J. Plasma Physics (2000), vol. 63, part 2, pp. 115–128.

Printed in the United Kingdom 115

# 2000 Cambridge University Press

Arc-liberated chemical energy exceeds electrical

input energy
P E T E R G R A N E A U," N E A L G R A N E A U,#
G E O R G E H A T H A W A Y$ and R I C H A R D L. H U L L%
" Centre for Electromagnetics Research, Northeastern University, Boston, MA 02115,
USA
# Department of Engineering Science, University of Oxford, Oxford OX1 3PJ, UK
$ Hathaway Consulting Services, 39 Kendal Avenue, Toronto, Ontario, Canada M5R 1L5
% Tesla Coil Builders of Richmond, 7103 Hermitage Road, Richmond, VA 23228, USA

(Received 11 December 1998 and in revised form 5 August 1999)

Abstract. This paper reports the first experimental results in which the kinetic
energy of cold fog, generated in a water arc plasma, exceeds the electrical
energy supplied to form and maintain the arc. The cold fog explosion is
produced by breaking down a small quantity of liquid water and passing a
kiloampere current pulse through the plasma. The 90-year history of unusually
strong water arc explosions is reviewed. Experimental observations leave little
doubt that internal water energy is being liberated by the sudden electro-
dynamic conversion of about one-third of the water to dense fog. High-speed
photography reveals that the fog expels itself from the water at supersonic
velocities. The loss of intermolecular bond energy in the conversion from liquid
to fog must be the source of the explosion energy.

1. Research motivation
Before describing the latest experiments concerned with cold fog explosions, the
motivation for this research will be explained. In 1994, we discovered that when
a small quantity of water (usually but not necessarily distilled) is converted to
high-density fog within microseconds, the fog explodes violently. Fog is defined
as a multiplicity of tiny water droplets that float in air. The discovery has
been fully described in a book dealing with pre-Maxwellian electrodynamics
(Graneau and Graneau 1996). More recent findings have been discussed in a
previous paper in this journal (Hathaway et al. 1998).
The fog generator is a small water-filled electric arc cavity to be described
later. In a typical explosion, the cavity receives less than about 50 J of
electrostatic energy from a high-voltage capacitor. Almost all of the input
energy is converted to low-grade heat, raising the water to a few degrees above
ambient temperature. This heat is incapable of raising steam or contributing in
any other way to the explosion. It seems inevitable that the fog is being
produced by electrodynamic forces in the current-carrying arc plasma. Such
forces can furnish the mechanical surface-tension energy required for tearing
bulk water apart into tiny fog droplets.
With 50 J of input energy, the quantity of fog produced is of the order of
0.75 g of water. To dissociate this amount of water into oxygen and hydrogen
would require 10 kJ of energy. Hence the fog explosion is unlikely to be caused
116 P. Graneau et al.

Atmospheric Cloud
vapour

Capacitor input energy

EC

Solar
energy Cold fog Rain-
Fog
accelerator water

Fog kinetic energy Low grade heat

Ek Eh

Measured : Ek < EC –Eh

Figure 1. Renewable water energy cycle.

by electrolytic dissociation of water molecules. Without this dissociation, the

most likely source of the explosion energy is that stored by hydrogen bonds
between the water molecules. This bond energy is said to be equal to the latent
heat of evaporation, and therefore could contribute up to 2200 J g−".
Figure 1 depicts the proposed renewable water energy cycle. The fog contains
less H O–H O bond energy per gram than a raindrop. In other words, its latent
# #
heat of evaporation is less than that of bulk water. This is known from the fact
that the equilibrium vapour pressure just outside a water droplet increases with
decreasing droplet diameter (Young 1993). The internal-energy difference
between the cold fog expelled from the accelerator must be made up by
atmospheric heat – that is, essentially by solar energy. No other energy source
appears to be available for replacing the extracted kinetic energy.

