Feasible homopolar dynamo with sliding liquid

Feasible homopolar dynamo with sliding liquid-metal contacts
Instituto Politécnico Nacional, Cerro Blanco 141,
Colinas del Cimatario, Querétaro, Mexico; [email protected]
2 Applied
Mathematics Research Centre, Coventry University,
Coventry, CV1 5FB, UK; [email protected]
Abstract: We present a feasible homopolar dynamo design consisting of a flat, multi-arm spiral
coil, which is placed above a fast-spinning metal ring and connected to the latter by sliding liquidmetal electrical contacts. Using a simple, analytically solvable axisymmetric model, we determine
the optimal design of such a setup. For small contact resistance, the lowest magnetic Reynolds
number, Rm ≈ 34.6, at which the dynamo can work, is attained at the optimal ratio of the outer
and inner radii of the rings Ri /Ro ≈ 0.36 and the spiral pitch angle 54.7◦ . In a setup of two copper
rings with the thickness of 3 cm, Ri = 10 cm and Ro = 30 cm, self-excitation of the magnetic field
is expected at a critical rotation frequency around 10 Hz.
1 Introduction
The homopolar dynamo is one of the simplest models of the self-excitation of magnetic field by
moving conductors which is often used to illustrate the dynamo action that is thought to be behind
the magnetic fields of the Earth, the Sun and other cosmic bodies [1, 2]. In its simplest form
originally considered by Bullard [3], the dynamo consists of a solid metal disc which rotates about
its axis, and a wire twisted around it and connected through sliding contacts to the rim and the axis
of the disc. At a sufficiently high rotation rate, the voltage induced by the rotation of the disc in
the magnetic field generated by an initial current perturbation can exceed the voltage drop due to
the ohmic resistance. At this point, initial perturbation starts to grow exponentially leading to the
self-excitation of current and its associated magnetic.
Despite its simplicity no successful implementation of the disc dynamo is known so far. The
problem appears to be the sliding electrical contacts which are required to convey the current
between the rim and the axis of the rotating disc. Electrical resistance of the sliding contacts,
usually made of solid graphite brushes, is typically by several orders of magnitude higher than that
of the rest of the setup. This results in unrealistically high rotation rates which are required for
dynamo to operate [4]. To overcome this problem we propose to use liquid-metal sliding electrical
contacts similar to those employed in homopolar motors and generators [5]. The aim of this paper
is to develop a feasible design of the disc dynamo which could achieve self-excitation at realistic
rotation rates.
2 Physical and mathematical models
The principal setup of the proposed disc dynamo shown in Fig. 1(a) consists of a stationary coil (1)
made of a copper disc sectioned by spiral slits and a fast-spinning disc (2) placed beneath it, which
Figure 1: Schematic view of the disc dynamo setup consisting of a stationary coil (1) made of a
copper disc sectioned by spiral slits and a fast-spinning disc (2), which is electrically connected to
the former by sliding liquid-metal contacts (hatched) (3) (a) and the experimental device (b).
is electrically connected to the former by sliding liquid electrical contacts (3). The coil is supported
by the holders which are not shown in this basic view. The liquid metal is hold in vertical state by
the centrifugal force. Such a compact and symmetric design not only minimizes the electrical
resistance in the system but also makes it amendable to simple analysis which is carried out in the
To simplify the analysis both discs are subsequently assumed to be thin coaxial rings of thickness d, the outer radius Ro ≫ d, and the inner radius Ri = λ Ro , where 0 < λ < 1 the ratio of the
inner and outer radii. The rings are separated by a small axial distance and connected to each other
at their rims through the sliding liquid-metal electrical contacts. The design of the stationary top
ring, which forms a compact coil consisting of spiral sections is described in detail below. The
bottom ring is mounted on an axle which is driven by an electric motor with the angular velocity Ω.
The electric current I0 is induced by the rotation of the bottom ring in the magnetic field generated
by the same current returning through the coil formed by the top ring. In the solid rotating ring,
the current is assumed to flow radially with the linear density Jr = 2Iπ0r , which decreases due to
the charge conservation inversely with the cylindrical radius r. Current returns through the top ring
where it is deflected by the spiral slits that produce an azimuthal component proportional to the
radial one:
I0 β
Jϕ = −Jr β =
2π r
where arctan β the pitch angle of the current lines relative to the radial direction. The shape of slits
following the current lines is governed by Jϕr = rddrϕ = −β and given by the logarithmic spirals
ϕ (r) = ϕ0 − β ln r,
where ϕ is the azimuthal angle. The electric potential distribution in the coil ring follows from
Ohm’s law
I0 (
−⃗er + β⃗eϕ = −σ d⃗∇φc ,
J⃗ =
2π r
as φc (r, ϕ ) = 2πσ
d (ln r − β ϕ ) . Thus, the potential difference along the current line between the
rims of the ring is
(1 + β 2 ) ln λ .
