Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Spinning an Unmagnetized Plasma for Magnetorotational and
Dynamo Experiments
Cary Forest , Mike Clark, John Wallace, Carl Wahl, Blair Seidlitz, Chris Cooper,
David Weisberg, Jason Milhone, Ken Flanagan, Ivan Khalzov, Fatima Ebrahimi
Princeton, NJ April 9, 2013
Cami Collins
University of Wisconsin - MadisonStability, Energetics, and Turbulent Transport
in Astrophysical, Fusion, and Solar Plasmas
Thursday, April 11, 2013
A New Frontier in Laboratory Astrophysics
2
Where do magnetic fields come from?
Why does matter rapidly fall inward in accretion disks?
dynamo? magnetorotational instability?
ρV2>>B2/µoRm=µoσVL>>1
Thursday, April 11, 2013
Why Plasma? (instead of liquid metal)
-Explore large Rm and destabilize at lower B
-Vary Pm = µoσν from the liquid metal/protoplanetary disk regime (Pm<<1) to black hole, neutron star disk (Pm>>1)
-Study effects beyond ideal MHD (two fluid Hall effect, plasma-neutral interactions, compressiblity)
3
Thursday, April 11, 2013
4
Plasma Couette Experiment Madison Plasma Dynamo Experiment
•Challenge: -Need confinement for plasma to be hot (σ) and dense (ρ) -Difficult to drive flow in an unmagnetized plasma
Laboratory Studies of Flow-Driven MHD Instabilities
Thursday, April 11, 2013
•Confining and stirring an unmagnetized plasma in the Plasma Couette Experiment (PCX)
•Creating flow profiles to study the magnetorotational instability, a mechanism thought to be important in accretion disks
•Application of this new stirring technique for dynamo studies in MPDX
Outline
5
Thursday, April 11, 2013
(Keplerian)
•Keplerian flows of conducting fluids are hydrodynamically stable
•Can be destabilized by a magnetic field: the magnetorotational instability (MRI)
(Couette)
Couette Flow of Plasma
6
!"
!
!
Bz
Thursday, April 11, 2013
+-
B
E
V
S N S N
S
The Concept: Create Plasma Rotation Using Biased Cathodes
7
Anodes
PCX Magnetic Geometry
0.6
0.4
0.2
0
0 0.1 0.2 0.3 0.4 0.5R [m]
Z [m
]
Hot Cathodes
Initial Setup for Spinning from Outer Boundary
•Toroidally localized electrodes are biased to create JxB torque
•Velocity couples inward through viscosity
•Rotation is axisymmetric
Cathode
Mach Probeunmagnetized
Thursday, April 11, 2013
8
He Velocity Profile2 Outer Cathodes Biased 550 V
(measured with Mach probe) ECH
CurrentPressure
Density
TeRotation Driven
At Magnetized Edge
Vϕ couples inwardthrough viscosity
The Rotation Scheme Works!
Pm~10 Rm~50Thursday, April 11, 2013
9
ECH
CurrentPressure
Density
Te
The Rotation Scheme Works!
