Final report Separation of mixed WEEE plastics

Final report

Separation of mixed WEEE plastics

A series of demonstration trials on novel techniques for the separation of mixed WEEE plastics.

Project code: MDD018 and MDD023 Research date: October 2008 - May 2009 Date: October 2009

WRAP helps individuals, businesses and local authorities to reduce waste and recycle more, making better use of resources and helping to tackle climate change. Document reference: WRAP, 2009, Separation of mixed WEEE plastics final report (WRAP Project MDD018 and MDD023) Report prepared by Axion Consulting

Written by: Mike Bennett, Leigh Edwards, Lidia Goyos-Ball, Robin Hilder, Dr Phillip Hall, Liz Morrish, Roger Morton and Nicola Myles,

Front cover photography: Mixed WEEE plastic fraction WRAP and Axion Consulting believe the content of this report to be correct as at the date of writing. However, factors such as prices, levels of recycled content and regulatory requirements are subject to change and users of the report should check with their suppliers to confirm the current situation. In addition, care should be taken in using any of the cost information provided as it is based upon numerous project-specific assumptions (such as scale, location, tender context, etc.). The report does not claim to be exhaustive, nor does it claim to cover all relevant products and specifications available on the market. While steps have been taken to ensure accuracy, WRAP cannot accept responsibility or be held liable to any person for any loss or damage arising out of or in connection with this information being inaccurate, incomplete or misleading. It is the responsibility of the potential user of a material or product to consult with the supplier or manufacturer and ascertain whether a particular product will satisfy their specific requirements. The listing or featuring of a particular product or company does not constitute an endorsement by WRAP and WRAP cannot guarantee the performance of individual products or materials. This material is copyrighted. It may be reproduced free of charge subject to the material being accurate and not used in a misleading context. The source of the material must be identified and the copyright status acknowledged. This material must not be used to endorse or used to suggest WRAP’s endorsement of a commercial product or service. For more detail, please refer to WRAP’s Terms & Conditions on its web site: www.wrap.org.uk

Separation of mixed WEEE plastics 1

Executive summary WRAP (Waste & Resources Action Programme) commissioned Axion Consulting to demonstrate a number of innovative techniques to tackle some of the more difficult separations encountered by primary and secondary waste electrical and electronic equipment (WEEE) processors. The aims were to trial and assess some of the most promising separation techniques that have emerged over the past three to four years since WEEE materials started to be reprocessed in significant quantities in the UK. The techniques selected for these trials were chosen because they may have potential to improve the effectiveness and commercial viability of WEEE recycling processes. It is envisaged that these techniques could be used either by primary WEEE processors, known as Authorised Treatment Facilities (ATFs), or by secondary processors, which take the plastic fractions produced by the ATFs and upgrade them further. During the selection process techniques were excluded which: have been tested and published in previous WRAP or Defra reports, particularly Defra project report

WRT 095 ‘Developing bulk WEEE polymer separation and analysis techniques’, June 2007; and

Axion has tested in-house and eliminated as they are not commercially viable.

The separation techniques were grouped into four categories according to their separation mode and application as follows: Sensor-based separators

o RTT Near infrared ‘UniSort’ sorter – for separation by polymer type TITECH Near infrared ‘Polysort’ sorter – for separation by polymer type

o Visys Spyder laser sorter – for separation of circuit boards o

Sha anpe d density separators o Allmineral wet jig - for separation of fine metal and glass from plastic and the separation of

plastics lass from plastic o Delft University Kinetic Gravity Separator – for separation of fine metal and g

o Holman Wilfley wet shaking table – for separation of fine metal from plastic o University of Nottingham dry pneumatic jig – for separation of metal, stone and glass from

plastic low classifier – for separation of fine metal from plastic o Allmineral ‘Allflux’ upf

Separation by impact milling o Pallmann PXL18 different

ial impact mill – for separation by polymer type

Alvan Blanch ‘Destoner’ air table – for separation of wood from plastic

e following techniques produced acceptable or nearly acceptable separations of their target

ed (NIR) ‘Polysort’ sorter;

r; nd

Allmineral ‘Allflux’ upflow classifier.

IR

ials, esting of this technology is required with

ultiple passes through the machine to establish if this is feasible.

PS / 10% PE and unscreened 90% PS / 10% PE) but it orked better with larger particles (greater than 6mm).

Effect of particle size distribution o

During the trials thWEEE materials: TITECH Near infrar Visys laser sorter; Delft University Kinetic Gravity Separato Holman Wilfley wet shaking table; a

The TITECH NIR sorter has good potential to separate acrylonitrile butadiene styrene (ABS) from polystyrene (PS) which is an important separation for WEEE processors. However the sorter was only able to identify and separate 50% of the material in the feed as the remaining 50% was too dark in colour to be identified by the Nsensors. For the 50% of the material which could be identified, the single pass separation used in these trials only produced 80:20 ABS:PS mixtures. To achieve a significant improvement in the value of processed materthe sorter would need to upgrade these mixtures to 95:5. Further tm The TITECH NIR sorter performed well overall and was able to separate PS and polyethylene (PE) across all three size fractions (sub 6mm 90% PS / 10% PE, 6-12mm 90%w


d quality metal concentrates: r;

d the Allmineral ‘Allflux’ upflow separator.

roduced a very high purity copper fraction but orked better on size reduced material (less than 5mm).

ilable as a production machine but offers e significant benefit of not requiring the material to be size reduced.

s were technically promising, but required the material to be ze reduced to below 2mm with a hammer mill.

niques which appeared to be technically viable: r;

r;

Gravity Separator, Holman Wilfley wet shaking table and Allmineral Allflux upflow separator) is valued etween £700-1,000/te. All three technologies are estimated to pay back their initial investment within two

the oided landfill cost of £60/te) is too low to justify

e investment cost of the machine at the present time. This situation could change with market price

r rated ABS and PS

om around £200/te for the unseparated styrenic chip mixture, to about £400/te for the separate ABS and PS

tain a reliable

allow pes for the plastic/metal mixtures caused

w polymer separation based on size after milling. The technique did show good potential for

eping the material dry. This avoids a drying step after the separation, however

cially

Alvan Blanch destoner air table – was able to separate wood from plastic. The trial showed that separation efficiency improved when tighter size ranges were fed to the separator.

Three of the metal separation techniques produced goo the Delft University Kinetic Gravity Separato the Holman Wilfley wet shaking table; an

The wet shaking table is a well established technology and pw The Kinetic Gravity Separator is at the research stage and is not yet avath The Allmineral Allflux upflow separator trial resultsi Axion evaluated the economic potential of the five tech TITECH Near infrared (NIR) ‘Polysort’ sorte Visys laser sorter; ratoDelft University Kinetic Gravity Sepa Holman Wilfley wet shaking table; and Allmineral ‘Allflux’ upflow classifier.

The metal concentrate product from the three more successful metal separation techniques (Delft University Kinetic byears. The Visys laser sorter produced a concentrated circuit board fraction with good separation efficiency butcurrent market value of the circuit boards (around £40/te plus avthfluctuations, which could make the investment more attractive. In addition, the TITECH NIR sorter was evaluated on the assumption that 95% pure ABS and PS products could be produced with four passes through the separator (a first pass for the feed material and then three passes fothe two products from the initial separation). Assuming a value upgrade of £200/te for the sepafrchip fractions, this machine should pay back its initial investment in approximately three years. Through the course of these trials, the following techniques were unable to produce technically viable separations with the target WEEE plastic mixtures: RTT UniSorter NIR sorter – was unable to distinguish ABS and PS from each other, or to ob

signal from relatively light coloured particles and currently it can only hold a library of eight polymer spectra at any one time. This is too few for sorting the complex mixtures found in WEEE;

Allmineral wet jig – the density differences between the particles to be separated were too small toit to make polymer separations. The wide range of shaparticle-particle entanglement and within the scope of this trial made it difficult to achieve a clean separation of fine metal and glass from WEEE plastic;

Differential impact milling – the differences in impact properties between the polymers tested were too small to allodelamination and separation of plastics and elastomers although further test work would be required to prove this;

Nottingham University dry jig and the Allmineral wet jig - the density differences between the particles to be separated were too small to allow it to make polymer separations. The dry jig appeared to be capable of separating fine metals from WEEE polymer mixtures in much the same way as the wet jig butwith the advantage of keconsiderably more experimentation would be required to scale up the dry jig process to a commeruseful throughput; and

Contents 1.0 Introduction ............................................................................................................................. 6

1.1 Background to the project......................................................................................................6 1.2 Aims and objectives of the project ..........................................................................................6 1.3 Structure of the report ...........................................................................................................6

2.0 WEEE plastics processing requirements .................................................................................. 7 3.0 Separation techniques ............................................................................................................. 9

3.1 Identification of techniques ....................................................................................................9 3.2 Methodology of trials ...........................................................................................................10

3.2.1 Trial plans ..............................................................................................................10 3.2.2 Preparation of samples............................................................................................10 3.2.3 Trial delivery...........................................................................................................11 3.2.4 Assessment of quality and purity..............................................................................11

4.0 Summary of techniques ......................................................................................................... 12 4.1 Sensor based sorting techniques ..........................................................................................19

4.1.1 How sensor based sorting works..............................................................................19 4.1.2 Where sensor based sorting worked well..................................................................21 4.1.3 Limitations of sensor based sorting ..........................................................................22 4.1.4 Economic assessment .............................................................................................23 4.1.4.1 Sensitivity analysis for TITECH NIR sorter ................................................................26 4.1.4.1 Sensitivity analysis for Visys laser sorter ...................................................................28 4.1.5 Conclusions ............................................................................................................28

4.2 Shape and density separation techniques..............................................................................29 4.2.1 How the shape and density separations work ...........................................................29 4.2.2 Where shape and density separation worked well .....................................................35 4.2.3 Limitations of shape and density separations ............................................................35 4.2.4 Economic assessment .............................................................................................37 4.2.5 Overall conclusions of the shape and density separation trials ...................................41

4.3 Separation by differential impact milling................................................................................42 4.3.1 How differential impact milling works .......................................................................42 4.3.2 Where differential impact milling worked well ...........................................................43 4.3.3 Limitations of differential impact milling ...................................................................43 4.3.4 Economic assessment .............................................................................................43 4.3.5 Conclusions ............................................................................................................43

4.4 Effect of particle size distribution ..........................................................................................44 4.4.1 How the Alvan Blanch Destoner air table works ........................................................44 4.4.2 Trial results ............................................................................................................44 4.4.3 Conclusions ............................................................................................................45

5.0 Overall assessment of economic and technical potential for WEEE processing..................... 46 6.0 Conclusions ............................................................................................................................ 47

6.1 Technically and commercially feasible techniques ..................................................................47 6.2 Technically feasible techniques.............................................................................................48 6.3 Unsuccessful techniques ......................................................................................................48

7.0 Recommendations.................................................................................................................. 49 Appendices ......................................................................................................................................... 50


