Biogenic carbon-enriched and pollutant depleted SRF from commercial and pretreated heterogeneous waste generated by NIR sensor-based sorting

Abstract

Mechanical processing using predominantly particle size and density as separation criteria is currently applied in the production of solid-recovered fuel or refuse-derived fuel. It does not sufficiently allow for the optimization of the quality of heterogeneous solid waste for subsequent energy recovery. Material-specific processing, in contrast, allows the separation criterion to be linked to specific chemical constituents. Therefore, the technical applicability of material-specific sorting of heterogeneous waste, in order to optimize its routing options, was evaluated. Two sorting steps were tested on a pilot and a large scale. Near infrared multiplexed sensor-based sorting devices were used (1) to reduce the chlorine (Cl) respectively pollutant content, in order to broaden the utilization options of SRF in industrial co-incineration, and (2) to increase the biogenic carbon (C_bio) content, which is highly relevant in the light of the EU emission trading scheme on CO₂. It was found that the technology is generally applicable for the heterogeneous waste fractions looked at, if the sensor systems are appropriately adjusted for the sorting task. The first sorting step allowed for the removal of up to 40% of the Cl freight by separating only 3 to 5% of the material mass. Very low Cl concentrations were achieved in the output stream to be used as solid-recovered fuel stream and additionally, the cadmium (Cd) and lead (Pb) concentration was decreased. A two- to four-fold enriched C_bio content was achieved by the second sorting step. Due to lower yields in the large-scale test further challenges need to be addressed.

Keywords

Solid recovered fuel biogenic carbon near infrared sensor-based sorting material-specific waste processing pretreated waste commercial solid waste refuse-derived fuel

Introduction

In terms of climate change impact, the use of waste-derived material for recycling and energy recovery is a preferable waste management option in comparison with pure disposal (e.g. Dehoust et al. (2010); Friege and Giegrich (2008); Ragoßnig et al. (2009)). This is underlined by the European ‘waste hierarchy’ (EP, 2008), which is aimed at prevention over reuse, recycling, recovery and finally disposal, in order to support environmental protection. Energy recovery is a relevant option for heterogeneous waste, which is characterized by a highly variable material composition and quality, and which can only be recycled with extensive effort.

The (co-) incineration of solid recovered fuel (SRF) – referring to a fuel meeting defined quality specifications (Rotter et al., 2011) – used as substitute for fossil fuel in industrial facilities (e.g. cement industry, pulp and paper industry) is a possible energy recovery option. In Austria, specifications to be met by SRF (e.g. heavy metal content) are officially defined in the waste incineration directive (BMLFUW & BMWFJ, 2010). Compliance with additional parameters, especially concerning the chlorine (Cl) content as well as fuel properties such as the lower heating value (LHV), is requested by the plant operators.

Despite recent changes in the waste incineration market (falling gate fees due to overcapacities) (see for example Friege and Fendel (2011)) the (co-)incineration of SRF in industrial facilities is still preferable over ordinary waste incineration from an economic point of view for the waste contractor (see Ragossnig and Faist (2009) for the situation in early 2009). This might be affected in the future by new regulations in the EU CO₂ emission trading scheme (EU-ETS) and its influence on the energy and resources market. In terms of SRF utilization, the EU-ETS is currently mainly affecting the cement industry (e.g. Pomberger and Abl (2008) and Pomberger et al. (2008)) but might have an influence on waste incineration and potentially other areas in waste management as well in the near future (Schlupeck, 2011). Consequently, the partly renewable nature of SRF and especially the technical ability to influence the proportion of biogenic carbon (C_bio) along with the Cl and heavy metal content will gain in importance in the processing of heterogeneous waste for the waste contractor. These interdependencies of the waste sector with the energy and raw material sector need a maximum of flexibility from waste contractors in order to be able to optimize the routeing of waste streams in terms of waste quality, specifications to be met for recovery and treatment options as well as the prevailing market situation.

Usually, basic mechanical processing such as crushing and shredding, classification and ballistic separation, using predominantly the particle size and density as separation criteria, is performed to split heterogeneous solid waste. Specific material characteristics, which can be linked to chemical constituents, are not yet widely applied as separation criteria although basic mechanical processing steps are not sufficient to influence the chemical characteristics of heterogeneous material, especially concerning the Cl content (Rotter et al. (2004)). The same is true for the C_bio content, as in order to selectively generate a C_bio-enriched waste stream out of a heterogeneous input, material-specific processing is required.

Material-specific characteristics can be used as separation criteria by sensor-based sorting – a single particle separation on the basis of identifiable criteria measured by suitable detectors. Consequently, in comparison with the above-mentioned separation technologies, completely new sorting tasks – increasing the functional efficiency of the separation – can be fulfilled (see Christiani (2006)). Additionally, this technology can be easily adapted for varying needs. First practical results show that sensor-based sorting, which is already a state-of-the-art technology for the processing of separately collected recyclables in material recycling, is potentially capable of splitting heterogeneous waste materials as well (Christiani, 2006; Nisters, 2006; Ragossnig and Faist, 2009; Titech, 2010).

