polarity charging on this mixture of HIPS and ABS. This is probably due to the similar chemical structure of the two plastics which are both butadiene modified styrenic polymers. The separation took place between the electrode and middlings collection zones (rather than belt / electrode – for opposite charged particles), indicating that small adjustments in the splitter plate would lead to large variations in the resultant split. Conclusions – Hamos Trial Results It is clear from both sets of trial on the EKS machine that the performance will vary widely with differences in the type of polymers in the mix. The ability to achieve a good separation depends to a large extent on the relative contact charging behaviour of the materials. If there is more than one plastic type in the feed mixture, then separation becomes more difficult. If it is found that a second pass is needed to generate an economically viable separation, then there will be a need to invest in additional equipment or to plan for extra running hours on the plant.


With this second pass in mind, the EKS machine can be purchased as either a single or duplex unit which gives either 750kg/hr or 1500kg/hr throughput. Capital Cost is in the range of €110k or €170k for these options. Running costs estimated at €20 per tonne for the above. Therefore the potential user of this technology must decide if the level of purity delivered by the machine can be justified in terms of increased value of the output polymer product. In the case of the KWS unit, it is clear that a process to separate, say, copper from PVC cable scrap, will deliver a much higher gross profit per tonne of material than a process aimed at removing unwanted wood residues from a mainly polymer stream. However, in the absence of other effective means to remove wood from plastics, it is clear form both the Hamos and Plas-Sep trials that electrostatics IS a viable technology for this vital plastic clean-up stage of the process.


9.7 Contact information

Plas-Sep Limited #7-98 Bessemer Court

London, ON, Canada, N6E 1K7 +1 (519) 686 2878

http://www.plassep.com

Hamos GMbH

Im Thal 17 D-82377 Penzberg

Phone: +49 (0) 8856 - 9261 - 0 Fax: +49 (0) 8856 - 9261 - 99 Email: [email protected]

www.hamos.com

Other Potential Suppliers of Electrostatic Separators

Eriez

Phone: 44-29-208-68501 Fax: 44-29-208-51314

[email protected]

www.eriez.com

Outotec - formerly known as Outokumpu Technology

Corporate Management Riihitontuntie 7 C, PO Box 86

FI-02200 Espoo, Finland tel. +358 20 529 211 fax +358 20 529 2200

www.outotec.com


http://www.plassep.com/


10 Analytical Methods This section gives a brief summary of the primary tools used for analysis of the plastic samples during the trials. Apart from standard tests, such as sieve analysis, we used a combination of 3 methods to check the performance of the separation technologies in the trial programme:-

• Density Analysis – to give the distribution of particles across a range of specific gravities associated with the particular trial separation

• Polymer Type Identification – using infra-red spectrography • Additive Content of Polymers – using X-Ray Florescence

10.1 Density Analysis Method As part of the DEFRA project Axion Recycling have developed a bespoke laboratory device for the rapid measurement of density of plastic particles. This instrument was designed around the need to make a fast assessment of the distribution of plastic densities in any given sample of an unknown mix of polymer materials. Some aspects of the design are subject to confidentiality of IP. The analysis makes use of increased settling force to bring about a separation of the material in the batch to be tested. Plastic is introduced into a range of very accurately prepared liquid solutions (accurate to 2 decimal places on S.G.). The elevated settling force is used to create the required separation at the set density, and the resultant split of material is then used for subsequent separations across the required density range. In this manner it is possible to build up a bar-chart of the distribution of mass across the chosen density fractions, as indicated below.