Tables Table 1 Summary of individual trials .............................................................................................................13 Table 2 Particle size distribution for Axplas PS07, a typical sample of mixed WEEE plastic................................22 Table 3 Payback calculation for the TITECH NIR sorter and Visys laser sorter .................................................25 Table 4 Sensitivity analysis for TITECH NIR sorter .........................................................................................28 Table 5 Sensitivity of the Visys laser sorter payback calculation......................................................................28 Table 6 Payback calculation for Holman Wilfley wet shaking table and Delft University Kinetic Gravity Separator...................................................................................................................................................................38 Table 7 Capital cost breakdown for Allflux upflow separator ...........................................................................39 Table 8 Payback calculation for Allflux upflow separator.................................................................................40 Table 9 Summary of the technical and economic feasibility of the demonstrated techniques ............................47 Figures Figure 1 Typical composition of WEEE ............................................................................................................7 Figure 2 Polymer types found in WEEE ...........................................................................................................8 Figure 3 Schematic of the TITECH NIR Sorter (courtesy of TITECH)...............................................................20 Figure 4 Schematic of the Spyder Laser Sorter (courtesy of Visys) .................................................................21 Figure 5 Process separation sequence for TITECH sorters with a base case separation efficiency of 80% .........24 Figure 6 Process separation sequence for TITECH sorters with an improved separation efficiency of 82%........27 Figure 7 Schematic of an Allmineral wet jig - front view of a side pulsed wet jig .............................................30 Figure 8 Cut-away sketch of Kinetic Gravity Separator...................................................................................31 Figure 9 Holman Wilfley wet shaking table....................................................................................................32 Figure 10 University of Nottingham pneumatic dry jig ...................................................................................33 Figure 11 Diagram of a full scale Allflux (courtesy of Allmineral) ....................................................................34 Figure 12 Process flow diagram for separation process using an Allflux upflow separator ................................39 Figure 13 Photograph of the Pallmann - PXL18 Mill .......................................................................................42 Figure 14 Photograph of the PXL18 Mill Rotor (8 elements on outer ring and 4 elements on inner ring) and Stator (8 element ring) .................................................................................................................................42 Figure 15 Alvan Blanch DS2 Destoner (photograph courtesy of Alvan Blanch).................................................44


Glossary ATF Authorised Treatment Facility for primary treatment of WEEE FTIR Fourier Transform Infrared Spectroscopy Technique KGS Kinetic Gravity Separator NIR Near infrared OEE Overall Equipment Effectiveness UHR Ultra High Resolution WEEE Waste Electrical and Electronic Equipment Plastics ABS Acrylonitrile butadiene styrene HIPS High impact polysytrene PA Polyamide PC Polycarbonate PCABS Polycarbonate acrylonitrile butadiene styrene PE Polyethylene PMMA Polymethyl methacrylate POM Polyoxymethylene PP Polypropylene PS Polystyrene PVC Poly Vinyl Chloride Eject A fraction removed from a bulk stream by ejection, typically using air jets, by a sensor based sorting

technique, depending if the sort is positive or negative the eject fraction can be a product or a waste stream.

Reject A fraction allowed to pass through the sensor based sorter and not be removed, depending if the sort is

positive or negative the reject fraction can be a product or a waste stream. Product separation efficiency (Q) the probability that the product is correctly sorted into the product stream. Reject separation efficiency (R) the probability that the secondary product/waste is correctly sorted into the secondary product/waste stream.

Acknowledgements Trial hosts: Allmineral Aufbereitungstechnik GmbH & Co, Delft University of Technology Recycling Laboratory, SGS Mineral Services UK Limited, Pallmann Maschinenfabrik GmbH & Co. KG, RTT GmbH, TITECH Visionsort Limited, Visys NV, University of Nottingham and Alvan Blanch.



1.0 Introduction 1.1 Background to the project The implementation of the UK Waste Electrical and Electronic Equipment (WEEE) Regulations in 2007 has stimulated large scale processing of a wide range of WEEE. There is a need to separate the plastics from this material stream, as well as contaminants such as metals, in order to maximise product value, minimise waste management and disposal costs and to demonstrate compliance with the recycling and recovery targets established by the WEEE Directive. In 2007 Defra commissioned research into the development of polymer separation and analysis techniques and published a report from this work1. This research identified and evaluated a number of techniques that were available at the time (June 2007). The report concluded that two main stages of processing were needed for WEEE materials. The first stage consisted of techniques to clean up the material by removing non-polymer contaminants and then secondly a polymer separation stage. The sector has developed and grown since the Defra report was published and there have also been developments of the techniques potentially available to the sector. WRAP now wishes to advance the techniques used by the WEEE recycling sector2 to ensure that state of the art technologies and processes are being adopted and used by the industry. In particular, WRAP wishes to help improve the quality of the plastic fractions produced by primary WEEE processors to enable the UK WEEE treatment industry to generate more value from the plastic supplied by the primary processors. In 2008 WRAP commissioned Axion to deliver a number of trials of techniques for separating WEEE plastics, which build on the techniques identified by the earlier Defra report and focuses on new techniques in the market place. This report summarises the conclusions of the project. 1.2 Aims and objectives of the project The objectives of this project are to: identify techniques that could be used to separate and process mixed WEEE plastics; demonstrate the techniques in terms of their ability to sort material by polymer type, colour and/or

remove impurities; assess the suitability and economic performance of the techniques; and produce reports with the detailed results of the trials to help both primary WEEE processors (ATFs) and

secondary processors, who upgrade the plastics fraction made by the ATFs, to identify technologies that could be used to improve their own operations.

1.3 Structure of the report

The project has delivered a number of trial techniques that could potentially be used by the WEEE recycling sector. It is envisaged that these processes might be used by ATFs to improve the quality of their plastic products, or by WEEE plastic reprocessors to supplement their existing separation techniques. This report provides an overview of the techniques and presents the key results and findings from the trials. A summary is provided for each group of how the technology works and the positive outcomes of the trials are presented. The report also discusses any limitations that were found with the techniques and provides an economic assessment for some of the techniques, for use within the WEEE recycling and reprocessing sector. Individual trial reports are provided as appendices to this main report and can be downloaded at www.wrap.org.uk/separationofWEEEplastics These provide more detail of the trials including sample material used, trial data, analytical results and conclusions of individual trials.

1 Defra project report WRT095 ‘WEEE Plastics Separation Technologies’ , Authors Keith Freegard, Gayle Tan, Sebastian Frisch, Axion Recycling, June 2007

2 Please note the report uses the term ‘WEEE recycling sector’. This is intended to cover treatment, reprocessing and recycling activities undertaken by the sector. The techniques presented in this report will be of interest to both the ATFs and the plastic reproccessors in the sector.

http://www.wrap.org.uk/separationofWEEEplastics


2.0 WEEE plastics processing requirements WEEE plastic processing plants in the UK take material from primary WEEE treatment plants (ATFs). These primary treatment facilities accept material from a range of WEEE collection schemes. The material includes separate streams of: fridges; televisions; IT equipment; large domestic appliances; and small mixed WEEE items.

Each stream is generally processed separately by specialist primary ATFs. Initial processing of each WEEE stream at the ATF will produce a number of separate material streams: metal, glass, circuit boards, plastics and impurities/waste and it is the plastics stream that is passed to WEEE plastics reprocessors which is the primary focus of this project. A WEEE plastic reprocessor will then perform further separation processes on the mixed plastic fraction in order to produce high grade plastics, as well as a metals stream and a residual waste fraction. The mixed plastic accepted by WEEE plastics reprocessors tends to be variable in composition and contains a significant proportion of non-plastics materials such as metals (ferrous and non-ferrous), rubber, paper, cable, glass and stones. Figure 1 shows the typical overall composition of WEEE whilst Figure 2 shows the polymers types found in WEEE. Therefore any techniques used by the sector need to be able to handle a mixed material stream of this kind and produce high quality separated product streams.

Figure 1 Typical composition of WEEE3

3 WRAP Project MDD009 ‘Compositional analysis of kerbside collected small WEEE’ Final Report, February 2009.


Figure 2 Polymer types found in WEEE4

4 WRAP Project MDD009 ‘Compositional analysis of kerbside collected small WEEE’ Final Report, February 2009.


3.0 Separation techniques 3.1 Identification of techniques The first task was to identify a number of techniques that could potentially be adopted and used by the WEEE recycling sector. Defra’s project 5 identified and trialled a wide range of separation techniques, which could be used by both ATFs and secondary WEEE plastic processors to separate and upgrade their materials. This report, together with Axion’s own experience in the WEEE processing sector was used as a starting point for compiling a list of potential techniques to be trialled in this project. Care was taken to ensure that none of the trials included in the Defra project were repeated in this project. Axion then used its knowledge of other research projects for the WEEE sector, its experience of separation techniques used in other processing industries, and of operating within the WEEE sector to identify a number of techniques that have potential for use to upgrade WEEE materials. This study particularly targeted separations for the following applications: Separation of particular polymer types from mixtures of WEEE plastic chips in the 2-12mm size range. WEEE contains a very wide range of plastics, many of which are incompatible with each other. Most WEEE processors are now able to separate mixed PS and ABS from polyolefins such as PE and polypropylene (PP) and from poly vinyl chloride (PVC) using techniques of the type tested in the Defra project. However separation of PS from ABS is difficult because they are very similar in their physical properties. The value of finished WEEE polymers is higher if ABS and high impact polystyrene (HIPS) can be separated from each other and from polypropylene (PP), polymethyl methacrylate (PMMA), polyoxymethylene (POM) and other smaller components present in the WEEE plastic mix; and Separation of fine non-ferrous metals, glass and stone from WEEE plastics. Primary WEEE processors use magnets and eddy current separators to remove the bulk of the metals from WEEE mixtures after shredding. Magnets work well for ferrous metals down to very small particle sizes but eddy current separators are unable to remove non-ferrous metals below about 2mm. These metals tend to be separated into a copper-rich plastic fraction by secondary WEEE processors. The plastic content of this material (90% +) is still too high for the metal to be of commercial interest to copper smelters.

This project also included trials to measure the effect of tightening the size distribution of the materials entering a typical separation currently used by WEEE processors on separation efficiency.

The technologies selected represent techniques that are currently used to reprocess other materials in different industries, particularly mining, agricultural processing and foods, but have potential to be transferred for use in the WEEE recycling sector.

5 Defra Waste Research Team, Developing bulk WEEE polymer separation and analysis techniques (Project code WRT095) July 2007


3.2 Methodology of trials The methodology for the trials in this project is summarised below: Trial host selection: once a technique was identified for a demonstration trial, a suitable trial host was

selected. For some of the techniques covered by this project there were a number of companies/organisations that could potentially supply equipment and host the trial. A host was selected based on a number of factors, including agreement that the results of the trial could be published as well as the suitability of the equipment to be used for a particular trial;

Trial objectives: once a trial host was chosen, a set of objectives for the trial was agreed with the host; Trial arrangements: arrangements for delivery of the trial were then agreed, including a date and

duration; Planning: a trial plan and risk assessment was prepared for each trial; Feed material sample preparation; statistically meaningful samples of feed material were prepared for

shipment to the trial site; Trial delivery: Axion engineers attended every trial in order to ensure that it was conducted according to

the trial plan and to ensure that samples were collected, recorded and properly labelled for shipment back to Axion; and

Analysis and reporting; post-trial analysis was conducted on the samples in an analysis laboratory and the results recorded in the trial report.

3.2.1 Trial plans Once the trial host and the objectives were agreed a trial plan was developed and details of the sample feed materials to be used in the trial were provided. The trial plan outlined the proposed procedure and the feed material samples and results to be obtained during the trial. Trial plans were agreed with hosts prior to delivery nd were then followed during the trial to ensure the plan was correctly carried out. a

3.2.2 Preparation of samples Sample feed material was discussed with each trial host to ensure the quantity and type of material was appropriate and whether any preparation of the feed material samples was required, to ensure a good understanding of the capabilities of the equipment and the type and mix of material that could be processed. In some cases, pre-trial feed samples were provided to the host in order for them to conduct preliminary testing and assessment before proceeding with a full trial. Some of the trials also required partial processing of sample material, for example size reduction or granulation of material. Trial material was then sourced, prepared, agged, labelled and shipped to the trial host. b

For the majority of the trials in this project, sample feed material was sourced from the Axion Polymers plant at Salford. Axion processes a range of different WEEE plastic sources including material from TV and monitor recyclers, fridge recyclers, producer take back schemes and bulk processors of both small and large domestic appliances. Most of the materials used in this trial were derived from bulk mixed small domestic appliances which re the most challenging feed stream faced by most secondary WEEE recyclers. a

The trial samples are representative of the mixed styrenic plastic and copper-rich materials produced by most European WEEE plastic reprocessors, which are typically produced in the size range 2-12mm, with the majority in the 8-12mm range. WEEE reprocessors usually granulate their feed material to this range as this is required to separate the polymer types and metals in the feed mix. Some of the polymer separations used by WEEE reprocessors to separate impurities such as wood, paper and other materials from the mix and to separate

yrenic polymers from polyolefines and non-styrenic polymers also work better with smaller particle sizes. st

etails of the samples used in each of the trials are provided in the individual trial reports (appendices). D


3.2.3 Trial delivery

anage

rial analysis. Any unused feed material om the trial was also returned to maximise recovery of the materials.