In the present study, the technical feasibility of the integration of near infrared (NIR) sensor-based sorting in a mechanical treatment (MT) plant in order to split heterogeneous – mainly commercial and pretreated – solid waste is analysed. The additional treatment steps should facilitate refining the utilization, treatment or disposal options for the output streams, by optimizing the (1) pollutant (Cl, heavy metals) and especially (2) the C_bio content of SRF. In this way, the marketing opportunities (flexible and economically optimized waste routeing) for the waste contractor, for example, based on the above-mentioned influences of the EU-ETS, should be improved.

Material and methods

The technical applicability of NIR sensor-based sorting in a MT plant for the splitting of heterogeneous solid waste was evaluated by processing trials.

The waste streams addressed in the processing trials were selected after a detailed analysis of a specific MT plant, processing mainly commercial and pretreated solid waste (60 000 t year⁻¹ in 2008). The waste streams were obtained from the input material after presorting, metal separation, air classification, screening and heavy fraction separation. The material composition (Figure 1) of the addressed waste streams was analysed four times during a 1-year period (November 2009 to August 2010) and additionally, the material used for the processing trials [three pilot-scale test series (→ ‘pilot’) and one indicative large-scale test run (→ ‘large’)], was characterized in order to confirm that the sorting trials were conducted with representative material.

The waste streams are classified in the following manner.

High calorific fraction (HC, LHV >20 MJ kg⁻¹_dry, particle size >120 mm, 20%_m of plant input): the HC contains mainly fossil material (bright and dark plastics, about 40–70%_m) and a smaller part of biogenic material (undefined organics, wood, paper and cardboard, about 20%_m) based on the wet mass (i.e. material as received) and is currently used for SRF production (SRF to be used as cement kiln burner fuel).

Medium calorific fraction (MC, LHV <20 MJ kg⁻¹_dry, particle size 20–120 mm, 55%_m of plant input): the MC contains biogenic material (undefined organics, wood, paper and cardboard, about 30%_m) as well as fossil material (bright and dark plastics, about 20%_m). Up to 30–50%_m of the material was classified as ‘fine fraction’ (<30–60 mm for the MC) and not distinctly characterized concerning its material composition. The MC is currently used directly as SRF for the cement calciner in the so-called ‘HOTDISC’ process (see Pomberger and Abl (2008) for a description).

The characterization of the selected waste streams indicated, that additional processing to optimize the (1) pollutant (Cl, heavy metals) and especially (2) C_bio content should allow for refining their utilization, treatment or disposal options. Based on the waste characteristics and the sorting task, a near infrared (NIR) sensor-based sorting technology was chosen for the trials.

Material-specific parameters for the identification by NIR are molecular characteristics and associated spectroscopic behaviour (see Eisenreich & Rohe (2006) and Günzler and Heise (1996) for further information). The principle of the NIR sensor-based sorting is explained schematically in Figure 2(a). In brief, a conveyor belt transports the material into the detection zone, where halogen lamps irradiate the material with NIR light. The light penetrates into the material, where it is scattered and partially absorbed at characteristic wavelengths, due to the interaction of the NIR light with molecular vibrations. Thus, the reflected light has been changed in its spectral distribution as a result of the material characteristics and can therefore be used as a separation criterion. Objects, recognized as the material to be removed, are ejected using a pneumatic pulse and are thrown over a separating edge into the ‘reject’ – others fall into the ‘passing’.

Polyvinyl chloride (PVC) was selected as target material to be separated from the waste stream in order to optimize the pollutant content, as PVC contains 30–50%_m Cl (Christiani, 2006) and in addition, (heavy) metals such as Pb, Cd, Sn, Zn, Ba and Ca, that are used as PVC stabilizers in a range of 2–6%_m (Domininghaus, 2008).

In order to separate components with an elevated C_bio content, material with NIR spectroscopic behaviour similar to different paper types was selected as target material in the second stage.

Extensive trials on a pilot scale and one indicative large-scale experiment were conducted, both accompanied by material and chemical characterization of the generated output streams.

Experimental procedure

Processing trials

The pilot-scale processing trials were conducted with a single, mobile, two-way NIR sensor-based sorting device (1 m width of conveyor belt) of the type ‘REDWAVE’ (BTW-Binder GmbH (ed.), 2010) with a multiplexed NIR spectroscope KUSTA 4004M/64 for a radiation wavelength of about 1400–1900 nm (see LLA (2010)) and 64 measuring heads. The input material was fed to the sorting device first to eject PVC-containing material from the waste stream. In a second step, the remaining material was fed again to the sorting device, now tuned to separate biogenic components – more specifically, material with a paper-like NIR spectrum, namely paper, cardboard, wood, cotton textiles, etc. – from this PVC-freed material stream. Consequently, the sorting steps resulted in three output streams: Reject 1, Reject 2 and Passing 2 (Figure 2(c)).

