IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 7 No. 5, October2000

625

Application of Pulsed HV Discharges to

Material Fragmentation and Recycling
H. Bluhm, W.Frey, H. Giese, P. Hoppe, C. SchultheiO, R. StraOner
Forschungszenttum Karlsruhe
Institut fur Hochleistungsimpuls- und Mikrowellentechnik
Karlsruhe, Germany

AB STRACT
The physical basis of electric impulse fragmentation and its applications to the recycling of
composite materials are reviewed. The method is based on the initiation of a pulsed electric
discharge inside the solid dielectric material. With pulse amplitudes of 300 ky material
layers of 2 cm can be punctured. Specific energy deposition, of 5 100 Jlcm at a GW power
level, leads to pressure buildup of 5 10" Pa in the discharge channel. Pressure waves and radially propagating cracks are launched into the solid body, which can lead to the separation of
inclusions from the matrix or to detachment at material boundaries. To induce the discharge in
the solid dielectric it must be immersed in a dielectric liquid with higher breakdown strength.
Most applications use water, which has excellent breakdown strength at fast ramp rates and,
due to its high permittivity, leads to field concentration in the solid dielectric. Electric impulse
fragmentation is a clean physical method without any environmental burden and therefore
well suited for recycling applications. In this paper we consider applications in the fields of
demolition debris, incineration ashes, contaminated surface layers, electric appliances, glass,
and elastoplastic materials. Finally, the economy and the scaling of the technique to large material throughput are discussed.
N

1 INTRODUCTION

ECYCLINC of

waste materials is only reasonable if certain economic

and ecological criteria are met. Economically it is advantageous if
the sum of earnings from the secondary raw materials and costs from
depositing, in case of non-recycling, is higher than the recycling costs.
Ecologically it is rational if the environmental alleviation by use of secondary raw materials is larger than the difference between environmental charges for recycling and depositing. Recycling of products like
concrete, electronic devices, electronic scrap, cables, electric appliances,
etc. requires separation into the constituents. In conventional recycling
plants this is achieved by shredding the materials into small pieces and
extracting the different components. Multiple steps are in general necessary to obtain pure materials. To recover the material components
with their original quality in general is not possible with mechanical
methods. Fragmentation of composites by initiating a pulsed HV discharge inside the solid material in some cases offers an effective method
to separate the material into its components without degrading their
quality

development was to apply it to the disintegration of rocks to obtain a

higher yield of precious minerals and crystals, while conserving their
original size and shape. However, the method was also applied in
drilling of wells and for the destruction of reinforced concrete plates.
Unfortunately, much of the original literature is not easily accessible.
Electric impulse fragmentation is a clean physical method without any
environmental burden and therefore certainly meets the ecological criteria for recycling. For which kind of applications it can also become
economically attractive is still under investigation. In this paper we review the physical basis of electric impulse fragmentation and its applications to the recycling of materials. In Section 2 we describe the most
important phenomena and the energy balance. Section 3 discusses the
basis for the selectivity of the destruction. In Section 4 we present a
typical pulse generator. Some applications are discussed in Section 5.
Finally some thoughts on scale-up and economy are given in Section 6.

ENERGY BALANCE OF
DISCHARGES IN SOLIDS

A 'buried' pulsed electric discharge in a solid dielectric, depositing

The destruction of solid material through pulsed electric discharges an energy of 10 to 100 J/cm within 1to 5 ps heats the spark channel
sometimes called 'electrodynamic fragmentation' originally has been to temperatures >IO4K and creates a pressure of lo9 to 10" Pa. The
investigated since the early sixties in the former Soviet Union, mainly spark channel, initially only 10 to 50 pm wide, expands and launches
at the Polytechnical University of Tomsk [I]. The principal goal of this a pressure wave into the surrounding solid material, which can lead to
N

1070-9878101 $3.00 02000 IEEE

Bluhm et al.: Application of HV Discharges to Material Fragmentation and Recycling

626

its disintegration. The energy balance equation for the spark channel
can be written as
pdV + dE = dW
(1)
where p is the pressure in the channel, and E = Ei El is the sum of
the internal energy E, of the channel products and the losses El due to
leakage at the channel ends, radiation and heat conduction. W is the
energy provided by the discharge.

+-

35

a
e!

25
2o
15
10

creases and its profile becomes triangularly shaped, which is important

for the destruction of composite materials.
To initiate the discharge, the arrangement schematically drawn in
Figure 2 is used: A capacitive energy supply delivers a fast rising voltage pulse of 5500 kV to a rod electrode touching the solid which rests
on a grounded plate elertrode, A dlschargo through the solid will occur if its breakdown voltage is lower than the applied voltage and if
the breakdown strength of any other path outside the solid is higher. A
necessary condltlon for this is that the local electric field inside the solid
body exceeds the breakdown field while it does not in the dielectric liquid. This always can be accomplished if the solid body is embedded
into a dielectric liquid with higher breakdown strength. A further possibility is to concentrate the electric field in the solid and to lower it
outside, This requires a liquid with much larger permittivity than that
of the solid. Finally the path length between the electrodes through the
liquid could be made much larger than that in the body, e.g. if the solid
body is spherically shaped, the shortest length outside the body is $ T X
larger.

Disiancs from spa& channel [mm]

Figure 1. Time dependent calculated pressure wave profiles origlnating from a discharge channel in a Plexiglas body [Z].

