Breakage characteristics of incinerator bottom ash in the HV pulse power
Alexander Weh & Abbas Mosaddeghi
SELFRAG AG, Biberenzelgli 18, CH-3210 Kerzers, Switzerland
Corresponding author: email@example.com
Recent research indicate the pre-concentration potential of the HV pulse power process by the accumulation of comminution
energy towards electrical conductive target phases. Bottom ash includes particles of various composition. Not only ferrous and
non-ferrous metals but also rubble and rubbish fragments can be trapped within melting products. This study presents breakage
characteristics of bottom ash in the HV pulse power process performed in specific single particle test setup. Several particles in
different sizes were treated at increasing energies. The resulting mass loss of the largest progeny sizes are used to quantify and
describe the breakage characteristics. The results confirm the preferred breakage of metal-bearing particles including a feed size
effect. The different breakage behaviour is dependent on how energy is deposited. When conductive phases are present the
discharge path is introduced into particles resulting in more efficient size reduction. In metal-free particles the discharge is
diverted along particle surface, achieving only limited size reduction in a specific fracture pattern. The analysis of progeny mass
loss vs energy distribution trends have the potential to identify and quantify active breakage modes and to evaluate the HV pulse
power process for the targeted application. The preferred liberation of included metals without overgrinding provides a
sustainable recycling solution for bottom ash, confirmed in industrial scale.
Keywords: bottom ash, HV pulse power, breakage characterisation, metal liberation, selective fragmentation
Waste incineration has increased worldwide. Advantage are the significant reduction of waste volume, the
production of thermal and electric energy and most important, the thermal/energetic degradation of organic
pollutants into minerals and metals. As with any anthropogenic deposit the composition of bottom ash varies
according to its origin and represents a strongly heterogenic mixture of various compounds that are related to the
introduced waste. Different authors (Lichtensteiner et al. (1993); Bunge (2010, 2016); Eggenberger et al. (2010))
describe the composition of bottom ash from central and northern Europe. The included metal and mineral content
have a considerable potential for recycling. The locking of metals and mineral fraction into melting products depends
on the conditions during incineration and the following storage. A summary of processes active are presented in
A detailed description of bottom ash processing is found in Bunge (2016, 2010), Pretz et al (1998) or Alwast et al.
(2010). Commonly, three dry processing steps are applied: (i) comminution for liberation; (ii) classification to
improve separation efficiency and (iii) separation of metals. Impact crushers are used to disintegrate slag-metal
composites. Differences in elastic modulus between slag and metal chip off the melting products. The main target
in processing is to recover ferrous metals (FE) and nonferrous metals (NF). The different size distribution of included
metals is problematic. While FE appear coarse in sizes +10 mm the more valuable NF like Al or Cu are present in the
finer -10 mm (Bunge 2010). Total metal recovery of current operational practice is limited to 60% at the economic
optimum (Bunge 2010) achieving concentrates quality of <80%. In the circuits 90% FE recovery is possible. NF metals
are reduced to 40% recovery due to its finer intrinsic size distribution. The remaining mineral fraction has limited
use as construction material in landfills, road construction or as underground filler. Most of it is still landfilled.
The high voltage (HV) pulse power process is an alternative comminution method to selectively liberate metals from
bottom ash. The wet process transmits highly energetic electrical pulses into a feed material along component
boundaries. The resulting discharges emits strong shockwaves into the feed. Commonly recognized analogues are
lightning and blasting. The selectivity of the process arises from the way the electricity and shockwaves interact with
electrical and acoustic material properties. Recent research from mineral processing (Zhou et al. (2014, 2015a), Shi
et al. (2015)) indicate a pre-concentration potential of the process by the accumulation of comminution energy
towards electrical conductive phases. The industrialization of the process is promoted mainly by SELFRAG AG in
Switzerland (Zhou et al. 2015b). The industrial adaptation of the process for bottom ash (Weh 2015) indicate a
significant increase of NF recovery and quality of concentrate. In addition, the mineral fraction is upgraded.
To confirm the superior performance of the process this study investigate the breakage characteristics of different
bottom ash particles in the HV pulse power process. The target is to identify how energy is introduced and distributed
between different material groups and to evaluate the resulting breakage characteristics for a better understanding
of process-material interaction.