In order to optimize the quantitative ejection of PVC (Reject 1) and the purity and yield of the biogenic reject (Reject 2), parameter configurations [identification scheme with underlying spectra database and decision trees used for the identification, sensitivity of identification, threshold (i.e. allowed noise in the signal), scanning speed of the sensor system, pressure and duration of the pneumatic pulse applied to eject particles] were varied in three pilot-scale test runs, which were conducted as three-fold replicates with around 20 kg_wet input material each. An additional aim was to keep the consumption of compressed air for pneumatic ejection and resulting operating costs small by using an optimized parameter configuration, without compromising the sorting efficiency. The throughput of the pilot-scale tests (100–500 kg_wet h-1 (1 m conveyor belt width)) was limited by the manual material feed to the conveyor belt.

The output streams of each test (Reject 1, Reject 2 and Passing 2) were treated as samples and used for material characterization as a whole. Thereafter, the material of Reject 2 and Passing 2 as well as input material were thoroughly mixed and shredded using a twin shaft shredder (2 cm), and a sample for further preparation in the laboratory (1–2 kg_wet) was drawn by coning and quartering.

Based on the results of the pilot-scale test in terms of optimized parameter configuration, one indicative test run on a large scale was conducted per waste stream. A mechanical treatment plant operating with NIR sensor-based sorting devices to split packaging waste (‘yellow bin’ waste in Austria) for material recycling was used for the large-scale test (Figure 2(b)). The test was conducted as a stationary operation during 1.5 h with 5400 to 7000 kg_wet test material and samplings of each output stream every 20 min yielding four samples. A first sorting device separated PVC and a second one separated material with paper-like NIR characteristics. Both sorting devices were of the type ‘REDWAVE’ (BTW-Binder GmbH, 2010) using the multiplexed NIR spectroscope KUSTA 4004M/64 for a radiation wavelength of about 1400–1900 nm (see LLA, (2010)), with a conveyor belt width of 2 m and 64 measuring heads. The throughput of the NIR sorting devices was 2500–3000 kg_wet h⁻¹ (2 m conveyor belt width). Additional processing steps (screening at 230 mm and 45 mm mesh size) had to be applied prior to the NIR sorting due to the concept of the large-scale processing plant.

The samples, which were drawn during the large-scale test, were divided into one part to be characterized manually for its material composition (5–15 kg_wet) and a second part for the preparation of chemical analysis (3–5 kg_wet). The material for chemical analysis was shredded using a twin-shaft shredder (2 cm), mixed and a sample for further preparation in the laboratory (1–2 kg_wet) was drawn by coning and quartering.

Evaluation and chemical characterization

The functional efficiency of the separation (yield, transfer coefficient) and the quality (purity, absolute concentration) of the output streams with regard to further treatment and utilization options were used to evaluate the processing trials.

Material characterization of the output streams and their mass proportions were used for a first evaluation. The material composition was analysed by manual sorting of the test material and the yield and purity were determined (Reject 2, Passing 2). The yield represents the ejected mass proportion of material supposed to be ejected. The purity of a specific output stream is the mass proportion of material sorted correctly into the output stream. The fine fraction represents a material mix. It could not be evaluated distinctly and was excluded from the evaluation. Textiles were assumed to be of biogenic origin if found in the Reject 2, and assumed to be fossil if found in the Passing 2, due to a lack of evaluation methods by manual sorting. It was shown by Pretz and Killmann (2007) that biogenic textiles are identified correctly by NIR sensors, making this assumption justifiable.

Chemical quantification of the C_bio and total carbon (TC) content, the Cl and heavy metal content (As, Pb, Cd, Cr, Co, Ni, Hg, Sb, Cu, Zn) as well as moisture, energy and ash content (see Table 1 for applied standard procedures) was used for further evaluation.

Table 1.

Parameters used for SRF characterization (threshold values according to the waste incineration directive (Limit 1, 2, 3 (in mg MJ-1 LHV,dry (Source: BMLFUW & BMFWJ (2010)).

Parameter		Austrian standard^a	Limit 1	Limit 2	Limit 3	Parameter		Austrian standard^a
Arsenic*	As	15411	2	2	1	Sulphur	S	15408
Lead*	Pb	15411	20	15	15	Hydrogen	H	15407
Cadmium*	Cd	15411	0.23	0.17	0.17	Nitrogen	N	15407
Chrome*	Cr	15411	25	19	19	(Total) carbon	TC	15407
Cobalt*	Co	15411	1.5	0.9	0.9	Biogenic carbon	C_bio	15440
Nickel*	Ni	15411	10	7	7	Higher heating value	HHV	15400
Mercury*	Hg	15411	0.075	0.075	0.075	Lower heating value*	LHV_dry	15400
Antimony*	Sb	15411	7	7	7	Lower heating value	LHV_wet	15400
Copper*	Cu	15411	–	–	–	Moisture content	H₂O	15414-1
Zinc*	Zn	15411	–	–	–	Ash content	Ash	15403
Chlorine*	Cl	15408	−#	–	–	Inert material	Inerts	–

