Transcript Boiling Basics

Target
Greg Smith (Jlab)
MOLLER collaboration meeting
September 18, 2009
Target Working Group:
• Greg Smith
• Silviu Covrig
• Mark Pitt
• Konrad Aniol
Outline:
• Performance scaling
• Cryo capacity
• Design concept
Summary of Target working group progress:
We are busy building a ½ power prototype target…
(aka the Qweak target)
Target Specifications
•
•
•
•
•
•
150 cm LH2 (17.5% X0) at 20K, 35 psia
5x5 mm2 raster area
85 µA beam current
Total cooling power required 5 kW
2 kHz helicity reversal frequency
Target noise contribution to asymmetry
width ΔA ~ 26 ppm
< ~ 5% contribution to ΔA
• Minimize window bkg
• Safe & reliable ops
Design by CFD
H2 Release/Safety
CFD calculations by S. Covrig (Jlab)
Heat Exchanger
Heater
Raster
Cell
Window
Dummy
GRS
Design Considerations
Knobs to turn:
• P & T
• Vflow
• Araster
• nhelicity
• nraster
• Intrinsic φbeam
• Cell/Flow design
• Window design
Constraints:
• Ibeam & Ltgt
• Window bkg
• Safety issues
• Available Pcooling
• Head
• ΔAstat
• Time available
• ASME compliance
6/24/2009
GRS
LH2 Targets for Parity Violation

p/T/m
L
P/I / E
psia/K/kg/s
cm
W / μA / GeV
sample
25/20/0.6
40
happex
26/19/0.1
pv-a4
Pv
W/cm3
dbeam
Δρ/ρ
δρ/ρ
mm
%
ppm
700/40/0.2
396
2
1
<1000
@60Hz
20
500/35-55/3
76
5x5
6x3
?
100
@30Hz
25/17/0.13
10
250/20/0.854
310
1.7
0.1
392
@50Hz
e158
21/20/1.8
150
700/12/48
467
1
1.5
<65
@120Hz
G0
25/19/0.3
20
500/40-60/3
346
2x2
<1.5
<238
@30Hz
Qweak
35/20/1
35
2500/180/1.1
245
5x5
<45
@250Hz
MOLLER
35/20/1
150
5000/85/11
120
5x5
<26
@2000Hz
5
Extrapolating Performance
Scaling the G0 Target Performance
Tgt
boiling Raster Target Beam Helicity Mass
width width length current Reversal Flow
(ppm) (mm)
(cm)
(uA)
(Hz) (kg/s)
G0:
238
2
20
40
30 0.29
Qweak:
34
5
35
180
250 1.08
Moller:
30
5
150
85
2000 1.08
Power
2
-1
-1
0.4
Fractional
Single
Full
Extra Increase
Octant StatisticalBeamtime
Above
Rate
width Required Counting
(MHz)
(ppm)
(%) Statistics
800
19125
140
81
6%
14%
1.03
1.07
1
Target Boiling Penalty
250%
Increase in running time (%)
Need similar
performance to
Qweak. Penalty
rises rapidly with
target noise &
with flip rate:
200%
150%
2 kHz
100%
500 Hz
50%
30 Hz
0%
0
20
40
Target Asymmetry Width (ppm)
60
Extrapolating Performance
Raster
Q weak  G0
2x2
5x5
Ltgt
150 cm
20 cm
Qweak = 238 ppm x 0.16 x
7.5
Ibeam
Massflow
85 μA
40 μA
x 2.1
4 l/s
15 l/s
x
0.27
nhelicity
 30 Hz 


 2000 Hz 
x
0.4
0.19 = 31 ppm
Note: G0 achieved σboil = 100 ppm with 3x3 mm2 raster.
Dependence
on ppm
G0 target
massflow
G0
achieved
σ
=
68
with
2*
the
pump
head.
boil
This dependence
determined
empirically
from
a
single
test
was
cubic!
Here
we
take
it
to
be
Would like more
flexibility
here!
which
n
flipping
using
gate widths,
and the
Note: mimicked
At 2 kHz
flip
rate,
expect
ΔA(stats)
= 78 ppm.
helicity
linear
(ultra-conservative).
WeC know
this knob works!
Hall
standard
is a runtime
bold extrapolation
given
Need σboilpivot
≤ 26 tgt.
ppm This
to keep
penalty < 10%
how little we Linear:
still 2understand
it…ppm) Not reliable.
0.27
(
31
Option 7x7
mm
? gain is purely statistical.