He Velocity Profile2 Outer Cathodes Biased 550 V
(measured with Mach probe)
Thursday, April 11, 2013
10
Ion Centrifugal Force is Balanced by Electron Pressure
Electron force balance:
Ion force balance:
Thursday, April 11, 2013
11
Radial Electric Field Measurement in Spinning Plasma
2 cm
Mach Probe
Tips to MeasureFloating Potential
Ar Floating Potential Profile
Location of Anode
Thursday, April 11, 2013
12
Electric Field and Velocity Measurements Agree
5.5 V/m 2 km/s
Ar Floating Potential Profile Ar Velocity Profile
Thursday, April 11, 2013
13
Viscosity Competes with Neutral Drag as Velocity Spreads Inward
edge-applied torque drag term, from ion neutral cx
viscosity couples momentum inward
Steady state velocity profiles can be calculated using measured plasma parameters (n, Te, neutral pressure) and modified Bessel functions:
where the characteristic momentum diffusion length scale is:
(without neutral drag, the solution is Couette flow)
Thursday, April 11, 2013
Viscosity Can Be Estimated By Fitting Velocity Profiles
14
Viscosity [m
2/s]
Future plans to measure Ti and confirm Braginskii viscosity
Ar Vϕ
Thursday, April 11, 2013
15
Viscosity [m
2/s]
Braginskii Viscosity Depends on Magnetization
Ar Vϕ
Ar νBraginskii, Sec. 4, (1965)
Thursday, April 11, 2013
16
Viscosity [m
2/s]
Helium is less viscous in the
magnetized region
Helium has much smaller
ionization fraction
Ar Vϕ
Ar ν
He Vϕ
He ν
Neutral Drag Modifies Profiles
[PRL 108, 115001 (2012)]
Thursday, April 11, 2013
17
Next Step: Rotate from the Inner Boundary
KeplerianSolid Body
Couette
Thursday, April 11, 2013
18
Pulley for raising and lowering center stack
into garage
Cathodes are heated tungsten filaments
Center Stack
Center Stack Assembly for Inner Boundary Rotation
Thursday, April 11, 2013
19
AnodesHot Cathodes
νum /10 [m2/s]
νm /10 [m2/s]
Bz [mT]
Cathode Placement is Critical
ωci τii
Thursday, April 11, 2013
20
Helium, 3 Inner Cathodes500 V inner, 400 V outer
Outer Cathodes (A)
Inner Cathodes (A)
ECH (kW)Pressure (Torr)
Te (eV)
n (m-3)
Spinning from the Inner Boundary
Thursday, April 11, 2013
21
500 V inner, 400 V outer
300 V inner, 400 V outer
Spinning from the Inner Boundary
Ti 0.11 eV 0.216 eV 0.25 eV 0.32 eVTe 5.9 eV 5.3 eV 6.5 eV 7.5 eV
n x1016 cm-3 1.7 1.8 2.4 2.3Ionization 0.35% 0.37% 0.74% 0.65%
Pm 0.2 0.88 1.3 2.9Rm 19 25 46 64
Rm_inner 5.1 10 11 17
575 V inner, 350 V outer
550 V inner, 400 V outer
3 Inner Cathodes 5 Inner Cathodes
Thursday, April 11, 2013
22
•To observe MRI, we simply need a weak seed field and Keplerian-like rotation that satisfies:
BRIEF ARTICLE
THE AUTHOR
(γ + Pm)(γ + 1) + S2
2(1 + 2) + 2ζR2
m(γ + 1)2 − 2(2− ζ)R2
mS2
+ηH
η
(γ + Pm)
2+
2ζR2m
(1 + 2)
(ζ − 2)Rm +
ηH
η
+ S2Rm(ζ + 2)
= 0
Pm =νη S =
kzVAηk2 Rm =
Ωηk2 γ =
γηk2 ηH =
Boµone
ζ =1
rΩ∂(r2Ω)
∂r =
h
(r2−r1)
kz =πh
kr =π
(r2−r1)
if vi = v, and j = e(nivi − neve), then ve = v − jne
Ohm’s law becomes E + v ×B = ηj +1ne
j×B
v =− ∂Φ
∂rBo
∂Ω∂r
< 0
∂(r2Ω)∂r
> 0
1
BRIEF ARTICLE
THE AUTHOR
(γ + Pm)(γ + 1) + S2
2(1 + 2) + 2ζR2
m(γ + 1)2 − 2(2− ζ)R2
mS2
+ηH
η
(γ + Pm)
2+
2ζR2m
(1 + 2)
(ζ − 2)Rm +
ηH
η
+ S2Rm(ζ + 2)
= 0
Pm =νη S =
kzVAηk2 Rm =
Ωηk2 γ =
γηk2 ηH =
Boµone
ζ =1
rΩ∂(r2Ω)
∂r =
h
(r2−r1)
kz =πh
kr =π
(r2−r1)
if vi = v, and j = e(nivi − neve), then ve = v − jne
Ohm’s law becomes E + v ×B = ηj +1ne
j×B
v =− ∂Φ
∂rBo
∂Ω∂r
< 0
∂(r2Ω)∂r
> 0
1
Profiles are Approaching MRI Unstable Regime
-Note: still need to turn on the Bz field (may change filament placement)
Thursday, April 11, 2013
23
HeliumVo=6.5 km/s, n=2x1018 m-3, Te=8 eV, Ti=1.5 eV,
F. Ebrahimi, 2011
•MRI growth rates can be estimated by a local stability analysis of incompressible, dissipative MHD with Hall term.