HV Electrode +
I.&
\

Figure 2. Schematic of a setup to induce an electric discharge through

a solid dielectric material.
To determine the channel expansion and the pressure field around
the spark, one has to add the momentum equation and the equation
of mass conservation, and solve the system with the appropriate equation of state both for the channel products and the solid, A complete
numerical simulation is very difficult and for heterogeneous composite
materials probably impossible. However, for initial guidance, B rather
simple hydrodynamic model using boundary conditions for the radius
of the expanding spark channel, derived from experimental observations, has been adopted in the literature [2]. Also losses from the channel are neglected and an equation of state of the form

is used [3],where y is the effective ratio of specific heat at constant pressure and constant volume. Results from this kind of analysis for spark
channels in PlexiglasTMsamples are shown in Figure 1 [2,3]. While
the pressure pulse propagates into the solid material, its amplitude de-

Figure 3, Dynamic breakdown strength of liquid, solid and gaseous

dlelertrlcs as a functioh of the voltage ramp rise time. Below a critical value of the voltage ramp time, the breakdown strength of water
becomes larger than that of most solld dielectrics

A suitable dielectric liquid is water, of which the breakdown

strength increases strongly if the risetime of the voltage pulse is reduced, This is schematically shown in Figure 3, where the breakdown
field strength of water is compared to that of solid rock material and
transformer oil, as well as gas. It is seen that at short voltage ramp risetimes, the breakdown strength of water becomes higher than that of the
solid material. The effect was first discovered in the late fifties both at
Tomsk Polytechnical Universlty [l]and at Aldermaston, where it was
utilized for the design of low impedance high power pulse forming
lines with water dielectric [4]. This can be understood by the streamer
mechanism of electric breakdown in liquids, where the streamer velocity only weakly depends on the macroscopic electric field strength,
but is determined by the field at the tip of the gaseous filaments (for
positive streamers) or by charge buildup (for negative streamers) at
the head of the streamer [5]. Nevertheless it is difficult to understand
why the breakdown strength of liquids can become larger than that of
a solid material, since it is well known that the intrinsicbreakdown

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 7 No. 5, October 2000

627

field strength (for thin samples and short periods of stress) of solid diSince breakdown is a stochastic process not every pulse will lead to
electrics is higher. Recently it has been suggested and experimentally a discharge in the solid. The probability can vary over a wide ran e,
demonstrated that the breakdown of gas filled pores inside the solid depending on the geometry, the dielectric properties of the material, &e
electrolytic conductivity of the water in the process chamber, etc.; howmay play an important role in this process [6].
ever in an optimized configuration 773 = 0.8 to 0.9. Only a fraction 7 4
Although transformer oil has a higher breakdown strength than wa- of the available electric energy is deposited in the discharge channel.
ter, it is not appropriate for many technical applications, especially for The rest is wasted in the generator, i.e. appears as ohmic and dielectric
recycling. Water does not only have a good insulation strength if it is losses. However, experience has shown that r)4 = 0.65 to 0.7 can be
stressed for short periods of time, but also has a very high permittivity achieved. Since the electrodes do not always touch the solid, part of
and thus pushes the electric field lines into the solid material, which in the arc channel can arise in the surrounding liquid, where it is less efgeneral has a much lower permittivity Of course this effect only occurs fective. Therefore, another efficiency coefficient 115 must be introduced,
if we have a layered distribution of liquid and solid materials between accounting for this effect ( r ) ~N 0.9).
the electrodes. However this is the most frequent situation in an operThe product of all these efficiencies leads to qt and the estimate
ating discharge vessel filled with pieces of material to be fragmented. that a fraction ~ 0 . of4 the stored electric energy can be released in the
useful part of the arc channel. This energy W splits into different forms
W = A + E, + El
(3)
A = [pdV
J

where A is the mechanical work performed by the expanding channel

in the surrounding solid. If losses are neglected due to the fast pulsed
character of the process one can estimate the thermodynamic efficiency
1)' = A/W = 1 - Ei/W. W is obtained from current and voltage
measurements. Using Equation (2) for E$withy = 1.1to 1.2 and deriving V from experimental observations of the channel radius and p
from the simplified hydrodynamic simulations mentioned above [l,21
one estimates 7 6 = 0.1 to 0.2. This relatively small value results from
the
large part of the internal energy that is spent for dissociation and
Figure 4. Electrode arrangement to achieve scraping of a surface. This
ionization.
Losses start to become important if the ratio of the channel
requires that the HV breakdown strength between the electrodes is
radius R, and the channel length L, > 0.1. Taking all efficiencies
larger outside the body than inside.
into account leads to the conclusion that qt = 4 to 8%of the electrical
Water also allows the realization of a configuration where both the energy is available for the destruction of the solid material. Part of this
HV and the grounded electrode contact the solid at the same side (Fig- energy is expended to deform the solid. If the main application is fragure 4). In this case the discharge can be carried through the solid and mentation one may consider only that fraction of energy beneficial that
blow off pieces from its surface. Scanning across the surface with this is used to create new surfaces. In this case the fragmentation efficiency
qf becomes
pair of electrodes, one can remove material layers from large areas.
ws

To assess the efficiency of material destruction by pulsed electric discharges one has to consider the following steps: Charging of the capacitor certainly can be carried out with efficiencies 71 > 0.95. For most
industrial applications tap water with a conductivity 0.6 mS/cm is
chosen as the dielectric liquid. In many cases the conductivity rises
during the process because of salt release from the fragments. Therefore electrolytic current losses occur before breakdown. To minimize
these losses, most of the electrode rod needs to be covered by a solid
insulating material. If the electrode tip does not contact the material,
the discharge may be delayed and the electrolytic losses can become
quite large before breakdown, depending on the water quality, and
the material filling factor in the interelectrode gap. Values as low as
0.2 to 0.4 have been measured for the pulse coupling efficiency 772 on
FRANKA-Stein (a semi-industrial prototype for concrete fragmentation,
see Section 5.1) in positive polarity In negative polarity 112 fell to 4l.1.
However,by controlling the water quality and filling fraction, the losses
can be reduced to 10% and 72 ---t 0.9. We have found that a conductivity of 5 2 mS/cm can be tolerated in the process water.
N

7lf = A 7 t

(4)

where w is the specific free surface energy, and S the area of the newly
created surface.
Using the assumption that most of the energy A is expended for
plastic deformation of the solid Epl and applying the approximate relation 111
= 9wSln

(z)

(5)

where T~ is the yield strength (3 to 300x10' N/m) and G the shear

modulus (1to 4 ~ 1 0 ~ ~ N / mOne
* ) .obtains w S x (0.013 ---t 0.047)A;
i.e. M 0.04 to 0.32%.
This value has to be compared with the corresponding value of mechanical fragmentation devices which is of the order of 0.002 to 1%,depending on the degree of fragmentation. We can therefore conclude that
electric impulse fragmentation is energetically comparable, but not superior, to mechanical fragmentation methods. Consequently one should
use the electric method especially for those applications where its technological benefits become obvious. Among these the smaller width of

Bluhm et al.: Application of HV Discharges to Material Fragmentation and Recycling

628

the grain size distribution curve, the relatively small amount of contamination introduced by the process, and the low heat transferred to
the milled material on the average. However its main advantage is the
high degree of selectivity

3 SELECTIVITY OF
DESTRUCTION
Recycling of composite materials (e.g. concrete, fiber glass enforced
plastics, etc.) or material composites (e.g. electrical apparatus, circuit
boards, etc.) requires separation into the basic components. Electric impulse destruction can produce separation at material or grain boundaries via three effects.