Example density distribution graph for two plastic samples


These density distributions are then used as a means to assess the performance of a particular separation device. This is a valid method of evaluation because a thorough check has been made of the required sample size and accuracy of liquid solution needed to perform each test. The separation technique used in the laboratory method shows good repeatability on separation of samples with known density composition. Further to these performance checks, the Fourth Progress report to DEFRA, from January 2007, explained the operating method used with the instrument. The exact detail of the design is the basis of some unique intellectual property developed by Axion during the project. 10.2 Infra-Red Analysis of Polymer type With the above method being used to give a quantitative measure of the mass percentage from any sample that falls within required density points, it was then necessary to have a technique which could give a fast identification of individual polymer chips. This would then enable the investigator to find out exactly which type of plastics were reporting to the different density fractions and thus evaluate the accuracy of any density separation method on a qualitative basis. Further to this, the technique was used to obtain an off-line evaluation of the separation performance for automated sorting equipment. An assessment was made of the different types of available testing instruments which employ infra-red light as the main identification technique. During this exercise the main choice fell between Near-Infra Red and Mid Infra-Red methods. After a detailed testing programme, it was decided to select the MIR method for polymer analysis, mainly because of the greater tolerance to dark coloured plastics shown by this technique. A Thermo-Nicolet model 380 machine was used for the majority of the individual plastic identification work. This instrument incorporates a diamond sensing ‘window’ against which the polymer sample if held by a thumb-wheel clamp mechanism with a pre-set pressure. This accessory if known as the ATR device (see below). The identification of an unknown sample material takes place by comparing the collected spectrum across the mid infra-red wavelengths with those of know polymers stored in the machine’s memory. In order to enable rapid processing of the collected data a method called Fourier Transform analysis is used. This gives a measurement of ‘best-fit’ between the unknown spectrum and those from the stored library of spectra. A more detailed description of the instrument can be found on the Thermo-Nicolet website, from which the following is quoted:- “Fourier Transform Infrared Spectroscopy (FTIR) is a method of molecular vibrational analysis. The sample in question is subjected to infrared radiation, as a result of which they undergo vibrations such as stretching, bending, rotating etc. These vibrations cause a change in dipole of the molecule which is tracked by the instrument. The frequency of these vibrations depend on the nature of the molecular bonds and the environment generating a spectrum that


is unique to the material and indicative of its chemical structure. Attenuated Total Reflection (ATR) is a quick method for sample analysis by FTIR that eliminates any need for sample preparation and facilitates analysis by non skilled personnel.” Quote ex - http://www.thermo.com/com/cda/article/general/1,,1115,00.html

Figure 10.1 Nicolet 380 with Smart Diamond ATR

10.3 X-Ray Florescence Analysis - XRF This analytical technique has been applied to identify individual elements that are present at parts per million level inside the polymer matrix. It can therefore be used to detect materials such as chromium (Cr) or bromine (Br), which are important elements in plastics due to the RoHS legislation, as they have been given maximum allowable levels within Europe for the electronics sector. The instrument uses the principle of energy dispersive X-ray florescence as the basis for detection of a wide range of elements. This relies upon exposing the unknown material to a low level of X-Ray radiation, which penetrates the sample to a depth of a few millimetres and impacts upon the individual atoms within the structure. When exposed to this external energy, the electrons surrounding some atoms are shifted out of their normal orbit and are replaced by electrons from a different energy level. This shift in energy levels at the atomic scale is signalled by an emission of a pulse of electromagnetic radiation, the wavelength of which is characteristic of individual atoms. This process is called florescence. The instrument detects the wavelengths of emitted energy waves and also quantifies the number of pulses at each level. By comparing these collected peaks across the full range of emitted energy with the known peaks for particular elements, it is possible to give a quantitative analysis of the elements detected within the sample.


A more detailed explanation of the EDXRF technology is given at this website:- http://www.innov-x-sys.com/technology/fluorescence The XRF method can only detect certain elements in the periodic table, these are mainly the heavy metals, metals and halogens. Atoms of low atomic number up to 12 can not be detected using this technique – so carbon, oxygen, nitrogen and hydrogen are not detected by the instrument. The hand-held instrument comes calibrated to detect a range of 20 common elements of interest to plastics technologists. (E.g. For plastics - Standard Elements: Pb, Cr, Hg, Br, Cd, Sb, Cl, P, Ti, Mn, Fe, Ni, Cu, Zn, Bi, Sn, Ag). Axion Recycling conducted a survey of commercially available units and selected an instrument supplied by Innov-X, as this most closely met their requirements. The hand-held device gives rapid measurement of elements found within plastic samples to a ppm level. It is thus very useful either for :-

• Screening unknown feed-stocks for unwanted elements • Proving that finished product comply with set legislative levels of

banned substances (e.g. chromium).