•The Hall term becomes important when ions are decoupled from the B field (ex: collisions with neutrals or finite device size)
•Result: Large growth rate appears only when B is antiparallel to Ω
MRI Growth Rate (global NIMROD simulation vs. local WKB)
-15 -10 -5 0 5[Gauss]
Local
-15 -10 -5 0 5[Gauss]
Local Global
What Would the MRI Look Like?
Thursday, April 11, 2013
24
Ion Neutral Drag term
Hall term
n=3x1010 cm-3
P=1x10-4 TorrTe=5.3 eVTi=.216 eV
How Does Ion-Neutral Drag Affect MRI?
Helium
Vinner=4.25 km/sVouter=5.5 km/s
Thursday, April 11, 2013
25
Velocity and Magnetic Field Structures Appear in the Saturated State, Which Depend on the Hall Term
MHD
HALL
[km/s]
F. Ebrahimi, 2011
Thursday, April 11, 2013
26
Hydrodynamic Stability of Viscous Couette Flow with Drag
Linearized momentum eq. + continuity eq. results in:
Chandrasekhar 1961Taylor 1923
Consider perturbations of the form:
Thursday, April 11, 2013
27
When is Viscous Couette Flow with Drag Unstable?
Rayleigh line
Marginal stability analysis with neutral-damping
unstable
stable
-Taylor vortices could occur at high Vinner, low viscosity-What are the boundary conditions?
Taylor (1923)
Reouter=VoL/ν
Re i
nner
=V
iL/ν
Thursday, April 11, 2013
28
Why Not Run Full Blast?
Before After
heating/arcing issues
Thursday, April 11, 2013
29
Laser Induced Fluorescence In Progress
Laser Excitation1047.00475 nm
Thursday, April 11, 2013
30
Laser Induced Fluorescence In Progress
Ti = 0.058 eV ± 4.4%
Laser Excitation1047.00475 nm
Thursday, April 11, 2013
31
The Madison Plasma Dynamo Experiment
•3 m diameter sphere•strong magnets = good confinement•fully ionized (200 kW LaB6 + ECH)•Goal: Rm ~ 2000
Thursday, April 11, 2013
32
Electrostatic Biasing Controls Edge Rotationpla
sm
aw
illrequir
e0.5
MW
.
S
N S N S N
+-
BE
V
Fig
ure
3:
Ele
ctrode
driv
efo
rflow
.A
lternat-
ing
posit
ive
and
negativ
eele
ctrostatic
bia
sis
applied
to
ele
ctrodes
betw
een
cusp
rin
gs.
The
result
ing
E×
Bvelo
city
ispurely
toroid
aland
controllable
.
Pla
sma
Rot
atio
nC
ontr
ol.
The
multid
ipole
geom
-
etry
allow
sa
sim
ple
and
robust
ele
ctrode
schem
e
to
driv
ethe
velo
city
field
inthis
geom
etry.
The
geom
etry
isshow
nin
Fig
.3.
The
concept
isto
use
toroid
ally
sym
metric
ele
ctrodes
betw
een
each
of
the
cusp
rin
gs,
and
to
apply
alternatin
gpos-
itiv
eand
negativ
epotentia
lsto
these
ele
ctrodes.
The
result
ing
ele
ctric
field
alt
ernates
inpola
rity,
whic
h,
together
with
the
alt
ernatin
gdir
ectio
nof
the
magnetic
field
giv
es
auniform
E×
Bvelo
c-
ity
inthe
toroid
aldir
ectio
n.
The
ele
ctrode
schem
e
mim
icsthe
boundary
conditio
nsthata
rotatin
gves-
sel
would
provid
ein
fluid
mechanic
sexperim
ents
(w
here
no
slip
condit
ions
can
be
assum
ed).
Sin
ce
the
UE×
Bvelo
city
can
be
estim
ated
on
the
surfa
ce
of
the
magnets
from
the
magnitude
of
the
potentia
ldrop,
the
dis
tance
betw
een
the
flux
surfa
ces,and
the
strength
ofthe
magnetic
field
on
the
surfa
ce
ofthe
magnets,it
can
be
proje
cted
into
field
lines
deeper
inthe
sphere.
We
pla
nto
insert
the
cathodes
into
magnetic
field
lines
whic
hare
separated
by
1cm
on
the
surfa
ce
ofthe
magnetic
s.