-I

4-HV

J7
Discharge Channel

\~.

_/

Weak

Medium
Pressure Wave

Strong

Figure 5. Mechanisms by which components in a composite material can be separated: Top: Metallic inclusions or inclusions with high
permittivity can attract the discharge track. Middle: A compression
wave can be transformed into a tensile and shear wave by reflection
and refraction at an inclusion and separate it from the matrix. Bottom: A crack propagating from the discharge channel into the solid
can branch around an inclusion if its mechanical properties are different from that of the matrix.

A second more important effect starts from cracks created in the

immediate surrounding of the channel. As can be concluded from Figure 1, the pressure exerted by the expanding channel almost always
exceeds the tensile strength of materials and leads to the formation of
cracks. If cracks have been formed in contact with the spark channel,
channel products can penetrate into them and exert force on the crack
walls. The character, dynamics and intensity of the crack formation
is determined by the rate of energy deposition in the channel and by
the properties of the material. Brittle materials show a large number
of cracks in a radial zone of 3 mm around the discharge channel,
created early in the discharge. During a later phase, a number of radially propagating cracks start to grow from this zone. The extension
and crack density around the channel correlates with the rate of energy
release [l].However, the number of cracks reaching the surface of the
probe depends more on the total energy released in the spark channel.
Consequently, one can conclude that to achieve comminution, a high
power of the pulse is required while the detachment of large fragments
is most effectively achieved with high pulse energies deposited over a
longer time interval. For the selectivity of fragmentation it is important to realize that material inhomogeneities in general, and acoustic
inhomogeneities in particular, influence the propagation of cracks in a
composite material. The reason for this is the existence of increased mechanical stress at the boundary of an inclusion. Stress waves reflected
from inhomogeneities or inclusions can interact with the growing crack
before the inclusion is reached [7-101. If cracks hit the inclusion they
can branch, depending on the angle of incidence, as schematically indicated in Figure 5 (bottom), and separate the inclusion from the matrix.
A third effect leading to separation at the interface of an inclusion
and the surrounding medium is connected with the action of an incident
compressive wave launched from the discharge channel [ll-131. This
is schematically shown in Figure 5 (middle). Initially, and in the immediate surrounding of the spark channel, the wave has the character of
a shock wave, while later it develops into a compression wave. It has
been shown [ll]that a compressive stress wave is converted into a tensile wave after refraction and reflection inside the inclusion. At small
amplitude, separation occurs first at the shadow side if the inclusion
has a higher acoustic impedance. Complete separation over the entire
interface of the inclusion and the matrix was observed at sufficiently
high wave pressures.
An important question is whether inclusions can be separated, but
remain unbroken, from the matrix. It is well known that a shock wave
arriving at a free surface or at a material interface with a strong jump in
acoustic impedance can lead to spallation [14].A strong advantage of
electric impulse destruction is that the energetic parameters of the pulse
power generator can be varied over a wide range, and adapted to the
physical, mechanical and acoustic properties of the composite. Because
of the complexity of composite materials, this has to be determined
experimentally for each material.
N

At inclusions where the dielectric properties are very different from

that of the matrix, the electric field intensity can be magnified and at4 THE FRAGMENTATION DEVICE
tract the discharge track to the inclusion, where it can continue to develop along the boundary This is shown schematicallyin Figure 5 (top),
Since gas bubbles, created during the HV discharge, must disapwhere a conducting sphere has been embedded in an insulating matrix. pear from the discharge vessel before the next pulse can be applied,
In this case separation of the inclusion from the matrix is caused directly the pulse frequency cannot be extended much above 10 Hz. Therefore
a material throughput per discharge section of 500 kg/h, which is
by the discharge channel.
N

IEEE Transactions on Dielectrics and Electrical Insulation

necessary for many industrial processes to become economic, requires
an interelectrode distance of 5 cm. To break down a material bed
of this size electrically, a voltage pulse of 500 kV is required. A fast
rising voltage pulse of this amplitude can be generated with the help of
a Marx generator [15]. This Marx generator should be able to produce
pulses with an energy content of 55 kJ and deliver a large fraction of
this energy to the discharge channel. This requires a low internal resistance of the Marx, which is also necessary to obtain a fast rising voltage
ramp. High power, typically 5 5 GW ,is needed if a large degree of
fragmentation is desired. Therefore, the Marx and its connections to
the discharge vessel also should have a low inductance.
N

Vol. 7 N o . 5, October 2000

629

Figure 7. Schematic of the wiring inside the Marx generator to minimize the magnetic stray field in the surrounding. In the right part of
the Figure the calculated magnetic stray field is presented.

