Transcript Document

MICE collaboration meeting
Osaka 1  4 August
2004
AFC Module update & Plan
Presented by
Wing Lau -- Oxford University
Items for discussion
•
Response to the Safety Review report;
•
Situation with the burst window test & window survey;
•
Progress on welded window R & D;
•
Progress on the Absorber development and the MTA test
•
Use of CryoCoolers for the absorber & magnets;
•
Alternative designs for the radiation Shielding valve;
Response to the Safety Review
A report was prepared and is ready to be sent to the Review Panel.
The following is a summary of actions taken against the comments /
recommendations made by the penal:• Hydrogen Gas Handling & Venting system
•
Remove buffer tank and vent the hydrogen out directly - implemented
•
Remove relief valves in the hydrogen vent lines and have burst disks only – retained
•
Completely separate vent system for the absorber and vacuum spaces -implemented
•
Detail specification of the Relief valve – work in progress
•
Is hydrogen detector appropriate in the vacuum line – still under consideration
•
Hydrogen detectors are needed in the ventilation system and in the personnel space around
the experiment – will be implemented
Response to the safety review
• Hydrogen Gas Handling & Venting system (cont.)
•
Examine the level to which piping should be Argon jacketed – will be addressed
•
Replacing the flame arrestor with a vent pipe with an inert atmosphere –
implemented
•
Adopt Fermilab requirement vacuum system volume 52 x LH2 volume – not
implemented
• R & D on the Metal Hydride system
• The use of hydride system requires active control.
• The panel suggested an scaled model test.
•
It also asked the group to examine the safety issues associated
with this system
--- R&D proposal defined and submitted
Safety Review panel – Additional Points
– response already
•
Practicality of using intrinsically safe electrical equipment
drafted
•
Pipe joints – will be as requested
•
Detection of Hydrogen in Personnel areas – agreed
•
Attention to Interlocks, alarms and control system - ongoing.
•
Continuation of HAZOP assessment – agreed
•
Response to Absorber system leak scenario – ongoing
•
Potential of liquid hydrogen sloshing in warmer part of the feed pipe – to be
addressed in level control.
•
Leak between the helium and hydrogen compartment in Absorber unit ongoing
Situation with the burst window test & window
survey
Burst Window test & window thickness measurement
The burst test conducted on a 21cm window in April this year had
mixed success.
• The recorded burst pressure was 144 psi which is 44% above the
required burst pressure for safety;
• No shrapnel was found and the window burst out from the centre;
However….
• No definitive record of how thick the window centre was;
• No Photogrammetry was taken on the window deformation during the
pressure test to enable a direct comparison with FEA results;
• The window did not leak before it broke which was predicted in the
latest FEA analysis;
What worries us is………..
Prior to the current window test, it was accepted that the Photogrammetry
technique would be a reliable way of taking the window thickness
measurement. This view is now somewhat in doubt. The reason:
The resource constraint
The technique depends heavily on the availability of a single source – John
Greenwood. The resource situation at NIU & Fermilab was such that John
is assigned to a number of projects which require equal or higher priority.
Dona Kubik who was deeply involved with the last Photogrammetry test
has moved on within NIU to pursue other interest.
In a multi-task environment, people’s dedication would inevitably be
affected. This may affect the individual’s level of attention and commitment
to the project and may have an impact on the quality of their work.
What worries us is………..
The quality issue:
1) The required minimum thickness is 105 microns;
2) The CMM measurement at the manufacture’s place put the window
thickness at 117 micron (reference from Ed Black’s email);
3) The Photogrammetry measurement gave a figure of 192 microns;
4) Thickness measurement of the fragmented piece after burst was
105mm;
Based on (4), it is reasonable to say that the minimum thickness would be
around 117 mm. This would imply that the CMM measurement made at
Mississippi was not an unreasonable yardstick.
So, why was there such a large discrepancy in the Photogrammetry results?
What worries is………..
The quality issue:
1) The required minimum thickness is 105 microns;
2) The CMM measurement at the manufacture’s place put the window
thickness at 117 micron (reference from Ed Black’s email);
3) The Photogrammetry measurement gave a figure of 192 microns;
4) Thickness measurement of the fragmented piece after burst was
105mm;
Based on (4), it is reasonable to say that the minimum thickness would be
around 117 mm. This would imply that the CMM measurement made at
Mississippi was not an unreasonable yardstick.
So, why was there such a large discrepancy in the Photogrammetry results?