Figure 10.2 Hand held XRF unit – Innov-X

The unit has been used in several tests during this work programme as a means to infer the presence of brominated flame retardants additives such as TBBA or deca-BDE. Typical detection limits are shown in the table below.


http://www.innov-x-sys.com/technology/fluorescence


Thermo – Nicolet www.thermonicolet.co.uk

Innov-X Systems

Helftheuvelpassage 20 5224 AP 's-Hertogenbosch

The Netherlands Tel: +31 (0)73-62 72 590 Fax: +31(0)73-62 72 599






11 Sample Preparation for Trials Where possible Axion carried out large scale trials of the separation technologies, using samples of material in the 500kg – 1 tonne size range. In these instances it was important to ensure that the samples being used to compare different trials were homogenous, so that valid comparisons could be made about the quality of the separation achieved. This also avoided any possible variation in the polymer mix as each trial progressed. Early on in the project, advice was taken from the University Of Southampton department of mathematics concerning the design of experiments and the sampling regime needed to make sure that the results obtained were statistically valid. This work was reported in earlier progress reports and also used in the design of the density analysis instrument developed during the project. The photos below show the procedure carried out to prepare bulk samples of well mixed and consistent plastic waste streams, before they were sent off for the liquid phase separation trials. In order to provide representative samples of 4 different WEEE streams for the polymer type separation trials the following procedure has been carried out for each fraction:

1."Bin-to-Bin" homogenisation. Discharging the upper bin. Material flows into the lower bin.

Material after 5 "Bin-to-Bin" runs. 2. Emptying the bin. 3. Filling into 25kg bags.


Each fraction is packed into different bags, these were then numbered randomly (not sequential). It must be noted that it was not possible to conduct large scale trials on all the technologies investigated in this report. In several cases the equipment suppliers simply did not have the capability to handle tonnage quantities of samples. These companies also claimed that their experience showed that demonstration trials carried out with 50 – 100 kilos of particulate material were able to be scaled up to tonnage quantities in industrial applications.


12 Life Cycle & Environmental Impact Assessment Conducting detailed environmental impact assessments for each of the individual technologies investigated during the trial programme was not considered a sensible approach. Each machine will form just a single unit operation within a complete mechanical recycling process for the conversion of WEEE plastic waste into replacement for virgin polymer. The processes investigated are all fairly low impact on the environment because they are essentially carrying out mechanical changes to the physical format of the input material stream. There is no chemical reaction taking place and no use of direct heat in the sorting and separation phases of the process. In the final conversion of sorted plastic chips into an extruded pellet for use by injection moulding operations, there is the application of indirect heating by external heaters on the extruder barrel. The level of direct emissions is also very low, being limited to local dust evolution and some waste water from liquid phase separation processes. There will be a level of noise associated with the equipment used in this study, in particular the size reduction machinery. There is, of course, some waste produced which goes to landfill. Thus the main impact on the environment from the actual processing of WEEE plastic waste is the consumption of electrical power represented by the combined motor drives used in the complete process. As mentioned earlier in the report, for a typical plastics recycling plant the largest electrical motor drives are associated with size reduction equipment and also the final extrusion stage. What really matters in terms of assessing the impact of recycling WEEE plastic by the mechanical methods studied in this report, is the benefit to the environment created by the substitution of virgin polymer material. Every tonne of waste plastic that is saved from landfill or incineration and recovered by a direct mechanical treatment process, can be considered to have saved the environmental impact associated with production of a tonne of similar virgin polymers. There have been detailed eco-profile studies published by the Association of Plastics Manufacturers in Europe (APME) for each of the primary virgin polymer types. This data is available for download at the following website :- http://lca.plasticseurope.org/index.htm the information can also be accessed from the APME website - http://www.plasticseurope.org/Content/. These eco-profiles give a detailed evaluation of the environmental impacts associated with the production of virgin polymers. As an example, a schematic of the process stages involved in ABS manufacture is shown below. It can be seen that this is a much more complex process than the mechanical recycling of waste plastic.