Parameters marked * have to be determined for SRF from solid waste according to BMLFUW & BMWFJ (2010). The presented limits stipulated by the waste incineration directive for selected parameters (BMLFUW & BMWFJ, 2010) are median values: limit 1 is valid for SRF to be used in co-incineration in the cement industry, limit 2: SRF to be used in power plants with a maximum contribution of 15% to the thermal output of the fuel, limit 3: SRF to be used in other co-incineration except power plants. Additional limits may be provided from the industry, # e.g. a Cl content of < 0.8%m,dry – 1%m,dry for SRF to be used in the cement industry (Christiani, 2006; Curtis, 2010).

15400, Austrian Standard CEN/TS 15400 (2006a); 15407, Austrian Standard CEN/TS 15407 (2006b); 15411, Austrian Standard CEN/TS 15411 (2006c); 15414-1, Austrian Standard CEN/TS 15414-1 (2006d); 15403, Austrian Standard EN 15403 (2009a); 15408, Austrian Standard EN 15408 (2009b); 15440, Austrian Standard EN 15440 (2009c).

Laboratory samples (1–2 kg_wet) of a particle size of 2 cm were prepared as described in the section ‘processing trials’. These samples were dried at 105 °C (drying oven, 24 h at maximum) and ground to 4 mm using a cutting mill and to 0.5 mm using an ultracentrifugal mill. Standard procedures that are suitable for application on waste materials (indicated in Table 1) were applied for chemical analysis. The parameters As, Pb, Cd, Cr, Co, Ni, Hg, Sb, Cu, and Zn were analysed using inductively coupled plasma optical emission spectrometry (ICP-OES) (Arcos, Spectro) after microwave-assisted acid digestion (Multiwave 3000; Anton Paar). Chlorine (Cl) was analysed by ion chromatography with suppressed conductivity detection (ICS-90; Dionex) after combustion in a bomb calorimeter and absorption into a diluted sodium hydroxide solution. Elemental analysis (high temperature combustion with subsequent gas analysis) was applied to determine S, H and N as well as TC (Vario Elementar). The higher heating value (HHV) was determined using a bomb calorimeter and the LHV was derived by calculation taking the S, H, N and H₂O content of the sample into account. The ash content was determined gravimetrically after heating to 550° C as described in the standard procedure. Inert materials (i.e. metal and stones) were determined gravimetrically.

The reliability of different available analytical methods addressing the C_bio content is challenged when analysing heterogeneous materials (e.g. Staber et al., 2008). In the present study, the C_bio content was determined according to Austrian Standard EN 15440 (2009c) Solid recovered fuels – Method for the determination of biomass content, using the ¹⁴C method, also described in Fellner and Rechberger (2009). In this procedure, the sample material is combusted in a calorimetric bomb and the formed CO₂ is absorbed in a cooled alkaline solution, which is mixed with a scintillation cocktail and measured by liquid scintillation counting. This procedure allows to determine the C_bio content as a function of the ¹⁴C to ¹²C ratio in a sample and gives a precision of 4% (n = 3).

In the results section, the mean values and the standard deviation of 3 to 4 separately drawn samples are shown. Consequently, the precision of these data reflect the sampling and not chemical analysis. The precision of chemical analysis is usually better than that of sampling and the error introduced by the sample preparation is the largest in an analytical process. Therefore, data presented in the results section are presented with the largest uncertainty in terms of their precision. In order to ensure the trueness of the chemical analysis, the performing laboratory takes part in round robin tests for all analysed parameters. Additionally, internal control samples and reference materials are in use for the analysis of the parameters TC, H, S, Cl and the heavy metals (As, Pb, Cd, Cr, Co, Ni, Hg, Sb, Cu) (information provided by Kienzl, 2011).

The concentration of chemical parameters and the mass output of a specific waste stream are used to derive transfer coefficients, which represent the yield of a chemical component (Equation 1: Transfer coefficient X (%) of a specific chemical parameter i in the output stream (reject or passing) of a sorting process (eq. 1).

X_{output} (i) = \frac{{mass}_{output} * {concentration}_{output} (i)}{{mass}_{input} * {concentration}_{input} (i)} * 100

(1)

Additionally, a specific chemical parameter’s concentration is used for evaluation of enrichment E (%) and depletion D (%) of the concentration c of a chemical parameter i in the reject (eq. 2a) and passing (eq. 2b) compared to the input material of a sorting step.