2. History of water arc explosions

It appears that the encounter of water drops with an electric arc produced the
first photograph of a cold fog explosion. This is shown in Fig. 2, which was
published by Trowbridge (1907). He must have realized that steam and water
vapour were invisible in air, which left him with no explanation of the white
explosion cloud. In his early high-voltage laboratory at Harvard University,
Trowbridge discovered another effect that has a bearing on fog explosions. He
stretched a sheet of paper along the path of his sparks of more than 50 cm
length, and found that at every corner and forking of the lightning-like
discharge, holes were punched through the paper without any burning or
charring. It took 80 years until experiments of this kind were repeated at MIT
and revealed that thunder was not the result of heat and the thermal expansion
of the lightning channel (Graneau 1989), as had been assumed during most of
the 20th century. It was shown instead that the shock wave in air was driven
by non-thermal forces, of which electrodynamic forces were then considered to
be the most likely. In view of the findings reported in the present paper, it is
possible that cold fog explosions also contribute to the mechanism that causes
thunder.
 Arc-liberated chemical energy 117

Figure 2. Trowbridge’s (1907) photograph of a cold fog explosion. The arc is " 40 cm
long.

During the Second World War, Frungel (1948) measured the unusual
strength of water arc explosions. He concluded that the explosions were not
caused by heat and steam, and admitted freely that he was unable to explain
the phenomenon.
Soon after Frungel’s publications, water arc explosions found applications in
electrohydraulic metal forming (Gilchrist & Crossland 1967) and underwater
pulse echo sounding (Frungel 1965). Not until the mid-1980s was the scientific
basis of the puzzling explosions more extensively researched at MIT (Azevedo
et al. 1986). It was then shown that the discharge of 3.6 kJ of stored capacitor
energy would create pressures in excess of 20 000 atm, in 7 ml of saltwater :
3.6 gm of water was ejected from the accelerator barrel with a velocity of the
order of 1000 m s−", and then punched a "-inch-diameter hole through a "-inch-
# %
thick aluminium plate (Graneau and Graneau 1996).
At the time, it was conjectured that the water was flying through the air as
118 P. Graneau et al.
a coherent liquid slug, since that would create the greatest impact. No evidence
of boiling and steam formation could be detected, and all the water found
after the explosion was cool. Accepting the general view that plasmas are
quasineutral and do not explode as a result of Coulomb forces, the available
evidence seemed to leave little doubt that the explosions were driven by
electrodynamic forces. This discovery motivated a 10-year investigation of the
electrodynamics of water arcs. The Lorentz force could not account for more
than a small fraction of the observed force. Ampe' re’s force law (Graneau and
Graneau 1996) fared better, but still fell short of predicting the measured values
by at least a factor of 10. The search for a new electrodynamic force was finally
abandoned in 1994.
Another report of electrically induced explosions in water came from Kansas
State University. Johnson (1992) claimed that the loudness was distinctly
greater than that obtained with an equivalent amount of gunpowder. He found
that the remaining water droplets were cool to the touch and that apparently
no steam had been produced. Johnson suggested that the explosions may have
been due to longitudinal Ampe' re forces, and were tapping a new source of
energy.
In 1994, the present authors took the first high-speed photographs of water
ejected from an arc accelerator, and discovered that the leading high-speed
component was not a coherent liquid, but actually very dense fog, which finally
expanded in the air under the laboratory ceiling. By trapping the fog in a balsa-
wood absorber and measuring its temperature, it was found that the fog was
still cold – at most a few degrees above ambient temperature. The discovery of
cold fog explosions changed our scientific outlook on the remarkable behaviour
of water arcs.