http://lca.plasticseurope.org/index.htm

http://www.plasticseurope.org/Content/

Natural gasextraction,processing &transport

Crude oilextraction& transport

Ammoniaproduction Cracking

Oilrefining

Reformingfor benzene

Ethylbenzeneproduction

Styreneproduction

Acrylonitrileproduction

Dehydrogenationof butenes

Polymerisation of butadiene

SANproduction

ABS Masspolymerisation

ABS Graftcopolymerproduction

Compounding

naphthanaphtha

air

propylene

ammonia

butenes

butadiene

benzene

ethylbenzene

ethylene

ABSpolymerisation

acrylonitrile-butadiene-styrene copolymer(ABS)

styrene-acrylonitrile copolymer (SAN)

styrene

styrene

acrylonitrilepolybutadiene

Aromaticsplant benzene

pygas

butadiene

polybutadiene

Schematic flow diagram of the principal operations leading to the production of ABS. Ack – APME – Ecoprofile ABS March 2005 – A detailed LCA study was carried out to compare mechanical recycling with a series of other recycling technologies under the WRAP BFR project in which Axion Recycling were involved. In this project, two different professional consultants were commissioned to apply LCA methods to compare a range of different recycling processes. In each case the study considered that the ‘Avoided Resource consumption’ was typically HIPS or ABS virgin polymer. From the APME Eco-profile reports the following has been extracted as an example of the amount of CO2 emission to atmosphere associated with the production of one tonne of virgin polymer material:- Polymer Type CO2 per Tonne ABS 3,100 kilo HIPS 2,800 kilo PP 1,100 kilo


The information taken from the WRAP report (REF:- Develop a process to separate brominated flame retardants from WEEE polymers, Interim Report 2, Project code: PLA-03) has given the following equivalent levels of CO2 produced for a mechanical plastic recycling process:- Mechanical Recycling of Plastics a) – to produce pure clean chip 90 kilos of CO2 b) – to produce extruded pellet 115 kilos of CO2 Taking an average level of CO2 produced for a mixture of typical WEEE plastics (HIPS, ABS, PP, PC) to be 2,600 kilos per tonne, it is clear that there are significant savings to be made by using mechanical recycling methods to recycle WEEE polymers. The WRAP study included the energy required to carry out a process step with removal of BFR additives from the plastic mixture. In summary it can be stated that the production of one tonne of high-grade recycled polystyrene using a mechanical process made up of unit operations similar to those studied in this project, will generate a 95% saving in CO2 impact upon the environment in comparison to manufacturing virgin polymer.


13.1 Conclusions This report presents a series of trials carried out on individual items of process equipment as part of the overall task of recycling WEEE plastic waste. In terms of configuring a total process to deliver high grade polymers as a product from the input waste material, there is no single answer to the question, “What is the optimum plant layout?”. The designer of a potential WEEE plastics process must first clearly define the limits of the range of raw materials that are going to be handled and their delivery formats. The degree of pre-processing that has taken place at the upstream treatment plant will affect the amount of processing needed at the plastics recycler. For example, finely shredded plastic from fridge treatment sites needs minimal size reduction in comparison to baled CRT casings. The exact choice of polymer processing technologies will also depend upon the required output material quality in terms of polymer purity. In general it can be stated that most WEEE plastics plant will require technologies that carry out the initial clean-up of the material, by removing the non-plastic contaminants as follows:-

• Size Reduction – primary • Metal removal – ferrous / non-ferrous / stainless steel • Fines and dust extraction • Wood and rubber removal • Stone and glass removal • Further Size Reduction - secondary

The above process steps should deliver a clean, accurately sized mix of granules into the polymer separation stages, at which point there is a choice of technologies to be used:-

• Increased-G Density Separation • Infra-Red Light sorting • X-Ray Transmission sorting • Electrostatics • Colour Sorting

The benefits and limitations of these technologies have been described in the trial reports. The following conclusions are highlighted as important findings from the project to be considered in overall process selection:-

• Size Reduction is important, it is needed to:- o liberate different materials joined together in one component o standardise the size of particles for the clean-up stages


o present a free-flowing granular product to customers, who need regular small pellets to feed into extrusion machinery

• Metal Removal is likely to be needed at more than one step in the process. All shredding and granulation equipment requires upstream metal removal to protect the machinery from damage. Final products will need to be metal-free to meet customer quality standards.

• Induction Sorting is seen to be a novel approach to removal of all metal types in one piece of machinery in comparison to traditional magnetic and eddy-current systems.