Apotentia
ldiff
erence
of
100
volt
sbetw
een
positiv
eand
negativ
esid
es
of
the
magnets
giv
es
a10
km
/svelo
city
whic
hshould
exis
tin
the
magnetiz
ed
regio
nofthe
pla
sm
a.
We
assum
ethatthere
will
be
avis
cous
coupling
of
the
magnetiz
ed
regio
nw
ith
the
unm
agnetiz
ed
regio
n.
The
peak
velo
city
UE×
B=
20
km
/s
isassum
ed
to
com
efr
om
a200
volt
diff
erence
betw
een
the
ele
ctrodes.
The
experim
ental
flexib
ility
of
the
rotatio
ncontrol
proposed
here
isrem
arkable
inthat
the
experim
enter
can
precis
ely
control
the
boundary
conditio
nΩ
(r
=a,
θ)by
adju
stin
gthe
volt
ages
betw
een
each
ofthe
ele
ctrodes.
So,fo
rexam
ple
,condit
ions
like
the
free-s
lip
surfa
ce
ofthe
sun
can
be
investig
ated
for
the
first
tim
e;at
the
equator
the
surfa
ce
of
the
sun
com
ple
tes
arotatio
nonce
every
25.4
days
while
near
the
pole
sit
’sas
much
as
36
days.
Rig
idrotatio
nshould
be
easy
to
accom
plish
by
adju
stm
ent
ofthe
ele
ctrode
voltages
at
each
latit
ude
to
impose
auniform
rotatio
n
rate
on
the
surfa
ce.
Assum
ing
that
the
magnets
are
uniform
lyseparated
polo
idally
and
that
they
have
identic
alstrengths,an
ele
ctrode
volt
age
∆U
(θ)∼
sin
θw
illgiv
euniform
rotatio
n.
One
mig
ht
ask
whether
more
com
ple
xflow
geom
etrie
sare
possib
lew
ith
such
asim
ple
flow
driv
e
or
ifthe
axis
ym
metric
geom
etry
preclu
des
more
com
plicated
flow
.T
he
flexib
ility
ofcontrollin
gthe
rotatio
nprofile
giv
es
aknob,how
ever,w
hic
hcan
be
used
to
driv
epolo
idalflow
s.
For
exam
ple
,ro-
tatin
gregio
nsnearthe
pole
sin
opposit
edir
ectio
ns,w
ith
rela
tiv
ely
little
rotatio
nnearthe
equatoria
l
regio
nis
exactly
the
type
ofgeom
etry
usin
gin
the
Von
Karm
an
flow
sregula
rly
investig
ated
influid
mechanic
sexperim
ents.
Inthose
experim
ents
dis
ks
at
each
end
ofa
cylindric
alvesselare
rotated
inopposit
edir
ectio
ns
makin
geach
hem
isphere
rotate
inopposit
edir
ectio
ns
and
strong
centrifugal
pum
pin
gthat
giv
es
polo
idalflow
sw
ith
inflow
at
the
equator
and
outflow
alo
ng
the
pole
s.
This
is
exactly
the
topolo
gy
used
inthe
liquid
sodiu
mversio
nofthe
Madis
on
Dynam
oE
xperim
ent
and
is
7
B
Arbitrary Vφ (r = a, θ)
Thursday, April 11, 2013
33
MHD Computation Predicts Laminar Plasma Dynamo
Rm=300, Re=100
Te = 9 eV, n = 8 x1017 m-3
Umax = 5 km/s, Helium
3
5 10 15Time [!"]
10-12
10-10
10-8
10-6
10-4
10-2
Energ
y [arb
]
Velocity
Magnetic
FIG. 3: Energy versus time for the boundary condition givenin Figure 1. For this run Rm = 400.
0.0 0.5 1.0 1.5 2.0 2.5 3.0Angle [Radians]
-1.0
-0.5
0.0
0.5
1.0
Toro
idal S
peed [arb
]
FIG. 4: Toroidal boundary condition based on odd-numberedspherical harmonic components, ! = 1, 3, 5, 7.
spherical harmonic components, ! = 1, 3, 5, 7, is pre-sented in Figure 4. In contrast to the boundary con-dition presented in Figure 1, this boundary conditionhas several large-amplitude sign changes. The result-ing steady state velocity field is presented in Figure 5.This boundary condition generates much more flow thanthe previous example. Its poloidal flow is considerablystronger, and toroidal velocity field fills whole sphere.With such a stronger velocity field it comes as no sur-prise that Rmcrit = 280 for this flow, much lower thanthe previous example.