3.2.4 Assessment of quality and purity

of a trial was to conduct analysis on the samples to determine the effectiveness of each chnology.

e

t. The technique is straightforward to use but time consuming as it

o

y

is added to the

and at break, impact testing

combination of these techniques enabled an objective assessment to be made of the success of each trial.

Axion representatives attended all trials to ensure the trial plan was applied, to oversee delivery and to many problems and variations. All relevant information from the trials including data, equipment settings, operating parameters and photographs of the samples and equipment were recorded. Sample material was collected, bagged and labelled before being returned to Axion for post-tfr

The final part te A range of analytical techniques were used during this project and are summarised below. The analytical technique chosen to assess a particular sample was determined by the type and mix of material in question. FTIR analysis: for sample material containing several different types of plastic, Fourier Transform Mid

Infrared Spectroscopy (FTIR) is used to determine the type of plastic of a polymer chip. Each type of plastic has a unique infrared spectrum which allows it to be identified by the analytical equipment. Thtechnique involves placing a single chip onto the FTIR machine and taking a reading. Software thencompares the spectrum from the chip to a library of spectra for known polymers and offers a set of possible matches in order of best fiinvolves handling individual chips;

Hand sorting: for samples consisting of materials which can be clearly distinguished by eye and in thesize range of 5mm or greater, a simple hand sort of the material can be conducted to determine the composition of the material. This technique involves taking a small sub-sample of the main fraction and sorting the sample into its respective material types. Typically hand sorting can involve separation intthe following categories; wood, plastic, rubber, copper wire, PVC coated wires, circuit boards, stone, glass and other metals. Hand sorting is a simple technique which provides comprehensive results;

Sink float separation: in the case of a PS and PE mixture a sink float in water is possible. For other morecomplex mixtures of materials, when the desired density separation is at a level higher than the densitof water, a high density solution is required such as sodium polytungstate. The high density solution can also be used to separate material which is too fine for the hand sorting technique outlined above. For example to separate copper from plastic at a particle size below 3mm, the samplehigh density solution - the copper fraction sinks whilst the plastic fraction floats; and

Other techniques involving the use of specialist laboratory tests to determine the physical properties ofsample material were also used. The plastic samples from the trials were extruded and moulded intoplaques for testing. Tests included tensile strength, elongation at yieldusing a notched IZOD impact test and the melt flow index (MFI) test.

Using a


4.0 Summary of techniques The techniques have been grouped into categories as follows: Sensor based sorting techniques

o TITECH Polysort UHR (Ultra High Resolution) NIR sorter; o RTT UniSort flake NIR sorter; and o Visys Spyder (multi-frequency digital laser sorter).

Shape and density separation techniques o Allmineral Alljig (wet jig); o Delft University of Technology Kinetic Gravity Separator (KGS); o Holman Wilfley wet shaking table; o Allmineral Allflux upflow classifier; and o University of Nottingham pneumatic dry jig.

Milling as a separation technique o Pallmann PXL18 differential impact mill

Effect of particle size distribution o Alvan Blanch Destoner air table

The separation trial results are summarised in the table below and described in more detail in the appendices.

Table 1 Summary of individual trials Trial host and equipment

Objectives Samples Technical results Economic assessment Conclusions

TITECH polysort NIR sorter

- Separate mixture of ABS and PS into distinct polymer types (PS-rich and ABS-rich streams). - Remove contaminants from bulk polymer stream. - Assess effect of particle size on sort efficiency. - Determine blackness detection limits of machine.

(1) Axplas PS11 (plastic from WEEE). (2) Axplas PS07 (plastic from WEEE). (3) Three size fractions of PS/PE mixture, size classified to <6mm, >6mm and unclassified. (4) Set of PS plaques covering a range of black shades.

- Produced ABS and PS fractions with compositions 80-85% each. Purity needs to be at least 95% to prevent compatibility issues during extrusion. - Unable to remove minor contaminants effectively. - All three fractions processed, >6mm fraction produced best results. - Blackness detection limit determined.

Assuming ABS and PS fractions with 95% compositions can be produced and with a capacity of 1 te/hr three units are required, the payback time is approximately 37 months.

NIR technology is widely used for sorting of whole packaging items and is very successful in that application. WEEE plastics present many extra challenges and test the technology to the limit. The TITECH polysort showed good technical potential and analysis of the initial trial results indicates that it should be commercially viable for bulk WEEE plastic separations, although multiple passes will be required and only about 50% of the material can be positively identified.

RTT UniSort NIR sorter

- Separate mixture of ABS and PS from each other. - Remove contaminants from bulk polymer stream. - Assess effect of particle size on sort efficiency. - Determine blackness detection limits of machine.

(1) Axplas PS11 (plastic from WEEE). (2) Axplas PS07 (plastic from WEEE). (3) Three size fractions of PS/PE mixture, size classified to <6, >6mm and unclassified (4) Set of PS plaques covering a range of black shades.

- Unable to distinguish between the ABS and PS spectra. - Unable to remove minor contaminants effectively. - Processed all three fractions, >6mm worked best, struggled to separate particles under 6mm. - Unable to determine detection limit for black as none of the test plaques were identified.

Not applicable as not technically viable.

Unlikely to be suitable for use within the WEEE recycling sector as equipment unable to produce products with required specification.


Trial host and equipment


Visys Spyder Laser sorter Remove circuit boards from shredded WEEE.

(1) Shredded WEEE from a primary recycler. (2) Shredded WEEE with extra circuit boards added. (3) Shredded WEEE magnetic fraction from rare-earth magnet. (4) Shredded WEEE non-ferrous fraction from eddy current separator.

- Successful removal of circuit boards. - Recovered up to 99% of circuit boards, typically at concentrations of between 50-60%.

Economics for the technique are weak due to the low market value for circuit boards recovered from WEEE. However removing circuit boards may add some value to the plastic fraction by making subsequent polymer separations easier.

Technically capable of recovering the circuit boards, but the low value of the circuit board fraction under current market conditions may make the machine uneconomical for UK WEEE recyclers. However the machine is already in use by other WEEE recyclers in Europe and as non-ferrous metal prices recover it is likely to become attractive.

Allmineral wet jig

- Recover copper from copper-rich WEEE plastic. - Separate plastic mixtures into density fractions.

(1) Copper-rich plastic fraction from WEEE. (2) Axplas PS07 mixed styrenic plastic chips.

- 77% recovery of copper at 10% concentration which is not commercially viable for copper smelters. - Unable to separate the bulk plastics mixture as the bed did not stratify, preventing the particles from settling to the correct density level.

Not applicable as technically unviable.

Further trials needed to demonstrate higher levels of metal purity and potential for use within the WEEE recycling sector.

Delft University Kinetic Gravity Separator

- Recover copper and other metals from metal rich WEEE plastic fractions.

(1) Copper-rich plastic fraction from WEEE. (2) Heavy copper/metal/plastic fraction from WEEE.

- Good separation for both samples. - 83% recovery of metal at 38% concentration for sample 1. - 95% recovery of metal at 76% concentration for sample 2.

Assuming 1 te/hr capacity and 15% metal recovery from feed to product fraction, with a 70% useful metal content, the payback time is ~15 months.

Technical results promising and economics are viable so has potential for use in the sector to recover metals and copper from WEEE streams.




Holman Wilfley wet shaking table

- Recover copper from copper-rich plastic.

Copper-rich plastic fraction from WEEE in six fractions: (1) ‘as it is’ (8-12mm). (2) 0-2.3mm (granulated). (3) 2.3-5mm (granulated). (4) 0-3mm (hammer milled). (5) 3-5mm (hammer milled). (6) +5mm (hammer milled).

- Technique worked best with size reduced samples. - 0-2.3mm: 95% copper recovered at 75% concentration. - 0-3mm: 56% copper recovered at 90% concentration. - Low levels of copper in other fractions. - Unsuccessful separation on 8-12mm fraction, particle size too large.

Based on a full size table processing 1 te/hr and a copper recovery of 10% of the feed, the payback time is ~15 months.

Technique successful on the size reduced material and demonstrated strong economic case with potential to be used within the WEEE recycling sector to recover copper from plastic.




Nottingham University – pneumatic dry jig

- Recover copper from copper-rich plastic derived from WEEE. - Produce a metal fraction containing less than 5% combustible materials.

Three size classified samples of copper-rich plastic material derived from WEEE: (1) Material granulated to -2.36mm.

) (2) Material granulated to between 2.36-5.0mm. (3) Material with average particle size of 12mm. Typical sample composition: plastic 70% copper/metal 1-5% coated wire ~5% rubber ~10% small quantities of glass, stone, wood and circuit boards.

-2.36mm material: - Separation effective over a range of different system settings. - Bottom jig layers had metal concentrations of between 44 and 52%. - Combustible content of the bottom layer for one of the samples was measured at 13%. 2.36-5.0mm material: - A range of different system settings tested. - Bottom jig layers had metal concentrations of between 7 and 15%. - Combustible content of the bottom layer for one of the samples was measured at 38%. Further testing of 2.36-5.0mm material: - Ten samples were jigged. - Bottom layers were combined and jigged as one. - Bottom fraction then had a metal content of 54%, compared to 7-15% previously. 0-12mm material did not stratify successfully.

Pneumatic jigs are used commercially in coal processing. However no economic assessment was conducted because, although the trial demonstrated that a metal separation could be achieved with careful feed preparation, the University has insufficient scale-up data to allow specification of a full scale production separator for this material. The unit tested in the trial was a small batch rig. Significant further design and test work on a larger continuous pilot rig would be required before this technique could be used at industrial scale for WEEE materials.

- Technical results from the trial were promising considering the experimental jig operated in batch mode, with only single passes. - Trial indicated that pneumatic jigging has potential as a means to recover metals from materials derived from WEEE but further development work is required. It appears that the technique is unable to resolve small density differences successfully and is therefore not suited to separation by polymer type on the basis of density.




Allmineral - Allflux upflow separator

- Recover copper from copper-rich plastic derived from WEEE. - Produce a high purity metal fraction with less than 5% combustible materials present.

Copper-rich plastic fraction from WEEE pre-processed into four samples: (1) Granulated with a 5mm screen. (2) Hammer milled to 95% less then 5mm. (3) -2mm sieved fraction of hammer milled material. (4) +2mm sieved fraction of hammer milled material.

- Granulated material blocked the system and was not suitable for processing. - Reasonably good copper separation achieved with the hammer milled material; however it also blocked the system. The heavy fraction contained 75% copper and 91% of the copper was recovered from the feed. - -2mm fraction separated very well with a copper recovery of 94%. The heavy fraction contained 95% copper with virtually no copper in the middle and lights fractions. - +2mm fraction separated initially but also blocked after a while. The heavy fraction only contained 36% copper with 61% copper recovery.

- Assuming only an inner separating section with a 2 te/hr capacity and 11-12% recover of copper from the feed gives a payback time of approximately 15 months which is economically acceptable.

-Trial results were technically promising but only with the material size reduced to below 2mm. - Economics are viable so the technique has the potential for use in the recovery of metals and copper from WEEE streams.

Pallmann differential impact mill

- Separate plastics into polymer types based on impact strength.

(1) Axion recycled (HIPS) grade PS02 (black) 400 kg. (2) Virgin (ABS) Taita 1000 (white) 150 kg. (3) Virgin (PP) K402 (red) 150 kg. The PP, ABS and PS were moulded and granulated to average particle size of 5mm. (4) Coat hanger clips (mixture of styrene and elastomer) 100 kg, granulated.

- Differential size reduction occurred, but degree of separation achieved was too low to be commercially significant for the sector. - Good at delaminating co-bonded PS and elastomer. - Unable to differentially size reduce the ABS/PP sample. - Possible to delaminate the elastomer from the PS for the coat-hanger samples.

