End-cap Overview - PPD - STFC Particle Physics Department

advertisement
End-cap
Closeout Review
Stephen Haywood
Rutherford Appleton Lab
 Introduction (10’)
 Support Structure (20’)
 Services (beyond Disc) (60’)
 Integration (30’)
 General (30’)
Stephen Haywood
End-cap Closeout Review
2
7 Dec 2007
Past Reviews



Support Structure Contract
Disc Assembly
Thermal Enclosures
Stephen Haywood
7 Sep 2004
27 Jun 2005
22 Sep 2006
End-cap Closeout Review
3
7 Dec 2007
Support Structure Contract
Stephen Haywood
End-cap Closeout Review
4
7 Dec 2007
Problems









Unforeseen problems led to delays and more resources.
Organising the contract through CERN proved difficult.
Two major failures in FWT of flat Panels associated with Cyanate
Esther – switch to FM73U film adhesive.
Serious problems placing Disk Fixation holes in Support Cylinder;
required 6 man-months of RAL effort to rectify.
Inappropriate use of 3D models by contractor.
Contractual problems with “build to print” design concerning
responsibilities.
We were poorly prepared to accept components (expected more
from Contractor; needs to be specified) – needed longer than 3
months.
Metrology can be much harder than envisaged !
Need greater clarity in who pays for what, including Tooling and
excess Materials.
Stephen Haywood
End-cap Closeout Review
5
7 Dec 2007
Successes


Probably correct not to impose penalties.
Load-test and Fit-check at company.
Stephen Haywood
End-cap Closeout Review
6
7 Dec 2007
Disks
Stephen Haywood
End-cap Closeout Review
7
7 Dec 2007
Issues
More Engineering effort should be put into
 Manufacturing processes
 Assembly
 Testing
Key design criteria should include
 Robustness – physical & ESD damage
 Serviceability, including Repair Scenarios
Secondary physics requirements should not compromise the
engineering principles.
Physicists’ requests should be questioned if they have significant
cost/time/design implications.
Stephen Haywood
End-cap Closeout Review
8
7 Dec 2007
Companies should be given QA requirements and acceptance criteria.
Regular checks should be made of series of components and care
should be taken to anticipate needs for raw materials.
Prototyping consumes many more raw materials than might be
expected.
Need more resources than expected ! (See later)
Stephen Haywood
End-cap Closeout Review
9
7 Dec 2007
Thermal Enclosure
Stephen Haywood
End-cap Closeout Review
10
7 Dec 2007
Issues













Is a TE needed? What is its role? (Thermal, Gas, Moisture, G&S)
Could a single TE be used?
Commonality between Barrel & End-cap
RAL avoid taking on too much work
Do not underestimate work/time/money, especially for trivial things
which are “just a …”
Ensure good connection with other aspects of project
Think abut G&S earlier on
Consider working with CTS/Lola
Sound out companies before tendering; engage in R&D
Consider acceptance criteria in advance
Insist on sensible envelopes
Avoid active cooling
Cu foil is much more flexible for G&S
Stephen Haywood
End-cap Closeout Review
11
7 Dec 2007
From Pippa
The hand over of the OTE to Valencia seemed to go smoothly, but the ITE at
CERN seemed to be almost too difficult (mea culpa).
Stephen Haywood
End-cap Closeout Review
12
7 Dec 2007
Completion



Jeff Bizzell – Heater Wiring
Jason Tarrant – Cooling
Jason Tarrant – Documentation:
SCT End-cap Documents & Drawings – ATL-IS-AN-0009
Abstract
This document is a guide to the many documents and drawings relating to the design of
the SCT End-cap. All the documents and drawings in this document are those
created by staff of the Science and Technology Facilities Council (STFC) in the UK .
This document can be used to locate any relevant information on any part of the SCT
End-cap that was produced by STFC.
Grounding
Drawings
G&S Parts
TD-1011-598 (ATLISES_0095) is the GRP Front Support cover assembly, this calls all piece parts required.
TD-1011-161 to 164 (ATLISES_0101 to 0104) are the 4 sections of Cu-polyimide Support Cylinder ground sheet.
Documents
ATL-IS-EN-0014 ATLAS SCT End-cap Grounding and Shielding, Engineering Implementation
Stephen Haywood
End-cap Closeout Review
13
7 Dec 2007
This Review
Questions:
 What have we learnt ?
Includes good lessons; Managerial & Technical
 What could be done better or differently ?
 What might we want to take forward for the future ?
& Celebration/Thank-you !
Stephen Haywood
End-cap Closeout Review
14
7 Dec 2007
Support Structure
Stephen Haywood
End-cap Closeout Review
15
7 Dec 2007
Stephen Haywood
End-cap Closeout Review
16
7 Dec 2007
Preparatory work:
 Prototyping … use of Contractors?
 FEA
Design of Support Structure:
 Was the concept of a Cylinder and Panels better than a space
frame?
Materials & Construction Techniques:
 CFRP, Cyanate-Esther adhesives, Korex
 Edge sealing
 Testing Support Cylinder bonding
Was RS4 CE film really OK? Unable to undertake FWT test
Contract:
 Should Support Structures and Disks have been in single Contract,
ensuring good fit?
Stephen Haywood
End-cap Closeout Review
17
7 Dec 2007
From Patrick
The problem was that PCI was not setup to machine cylinders within the
specifications. My experience with the disks was that PCI have made
excellent quality disks, but even machining the outer and inner diameter
was hard for them. Next time it is better to use PCI or any other composite
company where they are good in: making composite structures, and find the
right company for machining the parts if high tolerances are required. This
was done for the barrel cylinders.
The mistakes in the machining have caused several problems and confusions
throughout the project. At the end there was no alternative than to accept,
but everything became more difficult: measurements and supports that did
not fit and needed modifications
Edge sealing of the composite structures: This should have been done better,
at least for the wings. We found several Thermal Enclosure leaks through
these aluminium foil closures, and probably still we have leaks through
these.
Stephen Haywood
End-cap Closeout Review
18
7 Dec 2007
Services
Stephen Haywood
End-cap Closeout Review
19
7 Dec 2007
STFT
Stephen Haywood
End-cap Closeout Review
20
7 Dec 2007
Stephen Haywood
End-cap Closeout Review
21
7 Dec 2007
PPF1
Stephen Haywood
End-cap Closeout Review
22
7 Dec 2007
Services
Cylinder  Radial  Cryostat