III. MRI SIMULATIONS
Dynamo physics is not the only physics accessible withsuch an experiment. Because the velocity field, to someextent, can be fine-tuned, many velocity fields which re-quire dedicated experiments to be generated can be pro-duced. For example, a boundary condition that followsa Keplerian profile is plotted in Figure 6. This boundarycondition is Keplerian (v! ! "!
3
2 , where " is the cylin-drical radial coordinate) in the range 0.3 " # " $ # 0.3.
-0.5 -0.2 0.0 0.2 0.5V [arb]
V# Vpol
t = 2.62!"
FIG. 5: Steady state velocity field generated by the toroidalboundary condition given in Figure 4, before the magneticfield energy becomes large. The plotting convention is thesame as in Figure 2.
0.0 0.5 1.0 1.5 2.0 2.5 3.0Angle [Radians]
0.0
0.2
0.4
0.6
0.8
1.0
Toro
idal V
elo
city [arb
]
FIG. 6: Toroidal boundary condition which follows a Keple-rian flow profile. Odd-numbered spherical harmonic compo-nents, ! = 1 ! 17, are used to generate this profile.
However, the flow generated by the boundary, presentedin Figure 7 with Rm = 300 and Pm = 1, is not Kep-lerian. To show this, Figure 7 presents the toroidal ve-locity field at the equator, as a function of radius. It isclear that v!
!
"2
"
! r!0.76 for much of the radial range.Nonetheless, the flow satisfies the conditions required tobe unstable to the MRI in much of the volume of thesphere15.
Part of the time evolution of the instability is presentedin Figure 9, wherein is plotted the energy in the domi-nant toroidal velocity field modes of the simulation, asa function of time, in resistive units. Initially no ex-ternal field is applied, and the dominant velocity field
Spence, Reuter, and Forest, A Spherical Plasma Dynamo Experiment, The Astrophysical Journal 700 470 (2009).
Toroidal Velocity Simulation Input
Toroidal
Poloidal
Thursday, April 11, 2013
34
3000 4 kG SmCo Magnets Installed in MPDX
Thursday, April 11, 2013
Thursday, April 11, 2013
36
Current Setup
V! 0.0
+1.0
-1.0
North pole faces plasmaSouth pole faces plasma
LaB6 cathodeMolybdenum anode
90º N
90º S
0º
45º N
45º S
Coaxial LaB6 Cathode
Thursday, April 11, 2013
37
Even without microwave heating, LaB6 stirring cathodes create hot, dense plasmas
Discharge powers of up to 20 kW result in ne= 5x1017 m-3, Te = 5 eV plasmas.
LaB6 cathodes draw up to 50A each during each 10 second plasma discharge.
Thursday, April 11, 2013
38
Flow observed with cathodes withdrawn into magnetized region
Plasma spins up on neutral pump-out timescale.
3
(Te = 35 eV, Ti = 1 eV, Bcusp = 3000 G)
We = 2ρh = 4√ρeρi = 0.071 cm
1 >ωrf
Ωce=
ωrfme
eB
Re ∼ |(u·∇)u||ν∇2u| ∼ UL
ν
Rm ∼ |∇×(u×B)||λ∇2B| ∼ UL
λ = µ0σUL
∂B/∂t = ∇× (u×B) + λ∇2B
Pm = RmRe = µ0σν
Ro = U2LΩ sinφ
∂ω/∂t = ∇× (u× ω) + ν∇2ω
∂B
∂t= ∇× (u×B) + λ∇2B
mnduφdt
= (J×B)φ −mnuφτi0
+mnνii[∇2u]φ
τi0 =1
n0σcxu∂2uφ∂r2
+1
r
∂uφ∂r
−
1
L2u+
1
r2
uφ = 0
L2u = τi0νii
uφ(r) = AI1(r/Lu) +BK1(r/Lu)
Max velocity measured near anode location.Ve
loci
ty [
km/s
] fo
r Te=
5 eV
1
2
3
x1017
0
Neu
tral
Den
sity
[m-3]
1.0
0.8
0.6
0.4
0.2
Mach #
/ Ionization Fraction
Ionization Fraction
Ar Spin-Up at r=117 cm
Mach #n0
0 2 4 6 8 10 12 14 Time [s]
3
2
1
0
110 120 130 140 150 Radius [cm]
anode location
cathode location
Measured Velocity In Edge Region
Thursday, April 11, 2013
39
Energy Balance (He)
0 20 40 60 80 100Te [eV]
1
10
100
kW
QtotQionizationQcuspQradQion radQei
density: 5.0E+17 m -35
Loss Mechanism Expression [Energy/Time]
Ion losses at cusps !Acusp(e"V + 3
2Ti)
Electron losses at cusps !Acusp( 32Te)
Replacement ionization Qioniz ! !AcuspEioniz
Neutral radiation M(Te)Qioniz
Charge-exchange collisions 3
2nn0"!cxv#e(Ti $ T0)V ol
Ion radiation n2Ri(Te)V ol
TABLE I. Power loss mechanisms in magnetic bucket. Acusp
is the e#ective loss area at the magnetic cusps; M(Te) is theradiation coe$cient for neutral radiation; n0 is the neutraldensity; !cx is the charge-exchange cross-section; V ol is theplasma volume; T0 is room temperature; and Ri(Te) is theion radiation coe$cient.