the Marx, careful shielding is required. It turns out that shielding of

the magnetic stray field is difficult to achieve with non-ferromagnetic
materials. On the other hand p-metal may become too expensive for
some applications. However, skillful internal wiring of the Marx can
lead to compensating currents and thus reduce the outside magnetic
field. As shown in Figure 7, the field decays rapidly with distance from
the Marx. Another source of electromagnetic interference and noise is
the spark in the discharge vessel. Therefore, in the setup of Figure 8 the
discharge vessel has been surrounded completely by a Faraday cage
which at the same time serves as sound insulation.
In many cases the discharge vessel is built from PE, except for the
bottom part, to reduce electrolytic current losses. Since the process
water can become increasingly conductive during operation, the HV
electrode must be insulated anyway except for a small tip. In this case
Figure 6. Photo of a 400 kV, 1.8 kJ Marx generator designed to operit is acceptable to use a metallic vessel which then itself can shield the
ate at 10 Hz with a component lifetime of 10' pulses. The generator
electromagnetic noise. To relieve the electric field stress at the insuladischarge period is 2.5 ps.
tor/metal/water triole ooint. the tip at the end of the electrode has been
shaped like a mushroom. Depending on the process, the bottom part
The base line
Of a Marx-generator
with these re- of the vessel is built as a mesh, a grid, or as a closed half-sphere,
quirements is shown in Figure 6. This generator, built for recycling of
In Figure the
and current wave shapes and the derived
metals from dross, consists of 7 stages with two capacitors of 72 nF per
stage, Its inductance, including the contributions from the lead and spark channel resistance are presented for different conditions in the
the HV electrode,has been measured as 7,7 pH, The internal resistance discharge vessel. For this series of measurements an interelectrode dissipa1
shown h~Figure 9~
of this Marx is largely determined by the resistance of the spark gaps, tance Of 30mm has been chosen. The
which
has
been
measured
with
a
fast
resistive
voltage
divider, includes
used as closing switches in the configuration, and has been measured
80%
the
inductive
voltage
drop
at
the
electrode
and
spark
inductance. To
to be 0.5 fi for the present design. With this small resistance,
of the available energy can be deposited in the reaction chamber, Each calculate the spark resistance RE,it must therefore be corrected to
u ( t ) - L$
capacitor is charged to 60 kV Thus a 1.8kJ oscillating pulse train with a
RE =
(6)
maximum amplitude of >400kV and a period of 2.5 psis achieved. Dei(t)
spite the large voltage
a n high
~ r@%
alifetime of
If the material (concrete)fills the entire gap, a rapid breakthrough is
>lo8
Pulses has been rated the capacitor vendor AtesYs. The switch achieved during the initial rise of the voltage pulse and the mean value
are
tungsten
with a
Profile to
a of the spark resistance remains >2 fi. It is observed that the spark
homogeneous burn of the electrode material. PE is used as the switch resistance periodically rises near the zero-crossing of the current, we
housing material.
attribute this to an inflow of cold material from the channel wall and to
The Marx itself is housed in a thick walled metallic tank visible be- heat losses dominating over heat production at this time, both leading
hind the generator in Figure 6. Transformer oil is used in the tank for to cooling of the channel plasma. If a large fraction of the discharge
HV insulation and as a cooling medium. The Marx pulse is transmit- runs through water, an ignition delay occurs and the voltage begins to
ted through a Plexiglas interface to the reaction chamber using a large drop due to electrolytic current losses. Correspondingly the maximum
diameter flexible metallic tube.
attainable spark current and the energy deposited into the spark also
Important aspects of operation in an industrial environment are HV are
safety, electromagnetic interference, and noise protection. HV safety
Spark resistance RE,ignition delay, and the energy efficiency can
can be assured by standard regulations and will not be discussed here. be used to control the operation, determine the filling level, and the
However, to operate electronic devices safely in the environment of interelectrode distance.
I

630

Bluhm et al.: Application of HV Discharges to Material Fragmentation and Recycling

Material feed unit

HV processing unit

HV Generation
Material classification

Figure 8. Complete setup of an industrial demonstration facility for metal recycling with screen box and sound insulation

5 APPLICATIONS

concrete aggregates. Nearly all particles consist of single minerals. Under an optical microscope, the gravel fraction (>2 mm) is apparently free
5.1 RECYCLING OF BUILDING
from contaminants and baked particles. Spherical particles are domiMATERIALS
nant in the sand fraction (<2 mm, >0.5 nun)and seldom with cement
Approximately 1 m3 of concrete per inhabitant per year is used up adhesions. The total fraction of cement in the aggregate part (gravel,
for building purposes in Germany. Similar quantities are consumed in sand) is -1%.The recycled aggregates are not mechanically predamother industrialized countries. The raw materials gravel, sand and ce- aged and fulfill the increased demands of the frost-dew resistance acment are completely taken from natural resources. On the other hand cording to DIN 52104. Concrete made from these recycled aggregates
30 million tons residual masses of concrete, mortar, and brick are cre- has the same material strength as that from natural aggregates.
The radiography of the silt fraction (<0.063 mm) is free of quartz. It
ated per year in Germany The rate of reutilization of these building
materials is quite low and restricted to secondary constructional opera- consists of calcium-silicate-hydrate, ettringit (a Portland cement mintions like backfilling, noise protection dams, etc. Reutilization without eral), calcium-aluminum carbonate hydrate, and calcite.
degradation requires an improved separation into sand, gravel, and ceFrom a chemical point of view the cement fraction is very similar
ment. Crushing the material with multistage jaw breakers or impact to cement. On a laboratory scale cement clinker has been produced by
mills cannot separate its constituents and produces a large fraction of baking it at high temperatures. Recycling the cement fraction has the
dust and small particles.
following attractive advantages:
Concrete is a composite heterogeneous material and therefore well
1. It saves raw material (clay, lime).
suited for separation into its original components by electrodynamic
2. Milling of the raw materials becomes unnecessary
fragmentation. Microcracks between the aggregates and the cement
3. No decarbonization of the milled raw material is required. (The baking
matrix already exist in the unstressed concrete. Alternation of loads exprocess consumes 80% of the energy investment.)
4. Reduction of COz-emission is possible. (7% of the world wide CO2
pedites the detachment of aggregates from the cement matrix. Since the
release is due to cement production.)
acoustic inhomogeneities are rather large in concrete, ideal conditions
5.
No waste disposal of the cement fraction is required.
exist for separation by pressure waves. Also the pressure impulse at
the discharge channel mainly creates tensile and shearing forces, conOf course energy is necessary to dehydrate the cement fraction. If
ditions at which the strength of concrete is low. Therefore cracks will 50% of the raw materials used for the cement production are replaced
originate and spread from the channel.
by the recycled cement fraction, a 10% saving of the energy costs for
Figure 10 demonstrates that pre-broken concrete indeed can be sep- the burn process can be expected.
arated completely into its components.
In Table 1 the specific productivity of concrete fragmentation is
Figure 11 shows the achieved grading curves after different treat- listed. These data have been achieved at the lab facility FRANKA 0.
ment times, together with the initial grading curve of the natural ag- This generator is able to produce pulses with an amplitude of 250 kV
gregates for the specific concrete (according to DIN) [16]. It is striking rising in 200 ns, and with an energy content of 750 J. Operating at 4 to
that no coarse fragments appear in the grading curve of the separated 5 Hz its power consumption is 3 to 4 kW. In these experiments 2 kg
N

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 7 No. 5, October 2000

631

200
16

1W

12
0
8

-1CQ

L
a
203

16

1W
12
0
8

-1W

Figure 10. Pre-broken concrete piece before (right) and after treatment
in the discharge vessel of FRANKA 0. Sand and gravel are recovered
without degradation. The cement fraction (aside of the steel pieces)
can be baked to produce cement clinker.