(Cooling ) LMT  Opto  DCS  FSI

Routing & lengths

Caused a lot of worry for JT & TW
 Did we get it right (except for Heaters)

Physical routing of Heater Pad wires
Bends
Clips
Grounding

LMT’s: Al  Cu



Stephen Haywood
End-cap Closeout Review
23
7 Dec 2007
From Pippa
I worry a lot about the mixture of metals we've ended up with, and the problems
we had swapping from aluminium to copper C-seals.
The cooling layout has been the most recent headache, since everything had to
be redesigned to relocate the heaters. We are fortunate that the C-wheels
are not there, otherwise there's no way we'd have fitted everything in.
We have had the problems of dealing with overlength of a lot of the services,
but thankfully nearly everything has been long enough. There are just a few
heater pad cables that need extensions, due to a design change not getting
propagated through.
The routing of cables to PPF1 worked really smoothly for the end cap, and
benefitted a lot from adequate prototyping being done in advance. We had a
few issues with optimising the way the grounds were brought together at the
clamp, but all the horrible complications of different strip back lengths just
worked.
Dealing with the fibres as they came off the end cap was also rather easier
than for the barrel.
Stephen Haywood
End-cap Closeout Review
24
7 Dec 2007
And of course the evaporative cooling... more extensive and realistic testing
before hand might have found the problems with the control thermocouples
with more lead time. This still wouldn't have addressed the unreliability of
the heaters and the short to ground.
Stephen Haywood
End-cap Closeout Review
25
7 Dec 2007
STFT

Construction:

Foam & glue
 CFRP



Accessibility
(Sealing)
(G&S)
Stephen Haywood
End-cap Closeout Review
26
7 Dec 2007
From Patrick
What I can recall is that the STFT delivered to Nikhef was of very bad quality. The slots had to be remachined and repainted with araldite.
Also a mistake was made with rotational and mirror symmetry, which had to be fixed in situ.
I also think NOW that the parts should be been made much more precise to
make everything fit much better. At the end most of the following parts
needed adjustments to make them fit.
Stephen Haywood
End-cap Closeout Review
27
7 Dec 2007
Cable Trays
Radial  Cryostat



Complexity  Time (Design & Assembly) & Cost
Material
Assembly
Stephen Haywood
End-cap Closeout Review
28
7 Dec 2007
From Patrick
The CCT worked very well as well, except the Ribbon Guides which fitted very difficult on the tabs.
Especially at the top it took sometimes more than 1 hour to mount one!! Too reach too far inside
and only one hand available.
Mounting the RCT cover is also difficult and dangerous with the too long LMT's and R/O ribbons. Has
to fit on too many holes and screws. Need about 4 people in a too small area.
Stephen Haywood
End-cap Closeout Review
29
7 Dec 2007
PPF1
Cooling  LMT  Opto  DCS  FSI