energy, but also the energy they gain as they fall throughthe potential di!erence between the plasma and the cusp(Eq. 2). This gained energy is much larger than the initialion thermal energy. The e!ective cusp area for plasmalosses, which is used in calculating ion power losses, isnot well-understood, and empirically is di!erent for fastelectrons and for bulk plasma species [12]; here, we takethe loss width to be 4
!!i!e, proportional to the geomet-
ric mean of the ion and electron gyroradii at the mag-net faces. As each ion is lost, we assume there mustbe a new ionization to replace it and keep the densityconstant; each ionization reduces the plasma’s stored en-ergy by Eioniz. Power may also be lost due to ion andneutral radiation. The radiation coe"cients associatedwith these processes, M(Te) and Ri(Te), were providedby Dennis Whyte using the KPRAD code [13].The procedure for energy balance is as follows: we start
with given density n and heating power and we calculateTe by setting the heating power equal to the sum of thepower losses listed in Table I. We neglect ion radiationand charge exchange losses, and we also ignore Ti in theexpression for ion losses at cusps since it is small relativeto e#V . The results for helium and argon are shownin Fig. 4 for two di!erent heating powers. Comparisonwith Fig. 3 shows that we must assume only a fraction ofthe heating power couples to the plasma. For example,in Fig. 3, at n = 7 " 1016 m!3, we measure Te = 5eV with 4 kW of heating power, similar to the powerbalance calculation with only 1 kW of heating power (seeFig. 4). In other words, if we assume that only 1/4 ofthe microwave power couples to the plasma, we are ableto account for the electron temperature.The ion temperature is more di"cult to model. Ion
power loss is given by the last two terms in Table I, whileions are heated by collisions with electrons and by thecombination of ion acceleration through #V and ion-ioncollisions in the edge region. This last heating mecha-nism is thought to be the dominant one, but it is notwell understood. Instead, we may estimate the ion tem-perature [8] to be 0.25-0.75 eV, consistent with previousmulticusp plasma experiments.The implications of this power balance calculation for
1016 1017 1018 1019100
101
102
n [m−3]
Te [e
V]
1kW 10kW
HeAr
FIG. 4. Power balance calculation for helium and argon with1 kW of heating power; electron temperature decreases withdensity. Argon has no solution above certain density becauseof large radiation power losses.
1016 1017 1018100
101
102
103
n [m−3]
v!=3 km/s, Ti=0.5 eV, PECH=1kW
Rm
Pm
Re
HeAr
FIG. 5. Power balance calculation of Rm, Re, and Pm asfunctions of density for 1 kW heating power; Rm uses electrontemperature from Fig. 4.
Re, Rm and Pm are shown graphically in Fig. 5. Wetake the case of 1 kW heating power and use the re-sults of Fig. 4. It is seen that Rm can reach 103 at lowdensities, a value much larger than in liquid metal exper-iments. In addition, Pm can now vary over a large range,as opposed to being fixed at # 10!6 as in liquid metals.With extra power, the magnetic Reynolds number is ex-pected to reach even higher values, as Te is increased.This scaling will be tested in an upcoming experiment,which has recently been constructed in Madison [14].