2.1 HV pulse power process
All tests were done with a specific SELFRAG HV pulse power generator using a single-particle-test-vessel at the
SELFRAG pilot plant facility in Kerzers, Switzerland. The generator converts a continuous flow of electricity into pulses
by storing electricity in capacitors banks and spark gaps. By triggering the pulses with reduced pulse rise time (<500
ns, Bluhm et al. 2000) electrical breakdowns are forced within or near a feed material. This happens when a material
is neither conductive enough to allow electricity to flow through, nor resistive enough to fully inhibit the flow. As a
consequence, plasma channels of are formed. The following plasma collapse generates strong shockwaves, which
are responsible for the majority of fragmentation. The selectivity of the process arises from the way the electricity
and shockwaves interact with electrical and acoustic properties of material (van der Wielen et al 2013).
Figure 1 – Potential discharge path scenarios in dense, fine grained feed. Discharge path is guided by permittivity contrast (PC)
near or in particle dependent on composition.
How energy is transferred to the material is important. Tests from Zhou et al. (2014, 2015a) with dense, fine grained
material suggest that metallic inclusion influence the discharge path. Potential scenarios are presented in Figure 1.
A significant permittivity contrast within the material is favorable, when no textural planes like veins, coarse mineral
boundaries, cracks or pores are available. Such a situation induces massive field distortion, which initiates significant
leaders in the pre-plasma stage towards such areas (Situation I in Figure 1). The following internal discharge in the
material deposits most fragmentation energy directly around the metallic inclusion. The difference in the elastic
modulus between the different material components and/or the higher confining stress regime inside the material
supports the selective liberation and/or exposure of the conductive phase. The resulting size reduction is extended
with persuasive fractures and finer progeny sizes. If no internal permittivity contrast exists (Situation II & III), for
example when overall metal content is higher and/or distributed more homogeneously or no to very low amounts
of conductive phases are available, the permittivity contrast is largest between particle surface and the surrounding
process media. This leads the discharge towards the particle and along the particle surface with reduced size
reduction effect and larger progeny sizes. In general, the resulting mass loss of biggest progeny particle can indicate
the active discharge scenario. Zhou et al. (2014, 2015a) used the % mass loss of largest progeny sizes to describe the
body breakage probability (BBP). When largest progeny size show + 10% mass loss, it is defined as body breakages.
Less reduction is classified as surface breakage.
2.2 Test design & equipment
The test design of the study investigates the particle breakage characteristics to understand how electrical energy is
introduced (Figure 1). A representative selection of bottom ash particles in 4 close size fractions are treated at 4
energy levels in a single particle and single discharge approach. To have a minimum statistical impact, 10 to 20
particles per energy level and size fraction are treated. Energy is increased by manipulating the applied voltage
and/or the generator capacitance. To identify if a discharge is placed inside or outside a particle the largest progeny
size is weighed to calculate the weight% mass loss for the BBP analysis.
The used HV-Generator has an overall capacitance range between 15-37.5 nF. The pre-selected voltage levels from
low, intermediate and high are applied by manipulating the generators spark gaps. Plant and test specification are
summarized in Table 1. For the tests a specific single particle vessel was installed. Processing was done using a disk
working electrode and a plate counter-electrode. The process media used is demineralized water.
Table 1 – SELFRAG test-generator operating range and conditions for tests
Test operating conditions
90 – 250 kV
150; 180; 210 kV
1 – 25 Hz
Disk 100 mm diameter
Electrical field strength
2.25 – 25.0 kV/mm
2.3 Test material characterization
For the test a representative selection of coarse bottom ash particles were prepared in 4 size fractions (Table 2).
Material bulk density varies between 1.1-1.3 g/cm3, single particle density between 1.8-2.7 g/cm3 in relation to the
particle subgroup. Particle shapes are evaluated by the aspect ratio indicating very different appearance of discreet
Table 2 - Feed size characteristics of tested particles in average.
Average total sample weight
Average particle weight
For the tests only +20 mm bottom ash particle are classified into 5 homogeneous subgroups to work out differences.