Not applicable as technically unviable at its current stage.

Possibly some technical potential for the machine to be used to separate polymers but still appears unsuitable for commercial use in its current form.





Alvan Blanch - ‘Destoner’ particle size distribution

-Test the techniques ability to separate wood and plastic. - Investigate the effect of particle size distribution on the efficiency of the separation.

- Samples of wood and plastic (PS04) were granulated and sieved into two categories: 0-3mm and 3-8mm - Corresponding wood and plastic categories mixed together in a 10:90 ratio based on volume. Seven fractions then created:

(1) 100% 0-3mm material

(2) 100% 3-8mm material

(3) 90% 3-8mm and 10% 0-3mm

(4) 80% 3-8mm and 20% 0-3mm

(5) 70% 3-8mm and 30% 0-3mm

(6) 60% 3-8mm and 40% 0-3mm

(7) 50% 3-8mm and 50% 0-3mm

- Decrease in the product separation efficiency as the percentage of fines in the feed increased. Most apparent when fines were initially introduced. Effect of the fines on the separation efficiency reduced as more fines were added. - Small particles with a close size range, i.e. 0-3mm can be processed successfully. - Initially the efficiency decreased with the addition of fines, reached a minimum, and then increased again as more fines were added.

- No economic assessment was completed.

- Separation efficiency decreased as percentage of fines increased. - Increased particle size distribution also decreased separation efficiency. - Small particles with a close size range, i.e. 0-3mm, were processed successfully. - In addition material in the size range 3-8mm also processed successfully. Technique has potential to separate wood from plastic, depending on the material particle size and distribution. If the percentage of fines is high and the particle size distribution is over 8mm then not a suitable technique for wood removal.

4.1 Sensor based sorting techniques The project trialled three techniques that can be classified as sensor based sorting. The techniques are all similar in the way they split the feed material into two fractions: an eject fraction and a reject fraction. The following convention is used throughout this report for sensor based sorting trials: the eject fraction is the component of the mixture that is ejected by the air jets in the machine when it

is detected by the sensors; and the reject fraction is the component of the mixture that remains on the belt and is not ejected.

For some sorts the target product is the eject fraction, in other cases it is actually what is termed the reject fraction in this convention. The sensor based sorting techniques tested in this project are: TITECH polysort UHR NIR sorter; RTT UniSort flake NIR sorter; and Visys Spyder (Multi-frequency digital laser sorter).

4.1.1 How sensor based sorting works Sensor based sorting systems start with a feed section which distributes the material evenly in a single layer across a conveyor belt or chute. The material is then transported so that it moves under or in front of a light source. In the case of the NIR technologies, the illuminated object reflects a characteristic infrared spectrum which is unique for each type of material. The laser technique works in a similar way, in that a laser signal is reflected off the material. Different objects reflect different signals from the laser, therefore enabling them to be identified and separated. The next step in sensor based sorting is the detector which receives the signal from the illuminated object. This is then converted into a digital signal, which can be sent to a computer where software identifies the detected material and decides whether the object/particle needs to be removed based on what the system has been set up to eject. If it is decided that the object/particle needs removing, the computer activates the air jets in the machine and the chosen objects/particles will be ejected from the main stream and form the eject fraction. The material which is not ejected falls into the reject fraction. There is usually a splitter plate positioned between these two fractions to keep the fractions apart. Most sensor based sorting machines have been developed for a wide range of uses and applications and have varying degrees of operational flexibility. This type of machine is widely used in the plastic packaging recycling sector to identify whole packaging items by polymer type. The polysort UHR has been developed by TITECH to handle particles as small as 8mm and is set up with ultra high resolution sensors. This means that the equipment should be more suited to processing WEEE materials than previous TITECH machines, as it is generally necessary to reduce WEEE plastics to the size range 8-12mm in order to separate the different components found in WEEE. The TITECH NIR sorter uses a conveyer belt to carry the material from the feed point to the detection area of the equipment. The belt speed needs to be sufficient so that the material spreads out into a monolayer by the time it reaches the sensors, in order for all the particles to be detected and identified by the machine and therefore separated. This means the distance between loading material on to the belt and the detection station needs to be typically at least 3m. The material is carried underneath the NIR unit by the belt and the light source illuminates the material, reflecting the signal upwards. During the time between the particle passing under the NIR and it reaching the end of the belt, the software identifies the polymer type and decides whether to eject the particle or not. As the particles fall off the end of the belt, the air jets, which are positioned just beneath the belt, activate. The eject material is forced forward over the splitter plate by the air jet whilst the reject material falls from the belt onto a conveyor.



Figure 3 Schematic of the TITECH NIR Sorter (courtesy of TITECH)

s

ackground reference strip and the light source. At this point the NIR system detects

nd identifies the material type of the particle. The particles then fall in front of the air jets which activate to

, so

tect differences in the colour and surface finish of a particle n either side. Once the material has passed the sensor it falls in front of the air jets and the target particles are lown across a splitter plate into the eject fraction.

The UniSort flake sorter has been developed recently by RTT as their first near infrared sorter, designed for flakeof plastic. It is claimed by RTT that the machine can process material down to a particle size of 3mm. The UniSort flake sorter is different to the TITECH polysort in that it does not use a belt to carry the material to the NIR sensors. This means that the RTT machine is significantly smaller in size than the TITECH machine. The RTT machine has a vibratory feeder, with the material falling vertically from the feeder. Once the particles arefalling they pass between a baeject the material forward. The Visys Spyder machine is different to the other two techniques in this group, in that it uses scanning lasers in both the optical and infra red spectrum instead of fixed beam NIR technology to detect material types. It measures the reflectance of the fixed frequency laser sources from each individual particle. From this data it canmeasure the colour and surface roughness of the particles as they fall past the detectors. The Visys unit has a vibratory feeder, similar to that used by the RTT equipment, from which the particles slide down a chicane slide and then fall vertically between the lasers and detectors. The machine can be fitted with lasers on both sidesthat as a particle falls both sides can be scanned simultaneously. The system can use several laser frequencies todetermine whether to eject the particle, depending on the mix of signals received at each frequency. As the machine can scan both sides simultaneously it can deob


Figure 4 Schematic of the Spyder Laser Sorter (courtesy of Visys)

or further technical details of the trials, descriptions and photographs of each machine please refer to the

individual trial reports in the appendices.

lastic packaging (using NIR sensors to entify polymer type in whole containers), in metal separation (using induction coils to identify metals) and in

ties

end cted NIR

im that their detectors can differentiate between ABS and PS in the NIR nge. Previous trials by Axion, using laboratory NIR analysers, have confirmed that the spectra can be

product and also contain some valuable copper. Sensor based sorting

sing coloured laser light can be used to detect differences in surface colour and reflectance between circuit

NIR machine has a comprehensive library of material spectra which means it can identify most lastics and its software is particularly flexible. It allows a wide range of materials to be chosen for positive

ct th 80-85% ABS or PS content. Research conducted previously by Axion indicates that the ABS or PS

F

4.1.2 Where sensor based sorting worked well Sensor based sorting is well established in a number of sectors; for pidfood processing (using colour to identify impurities among grains). There are some polymer types in WEEE mixtures which are so similar to each other in their physical properthat they follow each other all the way through the separations normally found in WEEE processes. ABS and PS tend to do this, however unfortunately they are not compatible with each other once they are melted and extruded. If they can be separated then the resulting ABS and PS fractions are significantly more valuable to users. Sensor based sorting offers the potential to separate ABS and PS through differences in their reflespectrum. Both TITECH and RTT claradifferentiated in this spectral range. Circuit boards also have a tendency to follow the main plastic fractions through the WEEE separation process.They are a contaminant in the final plasticuboards and other articles in the mixture. The TITECHpselection. In the trials the TITECH machine was able to separate ABS and PS with some success. It produced produfractions wi


content needs to be at least 95% to prevent issues with brittleness and delamination when the plastic is

on the efficiency of separation was tested on both the TITECH and RTT achines. Both trials indicated that the machines were able to process particles with a size greater than 6mm

(most circuit boards are green on one side and brown on the other). The machine can be set up to ject when it receives both these signals simultaneously. The Visys machine performed well when making this paration.

ze distribution shown in Table 2 is for a sample of Axion grade Axplas PS07 mixed WEEE plastic, which was between 2 and 8mm.

Table 2 Particle size distribution for Axp pic WEEE plastic

extruded. The effect of particle size distribution mbut struggled with smaller particles. The Visys Spyder sorter originates from the food processing sector and it was tested for sorting circuit boards from shredded WEEE. The Visys machine is able to scan both sides of the circuit boards while they are falling from a chute ese

4.1.3 Limitations of sensor based sorting The main challenge for sensor based sorting in WEEE separation is that the particles which need to be separatedare much smaller than those found in plastics packaging recycling applications, typically 8-12mm. Plastic packaging can usually be separated as whole items while WEEE plastics must be size reduced before sorting inorder to separate small electrical components made with different polymer types from each other. The particle siused during the trials. The average particle size is 6mm, with approximately 95% by mass

las PS07, a ty al sample of mixed

Size range % distribution

8.0 - 11.2 5.1%

5.6 - 8 60.8%

4.0 - 5.6 25.1%

2.0 - 4.0 8.8%

<2.0 0.3% The resolution of the detection systems of commercial NIR sorters is still not advanced enough to cope with

s etected and ejected as they

ide down a chute, should be able to achieve higher throughput with less mechanical design complication.

is ifficult for NIR detectors to detect black chips reliably, resulting in detection of only the light coloured chips. The

nge require direct contact with the item to be analysed. FTIR of the spectral signal also requires much more

e he

r machine cost for small WEEE particles, if the detection and analysis speed and

fractions consisting of particles below 6mm; however this is currently being developed. Sorters such as the TITECH polysort unit which require the chips to be distributed in a monolayer on a conveyorbelt require very high belt speeds to achieve a commercially viable throughput when the chip size is small. As the particle size gets smaller this puts greater demands on the design of the belt and the system which distributechips onto the belt. In principle, sorters like the RTT Unisorter, where the chips are dslHowever they also require higher detection and ejection speeds from their software. A further fundamental problem for NIR sorting of WEEE plastics is that the carbon black pigment used to colour around 50% of styrenic plastics in mixed WEEE absorbs strongly in the NIR spectral range. This means that itdpresence of dark coloured chips can be detected but they cannot be separated on the basis of their spectra. Black carbon absorbs much less in the mid infrared spectrum but the FTIR based detectors available for this racomputing power. The result is that mid infra red detection is not practical for continuous sorting machines. In this trial it was found that the RTT sorter was unable to distinguish between ABS and PS so it was not possiblto test a separation of these two material streams. Axion believes that there are two reasons for this; firstly tRTT software contains a smaller library of spectra for styrenic polymers and secondly the fact that the particleshave to be detected and analysed as they fall, leaving little time to complete the spectral analysis before the eject/reject decision is made. In principle a chute-based system like the RTT model should be able to delivehigher throughput and lower


ot been removed. he toner particles create a potential explosion hazard and also coated the material in a layer of dust which may

addition to the ABS/PS separation the TITECH and RTT machines were both tested to remove a target of

and bounce off into the wrong fraction. A knife edge splitter with the split plate angle adjusted to intercept the falling stream of particles in line with their trajectory will improve the precision of the separation significantly. This was a particular issue with the Visys and RTT equipment, but the TITECH unit could also be

he TITECH separation trial demonstrated separation to 80% ABS and 80% PS with three passes. The first pass

three machines would be required to deliver the four pass sorting strategy shown in figure . In each case the reject fractions would be mixed back into other fractions with corresponding compositions,

and Visys sorters to demonstrate whether 95% purity can be

rts would be e volumes of material to be

se sorts if they 1 would require a dedicated machine for the initial split;

hours per year; the ejection jets;

valued at £400/te; overall recovery of about 40% high value material from the feed; and reject fractions retain the same value as the feed as they will still contain about the same proportions of

black styrenic material as the feed.

accuracy can be improved. RTT is working on these issues and expect to have an improved version of the software available in 2009. Dust in the feed material was an issue with both the RTT sorter and the Visys machines during the demonstrationtrials. With the RTT machine the dust became attached to the white background strip where the sensors detect the particles and started to burn under the halogen lights. The material processed by the Visys machine was derived partly from computer printers and it appeared that some of the toner cartridges had nThave affected the sorter’s ability to detect particles accurately. Dust removal should therefore be considered ahead of any sensor based sorter, for health and safety reasons as well as processing ability. Inaround 10% of minor polymer contaminants such as nylon and PMMA from a bulk stream of mixed WEEE plasticchips. Both of the machines performed poorly in this trial. An issue with all three sensor based sorting machines was the splitter plate design. Particles tended to hit the splitter plate

improved.