However,
part
of
the
Quadratic: 0.071 ( 8 ppm)
That is reliable!
Cubic: 0.020 ( 2 ppm!!!)
Msrd 30 Hz Δρ/ρ in Hall A
From Armstrong, Moffit & Suleiman (2004)
Machined 15cm LH2 beer can cells
Measured in Hall A with lumis
Confirms we win with Araster & νfan
G0 Raster & Pump Scaling
S. Covrig et al.,
NIM A551, 218 (2005).
31 Hz pump
42 Hz pump
Measured width
vs raster size
(stats & tgt noise
in quadrature)
Higher helicity reversal rates
The statistical width is given by:
2
2
stat  counting
 target
1. We can reduce the relative contribution of the target boiling term by
going to higher helicity reversal frequencies (increased counting).
2. Tests (VPI/Jlab/OU, June 2008) with a Hall C standard tgt indicate that
0.4
the boiling term drops with frequency as:
 30 Hz 
targ  30 Hz 

f


Measured
targ  constant
6/24/2009
80
60
40
20
μA
μA
μA
μA
10
GRS
Cryo re-summary
• New 4 kW ESR-II
–
–
–
–
Available 2013 – 2014?
Nominally 4.5 K, 3 atm supply
Return at 2.5 atm (only ½ atm ΔP!)
Possibilities for 6 kW at 15 K ?
• Old 1.2 kW ESR will survive
• Advised to plan for a hybrid HX ala Qweak
• Excess CHL capacity a possibility
(unofficially)
3 kW Hybrid Heat Exchanger
• Cooling Power >3000 W!
• Combine capabilities of both 4K
and 15K refrigerators  hybrid
HX
•
4 K: 2 layers, 2.4 kW @20 g/s
•
15 K: 1 layer, 900W @17g/s
•
24 liters of LH2.
•
CFD: head & freezing.
• Head: 0.6 psi @ 1 kg/s
• Doesn’t freeze despite 4K
coolant
• Basic design performance
calculated
analytically
(counterflow HX):
87.3 cm long, 27.3 cm diameter
Loads/Capacities: CHL 6GeV vs.12GeV
6 GeV
Unit Loads
Color key
6 GeV ops
12 GeV ops
Both
12 GeV
North Linac
2 K 50 K
(W) (W)
#
2 K 50 K
(W) (W)
Static loads
Transfer Line
530 6360
1
530 7000 0.57 228 3990 0.43 302 3010
Original CM’s
16
110 42.25 676 4648 21.25 340 2448
20
320 2200
12 GeV CM’s
50
250
5
250 1250
Dynamic loads
Original CM’s
72
12 GeV CM
250
Totals
#
5
42.25 3042
50
2 K 50 K
(W) (W)
South Linac
250 1250
21.25 1530
5
1250
#
2 K 50 K
(W) (W)
20 1440
250
5
1250 250
42.25 4248 11648 25.25 3598 7938 29.25 3562 6710
Capacities (W)
CHL#1 (W)
4600 12000
4600 12000
% of Full Load
92% 97%
78% 66%
CHL#2(W)
4600 12000
% of Full Load
77% 56%
From a talk by D. Arenius at ILC08, Univ. Illinois, Nov. ‘08
Viscous Heating
(v1  v2 )
hL 
2g
K L v2
hL 
2g
2
A1, V1
2
(Abrupt Enlargement)
A2, V2=V1*A2/A1
Flow
(Abrupt Contraction,
Commercial Fittings)
2
L v
hL  f
d 2g
Viscous Heating(W) 
(Circular Pipe)
Flow (l/s) Head(psi)
 6.89
pump efficiency
So, viscous heating v !
Ex: 15 l/s, 2 psi, 80%  250 W
30 l/s
 2000 W!
3
Note: ΔP = hL ρ g, Re = v d ρ / μ,
e ~ 0.0015 mm for Al pipes
Cooling Power Requirements
Pb(W) = Ib(μA) (g/cm3) t(cm) dE/dx(MeV/g/cm2)
With: Ib=85 μA, ρ=0.072 g/cm3, t=150 cm, 
Cooling Power Budget
60%
15
2
0.5
85
4562
150
75
345
50
Mass Flow (g/s)
Pump efficiency
Flow rate (liters/s)
Pump Head (psi)
Pump Power (hp)
Beam Current (uA)
Beam Power (W)
PID reserve (W)
Pump heat (W)
Viscouse heating (W)
Conductive Losses (W)
160.0
Pb=4.5 kW!