Figure 2 – Slag mix conglomerates with ceramic culets and metals inclusions before processing (A); spherical Fe-oxide inclusions
in pores after processing (B); Al inclusion in melting product after processing (C); Al layer intergrown with melting products after
120 140 160 180 200 220
Slag Mix (Figure 2): Conglomerates of glass/ceramic culets and metal within melting product matrix. The matrix is
dark grey to black with a glassy/amorphous and porous appearance. The pores are either empty, filled with a whitish
mineralisation or include spherical metals/metal-oxides inclusions. Inclusions can have different shapes and forms
(Figure 2). The melting product matrix dominates the particles with metal contents <20-30%. Average density is 2.1
g/cm3 and particles are elongated with an average ratio (AR) of 0.6.
Slag Dense: Melting products with less intense pores and very low metal/metal-oxide content. The particles appear
denser than slag mix. Rare conglomerates with mineral fractions are found. The shapes are less irregular and more
elongated (AR of 0.45) with densities around 2.2 g/cm3.
Sintered Grit: The sintered material is light brown, mineralized and homogeneous in appearance. The strong porous
texture results in low densities of -1.8 g/cm3 and elongated appearance of 0.38 AR.
Metal/slag Aggregates: Similar to Slag Mix but with +50% metal proportion. Often these are large metal parts with
attached melting products with high density +2.7 g/cm3 and AR of 0.51.
Mineral Fraction: The very heterogenic subgroup consist of single glass, ceramics, concrete, rocks or bricks
fragments or culets with different shape (AR = 0.45) but dense appearance.
3. Sample analysis and data reduction
3.1 Process conditions
The discharge probability indicates the ratio of electrical pulses released by the generator, and electrical discharges
where the energy caused damage to the particles. This ratio is a key aspect of HV process-efficiency and a basic
evaluation criteria. It is related to the electrical field strength – applied kV per mm - and relies on selected voltage,
electrode gap, pulse energy, process water conductivity and the water gap between material and electrode. The test
using a fixed electrode gap of 50 mm implies that elongated particles are processed at high water gaps. As a
consequence, at low voltages there is a lower discharge probability. This is found especially in subgroups with no or
low metal inclusions like the Mineral Fraction showed. The discharge probability increases again when higher voltage
are applied. Metal-rich subgroups indicate better discharge probabilities also at higher water gaps. At +180 kV
reliable pulsing is possible for all subgroups. This reflects a minimum field strength +3.6 kV/mm in average. Metal
bearing particles are more accessible for the process as only 2.7 kV/mm are required to have consistent discharges.
2.2 Product Evaluation - Body Breakage Probability (BBP)
In the BBP analysis largest product progeny sizes are recovered to discriminate body from surface breakage. Its
weight% mass loss of the parent particle is used as BBP value. Surface and body breakage events are considered to
be different in energy transfer and should show a different fracture pattern and size reduction. The herein presented
evaluation uses the BBP to characterize the respective breakage characteristics.
Figure 3 – Plot of BBP vs voltage of all tests (A) and specific energy of different fractions (B).
0 5 10 15
+ 45 mm
- 45+40 mm
- 40+30 mm
- 30+20 mm
In average, all tests beside those with full metallic parts indicate body breakage behaviour at BBP +10%. BBP
increases linearly towards higher applied voltages as total energy increases (Figure 3A). If BBP is plotted vs the
specific energy, ESC (Figure 3B), specific distributions appear. The BBP-ECS data from HV breakage indicate similar
trends as used for Axb ore grindability index (Shi et al. 2007) using following equation:
A and b are model parameters fitted to t10 breakage data and the specific energy applied. The A data is the average
value for the produced plateau. The b value is the steepness of the distribution to reach the plateau. The product of
both is used as ore grindability index in SAG application. The use of t10 values in the HV breakage single particle
approach is not useful as it is not sensitive to indicate what energy transfer scenario is active. The BBP data is more
useful as it includes indirect information to potential energy transfer and size reduction. As the BBP vs ECS
distribution indicates a suitable distribution, the above equation is adjusted to investigate the SELFRAG breakage
BBP data for all size fractions presented in Figure 3B were analysed and H & m values were modelled using above
formula for best fit. Table 2 and Figure 4 summarizes the data.
Table 2 – BBP index of different bottom ash fraction.
Figure 4 – Development of BBP index data for size fractions.