What worries is………..
The quality issue:
1) The required minimum thickness is 105 microns;
2) The CMM measurement at the manufacture’s place put the window
thickness at 117 micron (reference from Ed Black’s email);
3) The Photogrammetry measurement gave a figure of 192 microns;
4) Thickness measurement of the fragmented piece after burst was
105mm;
Based on (4), it is reasonable to say that the minimum thickness would be
around 117 mm. This would imply that the CMM measurement made at
Mississippi was not an unreasonable yardstick.
So, why was there such a large discrepancy in the Photogrammetry results?
John Greenwood’s
message showing
thickness at the window
centre (x=0) being
0.192mm
What worries is………..
The quality issue:
1) The required minimum thickness is 105 microns;
2) The CMM measurement at the manufacture’s place put the window
thickness at 117 micron (reference from Ed Black’s email);
3) The Photogrammetry measurement gave a figure of 192 microns;
4) Thickness measurement of the fragmented piece after burst was
105mm;
Based on (4), it is reasonable to say that the minimum thickness would be
around 117 mm. This would imply that the CMM measurement made at
Mississippi was not an unreasonable yardstick.
So, why was there such a large discrepancy in the Photogrammetry results?
……It is difficult to draw a conclusion to this. It could well be a
one-off event due to the lack of attention because the key
person was over-loaded with other project commitment. From
the project’s point of view, we do need to review our strategy
in future window thickness measurements
Is Photogrammetry still our preferred technique? If so, how do we
make sure that its quality is not operator dependent?
The current window is substantially stiffer than the previous design
(torispherical shape). This would allow the conventional CMM survey
back into the race again.
Other thoughts?
The FEA comparison
The earlier version of the window design predicted a bursting pressure of
around 100psi at the centre, and the actual burst pressure was recorded at
144 psi. So, why such a large discrepancy?
The accuracy of the FEA depends on a number of criteria such as:
• The accurate representation of the as-built geometry;
• The material properties;
• The loading;
The as-built situation:
• The required minimum thickness was 105 micron. The as-built
thickness was between 117 microns to 192 microns, a variance of
11% to 83%;
• The standard material property shows a yield stress of 273 MPa
and a UTS of 310 MPa. The material certificate shows a value of
between 272 to 293 MPa on yield, and between 298 to 315 on UTS.
• Similarly, the max. elongation was found to be only 11 - 12% while
the standard material shows a figure of 17%.
The bursting mode
Why didn’t it leak at a distance away from
the centre (as predicted) before it broke?
• In the light of the Safety review the
primary window requires to withstand an
external pressure of 1 bar without
buckling. This has lead to some 20%
increase in window thickness. This would
delay the burst of the window and shifts
the highest stress point away from the
centre.
High bending
stress area
• This thickness increase was however not
reflected in the window used in the burst
test.
Against this background and in the absence of a true record of
how the window deformed during pressure test, it would be
difficult to have a direct comparison between the test and the
FEA results.
Status of the welded window R & D test
WELDED WINDOW
• Objectives
– to investigate a practical weld seal solution for the vessel
windows
– incorporating a bayonet lock feature to react the internal
pressure force
– to have the capability of refurbishment and re-use of both
vessel and window
– compliant with construction codes and operational
requirements
Vessel Manufacture.