E_{reject} (i) = (\frac{{concentration}_{reject} (i)}{{concentration}_{input} (i)} - 1) * 100

(2a)

D_{passing} (i) = (\frac{{concentration}_{passing} (i)}{{concentration}_{input} (i)} - 1) * 100

(2b)

Results and discussion

Yield and purity of output streams

For both waste streams (high (HC) and medium (MC) calorific), an optimized parameter configuration for the quantitative ejection of PVC and the ejection of biogenic material with good quality and yield was found in the pilot-scale test run no. 2 as presented previously (Pieber et al., 2010). The adjusted parameter configuration combined basically an identification scheme appropriate for the material-specific properties of the waste particles, with adjusted pressure and duration of the pneumatic pulse, to keep the consumption of compressed air and consequently electrical energy low. In addition, the identification sensitivity and the allowed noise in the signal (threshold) were adapted. The large-scale test was conducted based on the information gained in the pilot-scale tests.

In general, 3–5% of the material was separated in the first sorting step on a pilot as well as a large-scale test, based on the input material mass of this sorting step. For the second sorting step, a material mass of 21–24% was separated from the input material of the pilot-scale sorting tests. On a large scale, the separated material mass was only around 5–6% based on the plant input. As a result of this low separated material mass, a decreased yield of 20% (HC) and 17% (MC) was observed on a large scale as compared to 90% (HC) and 80% (MC), respectively, in the pilot-scale trials (Table 2). The reduced yield in the large-scale test is attributed to the fact that the existing plant, which was used for the large-scale trial, was designed for a different type of waste and a different sorting task.

Table 2.

Mass balance (input = 100%), yield and purity achieved in the test runs.

			HC, pilot (n = 3)	HC, large (n = 1)	MC, pilot (n = 3)	MC, large (n = 1)
Mass	Reject 1	%_m,wet	5.11 ± 0.61	3.38	4.23 ± 0.50	2.65
Mass	Reject 2	%_m,wet	21.1 ± 6.7	6.20	23.8 ± 0.7	5.01
Purity	Reject 2	%_m,wet	88.1 ± 3.1	79.1	95.3 ± 0.4	87.4
Yield	Reject 2	%_m,wet	88.2 ± 4.3	20.2	79.4 ± 0.7	16.7
Mass	Passing 2	%_m,wet	73.8 ± 6.8	71.2	71.9 ± 1.2	51.8
Purity	Passing 2	%_m,wet	96.0 ± 2.8	72.7	84.4 ± 0.7	57.8
Mass	Oversize	%_m,wet	–	7.35	–	0.08
Mass	Fine fraction	%_m,wet	–	11.9	–	40.5

The mean value and standard deviation of n samples from pilot- and large-scale tests with the high (HC) and medium calorific (MC) waste stream are presented. Mass balances are based on 21.7 ± 2.1 kg_wet (HC, pilot), 38.6 ± 2.2 kg_wet (MC, pilot), 7030 kg_wet (HC, large), 5390 kg_wet (MC, large) and proportions refer to the total input. Yield and purity of Reject 2 and Passing 2 based on 20.6 ± 2.0 kg_wet (HC, pilot), 36.9 ± 2.1 kg_wet (MC, pilot), 180 kg_wet (HC, large), 170 kg_wet (MC, large).

Removal of pollutants (chlorine, heavy metals)

Chemical characterization of the output material in terms of its Cl and heavy metal content revealed that around 30% (MC) to 41% (HC) of the total Cl content were removed with only a very small material (around 3–5%_wet) and energy (3–4%_LHV,wet) loss in the large-scale test. This equals a transfer coefficient of 37 and 45% based on the input material of the specific sorting step (Figure 3). Additionally, up to 60% of the Pb and Cd content in the input for this sorting step was removed [these heavy metals were used as PVC-stabilizers (Vinyl, 2010) and are still highly relevant in waste management due to the lifetime of PVC-containing products].

Consequently, reductions (up to 70%) in the Cl as well as Cd and Pb concentration with very low Cl concentrations of 0.3–0.8%_m,dry in the Reject 2 and 0.4–1.7%_m,dry in the Passing 2 (dependent on the input concentration for sorting step 1 of 1.1 to 2.4%_m,dry) were achieved (see Figure 4 for the situation in the large-scale test). Simultaneously, very high Cl concentrations (12–15%_m,dry) were observed in the Reject 1 – a waste stream of, however, only very small mass. Similar reduction achievements also addressing the removal of PVC materials from a heterogeneous waste stream were previously reported (Christiani, 2006).

The mean concentrations (n = 3 or 4) of most heavy metals were below the specified median limits for SRF to be used in the cement industry (mg MJ⁻¹_LHV,dry) in the input as well as output streams (As, Cr, Co, Ni, Hg). For Cd, in contrast, the observed concentration was above the limit for SRF to be used in the cement industry in the Input, the Reject 1, Passing 1 and Passing 2 in most of the cases. Only the Reject 2 was below the limit in terms of its Cd concentration. Additionally, the Pb concentration in the Reject 1 (HC and MC, large scale) was above the limits as well as the Sb concentration (MC, large scale). The Reject 1 was not characterized chemically in the pilot-scale test, but it can be assumed that the concentration was above the limit as well. Consequently, in addition to the desired output streams Reject 2 and Passing 2, characterized by a large mass proportion and low pollutant concentrations, a small output stream, highly contaminated with Cl, Cd and Pb was generated, as previously reported (Pieber et al., 2011). Meaningful disposal or even utilization options for this PVC-enriched output stream still need to be identified.