∆φc = [φc (r, ϕ (r))]RRoi = −
2πσ d
The potential difference across the bottom ring, which rotates as a solid body with the azimuthal
velocity vϕ = rΩ, is defined by the radial component of Ohm’s law for a moving medium
= σ d(−∂r φd + vϕ Bz ),
2π r
where Bz is the axial component of the magnetic field. Integrating the expression above over the
ring radius we obtain
− ln λ = σ d (−∆φd + ΩΦd ) ,
where ∆φd = [φd (r)]RRoi is the potential difference across the rotating ring and Φd = Ri o Bz r dr is the
magnetic flux through it. Using the relation Bz = r−1 ∂r (rAϕ ), the latter can be expressed in terms
of the azimuthal component of the magnetic vector potential Aϕ as
Φd = rAϕ r=R
Jr =
In the stationary state, which is assumed here, the potential difference induced by the rotating ring
in Eq. (5) is supposed to balance that over the coil defined by Eq. (4) as well as the potential drop
over the liquid-metal contacts with the effective resistance R : ∆φd = ∆φc + RI0 . This equation
implicitly defines the marginal rotation rate at which a steady current can sustain itself.
To complete the solution we need to evaluate the magnetic flux (6) through the rotating disc.
The azimuthal component of the vector potential appearing in Eq. (6) is generated by the respective
component of the electric current which is present only in the coil. Thus, we have
Aϕ (r, z) =
0 Ri
Jϕ (r′ ) cos ϕ r′ dr′ d ϕ
r′2 − 2r′ r cos ϕ + r2 + z2
where z is the axial distance from the coil ring carrying the azimuthal current Jϕ defined by Eq.
(1). Note that the poloidal currents with radial and axial components circulating through the rings
and liquid-metal contacts produce purely toroidal magnetic field, which is parallel to the velocity
of the rotating ring and, thus, do not interact with the latter. In the plane of the ring (z = 0), the
double integral above can be evaluated analytically as Aϕ (r, 0) = µ80πβ2I0 [F(Ri /r) − F(Ro /r)] , where
the function F(x) = (1 − x)K(m+ ) + (1 + x)E(m+ ), +sgn(1 − x) [(1 + x)K(m− ) + (1 − x)E(m− )] ,
which is produced by the computer algebra system Mathematica in terms of the complete elliptic
integrals of the first and second kind, K(m± ) and E(m± ), of the parameter m± = (1±x)
2 [6]. Taking
into account that F(1) = 4, the magnetic flux (6) can be written as
Φc =
µ0 β I0 Ro ¯
Φ(λ ),
8π 2
¯ λ ) = F(λ ) + λ F(λ −1 ) − 4(1 + λ ) is a dimensionless magnetic flux and λ = Ri /Ro is the
where Φ(
radii ratio.
In the following, we assume the axial separation between the rings to be so small that the
magnetic flux through the rotating ring is effectively the same as that through the coil, i. e., Φd ≈
Φc . Substituting the relevant parameters into the equation defining the marginal rotation rate we
eventually obtain
4π (R¯ − (2 + κβ 2 ) ln λ )
Rm = µ0 σ dRo Ω =
¯ λ)
κβ Φ(
0.4 0.5 0.6
Radii ratio, λ
one ring sectioned
both rings sectioned
one ring sectioned
both rings sectioned
Dimensionless contact resistance, R
radii ratio
Dimensionless contact resistance, R
Optimal pitch angle, arctan(β) (degrees)
Optimal radii ratio, λ
Minimal magnetic Reynolds number, Rm
Magnetic Reynolds number, Rm
Figure 2: Marginal Rm versus λ for only one ring sectioned (κ = 1) at various dimensionless
contact resistances R¯ and the optimal β , (a). Minimal magnetic Reynolds number Rm (b), optimal
radii ratio λ , and the coil pitch angle arctan β (c) versus the dimensionless contact resistanceR.
which is the marginal magnetic Reynolds number defining the dynamo threshold depending on the
spiral pitch angle arctan β , the radii ratio λ , and the dimensionless contact resistance R¯ = 2πσ dR.
The case of solid rotating ring considered above corresponds to κ = 1, whereas κ = 2 corresponds to the rotating ring sectioned similarly to the stationary one except for the opposite direction
of the spiral slits. As seen from the expression
√ above, the latter case is equivalent to the former
with both Rm and β reduced by a factor of 2.
¯ In the
Now, let us determine the optimal β and λ that yield the lowest Rm for a given R.
simplest case of a negligible contact resistance, which corresponds
to R = 0, Eq. (8) yields Rm ∼
2/(κβ ) + β . It means that Rm attains a minimum at βc = 2/κ , which corresponds to the optimal
pitch angles of 54.74◦ and 45◦ for only one and both rings sectioned. The respective lowest values
of Rm, 34.63 and 24.49, are attained at the same optimal radii ratio of λ = 0.3602. (see Fig. 2(a)
¯ which also causes
for the case of only one ring sectioned). The minimal Rm increases with R,
a steep reduction of the optimal radii ratio and a comparably fast rise of the pitch angle (see Fig.