3. Accelerator and discharge circuit

Our experimental technique and many previous results were published in
Graneau and Graneau (1996) and Hathaway et al. (1998). Here we cite later
results together with brief details of the relevant accelerator and discharge
circuit as well as the diagnostics involved.
Figure 3 shows a cross-section though the Type A accelerator which has been
used extensively since 1984. During this period, it has been made of different
materials and has employed a variety of water charge volumes w and breech-
to-muzzle lengths l. For example, fog penetration of a "-inch-thick aluminium
%
plate was achieved with an accelerator using a steel barrel of l l 10 cm and
w l 7 ml of saltwater. Subsequent shots have not used saltwater. In the latest
experiments reported in this paper, the shaded metal components were made of
copper and the remainder was a nylon sleeve and a nylon block. The essential
dimensions are indicated on Fig. 3, with, typically l l 3.0 cm and w l 3.5 ml of
distilled water. The water level was 0.26 cm below the muzzle.
The secondary projectile stood on top of the muzzle, and consisted of a
2.65 cm diameter balsa-wood cylinder of 3.2 cm length pressed into a copper
can to add strength and weight, making the dry secondary projectile mass
approximately M l 65 g. Behind the accelerator stood a scale on which a video
camera recorded the height h to which the wet mass Mjm ascended, where m is
absorbed fog mass.
 Arc-liberated chemical energy 119

RC
L

W
l
S

ESV C HV
T

i
1 inch
4

1 inch DSO
2

Figure 3. Type A accelerator with discharge circuit.

In a typical experiment, the energy storage capacitor C l 0.565 µF was

connected between the barrel electrode and the baseplate-and-rod electrode.
The switch S was a triggered spark gap in atmospheric air, which trapped a
certain amount of residual charge in the capacitor when the switching arc
interrupted the oscillating discharge at low current.
In Fig. 3, the distributed circuit resistance and inductance are indicated by
lumped parameters RC and L. The discharge current i was measured with a
current transformer T and a digital storage oscilloscope DSO. The electrostatic
voltmeter ESV recorded the charging voltage V on the capacitor terminals with
!
the switch S open, and the residual voltage Vr after the switch had been fired
and then interrupted the current flow. With S open, the high-voltage power
supply HV was connected across the capacitor for each shot. After reaching the
desired voltage V , the supply was disconnected. Closing the switch S initiated
!
the fog explosion, with the underdamped discharge current ringing down over
a few microseconds.

4. Evidence of fog
That water arc explosions are in fact cold fog explosions was discovered with
high-speed photography at Oxford University (Graneau and Graneau 1996 ;
Hathaway et al. 1998). In the present paper, we are presenting further
photographic evidence of fog generation that reveals interesting properties of
the fog jets.
Figure 4 is a series of numbered photo frames showing the emergence of fog
from the muzzle of a water arc accelerator as the fog pierces the atmosphere.
The high-speed film was shot in the HCS Laboratory in Toronto at 35 000
frames per second. A powerful light flash, triggered by the switch S of Fig. 3,
was directed at the experimental apparatus. The camera formed images of the
travelling fog on a rotating mirror, which reflected them to a film adhering to
the inside of a rotating drum. The time interval between successive frames was
28.6 µs.
Water vapour is invisible in air. Relatively large water drops and films of
water are transparent. They show up on photographs only because of thin lines
120 P. Graneau et al.

11
31

51
1

2
12 32 52

3 13

33 53

14 34 54
5

15
6 55
35

16 36 56

Figure 4. High-speed photography of a fog jet, 35 000 frames per second.