Separation of biogenic components

The C_bio concentration was increased about two- to four-fold compared to the Passing 1 (pilot and large scale) by the sorting step. The concentration was additionally increased three- to up to 10-fold compared to the Passing 2 on a pilot scale (Figure 5). The final C_bio concentration in the Reject 2 was in the order of 62–77%_{m TC,dry} on a pilot as well as large scale; compared to around 16–47% in the input and 7–40% in the Passing 2. Consequently, around 62–77% of the CO₂ emissions resulting from energy recovery would be of biogenic origin, assuming that the TC content of this waste stream equals the total organic carbon (TOC) content. Simultaneously, the energy content (LHV) was decreased in the C_bio-enriched Reject 2 and increased in the C_fossil-enriched Passing 2.

The C_bio transfer coefficients of the processing were 49%_TC (MC) and 72%_TC (HC) on a pilot scale (Figure 6). On a large scale, the C_bio transfer coefficients were much lower, due to the low yield of the Reject 2 (below 20%) (Figures 6 and 7). This was found to be due to the present plant concept, as a MT plant which was designed and optimized for a waste stream with different physical characteristics concerning mainly the density, particle size and geometry (packaging waste) and a different sorting task (separating plastics) was used for the first indicative large-scale test. These circumstances led to technical limitations, which are discussed in the following and resulted in the largely decreased yield of the Reject 2 in the large-scale tests compared to the pilot-scale tests: Heavy (wooden) particles were correctly detected from the sensor system and a pneumatic pulse was applied. However, due to the present construction of conveyor belts, etc., which is optimized for material such as hollow bodies made out of plastics, these wooden objects often fell back into the passing. This specific issue was identified to be a common problem in sensor-based sorting: particles with an unfavourable ratio of weight to surface area are difficult to be ejected pneumatically (Nisters, 2006). This drawback for wood particles was increased by the fact, that reflection of the applied NIR radiation was received from the conveyor belt itself at higher proportions than in the pilot-scale tests. Consequently, the sensors could not be adjusted for dark material, which happened to be very often wood.

Also the number of measuring heads per band width of the conveyor belt (resolution) was reduced on a large scale by around 50% compared to the pilot-scale test: on a pilot scale 64 NIR sensors were applied per 1 m of band width – on a large scale in contrast, 64 sensors were applied per 2 m of conveyor belt width. Consequently, objects with too small a particle size could not be detected, which is relevant especially for paper particles.

Consequently, not only the sensor system needs to be adjusted for the waste stream, but very importantly, the whole plant concept. In order to evaluate the actual potential of the separation of biogenic components from the heterogeneous waste stream, a large-scale trial should be conducted with a plant which is optimized for the specific mechanical properties of the addressed material. The expected potential by implementing a sorting device with similar resolution as applied on a pilot scale and a powerful ejection unit should allow similar mass yields as those observed on a pilot scale. Based on the results from the pilot-scale test (assumption of a mass yield of 20% (HC) and 25% (MC) for the Reject 2) and a constant concentration of the C_bio in the Reject 2, the potential transfer coefficients and simultaneously decreased concentration in the Passing 2 were projected (Figure 7), in order to estimate the large-scale potential. It is shown that for instance for the HC waste stream the transfer coefficient of Reject 2 (elevated C_bio content) could potentially be increased from 13.5% (large-scale test run with non-optimized plant set-up) to 43.5% (forecast for achievable results with optimized plant set-up according to pilot-scale test results). Consequently, if the same mass yield was reached in a large-scale application as observed on a pilot scale with similar enrichment factors of around 2, a transfer coefficient of >40% could be expected. The potential based on the known composition of the input material is even higher, at least for the MC waste stream (Figure 1). In addition to the increased C_bio-content of the Reject 2 the LHV of the Passing 2 can be increased making it a valuable SRF for application as kiln burner fuel in the cement industry.

Conclusions and perspectives

The extensive NIR sensor-based sorting experiments on a pilot scale and indicative test runs on a large scale showed that the technology is generally applicable to (1) remove pollutants (mainly Cl and heavy metal bearing components) and to (2) generate a waste stream with increased C_bio content out of the heterogeneous waste fractions looked at, if the sensor systems are appropriately adjusted for the sorting task. Important in that context is that the plant construction concept needs to be adapted for the mechanical properties of the waste stream to be processed as well.