4
2

4llSlO

m~

...... .. ...

..

....

...~
......

24

[kVl
2w

015

2
4
Mesh-width [mm]

16

12

Figure 11. Grading curves of fragmented concrete after different numbers of pulses. DIN 1045 is the original grading curve of the aggregates. For comparison, the particle size distribution after heat treatment is shown also.

Table 1. Productivity of concrete fragmentation at FRANKA 0.

16

1CQ
0

0.2

-lW
4
2

Figure 9. Voltage u ( t )and current i(t) wave shapes and the derived
spark channel resistance RE for different conditions in the interelectrode gap. For this series of measurements an interelectrode distance
of 30 mm had been chosen. The voltage trace shown has not yet been
corrected for the inductive voltage drop.

Parameter
Productivity
AV. electric power

Value Unit
160

kg/h

on the water quality So far a throughput of 280 kg/h could be realized.

The limitations result from the strong quality requirements for the secondary aggregates. A closed water reprocessing circuit has been added
to keep the water conductivity low. The specific energy consumption
is similar to that achieved for the smaller facility FRANKA-0.
However
the value of 20 kWh/t includes a contribution of 6 kWh/t from water
reprocessing.

of concrete were destroyed within 45 s of operation at a repetition rate

of 4 to 5 Hz.
To demonstrate a larger throughput of 1000 kg/h the semi-industrial
prototype facility FRANKA 2 alias FRANKA-Stein shown in Figure 12
5.2 TREATMENT OF INCINERATION
was built. The concrete lumps are transported to the processing chamASHES
ber with the help of a vibratingconveyor. The material treatment time is
Thermal treatment of municipal solid waste is an effective method
controlled by the conveyor vibration speed and by gates at the entrance
and exit ports of the chamber. The Marx consists of 6 stages powered of waste disposal, which becomes increasingly important. It does not
from a 10 kW charging unit. At 60 kV charging voltage, the Marx out- only reduce the volume to be dumped but is a valuable source of enput rises to the 350 kV pulse amplitude within 0.2 to 0.4 ps, depending ergy and a resource for metals and mineral building materials. The

632

Bluhm et al.: Application of' HV Discharges to Material Fragmentation and Recycling

Figure 13.

FRANKA 1 before shipment to an incineration plant.

FRANKA 1is used to separate metallic components from the ashes and

to immobilize heavy metals in the ashes.

Figure 12. View of FRANKA-Stein, a semi-industrial prototype to

demonstrate concrete recycling with complete separation of the aggregates at throughput of ,$ lt/h. The Marx-generator is housed in
the top cylindrical case.
utilization of ashes as an aggregate for the production of concrete requires the separation of metal and the immobilization of heavy metals.
Heavy metals can be extracted from fresh ashes in contact with water.
An unavoidable content of anhydrous lime is responsible for the high
pH value (>U)
in ashes. Thus the elution of lead ions in water is increased by a factor of 1000 over that in a pH-neutral solution. Therefore,
fresh ashes must be stored for a period of at least three month before
utilization as a building material. During this time absorption of CO2
from the atmosphere reduces the pH value.

Figure 14. Comparison of heavy metal elution from treated and untreated fresh ashes. Also shown are the values of the German regulation LAGA 22.

5.3

REMOVAL OF SURFACE
LAYERS

The configuration shown in Figure 4 can be used to remove surface

layers contaminated by hazardous chemicals or radionuclides. Laboratory experiments have been carried out at Textron [17], Tomsk [l],and
at our own laboratory, where a device was built that can be moved in
all three dimensions above a 2x 3 m2 large water filled basin [16].
While we have tried so far only single pairs of electrodes, long parallel strips of electrodes have been used at Textron. It was expected that
We have found that under-water electrodynamic fragmentation in this configuration breakdowns would travel randomly along the elec(UWEDF) to separate the metal from the ashes also reduces the pH value. trode gap and that removal of concrete at one location would increase
This is attributed to the production of free OH radicals created in the the breakdown strength there and transfer the discharge to another podischarge channel and by the shock wave launched from it. Subsequent sition. By this kind of self-regulation, a uniform depth across the scabmeasurements of heavy metal elution show spectacular reductions, so ble path was predicted. The prototype system consisted of a 120 kV
that in principle storage can be replaced by an on line treatment with Marx delivering a 0.8 to 2 kJ pulse at a repetition rate of 5 to 40 Hz. The
average scabble speed was 5 to 20 cm/min, and thus a factor of 10
UWEDF directly after the furnace.
larger than that of a low voltage (20 to 30 kV) electro-hydraulic scabble
To demonstrate this process in an industrial environment, the fa- system based on water breakdown, although the energy consumption
cility FRANKA 1 was built and installed at a municipal incineration of the latter was a factor of 4x larger. The specific energy consumpplant. FRANKA 1 (shown in Figure 13 before shipment to the incin- tion of the HV system was 500 to 1000 J/cm3. Trials were also conducted
eration plant) can treat 2 tons of ashes per hour. FRANKA 1 operates at a US Dept. of Energy (DOE) test site where a uranium contaminated
with 7 Hz at a mean power of 10 kW and produces voltage pulses of concrete layer was removed from the plant floor. According to the ex350 kV amplitude. Figure 14 shows the values of heavy metal elution perience gained in these experiments, the technique can be used to defrom the ashes before and after treatment with the FRANKA 1 facility contaminate large floor areas. The radioactive products released during
For comparison the threshold values of the German LAGA 22 regula- the process are contained either in the concrete rubble or in the water
tion for reutilization as building material are shown also.
as fine suspended particles. These components must be removed from
A problem of this process is the strong enrichment of salts in the the water by filtering or evaporation of the water. Recontamination of
process water, leading to increasing electrolytic losses. An important the concrete surface laid open cannot be avoided completely, but can
be kept small by fast recycling of the water.
part of the process is therefore desalting of the process water.
N

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 7 N o . 5, October 2000

bioteshnici

633

msti ated

Figure 15. View of removed surface layer from a block of reinforced

concrete. Here the two-electrode system shown in the upper left corner of the photo was used.