of light that are reflections from the water surfaces. The uniform whitish-grey
appearance of fog and clouds is due to light scattering by a multitude of small
droplets. Hence the photographs in Fig. 4 conclusively prove the emergence of
fog from the accelerator barrel. Video camera records, obtained in the
laboratory of Tesla Coil Builders of Richmond (TCBOR), revealed that the fog
slowed down quite rapidly as it penetrated the atmosphere. On reaching the
ceiling, it rolled around like a cloud for a number of seconds while it evaporated.
When the accelerator depicted in Fig. 3 stood on an aluminium sheet spread
out on the floor, it was observed that small droplets of mist rained out of the
cloud. Fog droplets that float in air are said to be of diameters ranging from 1
to 100 µm. The fog recorded in Fig. 4 is likely to have contained some larger
 Arc-liberated chemical energy 121
droplets that fell in air, but, for the purpose of this paper, we shall call the
mixture of very small floating and falling droplets ‘ high-density fog ’. In other
photographs (Graneau and Graneau 1996), larger drops of water were seen to
follow behind the fog, some of which were millimetres in diameter and travelled
at much lower velocities.
Figure 4 shows a selection of the nearly 200 frames recorded by the camera
during a single shot. The fog column diameter just above the muzzle is equal to
the accelerator barrel diameter, which was 1.27 cm. This immediately proves
that the fog is cold and, unlike steam, does not expand laterally. A centimetre
scale stands to the left of the fog column, and, by knowing the camera speed,
the average velocity of the head of the fog column, taken over the first eight
frames, was measured to be approximately 200 m s−". The fog appears to be
coasting in the inertial mode. It is not understood why, initially, the column has
a pointed tip, which would normally indicate supersonic speed (
350 m s−").
Anyhow, frames 11–16 show air ablation of the tip. By frames 31–36, a
mushroom-shaped head has emerged with pronounced lateral fog spreading due
to atmospheric ablation. Mushroom heads usually result from subsonic
progression, and frames 31–36 demonstrate a velocity of 65 m s−".
The high-speed camera was triggered by the switch that initiated the arc
discharge. This occurred no more than 28.6 µs before the exposure of frame 1.
As the fog emerged from the muzzle, the explosion must have been over –
otherwise it would have resulted in lateral fog expansion. This finding suggested
that the explosion period was limited to the time during which the arc current
was flowing, which was of the order of 10 µs. Heat could not have been the cause
of the explosion, because this would have persisted for more than 28 µs.
The arc forms near the breech of the accelerator, located at the bottom of the
water column. This is where the fog is likely to be generated. The water column
in the barrel was 3 cm long. No water appears to have been pushed ahead of the
fog, which initially may have travelled with a velocity as high as 1000 m s−".
Other photographs have indicated that the fog breaks through the water
surface without causing as much as a ripple. The implication is that the leading
fog droplets are extremely small and spaced apart from each other. The smaller
the droplets, the more bond energy per unit mass has been liberated and the
more strongly the droplets should repel each other.
A transition from fog to transparent water begins to show in frames 51–53.
Frames 181–185 depict a liquid flow at 15 m s−". The water in the latter frames
must have been a hollow film tube with a few droplets in the surface, since
normal-density liquid water could have only accounted for a 3 cm long solid
column.
Figure 5 is a single frame taken with an ordinary video camera in the HCS
Laboratory. In this case, the liberated chemical energy is much greater than in
the high-speed film of Fig. 4. In Fig. 5, the fog plume is 2 m tall, and its shock-
wave-shaped pointed tip very clearly travels at supersonic speed. This
photograph reveals the light source about 50 cm to the right of the fog jet.
When a balsa-wood cylinder in a metal cup stood on the muzzle of the
accelerator, the fog penetrated deep into the porous material and transferred
momentum to the projectile, which was thrown vertically to a height h.
Weighing before and after a shot made it possible to determine the dry mass M
of the projectile and the absorbed fog mass m. Since the two masses travelled
122 P. Graneau et al.

Figure 5. Two-metre-tall supersonic fog jet. Approximate fog density 1 kg m−$.