• Polymer Type separation using increased-G density separation in the liquid phase works well, provided that the correct choice of centrifuge is made for the defined job. In these trials the Flotweg Sorticanter gave very accurate density cuts and did some partial drying of the plastic granules.

• Sorting for polymer type using accurate density splits only works if the infeed material contains a simple mixture of plastics across the target density range. Complex materials with high additive and filler content will move individual polymers outside of their normal ‘100% pure’ density range and make the separation task impossible under this method.

• NIR sorting technology can handle polymer type separation of the main plastic groups (e.g. styrenics sorted from polyolefins). In these trials the exact sorting of a specified type of polymer, such as ‘ABS-only’, from a plastics mixture proved less efficient. This sorting works best at larger particle size (30 – 50mm) and with a tight size distribution, this was not the case in these trials.

• X-Ray Transmission sorting showed good performance for the removal of brominated additives from a plastics mixture.

• Electrostatic separation works best for simple mixes of plastic and is dependent upon the charging properties of each individual polymer in the mixture. Fine metal removal using a Corona discharge unit works well. Good results were seen for removal of wood using these techniques.

• Colour Separation of small granules was seen to work well in these trials as a way to add value to finished products

• A partial success was seen on wood and rubber removal using air-blown Gravity Separators in this work programme. Although the separation effect is more one of concentration of the contaminant material, as opposed to a complete separation.

• The combination of Impact or Hammer Milling of plastic streams with a high residual metal content (>15% metal) with a Gravity Separator as the metal removal stage, was seen to deliver a high grade of clean metal spheroids from the waste stream. The effect on the plastic in this process was less beneficial. This approach would be best applied at a primary WEEE treatment plant as part of the metal recovery operation.


13.2 Recommendations The following recommendations are directed to process designers or plant operators who are considering investment in WEEE polymer separation plant:-

• Select the minimum number of different input plastic streams, in order to minimize the complexity of the separation task

• Check and define the format of the delivered-in plastic material from different suppliers, to make sure the plant can handle any variations.

• Ensure adequate metal removal steps are included throughout the process. Failure to provide good protection to size reduction machinery will cost dearly in terms of high wear and increase the likely occurrence of a catastrophic failure caused by metal ingress.

• Do not under-estimate the levels of dust, fines and other non-metallic contamination which will need to be removed from the input material.

• Make careful analysis of the mixture of plastic types found in all different input streams and use this information as the basis for the selection of polymer sorting technology. This analysis to include an evaluation of additive content, in particular flame retardants.

• Choice of the most suitable polymer separation technology should be based upon the above analysis. Density separations are unlikely to be successful with complex mixtures that include polymers with high filler and additive content. In those cases, a combination of the dry sorting technologies described in this report will be needed.

• Conduct full-scale trials of all main process steps before committing to purchase.


13.3 Main Implications of the Findings One of the main findings from this work that has clear implications in the treatment of E&E waste equipment, is the fact that the polymer streams arising from primary WEEE plants are highly co-mingled and contaminated mixtures of different materials. Even with the well defined streams of plastic arising from the treatment of CRT screens and refrigeration goods, there is still a lot of pre-processing to be carried out before the task of polymer separation can take place. When dealing with the mixed plastic from bulk treatment of small WEEE, the degree of variation in the plastic mix is even greater, as are the number and type of contaminants. This makes the processing task of recycling high-grade polymers from WEEE plastic waste even more difficult. The main result of this problem of highly mixed input waste material, is that the extraction of a useable fraction of high-grade single polymer type can only be achieved at low yield based upon the total input tonnage. There is then a large proportion of mixed plastic which cannot be easily recycled back into saleable product due to the incompatibility of the polymer types and the high incidence of fillers and additives (such as BFRs) found in the material. It is apparent that the majority of WEEE treatment plants in the UK will be operating on a mixed-bulk basis, with shredding or fragmentising of the whole electrical items as the leading technology approach. This means that the majority of the WEEE plastic arisings being created over the coming months will have the problems of contamination and mixing described above. Further work is therefore needed to understand how best to tackle the processing challenge represented by the residual fraction of plastic that is left-over after separation by the methods described in this report have taken place. In particular it has been found that the incidence of brominated flame retardants in some waste streams is very high (e.g. IT and computer VDU plastic scrap). In plastic waste from small household and large white goods, it has been found that there is a high level of heavily filled plastic and these material tend to broaden the range of densities associated with each common polymer type. It is not possible to separate these materials by density methods. If the recycling targets set in the WEEE Directive are to be achieved for the small household WEEE material, then some additional methods of processing this residual plastic mixture need to be found. This requires further work looking at the areas of:-