4.1.4 Economic assessment

Economic evaluations were conducted for the TITECH and Visys sorters. An economic evaluation was not done for the RTT unit because we were unable to demonstrate a significant ABS:PS separation. Tseparated around 50% non-visible black chips from the material. The second pass separated styrenic polymers from the visible fraction and the third pass separated PS from ABS within the styrenic fraction. Based on experience from the trials, the strategy in a production plant is likely to be to eject all visible styrenic polymers from the feed in a first pass, and then to separate just ABS or PS from the styrenic fraction in three further passes. From the throughputs measured in the trials a 2m wide unit should be able to process about 1te/hr of material and 5for example the eject from sort 3 which has an 50:50 mix of ABS to PS would be back mixed in with the visible material from sort 1. Further test work would be required on the TITECHachieved in this way. Specific assumptions for the TITECH machine are: 2m belt width for a capacity of 1te/hr with three machines required to achieve product specification in

four passes based on a separation efficiency of 80% for each pass. Note that a total of 6 sorequired as shown in figure 5Error! Reference source not found. but thseparated in sorts 2 to 6 mean that two machines would have the capacity to perform thewere operated in campaigns. Sort

plant operating 24 hours per day, 5 days per week, 6,000 power consumption about 100Kw, mostly for compressed air to supply mixed feed plastic value £200/te; separated 95% PS and ABS fractions

Figure 5 Process separation sequence for TITECH sorters with a base case separation efficiency of 80% 390 te/yr

384 te PS = 98.5%

6 te ABS = 1.5%

510 te/yr SORT 4

480 te PS = 94% Q = 80%

30 te ABS = 6% R = 80%

120 te/yr ‐ Recycled

750 te/yr SORT 3 96 te PS = 80%

600 te PS = 80% Q = 80% 24 te ABS = 20%

150 te ABS = 20% R = 80%

240 te/yr ‐ Recycled*

120 te PS = 50%

120 te ABS = 50% 390 te/yr

384 te ABS = 98.5%

SORT 2 6 te PS = 1.5%

1500 te/yr Q = 80% 510 te/yr SORT 6

Visible R = 80% 480 te ABS = 94% Q = 80%

Mixed WEEE plastic 50% ABS 30 te PS = 6% R = 80%

FEED 50% PS 750 te/yr SORT 5 120 te/yr ‐ Recycled

3000 te/yr SORT 1 600 te ABS = 80% Q = 80% 96 te ABS = 80%

50:50 split 150 te PS = 20% R = 80% 24 te PS = 20%


120 te ABS = 50%

1500 te/yr 120 te PS = 50%

Non visible

* Assumption that 50% of these streams is non‐visible and when recycled 50% is rejected to the non‐visible stream after sort 1 and 50% becomes product.



Specific assumptions for the Visys machine are: 1.4m chute width for a capacity of 4te/hr (throughput of 1te/hr was measured in the trial but Visys is

confident that this can be increased to 4te/hr in a production machine); plant operating 12 hours per day, 5 days per week, 3,000 hours per year; machine will separate a clean circuit board fraction from mixed WEEE plastic with the same efficiency

that was measured during the trials; power consumption about 30Kw, mostly for compressed air; and circuit board product value of £40/te plus avoided landfill cost of £60/te.

Other assumptions that are common to both evaluations are as follows: overall equipment effectiveness (OEE) of 70%, giving 4,200 effective running hours per year. Note that

OEE is defined in detail in the trial reports in the appendices; ting hour (full job cost of a single operator); and labour cost of £15/opera

power cost 10p/KW hr.

able 3 shows the TITECH and Visys payback calculations.

able 3 Payback calculation for the TITECH NIR sorter and Visys laser sorter

T

T

Trial Titech Visys

EquipmentNIR Sorter (3 units required for feed) Laser Sorter

Capacity te/hr 1 4Cost of plant £ 660000 430000Overall Equipment Effectiveness OEE % 70% 70%Basis of operation hr/yr 6000 3000Plant Input te/yr 4200 8400

Operating CostsPowerConsumption for 3 units kW 100 30Cost (assuming 10p/kW hr) £/hr 10 3

Power costs £/te of feed 10.00 0.75

Power costs £/yr 42000 6300

Labour costs 90000 45000

Total Operating Costs 132000 51300

Revenueall feed becomes oneof the product fractions

5% feed is PCBs

High grade product extracted (42%) te/yr 1764 420Value upgrade for high grade PS and ABS fractions from £200 to £400/te

£40 from PCB sale£60 saving of landfill cost

Value of product £/te 200 100

£/yr 352800 42000

Margin £/yr 220800 -9300

Payback time months 36 -555

Pay back calculation

e achieved in four passes using three machines then the TITECH NIR system

ould pay back in three years.

he payback time for the Visys sorter is not viable.

If the target purity of 95% can bsh T


f 95% ABS or PS would e achieved in three rather than four separation passes for each polymer, see Figure 7. In this scenario only o NIR separators would be required and payback time would reduce to 25 months.

able 4 shows the results of the sensitivity analysis conducted on the NIR sorter.

4.1.4.1 Sensitivity analysis for TITECH NIR sorter If further test work and more intensive ‘training’ of the TITECH NIR sorter was conducted and the separation efficiency was improved by only a few percent to at least 82%, the target specification obtw T

Figure 6 Process separation sequence for TITECH sorters with an improved separation efficiency of 82% 417 te/yr

413 te PS = 99%

4 te ABS = 1%

528 te/yr SORT 4

504 te PS = 95.5% Q = 82%

24 te ABS = 4.5% R = 82%

111 te/yr ‐ Recycled

750 te/yr SORT 3 91 te PS = 82%

615 te PS = 82% Q = 82% 20 te ABS = 18%

135 te ABS = 18% R = 82%


111 te PS = 50%

111 te ABS = 50% 417 te/yr

413 te ABS = 99%

SORT 2 4 te PS = 1%

1500 te/yr Q = 82% 528te/yr SORT 6

Visible R = 82% 504 te ABS = 95.5% Q = 82%

Mixed WEEE plastic 50% ABS 24 te PS = 4.5% R = 82%

FEED 50% PS 750 te/yr SORT 5 111 te/yr ‐ Recycled

3000 te/yr SORT 1 615 te ABS = 82% Q = 82% 91 te ABS = 82%

50:50 split 135 te PS = 18% R = 82% 20 te PS = 18%


111 te ABS = 50%

1500 te/yr 111 te PS = 50%

Non visible

* Assumption that 50% of these streams is non‐visible and when recycled 50% is rejected to the non‐visible stream after sort 1 and 50% becomes product.


Table 4 Sensitivity analysis for TITECH NIR sorter

3 passes 4 passes Separation efficiency (Q) 80% 82% 80% 82% No of machines required to process 0.5te/hr, 6000 hrs/yr

2 2 3 3

Yield 42% 43% 42% 43% Product purity 94.0% 95.5% 98.5% 99.0% Capex £500,000 £500,000 £660,000 £660,000 Pay back Product target

specification not reached so no price upgrade to PS/ABS fractions and no viable payback.

25 months 36 months 35 months

A small improvement to the separation efficiency has positive impact on the payback period for the TITECH NIR sorter.

4.1.4.1 Sensitivity analysis for Visys laser sorter The low value of mixed non-computer circuit boards, the relatively low circuit board content in mixed WEEE plastic and the relatively high cost of the Visys sorter gives it a very long payback time for the proposed application. If the sorter could be used for separation of a feed containing a greater concentration of higher grade circuit boards, for example when processing end of life computers in bulk, then the payback time improves greatly.

Table 5 Sensitivity of the Visys laser sorter payback calculation

Value of

PCB fraction per tonne

£40 £100 £250 £500

% of PCB's in feed

Payback time (months)

5% ‐ 325 65 28 10% 158 62 25 12 15%

69 34 15 8 From a technical point of view removing circuit boards from the WEEE plastic feed will improve downstream separations and also reduce wear in the size reduction equipment. However these benefits alone are unlikely to justify the additional cost of the Visys separator.

4.1.5 Conclusions The TITECH NIR machine performed well overall in terms of achieving separations of recoverable material. It was able to separate PS and PP on all three size fractions tested (0-6mm, 6-15mm and 0-15mm), although the separation efficiency for the 0-6mm material was not as good as for the larger particles. The most important aim to separate ABS and PS was also achieved, though the purity did not reach the 95% target with two passes through the machine. It is possible that processing each of the product fractions in a further one or two passes could achieve the required specification of at least 95% ABS or PS and Axion recommends that this should be tested. The economic evaluation includes the costs for four passes and assumes that this can be delivered in a


production plant using three machines because the processing time for the third and fourth pass will be relatively low. The RTT NIR machine was unable to remove the contaminants to a satisfactory level for this trial and the ABS PS separation was not possible at all, as the computer software could not distinguish between the two signals. The machine was able to separate mixtures of PS and PE. It performed better on the larger size fractions and it appears that its limiting particle size is 6mm. RTT plans further software development and other design adjustments. The Visys sorter is good at separating circuit boards from a mixture of plastics, circuit boards and metal. However, the value of circuit boards from televisions and other non-IT WEEE is low due to the low concentration of precious metals in the circuit boards. The economic evaluation of this equipment indicated that investment in the Visys sorter would not currently be viable for mixed plastics from general WEEE, but it may be viable for material from computer recycling, where the circuit board concentrations and values are higher. 4.2 Shape and density separation techniques The five shape and density separation techniques trialled were as follows: Allmineral Alljig (wet jig); Delft University of Technology Kinetic Gravity Separator (KGS); Holman Wilfley wet shaking table; University of Nottingham pneumatic dry jig; and Allmineral Allflux upflow separator.

The separations based on shape and densities were all wet techniques, apart from the pneumatic dry jig at the University of Nottingham. The following sections explain how the techniques work, the separations that worked well and the limitations of each technique. Economic viability estimates are included for the techniques which performed well.

4.2.1 How the shape and density separations work

The five techniques all exploit differences in shape, density or both in order to achieve a separation. Wet jigs, such as Allmineral’s Alljig, have been used in the mineral processing industry for many years. They separate based on differences in settling velocity. For particles of similar size (over about 1mm) settling velocity is related directly to density. Wet jigs can be used to separate a wide range of particle sizes from 150mm down to 1mm. The size of the particles which can be separated by a wet jig depends on both absolute densities of the particles in the mixture and the densities of the different material types in the mixture relative to each other. The potential attractions of a wet jig for processing plastics from WEEE are: they are high throughput devices, capable of processing many tonnes per hour of material; and they can potentially use water (rather than a heavy liquid medium) to separate materials with a density

greater than the density of water from each other. The need to use heavy media for density separation increases the operating cost and complication of WEEE polymer separation plants so the potential to use water presents a significant advantage.