Coolant Massflows for a 20K tgt
140.0
13K
120.0
100.0
15K
80.0
60.0
40.0
4K
20.0
0.0
Total Load (W)
5182
0
1000
2000
3000
4000
Cooling Power (W)
5000
5 kW He ΔP with existing Infrastructure
Effective
Pipe id
(in)
0.861
3.076
0.884
0.884
1.756
3.076
Pipe
Area
(in^2) P(atm)
0.582
3
7.433
2.5
0.614
3
0.614
2.5
2.421
3
7.433
2.5
T(K)
4.5
20
15
20
15
20
rho(He)
kg/m^3=g/l
129.7
6.1
9.9
6.1
9.9
6.1
mass
flow
g/s
47
47
183
183
183
183
Transfer Line Anatomy
volume ODH time
flow to 19.5% velocity
l/s
(h)
(m/s)
0.362
3.02
1.0
7.701
3.02
1.6
18.418
0.77
46.5
29.987
0.77
75.7
18.418
0.77
11.8
29.987
0.77
6.3
L (ft) dP (psi)
300
0.68
300
0.02
300 114.33
300 186.46
300
3.25
300
0.30
Pipe
4K supply
LN2 supply
15/20 K
15/20K
4&5K supplies
LN2 supply
LN2 Supply:
Inner pipe 5” IPS
Outer pipe 6” IPS
Both Sch-10 A=7.4 in2
15 & 20 K: ¾” IPS pipe, Sch-10
0.884” id, A=0.6 in2
Return: 1 ¼” IPS pipe,
Sch5 = 1.66” od,
.065” wall, A=1.8 in2
Supply: Annular space inside 2” od tube,
.065” wall, A=0.6 in2
ODH
• Last time relayed a potential ODH concern
– Because of addt’l coolant flow
• However:
– Hall engineer (Brindza) says Helium was never
an ODH concern  no restrictive orifice
• Cuz it rises, escapes hall thru dome vent
– ODH concern is on LN2 supply- it has a
restrictive orifice
• But we will not use the LN2 supply (as a LN2 supply)
• No ODH issue here. But may be a flow restriction.
Cryo Caveats:
• Both HRS’s (& septa) at 300K
• No LN2 usage (supply line hijacked)
• SC Moller solenoid a special problem
– Was a challenge to solve for Qweak
• Minimal loads from the other halls
– MOLLER will require ~all of the coolant
– This problem is scheme-dependent
• Some schemes impact other halls less
• No (low) losses in xfer lines
• Stay flexible. Meet with cryo early
E158 Liquid Hydrogen Target
Refrigeration Capacity
Max. Heat Load:
- Beam
- Heat Leaks
- Pumping
Length
Radiation Lengths
Volume
Flow Rate
1000W
500W
200W
100W
1.5 m
0.18
47 liters
5 m/s
Disk 1
Disk 2
Disk 3
Disk 4
Wire mesh disks in target cell region
to introduce turbulence at 2mm scale
and a transverse velocity component.
Total of 8 disks in target region.
Prototype for 11 GeV Møller Target Cell
Beam heating 4600 W @85 μA
Need δρ/ρ < 26 ppm @ 2000 Hz
Predicted ΔP = 0.5 psid
CFD by
S. Covrig, JLab
Beam
150 cm
Prototype: E158-type Target Cell
150 cm long, 3” diameter
Shows obvious areas where improvements can be implemented.
CFD: Disks do not seem to help!
Bulk Heating
• First CFD model has clear problems
at flow inlet. Still:
– ΔT(global) = 0.4 K
– ΔT(beam volume) = 1.2 K
• Δρ/ρ = 2%
• Clearly due to hot spot in the model
– ΔT = Q/(m CP) = 0.4 K (best you can do)
• Not an onerous situation
Film Boiling @ Windows
Threshold for
film boiling
Window Heat Flux
90
Heat Flux (W/cm2)
Total Heat Flux
(dE/dx) / Araster
80
70
60
50
40
30
20
10
0
Critical
• MOLLER looks
promising: careful
design may
eliminate film
boiling @ windows!
G0
Qweak
MOLLER
Convective part
Predicted by CFD
Two Phase CFD (window boiling)
CFD simulation by S. Covrig
Rastered Beam profile on 0.005” Al cell entrance window
Velocity Contours
Both Phases
Vapor Only
(BLUE
means no
vapor
there, ie
just liquid).