Body Bayonet thread
Assembly weld prep
Window Bayonet thread
Welded assembly
Design/Test overview
•
•
•
•
•
•
•
•
Material: AL 6061 for body and dummy window
Incorporate axial lock against internal pressure force
Minimise weld for vacuum/pressure sealing only
Thermal shock cycle on weld 3 cycles of RT>80k>RT
Vacuum leak check
Pressure test 6.8 barG (100psi)
Vacuum leak check
Refurbishment procedure
Welding parameters
Welding parameters
Pre Heat:
Current:
AC Balance:
HF:
Shield gas:
Back shield
Pre flow:
Post flow:
Filler wire:
Wire dia:
Tungsten dia
Manual TIG
AC
penetration
continuous
Argon
Argon
Initial weld
~ 80
240
7
6
none
3
9
AW 5356
5% mag
1.6
3.2
(no rotary manipulator)
C
amps
AC
l/min
sec
sec
(BS 2901 pt4)
mm
mm
Lathinated
Welding parameters
Pre Heat:
Current:
AC Balance:
HF:
Shield gas:
Back shield
Pre flow:
Post flow:
Filler wire:
Wire dia:
Tungsten dia
Manual TIG
AC
penetration
continuous
Argon
Argon
Re-weld after refurbishment
85
240
4
6
none
3
7
AW 5356
5% mag
1.6
3.2
(no rotary manipulator)
C
amps
AC
l/min
sec
sec
(BS 2901 pt4)
mm
mm
Lathinated
We should compare notes with KEK who supplies the
Absorber body which requires similar welding
Initial results
•
Weld test #1 (no thermo-couple used)
–
–
–
–
•
Leak test after thermo-cycle
Pressure test (6.8barG hold30 min)
Pressure cycling (no hold)
Leak test after pressure cycle
no detectable leak at 1x10-10 mbar Ls-1
no detectable leak
2x 0>6.8barG>0
no detectable leak at 1x10-10 mbar Ls-1
Weld test #2 (after refurbishment)
With thermo-couple
temp
C
Ambient
Pre heat
Initial running
Advancing to TC
Peak passing TC
Receeding from TC
Post weld
Window
17
85
75
95
160
140
95
54
Weld run
complete
0%
Start
20%
33%
60% < 30 sec
70%
100%
100% + 1min
Current Status
In system for vacuum
and pressure test
Comment on the progress made
• With the current budget and resource constraint, the
progress made by RAL is satisfactory
• Welding parameters that guarantee consistent
welding quality have been established
• Initial test result shows no detectable leakage
• Further work is continuing to establish if this
quality would be impaired by the refurbishment
and the re-use of both the vessel and the window
The results so far meet with our expectation
Progress on the Absorber and the MTA test
results
This talk is being covered by Shigeru in a
separate presentation shortly after this session.
Use of CryoCooler for the absorber and magnet
The technical justification for using CryoCoolers for the
LH2 Absorber & Coil cooling is outlined in a separate
document presented by Mike Green.
The following summarises Mike’s concluding comments
and looks at the engineering arrangement of fitting the
coolers within the AFC module
The Sumitomo SDRK 415-D GM Cooler –
as one of the possible choices
Characteristics of the 415D
GM Cooler
• 1.5 W is delivered at 4.2
K at the second stage.
• 18 W is delivered at 15 K
at the second stage.
• With 50 Hz power, the
cooler delivers 38 W at 50
K at the first stage.
• Cooling delivered at both
stages concurrently.
Date source provided by Mike Green
300 K
Attachment
Ring
Cryocooler
First Stage
T = 25 K to
T = 80 K
Cryocooler
Second Stage
T = 2.5 K to T
= 20 K
Cooler Requirements for MICE Magnets –
concluding comments from Mike Green’s talk
• The coupling coils have a single pair of 300 A leads.
Use a single 1.5 W cooler: 1 cooler T = 3.9 K
• The focusing coils have two pairs of 300 A leads.
Use two coolers: 1 cooler T = 4.7 K; 2 coolers T = 3.6 K
• The detector magnet has five coils. Each magnet coil has a
pair of 300 A leads.
Three coolers are needed: 2 coolers T = 5.1 K; 3 coolers T = 4.2 K
• In total as many as fourteen coolers may be needed to cool the
MICE magnets.
Connection of the Cooler
• If one wants to cool a magnet down with a cooler, the
cooler second stage must be connected directly to the
magnet with a flexible OFHC copper strap. The first
stage can be connected to the shields using a copper
strap.
• The temperature drop between the magnet high field
point and the cooler cold head has a negative effect on
magnet operation. Even a 0.4 K temperature drop will
affect the performance of the MICE coupling and
focusing coils.
• A gravity separated heat pipe can connect the cooler 2nd
stage cold head to the load with a very low temperature
drop (0.1 to 0.2 K) between the magnet hot spot and the
cold head.
Cooler Connection through a Flexible Strap
The temperature drop from
the load to the cold head is
proportional to the strap
length and inversely
proportional to the strap
area and the strap thermal
conductivity.
T = T3 - T0
L
T 
 Tc
kA
Tc = contact resistance
Tc is usually small.
Note: For T = 0.1 K, L = 0.15 m, and
k = 600 W m-1 K-1, then A = 0.0025 m-2
and Tc = 0.
In addition heat flow through 6061 Al
is quite poor at 4 K (k = 6 W m-1 K-1).
Note: For T = 5 K, L = 0.3 m, and
k = 1000 W m-1 K-1, then A = 0.00006 m-2
and Tc = 0.