Besides the optimization and evaluation of the NIR sensor-based sorting technology for application on the waste streams with current properties, possible changes in material composition, such as an increased proportion of dark plastics contained in the waste – which could significantly decrease the efficiency of the NIR sensor-based sorting technology – have to be kept in mind (dark particles show a very low reflection of NIR light and can therefore not be identified correctly). Additionally, the observed reduction of Cl, Cd and Pb is related to the identification of PVC – other, especially inorganic sources, of these pollutants would require an adapted identification system. The same is true for the separation of biogenic components, which is linked to paper-like NIR spectra and will potentially require an extended database for other biogenic components (e.g. polylactic acid materials) in the future. Therefore, it should be ensured that the applied technology can be adapted to changes in the input waste stream if required.

Anticipating a satisfactory result, this second sorting step will allow broadening future marketing opportunities for the SRF. On the one hand, this is true in terms of the EU-ETS on CO₂ as the achieved C_bio concentration on a pilot scale means that 62–77% of the CO₂ emissions in energy recovery would be of biogenic origin and therefore regarded as ‘CO₂ neutral’ – not requiring the emitters to hold certificates for the respective emissions. The generation of the C_fossil-enriched waste stream with increased heating value is relevant from a sustainability/resource conservation point of view as well: The SRF will need different qualities in the future, if assuming that the cement industry will show an increasing demand of SRF in order to further increase their substitution rate of fossil fuels. SRF of very high quality, namely low pollutants and high energy content, is required as kiln burner fuel. SRF of lower qualities can be used as calciner fuel: currently already a high portion of the energy consumption in the clinker burning process is introduced into the calciner, for which the quality specifications related to traditional fuel properties such as heating value and particle size are much more relaxed. The proposed separation allows generating one rather low calorific calciner fuel with high C_bio content and a high calorific kiln burner fuel according to the specifications needed.

Furthermore, even new utilization options for the biogenic material are conceivable, thinking of biomass power plants with a large demand of biomass currently already facing a shortage of wooden biomass – in the future even waste-derived biomass of high quality in terms of low pollutant concentrations could be an option as feedstock. Finally, the C_bio-enriched waste stream generated from the heterogeneous material could possibly be put back into material recycling, instead of energy recovery. This, however, requires very high purities of the separated biogenic-enriched waste fraction. Further development of sensor and separation technology will allow for increasing the performance of sensor-based sorting systems and will subsequently open ‘waste-to- (material) resources’ options in addition to the ‘waste-to-energy’ alternatives for that specific waste stream as well.

Footnotes

Acknowledgements

Special thanks are offered to Martina Meirhofer for supporting the practical investigations, Johann Felber and Anna Wobik for supporting the sensor-based sorting experiments and Norbert Kienzl for being responsible for chemical analysis.

This research within the K1-centre “BIOENERGY 2020+” was promoted by the COMET funding scheme executed by the Austrian Research Promotion Agency and financed by national Austrian funds as well as funds from the provinces of Burgenland, Lower Austria and Styria. The co-financing from industry to be provided within the COMET funding scheme was granted by BT-Wolfgang Binder GmbH, Saubermacher Dienstleistungs AG and Umweltdienst Burgenland GmbH.

References

Austrian Standard CEN/TS 15400 (2006a) Solid recovered fuels – Methods for the determination of calorific value. Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard CEN/TS 15407 (2006b) Solid recovered fuels – Method for the determination of carbon (C), hydrogen (H) and nitrogen (N) content. Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard CEN/TS 15411 (2006c) Solid recovered fuels – Methods for the determination of the content of trace elements (As, Ba, Be, Cd, Co, Cr, Cu, Hg, Mo, Mn, Ni, Pb, Sb, Se, Tl, V and Zn). Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard CEN/TS 15414-1 (2006d) Solid recovered fuels – Determination of moisture content using the oven dry method – Part 1: Determination of total moisture by a reference method. Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard EN 15403 (2009a) Solid recovered fuels – Determination of ash content. Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard EN 15408 (2009b) Solid recovered fuels – Methods for the determination of sulphur (S), chlorine (Cl), fluorine (F) and bromine (Br) content. Vienna, Austria: Austrian Standards plus G.m.b.H.

Austrian Standard EN 15440 (2009c) Solid recovered fuels – Method of the determination of biomass content. Vienna, Austria: Austrian Standards plus G.m.b.H.

BMLFUW & BMWFJ (ed.) (2010) BGBl. II Nr. 476/2010 (Waste Incineration Directive). Vienna, Austria.

BTW-Binder GmbH (ed.) (2010) Information and material provided by the BT-Wolfgang Binder G.m.b.H. Gleisdorf, Austria

10.

Christiani

(2006) Möglichkeiten und Grenzen der Sortierung von Abfällen. Gewerbeabfall- und Hausmüllaufbereitung [Options and limits of waste material sorting. Processing of commercial waste and domestic refuse]. Müll und Abfall 38: 532–536

11.

Curtis

(2010) Information provided by the Saubermacher Dienstleistungs AG Graz, Austria

12.