together at the initial projectile velocity v , momentum conservation required

!
that
muav l (Mjm)v , (1)
!
where uav was the average velocity of the absorbed fog. Then it follows from
energy conservation that
"(Mjm)v# l (Mjm)gh, (2)
or # !
v l N2gh, (3)
!
where g is the acceleration due to gravity.
From (1) and (3), the average fog velocity can be calculated as
Mjm
uav l N2gh. (4)
m
 Arc-liberated chemical energy 123
This average velocity gave the minimum kinetic energy of the fog as
Ek,min l "mu#av. (5)
#
The true kinetic energy of the fog depends on the root-mean-square (rms)
velocity of the droplets, because for n droplets of mass ∆m travelling at velocity
u, we have
Ek l  "∆m u# l "u#rms  ∆m l "mu#rms. (6)
# # #
n n
Sections cut from the balsa wood revealed that fog penetration varied widely
from place to place. However, in general, the velocity distribution has the form
of a half-cycle of a sine wave. The average ordinate of the sine wave is the
amplitude divided by "π and the rms value is the amplitude divided by 2,
#
yielding
urms l 1.1uav. (7)
The rms velocity is not very different from the average velocity, and (7) has
been used in the evaluation of the fog kinetic energy.

5. Energy flow diagram

The energy flow through the discharge circuit and accelerator to the fog jet can
be better understood with the aid of Fig. 6. The capacitor receives an energy E
"
from the high-voltage charging unit. Recognizing that a small amount of the
energy remains in the capacitor, E E is supplied to the discharge circuit,
# "
which is shown as separate from the accelerator. In the metallic components of
the circuit and in the switching arc plasma, part of E is dissipated as Ohmic
#
heat E , which is proportional to i# and can be measured. For the purpose of
$
Fig. 6, the ionization loss in the switch is added to the ionization loss in the water
to give E , which contributes to the measured damping of the ringing discharge.
&
In addition to the absorbed ionization energy, which does not heat the water
but is stored electrostatically until the ions recombine well after the explosion,
some Ohmic heating of the water plasma takes place. This has been measured
approximately with a thermocouple, and is labelled E .
'
Disregarding the fact that it is a plasma, the water itself acquires kinetic
energy E , which is evidenced by the explosion. We let E be the non-thermal
* (
mechanical energy required for tearing the liquid apart into fog droplets. Any
liberated internal water energy E may be added to E , and contributes to E .
) ( *
Before the kinetic energy of the fog jet E is measured, the fog has to penetrate
"#
a certain volume of liquid water, and is likely to dissipate some of its energy
owing to viscous drag. This loss component is denoted by E . Whatever forces
""
accelerate the fog in the water will also exert pressure on the cavity walls. The
resulting strain and deformation energy stored temporarily in the wall material
has been labelled E .
"!
The critical quantity of the energy flow diagram is E . Measurements
(
(Graneau and Graneau 1996) of E , E and E have established that E  E . In
$ & ' ( #
the relevant experiment, E was usually less than 50 J and the water quantity
#
was never less than w l 1 ml. It follows from these figures that the water
temperature could not exceed the ambient temperature by more than 12 mC.
Furthermore, it was established by experiment with a photodiode that the arc
124 P. Graneau et al.

E1 (40.68 J)
Capacitor
E2 (39.96 J)

L C RC E3 Circuit loss
i S (13.9 J)

E4 (26.06 J)
E5 Ionization loss

i (≈25 J)
W
E6 Joule heat
Accelerator

E7 Mechanical (< 5 J)
E8 Internal
Water
(≈95 J)
E9 (≈ 100 J)
E10 Deformation
Barrel
E11 Drag
E12 (63.62 J)
Fog jet
Figure 6. Energy flow diagram for shot TA70751.

current i was negligibly small until the water plasma was fully glowing in the
accelerator barrel. Hence no selective subheating of a subvolume (filament) of
w would have occurred. These facts determined once more that E was not heat
(
energy. It has to be electrodynamic energy associated with magnetic forces.
The numerical energy quantities in Fig. 6 apply to shot TA70751 of Table 1.
Where estimates have been made, the figure is preceded by the ‘ approximate ’
symbol $. In all other cases, actual measurements are presented. The methods
used for estimating and measuring energy losses are described more fully in
Graneau and Graneau (1996). The estimated value of E jE $ 25 J would
& '
produce a temperature rise in the 3.5 g of water in the accelerator of about
1.7 mC. This is consistent with temperature measurements made in other
experiments.
It should be noted that in shot TA70751, the amount of fog absorbed in the
balsa-wood secondary projectile was 0.537 g. If at any time this water mass had
existed as steam, it would have required the expenditure of 1213 J, an amount
equal to the latent heat of vaporization. However, the total energy discharged
from the capacitor was only E l 40 J. This was clearly insufficient to explain
#
the explosion by the adiabatic expansion of steam.
The most intriguing feature of the energy flow diagram is the large
magnitude of E as documented in the following section. This demands the
"#
supply of internal water energy E .
)