First Results
Make Predictions With Power Balance Calculations
Thursday, April 11, 2013
40
Operational space set by power, density, and ion species
!"#$%&%' %&%( %&%)
*"+,-./012003
%
%&
%&&
%&&&
45"6.78+09"2:"7;.<7"
&%&
=&
>&
?&@&
AB=&&0CD
!2#$C2#,
C2#,
E>
"$ heliumargon
heliumargon
Thursday, April 11, 2013
41
Soon (May, 2013) … eventually (Fall, 2013)
V! 0.0
+1.0
-1.0
North pole faces plasmaSouth pole faces plasma
LaB6 cathodeMolybdenum anode
90º N
90º S
0º
45º N
45º S
V! 0.0
+1.0
-1.0
North pole faces plasmaSouth pole faces plasma
LaB6 cathodeMolybdenum anode
90º N
90º S
0º
45º N
45º S
Thursday, April 11, 2013
42
Summary
•Controlled velocity profiles can be created in axisymmetric ring cusp geometry using biased cathodes in the magnetized edge to drive flows in the bulk, unmagnetized plasma
•Within a factor of Ti, the measured viscosity is consistent with Braginskii
•Rotation at the inner boundary has recently been achieved in PCX, which now allows us to explore hydrodynamic and magnetohydrodynamic stability
•MPDX is entering final phase of construction, expected to achieve Rm ~1000
Thursday, April 11, 2013
43
Further Reading
Stirring Unmagnetized Plasma C. Collins et al., Phys. Rev. Lett. 108, 115001 (2012).
Magnetic Bucket for Rotating Unmagnetized PlasmaN. Katz et al., Rev. Sci. Instrum. 83, 063502 (2012).
Global Hall-MHD Simulations of MRI in a Plasma Couette Flow Experiment F. Ebrahimi, et al., Phys. Plasmas 18, 062904 (2011).
Numerical Simulation of Laminar Plasma Dynamos in a Cylindrical Von Kármán FlowKhalzov, et al., Phys. Plasmas 18, 032110 (2011).
A Spherical Plasma Dynamo ExperimentSpence, Reuter, and Forest, The Astrophysical Journal 700, 470 (2009).
Resistive and Ferritic-Wall Plasma Dynamos in a SphereKhalzov, et al., Phys. Plasmas 19, 104501 (2012).
Optimized Boundary Driven Flows for Dynamos in a SphereKhalzov, et al., Phys. Plasmas 19, 112106 (2012).
PCX
MPDX
Thursday, April 11, 2013
44
! " # $ % &!!
"
#
$
%
&!
' ())*+,-./012
'345*+,-./012
Pm=2.1, α=6, ν=134
Pm=1.3, α=8, ν=90
n=2.5x1010 cm-3, Te=7 eV, Ti=.3 eV
n=2.4x1010 cm-3 , Te=6.5 eV , Ti=.25 eV
n=5x1010 cm-3, Te=6.3 eV, Ti=.25 eV
Helium, P=1x10-4 Torr
Pm=0.6, α=10 ,ν=44
The Goal: Achieve MRI Unstable Parameters
MRI stable region
Rayleigh line
MRI Could Be Unstable At Feasible Parameters
Thursday, April 11, 2013
45
500 V inner, 400 V outer
300 V inner, 400 V outer
How Close Are We To MRI Instability?
-Still need to turn on the Bz field (may change filament placement)-MRI analysis is local
Thursday, April 11, 2013
Thursday, April 11, 2013
47
Taylor Vortices Can Occur at High Vinner, Low Viscosity
-Marginal stability analysis is local-What are the boundary conditions?
unstable
stable
Thursday, April 11, 2013
48
Ti = 0.058 ± 0.0026 eV
Doppler Broadening Doppler Shiftλ0 = 1047.00475
Ti = 0.5 eV Vi = 5 km/s∆λ=0.009 nm 0.0175 nm
Preliminary LIF Temperature Measurements
Thursday, April 11, 2013
49
1700°C
~3 cm from magnet
Magnet Temperature (Front & Back)
Before After
Center Stack Temperature Reduced By Silvering
Thursday, April 11, 2013
50
Particle Balance
Vplasma = 10 m3
Aloss = Wcusp × Lcusp
Wcusp = 4√ρeρi ∼ 1 mm
Lcusp = 220 m
dndt = 1
2csnAlossVplasma
Thursday, April 11, 2013