In a wet jig the material to be separated is suspended in a stream of water and flows across a perforated bed. A pulsed water flow is forced up through the bed to agitate the suspended particles as they flow across it. The pulsed water flow helps the particles to settle past each other and stratify into layers of similar density. The lighter material flows off the end of the bed, via an overflow, and the heavier fractions are captured by splitter plates to produce either a heavy and a middling fraction or just a heavy fraction, depending on the application. The water below the bed may be pulsed by either a mechanical system using a piston or by a pulsating air flow bove the water reservoir. The Allmineral Alljig uses an air pulsation system. a


Figure 7 Schematic of an Allmineral wet jig - front view of a side pulsed wet jig

The Kinetic Gravity Separator from Delft University of Technology has been developed by the recycling laboratory at the University. The device was developed initially for separation of metals from incinerator bottom ash and for separating mixed non-ferrous scrap from eddy current separators into light and heavy alloys. The Kinetic Gravity Separator, like the other techniques tested in this group, has the advantage that it can perform a separation using water. This means it should be cheaper than heavy media or brine separations which require salt or media to form the solutions and washing of the final product to remove the media or brine. The Kinetic Gravity Separator separates on the basis of settling velocity. The machine consists of a vessel which has an internal rotating compartment containing vanes which create small individual compartments filled with water. Material is fed into the top of the vessel and falls vertically through the individual compartments. Dense particles with a high settling velocity fall down through the compartment fastest and exit the system part way round the circumference of the vessel as a heavy fraction. Less dense particles, which have a lower settling velocity, take longer to reach the outlet of the vessel and leave further round the circumference in middle or light fractions.


Figure 8 Cut-away sketch of Kinetic Gravity Separator

Copper, lead, zinc Plastic Aluminium & stone

In this trial the machine was set up to produce heavy, middle and lights fractions but it has the flexibility to separate up to six product fractions at the same time, as the equipment has six product collection compartments at the bottom of the vessel. The Holman Wilfley wet shaking table also originates from the mineral processing industries and has been used in this sector for many years.


Figure 9 Holman Wilfley wet shaking table

Head motion Tailings

(product outlet in buckets)

Product collection trough

Deck with riffles

Additional water can be added to the feed here

Water supply to table Feed trough Feed hopper

Material is fed onto a table which oscillates in a diagonal direction and is fed with a supply of water which flows across the full table in a film. The table has small longitudinal ridges moulded into the surface, known as riffles. The movement of the table and flow of water causes the material to separate. The dense compact particles move in one direction along the riffles whilst the lighter particles, with larger surface areas, flow over the riffles with the water. The table can produce up to five different product fractions simultaneously.


Figure 10 University of Nottingham pneumatic dry jig

At the University of Nottingham a re-appraisal of pneumatic dry jigging has led to the development of an experimental batch-scale jig. The main system component is the jig chamber where the separation takes place. The lower part of the separation chamber is fitted with a removable bottom screen with apertures that will retain the smallest particles of feed material placed upon it, yet will allow sufficient airflow through. During operation, the separation of feed materials is initiated by the flow of air, from an air supply unit (the blower) that runs into the separation chamber. The flow of air into the separation chamber is regulated and controlled through the combined actions of a variable speed control unit (inverter) and two pneumatic butterfly valves. The two pneumatic butterfly valves are time-controlled to open and close in alternating patterns thus giving the device the jigging effect leading to materials stratification. The full scale upflow separator is known as the ‘Allflux’. A smaller (1/8th section) test model was used in these trials and is known as the ‘Miniflux’. The term Allflux has been used throughout this report for easier recognition by readers. The principle of a fluidised bed is utilised by both models to achieve a separation. The advantage of the unit is that it can classify, separate, thicken and de-slime and do this at a high throughout. The original use for the technique was in the sand, ore, coal, heavy mineral sands and slag processing industries. It has not previously been trialled for plastics or WEEE recycling.


Figure 11 Diagram of a full scale Allflux (courtesy of Allmineral)

Valve which lifts to release middle

fraction

Light fraction

Heavy fraction

Middle fraction

The system has two separation stages. The inner section separates the heavy material from the middle/light material, whilst the outer section separates the middle and light materials from each other. The device produces three product fractions; heavy, middle and light. Feed material is suspended in water in a separate vessel. It flows into the top of the machine and down the central core. At the bottom the material encounters an upflow of water. The velocity of the up flowing water is set so that the lighter particles fluidise and are carried upwards whilst the heavy particles are able to settle downwards. During operation the water flow rate is adjusted in order to find the optimal setting for the separation. The heavy particles exit at the bottom of the machine.

The lighter material flows up and over into the outer section of the machine where it encounters another upflow of water. The water velocity in the outer separation section is less than in the inner separation section. The lightest material is fluidised but the middle fraction sinks, as its settling velocity is greater than the water velocity. The light material is carried over the weir and is collected. The middle fraction meanwhile collects in the bottom of the outer section and a valve periodically opens to allow the material to flow out of the bottom of the machine. The upward water flow rate in the outer separation chamber depends on the size and density of the particles to be separated. The machine can be supplied without the outer separation section if it is only required to separate two products (heavy and light). Full technical details of how each of the techniques work, the trial details, images and diagrams of the equipment are documented in individual appendices to this report.


4.2.2 Where shape and density separation worked well The Holman Wilfley wet shaking table was very good at recovering copper and producing a high purity fraction containing over 90% copper. Various size fractions of material were tested on the equipment during the trial. It was clear that the table produced better results with material which had been size reduced to below 5mm. Further size reduction of the material just to extract non-ferrous metal adds significant processing cost, of around £20/te of feed material, because the size reduction step consumes around 120KWh of power per tonne processed plus additional wear and equipment costs of about £8/te of feed. This cost must be balanced against the extra revenue from the copper fraction that is separated. The Delft University Kinetic Gravity Separator was able to process material in the size range 0-15mm, from Axion’s separation plant which had not been size reduced to produce a heavy fraction with a reasonable copper content. Therefore if size reducing material to 3-5mm adds more capital and operating cost than is justified by the material value (which is likely to be the case for metal concentrations in the feed below about 15%), the Kinetic Gravity Separator could be a more suitable option for the separation of metal from plastic. Trials with the University of Nottingham experimental pneumatic dry jig indicate that pneumatic jigging has potential as a means to recover valuable metal rich materials from WEEE streams. However, since the experimental rig was a batch mode system a multi-step processing approach was required. Historically pneumatic jigging units used in the coal and minerals industry had high capacities up to about 50-100 tonnes per hour per metre width. However this technology has not been tried at large scale on WEEE plastic mixtures. Further work would therefore be required to design and develop a continuous pneumatic jig. The Allflux upflow separator was successful at recovering copper from copper-plastic mixtures produced during WEEE recycling, if the particle size of the material was below 3mm. However the feed material must be size reduced in a hammer mill, with high power and wear costs, in order to make it suitable for processing. The copper products from the trial on the hammer milled material met the specification required by metal processors.

4.2.3 Limitations of shape and density separations

Both the Delft University and Holman Wilfley separators will be easier to install if they can be integrated into an existing wet processing system, however many WEEE processors currently operate all-dry separation systems. Substantial investment for water cleaning and containment would be required to add wet processing capability for an all-dry processor. It is difficult to accurately quantify this additional capital cost as it is dependent on the scale of the operation required and the water discharge consents available locally but is likely to be over £100,000. The limitation of the Holman Wilfley table in particular is that its separation is size dependent. The separation results for particles in the size range of 8-12mm were very poor. Size reduction by granulation or hammer milling to a maximum of 5mm is required for this separation to be effective. Size reduction equipment, power and wear costs, particularly as the feed mixture is likely to contain significant amounts of abrasive materials, such as glass, in addition to metal. The Allmineral wet jig trial concluded that it was unable to produce a concentrated copper product and unable to separate a mixed plastic fraction which contained different density plastics. The density differences between the plastic fractions to be separated from each other were too small to create sufficient separation force in the jig bed. In addition the shape of the particles caused too much particle-particle interaction and prevented the bed from stratifying, even with the larger density differences available when separating metal and glass from plastic. Allmineral commented after the trial that a continuous version of the wet jig could achieve a better metal separation than was achieved in the batch unit. This would need to be proven in a further large scale trial on a continuous test rig. The potential advantage of the continuous wet jig is that it should have very high throughput and relatively low operating cost per tonne processed if it can be made to work successfully. Allmineral have one unit working in a WEEE processing application in Europe. The Kinetic Gravity Separator produced a heavy fraction which contained copper but also contained significant amounts of glass, stone and plastic. The plastic content in the heavy fraction was mostly cable insulation material. Conventional copper smelters can handle glass and stone when reprocessing copper but organic materials such as plastic burn and cause excessive gas flows within the furnace. Ideally the combustible material content should be kept below 5%; the Kinetic Gravity Separator however achieved 8%. Further testing would be


required to establish whether the combustible content could be reduced consistently below 5%, perhaps by using a second pass through the machine or by treating the copper-rich fraction using a different process. Another limitation of the trial Kinetic Gravity Separator machine related to the current design of the product offtake system. The pipes which carried the product from the main vessel to the wash down sieves regularly blocked with the materials which were processed during the trial. This severely limited the throughput of the device during the trial. There were a number of reasons for this: the water velocity in the product offtake pipework was too low to convey the heavier metal fractions; the system was constructed from corrugated flexible plastic pipe which generated high pressure drop

and created numerous trap points for material; and there were too many bends and low points in the pipe work, which created further opportunities for

blockages to develop. It should be relatively straightforward to overcome these problems in a production version of the separator. The dry jigging trial was carried out on the experimental Nottingham University dry jig. Dry jigs are used in a limited number of applications in the coal and minerals industry. However scale-up and design information for other applications is currently uncommon. The main operating costs are likely to be energy used for vibration, air supply plus dust extraction. Further work would be required to design and develop a continuous pneumatic jig. The Allflux separator was not suitable for processing granulated WEEE material that had been size reduced by granulation as it blocked the system very quickly when wire fragments tangled in the separation ducts. The hammer milled material, in its original unsieved form, initially worked but then blocked the system. When the hammer milled material was sieved the -2mm sieved fraction produced the best results of the samples trialled. An Allflux separator can therefore be used to recover copper from copper-plastic mixtures produced during WEEE recycling if the feed material is size reduced in a hammer mill to a particle size of the material below 3mm. Accurate density separation of WEEE plastics requires a density resolution of better than 20kg/cubic metre. None of the shape and density separation techniques that were trialled in this project proved to have the ability to resolve density differences that were small enough to allow separation of WEEE plastics by polymer type. If this could be achieved it would create a major cost and throughput advantage for WEEE processors because both separation techniques are low cost and high volume and avoiding un-necessary size reduction of plastics cuts power and wear costs.



4.2.4 Economic assessment Technical results for the wet shaking table, Kinetic Gravity Separator and Allflux separator were promising. Economic evaluations were therefore conducted for these units – the wet shaking table and Kinetic Gravity Separator are presented together, followed by the Allflux unit assessment. No economic assessment of the wet jig was conducted because the separation trial results for WEEE material were not considered acceptable. As the Nottingham University dry jig was a batch rig, for the technique to have industrial relevance and to be able to conduct an economic assessment, work is required to design and develop a continuous pneumatic jig. It was not possible therefore to make an economic assessment for this unit. Specific assumptions for the Holman Wilfley machine are: table sized for a capacity of 1te/hr (one full size Holman Wilfley table); plant operating 12 hours per day, 5 days per week, 3,000 hours per year; power consumption about 10Kw, for table drive and water pumping plus 40Kw for granulation of the

feed; capital cost includes a dedicated granulator with a fine screen, wet shaking table, associated tanks,

screens and water cleanup system; mixed metal/plastic feed sent to landfill for final disposal at present at a cost of around £60/te; separated metal-rich fractions valued at £1,000/te; overall recovery of about 10% metal-rich from the feed; and reject fractions sent to landfill for disposal at around £60/te or sent for further processing to recover

separate glass and plastic fractions.

ravity Separator are:

nd osal at around £60/te or sent for further processing to recover

ing hours per year; job cost of a single operator); and

power cost £10p/KWhr.