The overall data separated in single feed sizes indicate that modelled H plateaus data achieve similar ranges around
70% BBP with a hinted reduction towards finer and larger sizes. Most obvious, towards finer feed sizes the steepness
m reduces significantly. As a consequence, larger particles seem to be more accessible, and confirms the previous
known particle size effect for the process. Limitations are given by the used electrode size and gap. Any larger
material will be out of the active zone resulting in reduced performance. Towards finer sizes available surfaces and
particle strength increases, which seems to influence the HV breakage situation.
Figure 5- BB P vs ESC plot different voltage test for total sample, slag mix (SM) and mineral fraction (A); development
of Slag Mix BBP vs ESC of different fractions.
0 5 10
0 5 10 15
Total Mineral Fraction
10,00 20,00 30,00 40,00 50,00 60,00
mean feed size mm
To further investigate if the H, m and the H*m product are useful indicators to differentiate the breakage behaviour
in the HV pulse power process selected subgroup trends are evaluated. For that the Slag Mix and Mineral Fraction
are selected as they include largest differences in composition and appearance. In addition they include the most
valuable components for recycling. Relevant BBP vs ECS trends are presented in Figure 5 A &B.
Figure 5 A compares the development of BBP data towards increasing ESC of the total sample, Slag Mix and Mineral
Fraction subgroups. The data combine all treated fractions of the relevant subgroups. Different plateaus H are
reached according the plotted subgroup, while steepness m changes only moderately (Figure 5 A, Table 3). The total
sample confirms plateau 70% H similar to Figure 3. Slag Mix data the plateau H increases to 88% BBP indicating a
more intense overall size reduction. The Mineral Fraction shows the lowest BBP plateau data H around 63%. Figure
5 B & Table 3 presents the Slag Mix BBP data of different fractions. BBP plateau data H are stable around 80-90%
confirming data when all fractions are combined. Still declining steepness data m towards finer fractions appear.
Table 3: Data to BBP H*m index at increasing energies and different composition
Slag Mix fractions
The H, m and the H*m data have the potential to indicate and quantify how HV pulse power breakability is influenced
and what breakage mechanism is active. The analysis point to several factors. The BBP plateau value H show how
intense fragmentation can be. Current data confirms the influence of the material composition. Metallic inclusions
support more intense fragmentation, while reduced plateau are indicated in metal-free particles. The process energy
transfer by body and/or surface discharges seem to be the controlling factor which plateau levels are achievable.
High plateaus +70% BBP indicate a dominance of body breakage. Values -70% BBP point to an increasing influence
of surface breakage. As a second factor the steepness m controls how fast the plateau is reached with declining
trends towards finer feed sizes. This effect appears less material dependent (Figure 5A & Table 3), but still indicates
a significant change in the fragmentation regime. The increased surface but also higher strength towards finer feed
sizes seems to promote surface discharges and/or reduces the effects of body breakage by the higher strength. Both
values in combination can therefore describe and quantify HV breakage. The product of both values H*m involves
both material and feed size effects and is a useful indicator of the overall HV breakability of a material in a given size
range in the HV pulse power process.
The BBP data are useful to quantify characterize the breakage results. For the bottom ash an average value of 46.41
H*m is achieved with H around 70% suggesting that body breakage is the dominating mechanism for the material.
Within the bottom ash the Mineral Fraction subgroup indicate surface breakage results shown in low plateau H data
of 63%. The low H*m of 36.17 confirms the reduced HV breakability of the subgroup. Slag Mix subgroup reacts most
efficient with 66.46 H*m. The dominance for body breakage is given by the high plateau values H of 88%. Both
subgroups experience declining trends towards finer feed sizes. This shows that independent on the material
composition the active breakage mechanism will gradually increase towards surface discharges. However, how fast
this will happen still depends on the composition.
To verify the findings above, the resulting progeny particles of the relevant subgroup are further evaluated. Progeny
particles indicate specific fragment characteristics, which supports above findings of how the energy is applied
besides the BBP. In fracture mechanics, various effects occur from elastic deformation to brittle and ductile failure.