6. Fog kinetic-energy results

Table 1 lists the measured kinetic energy of the fog jets in eight shots fired
under identical conditions with the Type A accelerator of Fig. 3. As already
explained, almost all of the fog kinetic energy is derived from internal water
 Arc-liberated chemical energy 125

Table 1. Results of eight shots with the Type A accelerator, with V l 12.0 kV,
!
Vr l 1.6 kV, C l 0.565 µF, E l 39.96 J l l 3.0 cm and w l 3.5 ml (distilled water).
#
Shot M (g) m (g) h (m) E (J) E \E
"# "# #
TA70564 69.278 0.718 0.57 40.54 1.015
TA70565 65.363 0.608 0.58 50.18 1.256
TA70567 64.548 0.632 0.48 39.00 0.876
TA70568 64.200 0.621 0.56 45.80 1.146
TA70569 63.588 0.843 0.59 35.12 0.879
TA70570 65.481 0.659 0.58 46.54 1.165
TA70571 64.905 0.537 0.66 63.62 1.592
TA70572 63.660 0.823 0.54 32.98 0.825

energy, and E  E . The results in Table 1 show that in five of the eight shots,
( #
the kinetic energy of the fog was greater than the energy supplied by the storage
capacitor. In other words, we achieved E \E
1. The ratio E \E is one of
"# # "# #
several markers indicating the release of internal water energy.
To appreciate how small the thermal action of the 40 J input energy is, it
helps to think of an ordinary match, which liberates between 100–200 J of
thermal energy. If this energy is used to heat 3.5 ml of water in a test tube, it
will certainly not cause a water explosion, nor will it produce a noticeable lift
of the water surface. This immediately suggests that E , which must be
(
responsible for the fog explosion, has to be mechanical energy. In any case, the
measured low-grade heat losses E jE jE account for almost all of the input
$ & '
energy E (Graneau and Graneau 1996), leaving very little for E .
# (
Dielectric breakdown and ionization do not result in an explosion unless
subsequently a significant amount of current flows through the arc plasma. This
is supported by the fact that breakdown and ionization are energy-absorbing
phenomena. The strong current dependence of the explosions was already
demonstrated by Azavedo et al. (1986). In their work, lowering the maximum
pulse current from 25 to 12 kA reduced the explosion impulse from 7.0 N s to
1.5 N s. Currents of less than 1000 A yielded negligible explosion forces. The
ionization of water during the breakdown and plasma generation phase, as well
as any Coulomb forces associated with charge separation, do not appear to be
responsible for the arc explosions. This leaves only electrodynamic (magnetic)
forces between current elements as the possible cause of the explosions. The
strength of the explosions undoubtedly increases with the current i, and this is
in accordance with electrodynamic processes.
If F is the time-varying force that accelerates the fog droplets then
Newtonian mechanics requires that the impulse of this force, F dt, has to
generate the initial fog momentum muav. Let a constant effective force Fe be
defined, acting for an interval τ, such that it is equal to the impulse F dt. Then

&F dt l Fe τ l muav. (8)