Specific assumptions for the Delft University Kinetic G 2m diameter unit with a capacity of 1te/hr; plant operating 12 hours per day, 5 days per week, 3,000 hours per year; power consumption about 15Kw, mostly for rotor drive and water pumping; mixed metal/plastic feed sent to landfill for final disposal at present at a cost of around £60/te; separated metal-rich fractions valued at £700/te due to lower metal content of the heavy fraction; overall recovery of about 15% metal-rich from the feed. The higher recovery percentage of heavy

fraction than the Holman Wilfley table is due to the greater proportion of non-metals in the product; a reject fractions sent to landfill for disp

separate glass and plastic fractions. Other assumptions that are common to both evaluations are as follows: overall equipment effectiveness (OEE) of 70%, giving 2,100 effective runn labour cost of £15/operating hour (full


Table 6 Payback calculation for Holman Wilfley wet shaking table and Delft University Kinetic Gravity Separator

Trial Holman Wilfley DelftEquipment Wet Shaking Table Kinetic Gravity Separator

plus dedicated granulator for fine size reduction

(does not require further size reduction)

Capacity te/hr 1 1Cost of unit £ 180000 200000

Basis of operation hr/yr 3000 3000Overall Equipment Effectiveness OEE % 70% 70%Plant Input te/yr 2100 2100

Operating CostsWaterConsumption kg/hr 100 100Cost (assuming £2/te) £/hr 0.20 0.20

PowerConsumption KW 50 15Cost (assuming 10p/kW hr) £/hr 5 1.5

Water and Power costs £/te of feed 5.20 1.70Water and Power costs £/yr 10920 3570

Wear costs for granulator £/te feed 6£/yr 12600

Labour costs (assuming £15/hr) 45000 45000

Annual process licence costs 0 10000

Total Operating Costs 68520 58570

Revenue

Assuming 10% of feed is separated as product containing 95% metal

Assuming 15% of feed is separated as product containing 70% metal

Product extracted te/yr 210 315Assuming only 70% of the product is useful metal with a value of £1000/te

Value of metal product £/te 1000 700

£/yr 210000 220500

Margin £/yr 141480 161930

Payback time (months) 15 15

Pay back calculation

Both the wet shaking table and the Kinetic Gravity Separator are projected to pay back their investment in just over a year, provided they can process at least 2,000te/yr of feed material containing at least 10% useful fine

ty

hich n the

romising technical performance the economic potential was assessed using a payback calculation.

non-ferrous metal. The payback period increases to three years for the Holman Wilfley table and two years for the Delft UniversiKinetic Gravity Separator if the useful metal content in the feed drops to 5% from the current level of 10%. The results from the Allflux trial show that this separation technique has the potential to create a product wmeets the desired specification of less than 5% combustible material from copper smelters. Based op

The Allflux separator can be built without the outer separation chamber in which the lights and middles are separated from each other. The separation performed in the outer section ultimately has no commercial benefit because the mixed heavy plastics are worth very little once the copper fraction has been removed. The trials demonstrated that the feed material must be size reduced to at least 3-4mm. Therefore size reduction prior to the Allflux is required, ideally by a hammer mill. After the Allflux the material must be dewatered. Figure 12 shows a flow diagram for the full separation system that is modelled in this economic analysis.

Figure 12 Process flow diagram for separation process using an Allflux upflow separator

The payback calculation includes the equipment costs and installation of a hammer mill, 2mm sieve and dewatering screens along with the Allflux separator. Table 7 shows a breakdown of the capital cost for the various pieces of equipment required, along with the installation factor.

Table 7 Capital cost breakdown for Allflux upflow separator

Equipment Cost (£) Installation factor (%)

Hammer Mill 180,000 50 Allflux Separator 45,000 100 Sieves (2mm x 1, dewatering x 2)

45,000 100

Total 450,000 Two scenarios are modelled for treatment of the +2mm fraction generated by the sieving step after the hammer mill. Either: Scenario 1: The sieve size after the hammer mill could be increased to 2.5mm and the +2.5mm fraction

could be discarded without further processing because the copper content of this fraction is relatively small; or

Scenario 2: The +2mm fraction could be re-circulated round the hammer mill to be sized reduced further and then processed in the Allflux.

Water recycle

Hammer mill

Allmineral Allflux

Dewatering screens

2mm sieve PRODUCT FEED



Table 8 Payback calculation for Allflux upflow separator Pay back calculation

TrialEquipment

Scenario 1 Scenario 2

process only the ‐2.5mm in Allflux

recycle +2mm fractionthrough mill and process in Allflux

Capacity te/hr 2 2Cost of equipment £ 450000 450000Basis of operation hr/yr 3000 3000Overall Equipment Effectiveness OEE % 70% 70%Plant Input te/yr 4200 4200

Operating CostsWaterQuantity kg/hr 150 200Cost (assuming £2/te) £/hr 0.3 0.4

PowerQuantity ‐ Upflow Separator kW 20 20Quantity ‐ Hammer mill kW 200 260Cost (assuming 10p/kW hr) £/hr 22 28

Water and Power costs £/te of feed 11.15 14.2Water and Power costs £/yr 46830 59640

Wear costs for hammer mill £/te feed 6 8£/yr 25200 33600

Labour costs (assuming £15/hr) 45000 45000

Annual process licence costs 0 0

Total Operating Costs 117030 138240Revenue

Product extracted te/yr 462 504

Value of product £/te 1000 1000

£/yr 462000 504000Margin £/yr 344970 365760

Payback time months 15.7 14.8

AllmineralAllflux Upflow Separator

Assume 11%copper recovered

Assume 12% copper recovered


For both scenarios the calculation assumes a capacity of 2 tonnes per hour, using only the inner separating

ction, and a total installed cost for all the equipment of £420,000.

an overall equipment effectiveness (OEE) of 70%, where e OEE is defined as follows:

se The plant is assumed to operate 12 hours per day, 5 days a week, 50 weeks a year, giving a total of 3,000 hours of operation per year. The payback calculation assumes th

he overall plant throughput will be 4,200 tonnes per year.

that

hr. The calculation assumes that the system will require one operator at a total job cost f £45,000 per year.

000 per tonne. The revenue is £462,000, ith a margin of £344,970, which gives a payback time of 16 months.

is produced which gives a revenue of

504,000 and a margin of £365,760. The payback time is 15 months.

nd the shorter payback time takes effect when the versized fraction is recycled through the hammer mill.

4.2.5 Overall conclusions of the shape and density separation trials

aterial. Further demonstration trials ould be required to show achievement of higher levels of metal purity.

y of the non-ferrous metal product is igh and the economics for the separation are promising.

sor. unit would need to be redesigned for a full scale production unit, particularly the product

fftake system.

Since

in from WEEE streams, further work would be required to design and develop a

ontinuous pneumatic jig.

T Operational costs The feed material is currently sent to landfill and once the copper has been recovered the residual material will still be sent to landfill, so there is no net cost or benefit from disposal of the residue fraction. It is assumed10% fresh water is lost per tonne of material sent to landfill with a water cost of £2/te. The power cost is assumed to be 10p/kWo Revenue, margin and payback time For scenario 1, where the oversized sieve fraction is discarded, the revenue estimate assumes that 11% of the copper is extracted from the feed. This produces 462 tonnes of copper per year. The value of this material will fluctuate with market conditions but is assumed to be approximately £1,w In scenario 2, where the oversized sieve fraction is recycled through the hammer mill, it is assumed that 12% ofthe copper is recovered from the feed. 504 tonnes of high purity copper£ The difference between the two scenarios is very little ao

The Allmineral wet jig technique did not work well on the mixed plastic mw The wet shaking table is a well-established technology and can be used to recover metals from a metal/plastic mixture from WEEE, provided it is size reduced to below 5mm. The purith The Kinetic Gravity Separator was still in the research stage but it achieved a reasonable separation without needing the feed to be size reduced below 5mm. This would save on cost and complexity for a WEEE procesSome parts of theo The University of Nottingham dry jig is very much at the experimental stage but the work indicates that pneumatic jigging has potential as a means to recover valuable metal rich materials from WEEE streams.the test pneumatic jigging unit involved a dry process there would be no associated water recovery and treatment costs, but dust extraction costs are a potential factor. For industrial and commercial relevancerecovery of metals/copper c

The Allflux separator is a high throughput low operating cost device that can be supplied without the middlings/lights separator, lowering the overall cost. This arrangement would be more suitable for processing copper-plastic mixtures from WEEE. Based on the trial results, the Allmineral engineer estimated that a full scale unit with this configuration should be able to separate up to 12te/hr of feed material. 4.3 Separation by differential impact milling Using differential impact milling as a polymer separation technique is a new concept which has not been trialled before.

4.3.1 How differential impact milling works Differential impact milling aims to exploit differences in the impact strengths of the materials being separated. Materials with high impact strength are less likely to shatter in a high intensity impact mill than materials of low impact strength. This should create a difference in particle sizes between the different polymer types and allow them to be separated from each other by sieving after the mill. The impact mill used for the trial was a Pallmann PXL18.

Figure 13 Photograph of the Pallmann - PXL18 Mill

This is a rotor-stator mill with two rings of elements on the rotor, four on the inner ring and eight on the outer ring. There is a single ring of eight elements on the stator. As the parts of the mill rotate, the pins pass each other and pulverise the material. The speeds at which the rotator and stator can rotate are limited by the strength of the pins and the potential for blinding of material between the closely spaced pins.

Figure 14 Photograph of the PXL18 Mill Rotor (8 elements on outer ring and 4 elements on inner ring) and Stator (8 element ring)


The maximum throughput which can be achieved is highly dependent on the material type and particle size distribution of the feed. This trial tested several different polymer mixtures in order to screen a range of possible applications for the separation technique. Each type of material was made of a different colour so as to make subsequent identification of material type simpler. Two groups of materials were trialled: pairs of thermoplastics with different impact properties; and thermoplastics and elastomers.

This technique is potentially interesting for ABS and PS, which are often found as comingled materials and are difficult to separate with established techniques. The aim was to find combinations of polymer types where the difference in impact properties was sufficient for impact milling to create a significant difference in particle size between the two polymer types such that they could be separated efficiently by sieving. For the differential impact milling technique to be technically viable it is a prerequisite that one of the components fractures more readily than the other. Please refer to the separate trial report appendices for full technical details.

4.3.2 Where differential impact milling worked well The trials demonstrated that a very large difference in impact strength is required in order to achieve viable separation efficiency. Most of the trial runs demonstrated only a modest separation on the basis of impact strength. However the trial that separated a rubber/HIPS composite material showed promising results.

4.3.3 Limitations of differential impact milling

For most polymer types likely to be of commercial interest in WEEE separation, this technique did not create polymer fractions which were sufficiently different in particle size to allow an effective separation to be possible.

4.3.4 Economic assessment The trial results were not conclusive enough to justify conducting an economic evaluation of this separation technique.

4.3.5 Conclusions

Differential impact milling is not likely to be a useful technique for separation of most WEEE polymer mixtures; however it may have potential for separation or at least delamination of mixed thermopolymers and elastomers. One of the trials compared the number of free elastomer particles to the number of elastomer particles that were still bonded to a piece of PS after each pass, showed the mill’s ability to delaminate the elastomer from the PS.


4.4 Effect of particle size distribution This work describes trials on a destoner, sometimes referred to an air-table, held at Alvan Blanch, Wiltshire. The purpose of the trial was to monitor the effect of differences in particle size distribution on the efficiency of separating wood particles from plastic. The need to separate these two fractions from one another is commonly encountered in WEEE recycling. Air tables were tested successfully for this application during the previous DEFRA project WRT095. The aim of this trial therefore was not to prove the capability of an air table, but to investigate how its separation performance changes with different feed size distributions.

4.4.1 How the Alvan Blanch Destoner air table works

Figure 15 Alvan Blanch DS2 Destoner (photograph courtesy of Alvan Blanch)

The Destoner is designed to separate stone and metal particles from lower density and/or longer/thinner feed materials. The separation is possible due to the differing density and shape of each type of material. Lighter, smaller particles are generally fluidised by the air flow and hence are separated from the heavier, larger particles. Using a vibrating bed the feed material is stratified through a combination of the vibrations and the air flow from beneath the bed. The bottom layer is in contact with the bed resulting in an uphill throw motion which causes the heavier particles to climb up the deck until they are discharged. The effect of gravity causes the light particles to travel down the vibrating bed. Examples of materials which are regularly processed on destoners to remove stones and other heavy particles include cereals, peas, beans, lentils and coffee beans.