Dehoust

Schüler

Giegrich

. (2010) Climate protection potential in the waste management sector examples: municipal waste and waste wood. Umweltbundesamt Texte 61/2010, ISSN 1862-4804, online at: http://www.uba.de/uba-info-medien-e/4049.html (accessed 25 June 2011).

13.

Domininghaus

(2008) Kunststoffe, Eigenschaften und Anwendungen [Plastics, properties and applications]. In: Elsner

Eyerer

Hirth

(eds). Berlin/Heidelberg, Germany: Springer Verlag.

14.

Eisenreich

Rohe

(2006) Infrared Spectroscopy in Analysis of Plastics Recycling. Encyclopaedia of Analytical Chemistry. DOI: 10.1002/9780470027318.a201110.1002/9780470027318.a2011.

15.

EP (European Parliament) (ed.) (2008) Directive 2008/98/EG (Waste Framework Directive). Brussels, Belgium: European Union.

16.

Fellner

Rechberger

(2009) Abundance of ¹⁴C in biomass fractions of waste and solid recovered fuels. Waste Management 29: 1495–1503.

17.

Friege

Fendel

(2011) Competition of different methods for recovering energy from waste. Waste Management & Research 29: 30–38.

18.

Friege

Giegrich

(2008) Climate effects of a local waste management system. conclusions drawn from a case study. In: Conference Proceedings (CD-ROM) ISWA/WMRAS World Congress 2008, Singapore. Copenhagen: ISWA.

19.

Günzler

Heise

(1996) IR-Spektroskopie, eine Einführung [IR-Spectroscopy, an Introduction]. Weinheim, Germany: VCH.

20.

Kienzl

(2011) Information provided by the Working Group ‘Data, Analyses & Measurement Techniques’ of the BIOENERGY 2020+ GmbH, Graz, Austria.

21.

LLA (ed.) (2010) LLA Instruments GmbH, online at: http://www.lla.de/ (accessed 29 December 2010).

22.

Nisters

(2006) Ersatzbrennstoffherstellung mittels NIR-Technologie [Producing waste-derived fuel with NIR technology]. In: Conference Proceedings Sensorgestützte Sortierung 2006, Aachen: Taschenbuch, pp. 43–44

23.

Pieber

Ragossnig

Brooks

. (2010) Advanced waste-splitting by sensor based sorting on the example of the MT-plant Oberlaa. In: Lorber

Adam

Aldrian

. (eds.) Conference Proceedings DepoTech 2010, Leoben: Montanuniversität Leoben, pp. 695–698.

24.

Pieber

Ragossnig

Sommer

. (2011) Splitting of heterogeneous waste by sensor-based sorting as a basis for ptimized material-specific waste-routing. In: Kühle-Weidemeier

(ed.) Conference Proceedings Waste-to-Resources 2011, Göttingen: Cuvillier, pp. 501–513.

25.

Pomberger

Abl

(2008) Mittelkalorische Ersatzbrennstoffe für das HOTDISC-Verfahren [Medium Calorific Solid Recovered Fuel for the HOTDISC Process]. In: Lorber

Kreindl

Menapace

. (eds) Conference Proceedings DepoTech 2008 Essen: VGE Verlag, pp. 275–280.

26.

Pomberger

Curtis

Scherübl

(2008) Climatological relevance of waste management as seen by a collector. Österreichische Wasser- und Abfallwirtschaft 07-08: 113–122.

27.

Pretz

Killmann

(2007) Innovative Sortiertechnologien zur Gewerbeabfallsortierung [Innovative sorting technologies for commercial waste]. Müll und Abfall 11: 546–551.

28.

Ragossnig

Faist

(2009) Destination-process-specific optimization of waste processing using innovative treatment technology. In: ISWA (eds.) Conference Proceedings ISWA World Congress 2009, Lisbon, Portugal. Lisbon: ISWA.

29.

Ragossnig

Wartha

Pomberger

(2009) Climate impact analysis of waste treatment scenarios – thermal treatment of commercial and pretreated waste versus landfilling in Austria. Waste Management and Research 27: 914–921.

30.

Rotter

Kost

Winkler

. (2004) Material flow analysis of RDF-production processes. Waste Management 24: 1005–1021.

31.

Rotter

Lehmann

Marzi

. (2011) New techniques for the characterization of refuse-derived fuels and solid recovered fuels. Waste Management & Research 29: 229–236.

32.

Schlupeck

(2011) Im Eiltempo. Recycling Magazin 10/2011, Munich, Germany.

33.

Staber

Flamme

Fellner

(2008) Methods for determining the biomass content of waste. Waste Management and Research 26: 78–87.

34.

Titech (ed.) (2010) TITECH sorting systems specified at world’s most modern municipal solid waste plant in Larnaca, Cyprus. Online at: http://www.titech.com/news/titech-sorting-systems-16281 (accessed 29 December 2010).

35.

Vinyl (2010) Vinyl 2010: ten-year voluntary programme on Sustainable Development by the whole PVC industry. Online at: http://www.vinyl2010.org/ (accessed 29 December 2010).