The average velocity can be calculated from (4). The time constant τ of the
underdamped discharge current will be assumed to be the interval during which
Fe is active. In the absence of steam, it is difficult to see how the force pulse
126 P. Graneau et al.
could be stretched out beyond the length of the current pulse. The discharge
current has the form
i l I e−t/τ sin ωt, (9)
!
where ω is the angular ringing frequency. From current oscillograms, a typical
value of the time constant was τ l 4 µs. This resulted in Fe l 52 400 N for the
weakest shot (TA70752) of Table 1, or an effective pressure over the cross-
sectional area of the accelerator barrel of 4.14i10) N m−#, which is equal to
4220 atm. This would appear to be sufficient pressure to achieve the observed
fog acceleration.
Maxwell’s electromagnetic field theory indicates that by far the strongest
electrodynamic force on the plasma arc column is the magnetic pinch exerted
by radial Lorentz forces. This produces an axial thrust that could expel water
from the barrel. To make this force a maximum, we take the optimistic view
that the current flows all in the axial direction. Northrup (1907) showed that
the total axial thrust on a cross-section of a cylindrical current distribution is
given by
µ i#
Fl ! , (10)
4π 2
where µ is the magnetic permeability of free space. The impulse of this pinch
!
thrust can again be expressed by an effective force Fe such that

& F dt l Fe τ l 4πµ! 12 & i# dt. (11)

The action integral i# dt was measured for many discharges, and a typical value
was 130 A# s. With this information, the effective Lorentz thrust of (11)
becomes Fe l 1.6 N, as compared with the force of 52 kN required to accomplish
the observed fog acceleration.
Newtonian electrodynamics (Graneau and Graneau 1996), which was widely
taught and used in the 19th century, gives a larger effective electrodynamic
force, possibly as high as 60 N. This is, however, still far too low to explain the
fog acceleration. The chance of accounting for the observed fog accelerations
with electrodynamic forces is even worse than these figures indicate, since E ,
*
the true fog acceleration energy, must be greater than E because of
"#
deformation and drag losses, and could even be twice as large as the measured
energy.
The considerations outlined in this section demonstrate that, quite apart
from the fact that E \E
1, the observed fog explosions demand the supply
"# #
of internal water energy.

7. Water-to-fog conversion
Condensation of water vapour in the atmosphere is the greatest fog producer on
Earth. It is a relatively slow process, which cannot generate a sufficient amount
of fog in a few microseconds to cause an arc explosion. In the early formative
stage of a water arc, the fog density must approach that of liquid water, and is
therefore up to 300 000 times as great as the fog density of clouds (approximately
3 g m−$).
 Arc-liberated chemical energy 127
It is difficult to think of any other way of creating the dense fog than by
mechanically tearing the liquid apart into tiny fragments. These are the fog
droplets, ranging in size from 1 to 100 µm in diameter or smaller. As discussed
previously, the tearing force must be an electrodynamic force, but the direction
of the Lorentz force is not such that it could split water into droplets. On the
other hand, the Ampe' re force, or Ampe' re tension as it is often called, is well
qualified to describe the break-up of the liquid.
The most well-known consequence of Ampe' re tension in metallic conductors
is the phenomenon of wire fragmentation. In liquids and plasmas, it leads to
plasma bead formation, which has been observed in plasma focus fusion,
deuterium-fibre fusion and capillary fusion experiments. These phenomena are
fully reviewed in Graneau and Graneau (1996). Hence it is not only possible but
likely that Ampe' re tension will generate fog in water arc explosions.
Surface tension requires that tearing forces be involved in the breakup of the
liquid. The surface tension γ of water at 20 mC is 72.75 dyn cm−". It turns out
that surface tension energy per unit area has the same dimension as surface
tension per unit edge and is numerically equal to it, so that
γ l 72.75 dyn cm−" l 72.75 erg cm−# l 72.75i10−( J cm−". (12)
The fog mass generated in the experiments of Table 1 varied between
0.537 g and 0.843 g. Let us evaluate the additional surface-tension energy
required to convert 1 g of water into fog. If the drops are all of the same
diameter d then the number density n of droplets generated will be
6
nl . (13)
πd$
Fog droplets are said to be between 10−% and 10−# cm in diameter. Hence the
number of droplets lies in the range from 1.9i10' to 1.9i10"# per gram. The
total new surface energy Es for drops of the same size is
Es l γnπd#. (14)
For the two extreme droplet diameters of 10−# and 10−% cm, this energy comes
to 4.37 and 437 mJ respectively. Hence the part of E in the energy flow
(
diagram of Fig. 6 that must supply surface tension energy is only a fraction of
one joule. The Ampe' re tension forces can comfortably meet this energy
demand, and may have some strength left over for fog acceleration. These
findings lead us to believe that Ampe' re tension is responsible for the conversion
of liquid water into dense fog.