In uniaxial state of stress the Young Modulus E and the Poission’s ratio characterize the elastic behaviour of isotopic
materials like rocks. Increasing axial stress ultimately leads to failure or breakage, when material cannot compensate
the increasing stress by elastic deformation. The stress level when it fails is a measure of material strength. Brittle
failure occurs with the formation of a fracture, which is a surface where material has lost cohesion. In general two
types of fractures occur: extensional and shear fractures. Both include different orientation against the direction of
applied stress. Extension fractures are normal to lowest and parallel to maximum principal stress direction. Shear
fracture include a specific angle to the main stress direction. In uniaxial compression when minimum stresses are
near 0 and compression is normal to the surface, tension fractures form in longitudinal splitting. The energy required
depends on the material strength.
The stress regimes for the creation of fractures within the HV pulse power process can be very different and strongly
depend where and how the energy is deposited (Figure 1). In the body breakage situation energy is brought into the
materials. The resulting stress situation causes intense fragmentation due to the higher confining stress regime
inside a particle. Different oriented fracture planes including shear fracture occur, resulting in fine progeny sizes and
very irregular and angular progeny particle shapes (Figure 3 B-D). When the discharge energy is transferred across
the material surface, the breakage mode is comparable to uniaxial compression. Along with a restricted size
reduction, a very specific fracture pattern also occurs during surface breakage events presented in Figure 6.
Figure 6 – Developing fracture pattern according applied stress regime: (A) Surface discharge marks on brick not
broken at 180 kV; (B) Surface discharge marks on a rock with starting perpendicular crack system at 150 kV; (C)
extensional fracture pattern (longitudinal splitting) of brick by surface discharge with perpendicular fracture planes
treated at 210 kV.
Figure 6 A shows two surface discharge marks on a piece of brick at an intermediate test voltages of 180 kV and low
generator capacity without visual fracture formation. The discharges are not able to enter the brick and the
transferred energy via the surface is not strong enough to overcome the elastic behaviour of the material. Figure 6B
presents a fine grained rock sample treated at 146 kV. The discharge mark is more intense and first perpendicular
cracks developed without being persuasive enough. Figure 6C shows a rock brick treated at 211 kV. The processing
result indicate a very distinct fracture pattern comparable to longitudinal splitting with two perpendicular fracture
systems in a wide spacing delivering large progeny sizes. The achieved progeny shapes after surface discharges are
therefore very different to body breakage results.
4. Discussion & Conclusion
The breakage characteristics of bottom ash particles in the HV pulse power process are different due to the
heterogenic mass flow. The separation of bottom ash into subgroups results in a more homogeneous material flow
to detect differences. The herein presented study confirms the influence of electrically conductive material
inclusions in the HV pulse power process. Material composition has a large influence to the breakage results and
confirms the findings of Zhou et al. (2014, 2015a) in a grade splitting application. Feed bottom ash material that
contain metallic inclusions are processed more intensely, resulting in a finer product size, while metal-free particles
stay coarser. The origin of this behaviour is how and where fragmentation energy is deposited. The test could predict
how material composition triggers surface or body discharges. The analysis using a single particle/single discharge
approach at increasing energy level is most useful to quantify the HV breakage and to identify body and surface
breakage events. For that the BBP approach of Zhou et al. (2014, 2015a) was modified. To use only the +/- 10 weight
% mass loss to discriminate body from surface breakage is not reliable as at increasing energy input surface breakage
can also produce finer progeny sizes (Figure 3A). In addition it does not account the effects of a changing feed sizes.