As for the case of concrete decomposition the conductivity of the

process water rises with time and needs to be controlled. This is especially important for the parallel strip electrodes used in the experiments
at Textron. We therefore preferred rod electrodes that can be enveloped
entirely by an insulator except for a small section at the tip. Also the
pulse amplitude should be raised to >400 kV. At 350 kV the specific energy consumption was as low as 70 J/cm3 in our experiments. A result
from these trials is shown in Figure 15, where the concrete had been
removed up to the first grid of reinforcement steel. Even lower specific energy consumption has been found by Kurez et d. They used a
powerful 420 kV, 19 kJ per pulse Marx generator and a comb-like electrode system to destroy reinforced concrete plates. The electrodes were
connected sequentially to the generator. With this system they found
values between 7 and 30 J/cm3, depending on the number of reinforcement grids.

5.4 RECYCLING OF OTHER

PRODUCTS
Besides building materials, numerous other composite materials
have been explored. We can divide these into two groups: Material
composites containing metallic and dielectric components like electric
appliances, spark plugs, circuit boards, etc.; and brittle homogeneous
materials like glass, silicon, coloring pigments, minerals etc.
Easy separation of metallic and nonmetallic components can be
achieved directly for small electric appliances as shown in Figure 16.
Large electric appliances need coarse crushing before processing. Unlike in conventional recycling, the metallic components can be retrieved
easily as complete parts, enabling much easier recovery of precious
metals. It is assumed that the discharge occurs at material interfaces
and thus detaches the bonding between components.
The interest of using electrodynamic fragmentation to mill or destroy homogeneous materials is based on the observation that relatively
small amounts of contamination are introduced by the process and that
more favorable grain size distributions, without a large fraction of fines
can be achieved, e.g. contamination poor milling of borosilicate glass for

Figure 16. Pulsed electric discharges can be used to separate metallic

and dielectric components in electric appliances (a). The result of a
razor treatment after a few pulses (b),

The glass particles suspended in the process water were sucked off
continuously through a filter of suitable pore size. Iron contamination
from electrode burn was further reduced through magnetic separators
at the entrance to the sedimentation pit. Compared to the conventional
milling process not only the amount of contamination was reduced,
but also the grain size distribution was much smaller and thus the useful yield was increased. Another advantage is that a relatively small
fraction of the material came into contact with the hot channel products and became molten. Thus the porous structure on the surface of
the borosilicate grains, an important feature for biological applications,
was preserved for a larger fraction of particles than with conventional
milling.

5.5 RECYCLING OF
ELASTOPLASTIC MATERIALS
Electrodynamic fragmentation turned out to be less successful for
elastic or impact resistant materials. In an attempt to improve the performance of the method in this field of application, investigations were
launched at Tomsk Polytechnical University [NI, in which elastic materials (in particular rubber) were immersed into liquid nitrogen (LNz)
to increase their brittleness.
However, the success of this approach does not only depend on the
increased brittleness of the material but also on the dielectric properties
of LN2. A real drawback of LN2 as a dielectric liquid is its small permittivity of E, = 1.454. Thus field intensification inside a solid material
immersed in the liquid cannot be expected. On the contrary, since many
solid materials have a higher permittivity, a weak field enhancement in
the liquid may occur.

634

Bluhm et al.: Application of HV Discharges to Material Fragmentation and Recycling

Another reason for weakening of the electric breakdown strength
in LN2 can be the accumulation of ice crystals within the liquid, originating from frozen out air humidity if the surface is exposed to normal
atmospheric air. Of course these problems could be reduced significantly by pressurizing the LN2 and by minimizing the contact of its
surface with atmospheric air.