• Extraction of Brominated Flame Retardants from the plastic matrix • Development of compatibliser additives for thermoplastic mixtures to

produce viable polymer blends with acceptable physical properties to suit market applications

• Recovery of extracted fines, wood, rubber and minority plastic types by environmentally sound methods such as incineraton or feedstock recycling.


It is important that the primary WEEE treatment processes being implemented across the UK are properly controlled and audited to make sure that the principles of ‘Best Available Treatment and Recycling & Recovery Technology’ (BATRRT) are being put in place. This can only occur with proper enforcement of the legislative requirements, without which the WEEE plastics stream may well become the best method of waste disposal for the primary processor. Finally it is essential to level the playing field for European WEEE polymer processors by ensuring that exports of mixed WEEE polymers and streams that contain high levels of legacy additives such as BFRs are strictly controlled to ensure that they are subject to the same controls and processing requirements as material processed within the EU. If this does not happen then European recyclers will continue to be disadvantaged and both innovation and investment will be restricted. Axion Recycling gratefully acknowledge DEFRA funding support for this research project. The use of brand names for some of the commercial equipment tested during the work programme does not represent any endorsement of particular brands by either Axion or DEFRA.


Appendix 1 – Polymer Types in WEEE Plastic & Glossary The mix of the primary polymer types found in WEEE plastic waste has been widely reported, however in summary the main groups of plastic types are as follows:- Polyolefins – this term is used to described the thermoplastic polymer materials which are manufactured from oil-derived monomers, mainly ethylene and propylene. The polymerisation of these monomer compounds into long chain structures produces a range of plastics, which are dominated by the two materials:- Polyethylene - abbreviated to PE Polypropylene – abbreviated to PP In a typical sample of mixed WEEE plastic one would expect to find between 5 – 20% of these polymers. They are usually identified by their ability to ‘float’ in water as the pure plastics have densities less than 1.0 gram/cc. Styrenics – The styrene monomer is the basic building block of this group of plastics. It includes a benzene ring grafted onto the base ethylene monomer molecule. Polymerisation of this produces polystyrene plastic (PS) , often associated with its expanded form as packaging foam – expanded polystyrene (EPS). There are many developments of engineering plastics based around the styrene building-block, these include more complex blends or alloys of different polymer types which can also include ‘rubber-like’ elements to increase flexibility and toughness of the final material. Some of the polymers mentioned in this report include:-

ABS - Acrylonitrile-Butadiene-Styrene PC/ABS - a blend of polycarbonate with ABS HIPS - high impact polystyrene (contains some butadiene)

Typically the styrenic fraction of any WEEE plastic sample will account for 30 – 70% of the mixture. Other plastics in WEEE Some of the other polymers mentioned in this report and found as minority fractions of the mix are :- PVC – polyvinylchloride PU – polyurethane – mostly from foamed insulation in fridges PA – polyamide – commonly called ‘nylon’ PMMA – polymethyl methacrylate – acrylic or Perspex POM - acetal PET - polyethylene terephthalate More detail can be found at websites of the APME http://www.plasticseurope.org/Content or BPF http://www.bpf.co.uk/bpfindustry/plastics_materials.cfm Useful information on the properties of commercial grades can be found at www.matweb.com


http://www.plasticseurope.org/Content

http://www.bpf.co.uk/bpfindustry/plastics_materials.cfm

http://www.matweb.com/

References 1 – Report for Hampshire Natural Resources Trust – Sorting and Identification of Polymers from WEEE – K Freegard – Axion Recycling Ltd Feb 2006 2 - Develop a process to separate brominated flame retardants from WEEE polymers - Final Report WRAP Project code: PLA- 037, November 2006 ISBN: 1-84405-315-6

(end of doc) ______________________________________


title :- weee plastics separation technologies

Documents