MLI
Details of the Copper Strap
Arrangement
Cooler Connection through a Heat Pipe
T  Tb  Tf  Tc

Tb = Boiling T Drop
Tf = Condensing T Drop
Tc = Contact Resistance
These can be made small.
T = T3 - T0
The temperature drop
from the load to the cold
head is independent of the
distance between the load
and the cooler cold head.
Liquid He distributes the
cold around the coil.
Cooler Requirements for the Absorbers
• The absorber total heat leak should be 10 W or less.
Beam heating and dark current are not a factor in
MICE.
• A single cooler should be capable of holding the intrinsic
heat load into a MICE liquid hydrogen absorber. Direct
cool down of a MICE liquid hydrogen absorber using a
cooler may be difficult. The cooler first stage plays
almost no role in cooling the absorber.
• A liquid helium absorber can not be cooled.
Absorber Cooling with a Small Cooler
General arrangement of the CryoCoolers for
the LH2 Absorber and the helium bath for the
magnet in the Focus Coil Module
24-M4
24-M4
4-M6
24-M8
4-M6
24-M8
24-M4
24-M4
4-M6
4-M6
24-M8
24-M8
Concluding Comments
• It appears that a coupling magnet can be cooled with a single
cooler. The use of a heat pipe is advised to keep the T between
the far side of the magnet and the cold head down to 0.1 K.
• The focusing magnets may require two coolers to cool the
magnet and its leads. The leads are the dominant reason for
needing a second cooler. The coolers may be connected to the
magnet directly and through a heat pipe so that T < 0.1 K.
• The detector magnet requires three coolers to cool the magnet.
The dominant heat load is the leads on both stages of the
coolers. Direct conduction cooling is precluded by the INFN
magnet design.
Concluding Comments (cont.)
• It is unlikely that small coolers will be used to cool down the
magnets to 80 K. Using coolers to cool down some of the
magnets from 80 K to 4 K is possible, but it is probable that
liquid cryogens will be used to cool down the magnets over the
entire range of temperature. Recent results from the MTA
absorber cool down will be studied to see what we can learn
from there
• It appears that the liquid hydrogen absorber can be cooled
using a small cooler. The total heat leak into the absorber must
be less than 10 W. A heat pipe connection between the 2nd stage
cold head and the absorber is probably mandatory
• Direct cool down of the absorbers may be possible, but the
cooling strap length is long and the cross-section area must be
kept small. Liquid cryogen cooling using the absorber heat
exchanger will be the most likely absorber cool down scenario.
Alternative designs to the radiation
shielding
Background
• The equivalent of 50mm thick lead is needed to protect the
fibre trackers during the tuning of the RF cavities to full
gradient.
• Ed Black suggested using commercial gate valve , which
requires a minimum space of 370mm when all the flanges and
connecting plates are included. As a result, the space between
the final focus coil and the first match coil was set at about
600mm.
• The physics and engineering benefits of reducing the gap
spacing from 600mm to 450mm have already been discussed
extensively in several talks (Green & Bravar) in this meeting
Radiation Shielding Design Options
As a result, a design study was carried out to reduce the
thickness of the current ”Gate Valve”, proposed by Ed Black, to
220mm, or less, to fit in a space of 450 between the final focus
coil and the first matching coil. The new design will be presented
by Stephanie Yang at a separate session this afternoon.
•
Rotating shutter with hydraulic motor design
•
Double shutter with linear actuator design
•
Double shutter with ‘bicycle chain’ design
•
The ‘sash window’ shutter type design
In this talk, we shall address some of the issues
surrounding the interface between this module and the
modules that are connected to it.
Connection between the radiation shield module, AFC module and the
Detector modules
Note that the stud bolt is not connected to the
Radiation Shield module flange to isolate any
transmission of forces from AFC
Bellow flange is
threaded and fitted
with vacuum seal
Magnet forces
Stud bolts are used in compressing the bellow to
create a gap during the insertion of the AFC module.
Remarks
• Reducing the gap between the coils from 600 mm to 450
mm is acceptable. A gap less than 450 mm may increase
detector magnet heat load.
• More than one shield design will fit into a 450 mm gap
between the coils.
• Rigid & leak tightness connection between the Radiation
shield module & the detector module
• Flexible connection to the AFC module using bellows
isolates any magnet forces transmission from the AFC /
Coupling module to the Detector module. This protects the
Radiation Shield module casing from any excess bending
load