The removal of bubbles from the LN2 bath also can limit severely the
achievable repetition rate. If typically 125 J of energy are deposited in
the discharge channel and converted into heat, -0.5 1 of N2 gas must
be removed from the bath between shots. Although the viscosity of
LN2 is relatively small, O . 2 l ~ l O -Pas
~ at 77K, as opposed to lop3Pa
s for water at 20T,which facilitates bubble movement to the surface
Sample Thickness [mm]
under buoyancy force, corresponding experiments 119,201 have shown
that N2 gas bubbles in LN2 submitted to buoyancy force only, move
Figure 17. Dependence of the breakdown voltage in LN2 on sample
thickness and for different ramp rates. Also shown is the breakdown
upward with a speed of only 0.2 m/s. Given a depth of the LN2 bath
voltage of rubber at cryogenic temperatures. 1181.
of 20 cm, the pulse rate will be limited to <1Hz if complete removal of
bubbles between shots is required. Only practical experience will show
Although a considerable amount of information is found in the liter- however, what the real repetition rate limitations are.
ature on the breakdown characteristics of LN2, very little is of relevance
for the dynamic stress situation typical for electrodynamic fragmentation. A summary of the results obtained from experiments conducted
at Tomsk is shown in Figure 17 [lS]. Here the change of breakdown
voltage has been depicted as a function of sample thickness in the case
of LN2 for three different pulse rise times: 200 ns, 1ps and a 5 ms sine
wave, and for two different rubber samples: car tire rubber and vacuum
seal rubber. The risetimes always refer to a pulse amplitude of 250 kV
and thus can be expressed as ramp rates of 1.25 MV/ps, 250 k V / p and
50 V/ps. It was observed that the breakdown behavior of LN2 differs
significantly from that of other dielectric liquids, like transformer oil,
glycerol, ethanol, or water (at normal conditions).
Only for LN2 layer thicknesses (gap widths) 120 mm the breakdown
voltage rises linearly The increase becomes much slower, >20 mm. This
behavior probably is connected with a change of the breakdown mechanism from an area to a volume effect. The probability for the appearance of bubbles grows if the stressed volume increases. Such bubbles
promote the formation of streamers inside the volume and reduce the
macroscopic breakdown field strength. The formation of bubbles in
the LN2 bath is of particular relevance since the temperature of liquid
gases under atmospheric pressure stabilizes close to the boiling point.
In such liquids, even the minute heat production inferred by prebreakdown currents can lead to the appearance of bubble chains, along which
premature breakdown can occur, before a discharge path through the
Figure 18. FRANKA 3 built to explore the applicability of electrodysolid material has been established.
namic fragmentation under LN2. The Marx-generatoris located in the
It can be concluded also from Figure 17 that the variation of breakbarrel hanging from the roof.
down strength with pulse rise time is much less pronounced than for
water. While in water, the breakdown strength increases by a factor of
To explore the applicability of electrodynamic fragmentation under
10when passing from a pulse rise time of 1ps to 200 ns, the correspond- LN2 for industrial processes the facility FRANKA 3, shown in Figure 18,
ing increase in LN2 is merely 25%. On the other hand the investigated was built. A Marx generator of similar design as for the other FRANKA
rubber materials showed a linear increase of breakdown voltage with facilities allowed to produce pulses of 300 kV with a ramp of 2 MV/ps
sample thickness.
and an energy content of 440 J. It was contained in a barrel hanging
From these results, the authors of I181 concluded that to induce the from the roof above the LN2 process vessel and was isolated with SF6.
discharge in a solid material like rubber with greater probability than The electrode gap was adjusted at 30 to 40 mm. The stainless steel LN2
in LN2, one should apply pulses with risetimes of (0.2 to 0 . 5 ) ~ 1 0 -s ~ process vessel is presented in Figure 19. It was placed on a scissors-type
elevating platform below the Marx generator. Using this facility, a large
and pulse amplitudes of 200 kV for samples to 30 mm thick.

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 7 N o . 5, October 2000

635

6 SCALING AND ECONOMIC

CONSIDERATIONS

Figure 19. Details of the LN2 discharge vessel of FRANKA 3.

number of components was treated:

1. Components from motor vehicles containing rubber or other plastic
2.
3.
4.
5.

parts,
Optical fiber wave guides,
Printed circuit boards,
Laminated plastics, and
Pressure tubing, vacuum seals, etc.

Gross separation of metal-rubber and metal-plastic composites was

easily achieved. However, small traces of rubber remained on the
metallic surface and could not be removed even by prolonged treatment. Also little craters appeared on the metallic surface at arc foot
points.

The separation of components from thin sheets of material, like

printed circuit boards, could not be achieved efficiently In this configuration the channel products can escape rapidly from the channel
and prevent the buildup of any significant pressure.
Also it was found that many organic materials were decomposed
at the high temperatures in the discharge channel and, during cooling,
lead to new uncontrollable substances. Although the quantities were
small and may be tolerable in some processes, they are certainly unacceptable in the treatment of pharmaceutical and food products. Therefore, and because of the appreciable costs of LN2, we have been unable
so far to identify an industrial application that justifies electrodynamic
fragmentation or milling under LN2.

The economy of any recycling technique is determined by the machine price, the specific energy consumption, operating and maintenance costs and by the number and quality of personal to run the facility. Most industrial applications need large throughput to become
economical and therefore considerable extrapolations from present laboratory type electrodynamic fragmentation devices to industrial size
facilities are necessary, leading to big uncertainties in cost estimates.
It is obvious that the quantity of material that can be treated per
arc channel is limited and cannot be increased much above that of an
optimized laboratory device. The repetition rate must stay 5 15 Hz to
remove the gas bubbles between pulses. Also it does not seem reasonable to raise the pulse amplitude appreciably above 500 kV, since the
expenditure for insulation may become prohibitive. The pulse amplitude determines the possible length of the discharge channel and thus
the accessible volume in the treated material. Other parameters that
have an influence on productivity are the power of the pulse and its
energy, Augmenting the energy per pulse is however counteracted by
a reduction of the discharge channel resistance leading to smaller efficiencies. Nevertheless, raising the power may lead to a certain gain in
productivity, especially if milling is the main task of the device. The
achievable increment in productivity will of course depend on the specific product. A systematic study on the disintegration of granite samples has been carried out in [21]where an optimum set of parameters
in an energy-field plane was derived. However, the term 'disintegration' was not specified and therefore it is difficult to relate these data to
specific productivity Considering concrete fragmentation for complete
separation of the aggregates, we expect that a factor of 3 to 5 increase
in throughput over that of our FRANKA facility can be achieved for an
optimized discharge channel, i.e. a throughput of 1000 to 3000 kg/h
of completely separated concrete may be produced per channel with an
average power of 30 kW and a pulse energy of 2 kJ operating at 15 Hz.
Thus to realize an industrial facility with lo5 kg/h, 100 arcs operating
simultaneously are required. Every arc needs a certain process space
so that it does not interfere with its neighbors. Either parallel or serial
arrangements of the active arc zones are conceivable.
To produce 100 arcs simultaneously does not necessarily mean that
all components need to be multiplied by this number, e.g. the Marx
generator can run at a higher frequency and distribute its pulses alternately to different discharge sections. Also a capacitor charging unit of
sufficient power can supply several units in parallel. Since the price of
a power supply does not increase proportional to the power there is a
large saving potential in the scale-up of a facility
Component wear and lifetime are further important economic factors. The components with the largest wear are the HV electrodes in
contact with the material to be fragmented and the switch electrodes.
We have found that the material loss from the steel electrode used for
concrete fragmentation amounted to 10 pg per shot. Consequently an
electrode with 1 cm cross section consumes z
lcm of its length per
week of full operation (8h working day). Therefore, provisions must be
made to adjust the electrode, and replacement becomes necessary from
time to time. However the wear is sufficiently small to be economically
N