6/24/2009
Qweak Lessons
• ASME compliance has been a nightmare
– Should be less onerous for Moller.
• Biggest problem: lack of management
support for early testing
– This will not change. Priority goes to “next
experiment”, & polarized targets.
– Only solution I see is to build offsite, then
test here (ala G0).
• We can build on-site. But then forget early testing.
• ASME complicates this, but it’s still possible
• Hold initial design review early
The End
ASME
Qweak target design authority: D. Meekins
Target Cooling Power Loads
• Beam: Pb(W) = Ib(μA) (g/cm3) t(cm) dE/dx(MeV/g/cm2)
– With: Ib=85 μA, ρ=0.072 g/cm3, t=150 cm,  Pb = 4.5 kW!
• Viscous Heating: Pv(W) = 6.89 Flow(l/s) Head(psi) / ε
– With: Flow 15 l/s, Head 1.3 psi, ε=60%  PV = 225 W
• PID Loop (feedback): need heater power to control T
– Reserve ~ 150 W
• Pump heat: Pp (W) ~ 20% (Pump power (hp) * 745.7)
– With: pump power = 0.5 hp, Ppump ~ 75 W
• Conductive losses:
– Guess, 50 W
2004 Cryo Agreement
Cryo Systems Capacities and Loads in equivalent g/s
Rao
Assumptions:
Confirmed during spring,
‘09 tests: See TN-09-041
1. No degradation in the CHL & ESR Cryo Capacities
2. No increase in cryomodule static (vacuum) and dynamic loads
3. No increase in Hall magnet and transfer line static (vacuum) loads
Loads
Capacity Present
Cryo loads
Nov-10-04
Near term
Expected
Option -1
FEL @ Present
Option -2
FEL_Off
Option -3
Hall-A_Off
Option - 4
SBR_On
Option - 5
4_Kw_On
FEL @ Present
CEBAF Linac 5.5 GeV
CEBAF Linac 5.8 GeV
188
196
188
FEL Linac
FEL FL03 full power
FEL new Injector
20
10
5
20
20
10
5
20
Halls Base loads on ESR Ref.
Halls Base load on CHL
Hall-C Moller
Hall-A Septa
11
5
2
5
11
5
11
5
2
11
5
2
CTF load on CHL
5
224
241
235
11
235
11
Total Cryo Capacities
235
11
20
40
190
188
188
20
20
10
5
20
10
5
11
5
2
11
5
2
11
5
2
11
5
2
226
208
226
241
249
235
11
235
11
235
11
235
11
20
235
11
40
246
246
Shutoff Hall-A (Credit)
Shutoff Hall-C magnets (Credit)
Shutoff Hall-C Target (Credit)
Available for Targets
188
196
Total Cryo loads
Cryo Capacities
CHL Capacity
ESR Capacity
SBR (estimate)
4 KW_Capacity_if installed
188
22
5
246
246
246
266
286
2
2
2
2
7
2
2
2
2
2
2
24
42
31
29
41
Closest Comparison: Qweak
•
•
•
•
Still virtual, but many lessons learned
Novel, dual HX technique & design approved
Use large Araster & vflow (viscous heating limit)
Cryo-agreement negotiated fall 2004
– thru JROC: all ADs, cryo, tgts, Qweak
– Coolant supply methods identified
• High pressure loop  higher T, more cooling
power, more sub-cooling
• CFD calculations steering cell design
• Fast (~300 Hz) helicity reversal
Pmax Considerations
Lower P:
–
–
–
–
Higher P:
Don’t go sub-atmospheric
Thinner windows = less bkg
Lower warm gas storage P
Less gas inventory
– More cavitation headroom =
Pop – PVP . Cavitation occurs
at trailing edge of pump
blades when P < PVP . For
LH2 PVP(19K) ~ 10 psia.
– Higher boiling temps
• Run at higher T 
more cooling power
• Run at fixed T 
more subcooling
– Less film boiling at
windows?
» No (App. 9.1)
6/24/2009
Settled on 35 psia &
GRS
Comparisons
Moller
• 2.4 times Qweak
• 17 times G0 forward
• 20 times E158
Energy Loss (11 GeV, 150 cm LH2)
• Ionization Energy Loss
– 4.995 MeV/g/cm2
– ~10% Higher than at lower energies
– 54 MeV total (what counts for heat load)
• Bremsstrahlung Energy Loss
– 1.74 GeV ! total
– That’s 16%!
Forget your focus!
The G0 Target Loop
CFD calculation by
S. Covrig, UNH