4.4.2 Trial results The full trial report is included as a separate appendix. The main conclusions of the work are: particle-particle interactions play an important role in the air table separation (as for many other WEEE

separations). For material with a narrow particle size distribution interactions between particles (causing heavy particles to be trapped on top of light particles and vice versa) have less of an effect on the overall separation. Particles with a close size range were separated successfully;

the trial showed a decrease in the separation efficiency for the product fraction as the percentage of fines in the feed increased;

initially the separation efficiency decreased with the addition of fines until it reached a minimum. It then increased again as more fines were added. This is likely to be because fines tend to be carried on larger particles initially, but once they reach a critical percentage (which will depend on the relative size and


if the percentage of fines is high and the mean particle size is over 8mm then the Destoner is not a suitable technique for wood removal.

4.4.3 Conclusions The aim of this trial was not to test the effectiveness of an air table to separate wood from plastic as this had previously been demonstrated in Defra project WRT095, but to study the effect of particle size distribution on separation efficiency for a typical WEEE separation. Therefore no economic assessment was performed on the technique. As the percentage of fines increased, the separation efficiency decreased. The increased presence of smaller sized particles, and the wider particle size distribution, had a negative effect on the separation ability of the Destoner. Particles with a close size range, i.e. 0-3mm or 3-8mm were processed more successfully than material in the wider size range 0-8mm.


5.0 Overall assessment of economic and technical potential for WEEE

processing Five of the processes assessed in this project have demonstrated technical potential for use in WEEE recycling and could be integrated into existing WEEE separation systems. The TITECH NIR sorter was unable to produce fractions of PS and ABS suitable for commercial reprocessing from mixed WEEE polymers with two separation passes. However, if it can be demonstrated it can produce 95% PS and ABS products with three passes, then there is a strong economic case for investment in this technology, with a payback time of about two years. The Visys laser sorting machine efficiently separates circuit boards from WEEE plastic mixtures but it is difficult to justify investment at the current market value for lower grade circuit board material. However, the separator could be of interest to larger scale recyclers of computer equipment. The Delft University Kinetic Gravity Separator has the potential to separate metals, glass and wire from WEEE plastic with only a few relatively minor modifications to the design. Provided the technical issues can be resolved there is a good economic case for investment in this separation technique, with a forecast payback time of about one year. A particular advantage of the Kinetic Gravity Separator is that it produced a reasonably good non-ferrous metal concentrate without the need for further size reduction of the feed material. The Holman Wilfley wet shaking table and the Allmineral Allflux upflow separator have the strongest technical and economic cases of all the techniques evaluated. The trials of both these techniques demonstrated that they can produce good quality copper concentrates from mixed WEEE plastics and fine metals, provided the full mixture is size reduced to below 3-5mm. Both have predicted payback times of 15 months, even when the additional capital and operating costs of size reduction of the feed material are taken into account. The TITECH and Visys sensor based sorters are particularly simple to integrate into existing plants because they are dry separation techniques. The Holman Wilfley, Allmineral Allflux and Delft University Kinetic Gravity Separator are wet techniques which would fit best into processes where the material is already wet from other upstream processes. The Visys machine would fit in either primary WEEE processors or secondary WEEE plastic processors in order to upgrade their plastic fractions and extract value from the circuit board component applicable to IT derived WEEE, which is otherwise a contaminant in the plastic stream. The TITECH machine is more suited to a secondary processor’s operation where the WEEE material has been shredded into chip form and has undergone various separations to produce a concentrated styrenic polymer stream. This would upgrade the ABS/PS mixture into two higher-performing separate polymer grades. The Holman Wilfley, Allmineral Allflux and Delft University machines were used to test the separation of metals from WEEE plastics. These techniques would fit within a secondary process to recover fine copper wire and glass from mixed WEEE plastics. The Holman Wilfley table requires additional size reduction to below 5mm to be effective. The Kinetic Gravity Separator technique has also been used at Delft University to separate light and heavy non-ferrous metals from each other, so would possibly fit with processors who produce mixed non-ferrous metals from eddy current separators. The Nottingham University dry jig showed good potential for dry separation of fine metals from WEEE plastics. Further test work would be required at a larger scale to optimise the design and adapt it for WEEE material separation.


6.0 Conclusions Overall the project trialled ten potential techniques for separating materials produced during WEEE recycling. The trials showed that four of these techniques have both good commercial and technical potential for use in recycling WEEE. At the time of delivering the trials the other techniques did not demonstrate sufficient technical capability to be potentially useful to the WEEE recycling sector but this could change in future with further improvements to the separators and changes to material prices and landfill costs.

Table 9 Summary of the technical and economic feasibility of the demonstrated techniques Technique Technical feasibility Economic feasibility TITECH polysort NIR sorter Yes Yes RTT Unisort Flake NIR sorter No No Visys Spyder laser sorter Yes No Allmineral wet jig No No Delft University Kinetic Gravity Separator Yes Yes Holman Wilfley wet shaking table Yes Yes University of Nottingham dry jig Yes No Allmineral Allflux upflow classifier Yes Yes Pallmann differential impact mill No No Alvan Blanch Destoner Yes Not applicable 6.1 Technically and commercially feasible techniques The four techniques that have been evaluated as being both technically and commercially feasible are: TITECH Polysort NIR sorter; Delft University Kinetic Gravity Separator; Holman Wilfley wet shaking table; and Allmineral ‘Allflux’ upflow classifier.

The TITECH Polysort NIR was able to handle particles in the 8-10mm size range commonly produced during WEEE recycling. The machine is programmed with a comprehensive library of polymer types and the software can be adjusted to give a wide range of sorting options. The trials showed that the machine was able to identify and distinguish between ABS and PS; however the separation is not good enough in a single pass. The trials achieved 80:20 ABS:PS and PS:ABS mixtures in two passes, starting from roughly 50:50 feed material. The machine needs to produce fractions with at least 95% ABS or PS in order for the product to have significantly better physical properties than the feed material. Assuming that this can be achieved in three or four passes, the economic analysis indicates a healthy investment payback time of about two to three years. The Delft University of Technology Kinetic Gravity Separator is a novel design which is based on a robust separation principle. The machine is still at the research and development stage but is close to commercial viability. Redesign of parts of the system would be required before it can be used commercially on WEEE materials, specifically the product collection system. It was able to concentrate the copper into heavy fractions at a concentration close to the trial target with no pre-treatment of the feed. There is good potential to increase the purity of the metal fraction with further optimisation of the machine settings. Estimated payback time for this machine is just over a year. The Holman Wilfley wet shaking table is a well-established technique that is used widely in the mining industry. The trial showed that the machine was able to recover a good quality copper concentrate from WEEE plastic mixtures, provided the material was size reduced to below 5mm. The machine would fit well as a final recovery unit for a secondary processor to recover the fine copper that would normally be sent to landfill for final disposal. The economic assessment is very promising as the estimated payback time is approximately 15 months. The Allmineral Allflux upflow classifier trial results were technically promising, but only with the material size reduced to below about 2mm with a hammer mill. Even including the cost of a hammer mill, the economics are viable so the technique has potential for use in the recovery of fine metals from WEEE streams. The unit could


be used without the outer section – reducing the capital cost – and assuming a 2 te/hr capacity, plus 11-12% copper in the feed material, the payback period is acceptable at approximately 15 months. 6.2 Technically feasible techniques The Visys Spyder laser sorter is technically feasible and separated circuit boards from WEEE efficiently. The machine allows for flexibility and can be tuned to allow different laser signals to trigger ejection of the target material, depending on which combination of signals is strongest or most distinctive. The machine costs between 180,000 - 250,000 Euros, which is higher than other pieces of equipment tested in this project. The economic evaluation shows that the payback time is highly dependent on the concentration of circuit boards in the feed and the prevailing market value of the circuit board fraction. At current circuit board values the payback time is too long to be commercially viable for most WEEE treatment facilities or secondary WEEE processors, however the separation may be commercially viable for ICT recyclers, whose feed material tends to contain a greater percentage of high value circuit boards. Trials for the University of Nottingham dry jig indicate that pneumatic jigging has potential as a means to recover valuable metal rich materials from WEEE streams. However, since the experimental rig was a batch mode system a multi-processing approach was required. Historically pneumatic jigging units used in the coal and minerals industry had high capacities up to about 50-100 tonnes per hour per metre width. As this is a dry process there are no water treatment and recovery costs. Further work would be required to design and develop a continuous pneumatic jig suitable for WEEE separation. With the Alvan Blanch Destoner, small particles with a close size range i.e. 0-3mm were processed successfully, as was material in the size range 3-8mm. However, wider size distributions from 0-8mm were less successful. 6.3 Unsuccessful techniques Three of the techniques demonstrated were found to be technically unfeasible for the WEEE recycling sector at the time of trialling. The RTT Unisort NIR flake sorter did not perform adequately on the WEEE sample materials. In its current form it is unable to identify and distinguish between ABS and PS. It was able to handle the 8-10mm WEEE particle size samples and performed well on separating the PS PE mixture. The machine however, is still being developed so it may soon have the capacity to separate ABS and PS but this would require further testing to confirm. Use of the Allmineral wet jigging machine is well established in the mining sector and it has potential to allow separation of metals and plastics with a density greater than 1, using water. However, laboratory scale trials showed that the relatively small density differences between polymer types create insufficient driving force to allow plastic from plastic separations. Differential impact milling is a novel separation application for WEEE plastics which was tested for the first time in this project; however the trial was largely unsuccessful. A limited degree of size differentiation was achieved with mixtures of polystyrene and elastomer, but the differences were insufficient to allow a useful separation by size classification after milling.


7.0 Recommendations The recommendations from the results and findings through the course of the trials are as follows: Further testing should be conducted with mixed WEEE plastic with NIR sorters in order to confirm whether the target of 95% ABS or PS in the product streams can be achieved in two or three passes by re-sorting the product fractions from the primary sort. Separation of ABS from PS is of great interest to WEEE recyclers as it can improve the physical properties and value of both materials significantly. Further trials to build on the promising results achieved in the trials at TITECH would be very helpful. The core separation element of the Delft University Kinetic Gravity Separator worked well but the system for removing and drying the product fractions would need to be improved for continuous operation on WEEE materials. The design of the product offtake system is also a key factor and would need modification in a production version of the equipment. Further trials on this separator would be valuable, particularly as the trials demonstrated the technique does not appear to require further size reduction of the metal-containing material prior to the separation. Further trials should be carried out on the design and development of a continuous pneumatic dry jig, based on the University of Nottingham’s batch test rig. A dry separation technique for the recovery of metal from copper-rich WEEE streams would provide significant advantages over a wet technique by eliminating the need for drying steps.



Appendices Appendices are available to download with this report on WRAP’s website at www.wrap.org.uk/separationofWEEEplastics Appendix 1 - TITECH Polysort UHR NIR sorter

Trial plan Trial report

Appendix 2 - RTT Unisort NIR Flake sorter


Appendix 3 - Visys Spyder laser sorter


Appendix 4 - Allmineral Alljig


Appendix 5 - Delft University of Technology Kinetic Gravity Separator


Appendix 6 - Holman Wilfley wet shaking table


Appendix 7 - Pallmann Differential Impact Mill


Appendix 8 – Allmineral Allflux upflow classifier


Appendix 9 – University of Nottingham pneumatic dry jigging


Appendix 10 – Alvan Blanch Destoner air table


http://www.wrap.org.uk/separationofWEEEplastics

www.wrap.org.uk/separationofWEEEplastics

Final report Separation of mixed WEEE plastics

Documents

Transcript of Final report Separation of mixed WEEE plastics