8. Energy conservation
If energy is to be conserved then the internal energy per unit mass of fog (see
Fig. 1) has to be less than that of rainwater. During the explosions, the water
molecules do not appear to undergo chemical changes ; so the energy difference
must be explained by a change in the intermolecular bond energy per unit mass.
This reserve of energy is believed to be equal to the latent heat of evaporation,
and, in the case of rain water, it comes to 2259 J gm−".
In shot TA70751 (see Table 1), 0.537 g of water was converted to fog. The
latent heat of this amount of water is 1213 J. The measured fog kinetic energy
128 P. Graneau et al.
came to E l 63.62 J. This is just over 5 % of the latent heat. Furthermore,
"#
extensive analysis presented in Graneau and Graneau (1996) and Azevedo et al.
(1986) has demonstrated that electrodynamic forces could only amount for a
negligible portion of the kinetic energy developed in arcs. Therefore almost all
of the input energy E (EC in Fig. 1) is converted to low-grade heat, so that
#
nearly all of E must have come from the 1213 J stored in the bulk water.
"#
This calculation has ignored the E and E losses shown in Fig. 6. Hence the
"! ""
energy liberated in the water-to-fog conversion is greater than 5 % of the stored
energy, implying that the latent heat of the fog created by the accelerator
contains less than 95 % of the latent heat of bulk water. As demonstrated by
(14), less than 1 J (0.1 % of the latent heat) of the stored energy loss can be
associated with additional surface tension energy acquired by the fog. Therefore
the remainder must have resulted from differences in the water structure. How
this average structure and the intermolecular bonding changes with the number
of molecules in the droplet is a topic of physical chemistry that is a subject of
intense research (Postorino et al. 1993 ; Li and Ross 1993 ; Li 1996). The
difference in the latent heat between fog and bulk water is eventually restored
by heat in the atmosphere, which allows the fog to condense and return to
earth. Our investigation has not yet extended into the details of water
structure, but the discovery of cold fog explosions will help in this direction.

References
Azevedo, R., Graneau, P., Millet, C. and Graneau, N. 1986 Phys. Lett. 117A, 101.
Frungel, F. 1948 Optik 3, 125.
Frungel, F. 1965 High Speed Pulse Technology, Vol. 2. Academic Press, New York.
Gilchrist, I. and Crossland, B. 1967 IEEE Conference Publ. No. 38 (December), 92.
Graneau, P. 1989 J. Phys. D : Appl. Phys. 22, 1083.
Graneau, P. and Graneau, N. 1996 Newtonian Electrodynamics. World Scientific, New
Jersey.
Hathaway, G., Graneau, P. and Graneau, N. 1998 J. Plasma Phys. 60, 775.
Johnson, G. L. 1992 In : Intersociety Energy Conversion Engineering Conference Proceedings,
San Diego, CA, p. 335. IEEE Press, New York.
Li, J. and Ross, D. K. 1993 Nature 365, 327.
Li, J. 1996 J. Chem. Phys. 105, 6733.
Northrup, E. F. 1907 Phys. Rev. 24, 474.
Postorino, P., Tromp, R. H., Ricci, M. A., Soper, A. K. and Nellson, G. W. 1993 Nature 366,
668.
Trowbridge, J. 1907 Mem. Am. Acad. Arts Sci. 13(5), 185.
Young, K. C. 1993 Microphysical Processes in Clouds. Oxford University Press.