3 Feasible setup
Finally, let us evaluate the rotation rate required for self-excitation in a setup with the outer radius
of Ro = 30 cm and the ring thickness of d = 3 cm. First, we need to estimate electrical resistance
of sliding liquid-metal contacts. A suitable metal for such contacts may be the eutectic alloy of
GaInSn [5], which is liquid at room temperature with the kinematic viscosity ν = 3.5 × 10−7 m/s2 ,
electrical conductivity σGaInSn = 3.3 × 106 S/m. Assuming the contact gap width of δ = 0.5 cm and
δ σ −1
the inner radius Ri ≈ 10 cm, we have Ri ≈ 2πGaInSn
dRi ≈ 0.02 µ Ω. The resistance of the outer contact
is by a factor of λ = Ri /R0 = 0.33 lower than Ri . Then the dimensionless contact resistance can be
σCu δ
estimated as R¯ = 2πσCu dRi (1 + λ ) ≈ σGaInSn
Ri (1 + λ ) ≈ 0.2. If only one disc is sectioned, which
is easier to manufacture, the respective magnetic Reynolds number is Rm ≈ 40. This corresponds
to the rotation frequency f = 2Ωπ = 2π µ0Rm
σCu dRo ≈ 10 Hz, which is well within the operation range of
standard AC electric motors. The respective linear velocity of the outer edge of the ring is around
v ≈ 20 m/s. At this velocity the tensile stress at the rim of the ring, ρCu v2 ≈ 4 MPa, is more than by
an order of magnitude below the yield strength of annealed Copper [7]. The optimal inner radius
Ri ≈ 0.3R0 ≈ 9 cm following from Fig. 2(c) is not far from the value assumed above. The respective
pitch angle for β ≈ 1.6 is about 58◦ .
The number of spiral arms is determined by the following arguments. The current distribution
defined by Eq. (1) can hold only in the inner parts of the ring which are radially confined between
the spiral slits. This ideal distribution is expected to break down at the rims, which are radially
exposed to the edges of the ring located at the nearly equipotential metal liquid contacts. In order
to confine this perturbation to the outer rim with r/Ro ≳ 0.9, Eq. (2) suggests that β−2
ln 0.9 ≈ 40
equally distributed spiral slits are required.
The last critical issue is the viscous power losses associated with the turbulent drag acting on
the outer sliding contacts at high shear rates. These losses can be estimated as Q = Sτ v ≈ 7 kW,
where S ≈ 2π dRo the area of the outer sliding contact, τ = 4c ρ2v is the turbulent shear stress, and
c ≈ 0.02 is the Darcy friction factor for turbulent pipe flow with the Reynolds number Re ∼ 105 [8].
Note that the relatively large setup size is due to the turbulent energy dissipation which scales as
Q ∼ (dRo )−2 ∼ Ω2 . Namely, reducing the system size by one third would require about five times
higher power input to achieve self-exitation of the magnetic field.
In conclusion, the proposed disc dynamo design appears feasible in terms of both the disc
spinning rate and the power required to drive it. A corresponding device, which is shown in Fig.
1(b), has been built and is to undergo first tests in May 2014.
[1] Moffatt, H.K.: Magnetic field generation in electrically conducting fluids. Cambridge, 1978.
[2] Beck, R.; Brandenburg, A.; Moss, D.; Shukurov, A.; Sokolov, D.: Galactic magnetism: recent
developments and perspectives, Annu. Rev. Astron. Astrophys. 34 (1996) 155–206.
[3] Bullard, E.C.: The stability of a homopolar dynamo, Proc. Camb., Phil. Soc. 51 (1955) 744–
[4] Rädler, K.-H.; Rheinhardt, M.: Can a disc dynamo work in the laboratory? Magnetohydrodynamics 38 (2002) 211–217.
[5] Maribo, D.; Gavrilash, M.; Reilly, P. J.; Lynch, W.A.; Sondergaard, N.A.: Comparison of Several Liquid Metal Sliding Electric Contacts, Proceedings of the 56th IEEE Holm Conference
on Electrical Contacts (HOLM) (2010) 1–7.
[6] Abramowitz, M.; Stegun, I.A.: Handbook of mathematical functions, Dover, New York, 1972.
[7] Li, M.; Zinkle, S.J.: Physical and Mechanical Properties of Copper and Copper Alloys, In:
R.J.M. Konings (ed.), Comprehensive Nuclear Materials, vol 4, Elsevier, Amsterdam, 2012,
pp. 667–690.
[8] Maribo, D.; Sondergaard, N.A.: Further studies of a low-melting point alloy used in a liquid
metal current collector, IEEE Transactions on Components, Hybrids, and Manufacturing Technology 10 (1987) 452–455