Only the analysis using increasing energy levels to several close feed size ranges similar to the drop weight tests
procedure can provide a realistic evaluation of the HV breakage characteristics. Data analysis of the resulting
distribution (BBP vs ECS) adapting the Axb formula of Shi et al. (2007) provide a suitable approach. The value H
representing the achieved plateau gives the most information to body or surface breakage characteristics. For
bottom ash +70% BBP data H point to body breakage domination. BBP H data -70% point to coarser product sizes
and increasing surface breakage influence. The change from body to surface breakage leads to different fracture
mechanics and changed stress regimes. This is confirmed by the apparent size reduction and progeny shapes. The
steepness m of the resulting BBP vs ECS distribution illustrates how fast the plateau H can be reached and relates to
the feed size distribution. The data points to a significant change in the fragmentation regime towards finer feed
sizes probably linked with increased surface areas and higher particle strength. Dependent on material composition
finer feed sizes seem to promote surface discharges and/or reduces the effects of body breakage by the higher
The applied procedure consider material and feed size effects in the HV breakage evaluation and combines it in the
H*m data. Similar to Axb grindability index in mineral processing, the product of H*m quantifies how easy a size
range of a specific material is to treat in the HV pulse power process. When linked with size reduction, liberation,
weakening, feed composition or target grade data the assessment has the potential model process requirements to
achieve target liberation, pre-weakening or grade splitting applications. The herein presented approach has
potential to be transferred to other materials for the characterisation in the HV pulse power process. The analysis
of breakage characteristic for bottom ash shows that the SELFRAG process delivers its fragmentation energy
preferentially to metal-bearing particles and avoids overgrinding of the mineral fraction. The resulting improved
metal liberation allows high metal recoveries at best concentrate quality (Weh 2015). The further potential usage of
the remaining mineral fraction provides a sustainable recycling strategy for the difficult mass stream.
Alwast, H. & Riemann, A (2010). Verbesserung der umweltrelevanten Qualitäten von Schlacke aus Abfallverbrennungsanlagen,
Studie des Umweltbundesamtes,
Bluhm, H. et al. (2000). Application of pulsed HV discharges to material fragmentation and recycling, IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 7, Nr.5, P. 625-635
Bunge, R.: Wertstoffgewinnung aus KVA-Rostasche. In Schenk, K. (Ed.) 2010: KVA-Rückstände in der Schweiz. Der Rohstoff mit
Mehrwert. Bundesamt für Umwelt, Bern, P. 170-182
Bunge, R. (2016). Recovery of metals from waste incineration bottom ash, online publication UMTEC, HSR Rapperswil
Eggenberger, U. & Mäder, U. Charakterisierung und Alterationsreaktionen von KVA-Schlacken. In Schenk, K. (Ed.) 2010: KVA-
Rückstände in der Schweiz. Der Rohstoff mit Mehrwert. Bundesamt für Umwelt, Bern, P. 104-115
Lichtensteiner, Th & Zeltner, Ch. Wie lassen sich Feststoffqualitäten beurteilen. In: Baccini, P. & Gamper, B. (Ed) 1993.
Deponieriung fester Rückstände aus der Abfallwirtschaft – Endlagerqualität am Beispiel Müllschlacke. Verlag vdf Hochschulverlag
Zürich, P. 11-33
Pretz et al (1998). Aufbereitung von Müllschlacken unter besonderer Berücksichtigung der Metallrückgewinnung. In Tagungsband
Rohstofftechnik im Wandel; Aachener Umwelttage der Fakultät für Bergbau, Hüttenwesen und Geowissenschaften / Aachener
Shi, F. & Kojovic, T. (2007). Validation of a model for impact breakage incorporating particle size effect. Int. J. Miner. Process. 82.
Shi, F., Zuo, W., Manlapig, E., (2015). Pre-concentration of copper ores by high voltage pulses. Part 2: opportunities and
challenges. Miner. Eng. 79, 315–323.
van der Wielen, K.P. et al. (2013). The influence of equipment settings and rock properties on high voltage breakage. Mineral
Engineering; 46-47. P. 100-111
Weh (2015). Effiziente Metallrückgewinnung aus Kehrichtverbrennungsasche mittels Hochspannungsimpulsverfahren. In:
Thomé-Kozmiensky, K. J. (Hrsg.): Mineralische Nebenprodukte und Abfälle. Neuruppin: TK Verlag Karl Thomé-Kozmiensky, S. 167-
Zuo, W., Shi, F., Manlapig, E., 2014. The effect of metalliferous grains on electrical comminution of ore. In: International Mineral
Processing Congress, Santiago, Chile.
Zuo, W., Shi, F., Manlapig, E., (2015a). Pre-concentration of copper ores by high voltage pulses. Part 1: principle and major
findings. Miner. Eng. 79, 306–314.
Zuo, W., Shi, F., van der Wielen, K.P. & Weh, A (2015b). Ore particle breakage behaviour in a pilot scale high voltage pulse machine.
Mineral Engineering, 84. P. 64–73