636

Bluhm et al.: Application of HV Discharges to Material Fragmentation and Recycling

irrelevant. Concerning the lifetime of the insulator enveloping the steel

rod, insufficient experience exists for operation under industrial conditions. Damage of the insulator can occur through material fragments
impacting on its surface. Laboratory experience shows that it is most
important to relieve the triple point at the electrode tip.
Of more concern than the erosion of the operating electrode is the
wear of the switch electrodes. We have measured a loss of 3 g after lo6
pulses from spherical steel electrodes. Using CuW with Borda profiles
and sufficiently large diameters should lead to acceptably small wear
at the power densities involved. Nevertheless, adaptation of the switch
gas pressure will be routinely necessary and the electrode gap distances
probably must be readjusted at maintenance intervals of the order of ca.
a week.
Capacitor lifetime has been guaranteed for >lo8pulses even for the
conditions of large voltage reversal occurring in the electric discharges
and is presently not considered to be an economic limitation.
Depending on the process, a strong liberation of salts can occur and
increase the conductivity of the process water. At a conductivity level
>1500 pS/cm, efficient operation becomes impossible and the water
must either be replaced or conditioned, which can become an important
factor for cost effectiveness.

REFERENCES
[l] B. W.Sjomkin, A. E Ussow and V I. Kurets, "The Principles of Electric Impulse Destruction of Materials", Russian Academy of Sciences, Kola Science Center, Editor
N. E Tusow, 1995 (in Russian)
[2] V. V. Burkin, P. V. Makarov, B. V. Semkin, and B. G. Shubin, "Pressure Field around
a Spark in a Solid Dielectric", Sov. Phys. Tech. Phys., Vol. 20, pp. 1490-1492,1976.
[3] N. T. Zinov'ev, B. V Semkin, "Energy Balance tor a Spark in a Solid Dielectric", Sov.
Phys. Tech. Phys., Vol. 23, pp. 369-370,1978,
[4] T. H. Martin, A. H.Gnenther, M. Kristiansen, Editors, J. C. Martin, On Pulsed Poiuer,
Plenum Press, New York, 1996.
[5] J. C. Devins, S. J. b a d , R. J, Schwabe, "Breakdown and Prebreakdown Phenomena
in Liquids", J. Appl. Phys., Vol. 52, pp, 4531-?4545,1981.
[6] V Lisitsyn, H. Inoue, S. Katsuki, H. Akiyama, I. Nishizawa, "Drilling and Demolition of Rocks by Pulsed Power", Proc. 12thIEEE Int. Pulsed Power Conf., Monterey,
USA, June 1999, p. 169.

[7] I. S.Guz, "Interaction betweenstress Waves and a Growing Crack, lourn. of Appl.
Mech. and Techn. Phys., Vol, 15, pp. 726-730,1974.
[8] E. N. Sher, "Investigation of the Dynamics of the Development of Cracks by the
Method of Photoelastocity", Journ. of Appl. Mech. and Techn. Phys., Vol. 16, pp.
858-864,1974.
[9] E A. Martynyuk, E. N.Sher, "Elastic-Wave Diffractionby a Finite Crack under Antiplane Strain Conditions", Jonrn. of Appl. Mech. and Tech. Phys., Vol. 13, pp,
406410,1974.
[lo] Sh. G. Volodarskaya, I. S. Gm',I. A. Knmkin, V M. Finkel, "An Investigation into
the Interaction of a Stress Wave with a Stationary Macrocrack in Elastic-Plastic and
Quasibrittle Materials", Journ. of Appl. Mech. and Techn. Phys., Vol. 13, pp. 432437,1975.
Ill] J. D. Achenbach, j. H. Hemann, E Ziegler, "Separation at the Interface of a Circular
Inclusion and the Surrounding Medium under Incident Compressive Wave", Transactions of the ASME, Joum. of Appl. Mech., Vo1.E 37, pp. 298-304,1970.
[12] M. L. Baron, A. T.Matthews, "Diffractionof a Pressure Wave by a Cylindrical Cavity
in an Elastic Medium", Joum. of Appl. Mech., Transactions of the ASME, Vol. E28,
pp. 347-354,1961.
[13] Xh-Hsing Pao, C. C. Mow, "Scattering of Plane Compressional Waves by a spherical
Obstacle", Journ. of Appl. Phys., Vol. 34, pp. 493-499,1963.
[14] G. 1. Kanel, S. V Razorenov, A. V Utkin, Spallation ofsolids under shock waue loading, in
High Pressure Shock Compression of Solids 11, L. Davison, D. E. Grady M. Shahinpoor,
Eds.
[15] E. Marx, "Versuche iiber die Priifung von Isolatoren mit Spannungsstossen", Elektrotechn. Z., Vol. 45, pp. 652-654,1925.
1161 C. SchultheiP, R. StraUner, H. Bluhm, E Stemmermann, "Concrete recycling and
scraping by electrodynamic fragmentation, in Physics of Intense Light Ion Beams,
Production of High Energy Density in Matter, and Pulsed Power Applications",
Annual Report 1996/97, H. Bluhm, Ed., FZKA 6205.
[17] V Goldfarb, R. Bndny, A. Dunton, 6. Shneerson, S. Krivosheev, Yu.Adamian, Removal of Surface Layer of Concrete by a Pulse-Periodical Discharge", Proc. 12th
IEEE Int. Pulsed Power Conf., Monterey, USA, June 1999, p. 1078
[18] V I. Knrets, E. N Tarakanovskii, G. P. Filatov, "Electric Pulse Resistance of Liquid
Nitrogen and Rubber", Russian Physics Journal, Vol. 38, pp, 455457.1995.
[19] Y. Takahashi and K.Ohtsuka, "Corona Discharges and Bubbling in Liquid Nitrogen", J. Phys. D Phys., Vol. 8, pp. 165-169,1975,
[20] Y. Takahashi, "Koronaentladungen in fliissigem Stickstoff", Bull. Facul. Sci. & Eng.,
Chuo University, Vol. 20, pp. 247-255,1977.
[21] U. Andres, "Parameters of Disintegration of Rock by Electrical Pulses", Powder
Technology, Vol. 58, pp, 265-269,1989.
Manuscript was received on 19June 2000.