Accessibility
Interaction with other detectors
Stephen Haywood
End-cap Closeout Review
30
7 Dec 2007
From Patrick
The PPF1 design did not work out very well in the beginning, also because
Jason was too busy with the HEX redesign. Too late and too many
problems. At the end all is working, but could have been a lot better.
We should have had proper protection taken into account during the assembly.
The electrical PPF1 were too fragile. In ATLAS there is very little respect for
somebody else his parts. People leaning against or on PPF1's to reach
somewhere. We have really seen non SCT people abusing our PPF1's. Of
course this caused many faults, and damages.
Also the Cooling PPF1's are not very well supported. With making the Serto
connections too much force is needed and move around. One patch panel
has M3 and the other M4 screws, very annoying.
Stephen Haywood
End-cap Closeout Review
31
7 Dec 2007
From Patrick
The Type II cable clamp should have been wider closing the PPF1 box. This would prevent dirt getting
into the box and solve problems we had with routing the R/O ribbons and LMT cooling pipes. LMT
cooling should have had a proper patch panel at the ECex1 connector. The pipes coming of the
flange were not properly positioned and too stiff to reroute them.
The whole installation and connection testing of LMT's went rather well.
The R/O ribbons below the LMT's should be changed for next time. We had to
disconnect & connect R/O ribbons several times when LMT's were already
connected. This was rather risky with violating all the rules about bend radii
for optical ribbons. The Cu/Kapton foils caused us a lot shorts etc.
Next time some risk analysis / review should be done for PPF1's, and the
person designing should be expecting the worst.
Stephen Haywood
End-cap Closeout Review
32
7 Dec 2007
From Pippa
I'd also emphasise the importance of prototyping and system testing, with as
close to final parts as possible. A few examples:
There was a genuine effort to prototype barrel PPB1, but the final cables were stiffer, and it was a real
problem to fit 6 of them across the width available. We had to modify the PPB1 brackets on the fly,
even though they were partly glued to the cryostat wall, giving few options.
For the end caps, we tried out the real cable in the clamp ahead of time, and were able to redesign it in
a calm way, before the cable installation in the pit.
There has been lots of system testing of modules, but in the last round of power supply firmware
updates, there were changes made which were OK with shorter test cables, but failed to work
when combined with long cables and PP3.
Stephen Haywood
End-cap Closeout Review
33
7 Dec 2007
Cooling







Use of CuNi
X0
Bending
Connectors
Were we right to design our own?
Accessibility of Connectors
Flexibility for connection
HEX




Design
Communications (CERN, RAL, QMW)
Manufacture
Soldering problems
 Tony Feedback
Stephen Haywood
End-cap Closeout Review
34
7 Dec 2007
Assembly
Sealing
 Use of Foils & Membranes (Cu-Polyimide)
Adhesives
 Araldite 2011
 Techsil
 Templflex
Stephen Haywood
End-cap Closeout Review
35
7 Dec 2007
From Patrick
Parts received at CERN:
Most of the parts we received at CERN did not fit and needed modifications.
The last known are the metal N2 pipes which did not fit into the Elbow
adapters. We had to re-machine all the pipe ends to make everything fit.
Another example are holes in the Heater Pad CCT with were we have
7.5mm holes while M3 screws needed to be used. I think at the end we
spend a couple of man years modifying parts here at CERN.
Parts we received at CERN for an assembly we often received in different
batches. We did not have a good system at CERN to receive and storage
parts so often parts were lost for some time. Also due to time pressure, the
parts were usually not taken out of there boxes until we needed them and
then we found assemblies not complete or parts would not fit, causing the
next crises.
Often parts we received still needed some way of assembly: for example
soldering of some copper pipes for the N2 system. This would have been a
lot easier for us if these had already been soldered and there are several
other issues where this was the case.
Stephen Haywood
End-cap Closeout Review
36
7 Dec 2007
From Patrick
The SCT had a particular problem that all most all of our manpower at CERN
was travelling. If parts are shipped to CERN these should be shipped
complete, assembled and checked before shipping. Do not ask people at
CERN to fill in the blanks. If you for what ever reason want to ship parts to
CERN not assembled also deliver the right manpower to prepare the parts
at CERN as soon as they arrive.
We should have invested from day 1 much more in storage and cupboards at
CERN and should have given some sort of priority to checking parts as
soon as they arrive.
We were never very good in checking and approving drawings, which caused a
lot of problems.
Stephen Haywood
End-cap Closeout Review
37
7 Dec 2007
From Patrick
During the assembly of the EC the services TPPF1's should have been
positioned at there final location and moved inward for transport only.
The too long LMT's and ribbons caused several problems and cables are now
under tension, which may cause us some problems in the future.
The HEX assemblies should have been part of the assembly and testing in
NIKHEF/Liverpool or at least in SR1. Only simple pipe connections should
be allowed next time during installation. I know this has a big history with
Marco, but next time we should not allow this to happen again.
The Thermal enclosure should be sealed in SR1. This we should have never
allowed to happen.
Stephen Haywood
End-cap Closeout Review
38
7 Dec 2007
From Pippa
I think that having a design that didn't allow closing of the thermal enclosure in
SR1 was a big mistake. I do realise that this may have been the only
solution consistent with the envelopes originally imposed. But the detector
has been in the pit for months now without being finally closed at the end,
which is risky. It's also an interim problem for installation and testing, in that
we can't run cooling at all until the entire plumbing installation is finished.
The space constraints also pushed us to use lots of specially designed nonstandard connections, which was tricky, and in the end forced some things
with the installation sequence that were a problem for the schedule.
Techsil... pah!
Stephen Haywood
End-cap Closeout Review
39
7 Dec 2007
Grounding & Shielding
Stephen Haywood
End-cap Closeout Review
40
7 Dec 2007
Schematic
Stephen Haywood
End-cap Closeout Review
41
7 Dec 2007

Design

Retrospective
 Ground all metallic parts


Groundsheet on Support Cylinder
Membranes


STFT


Tabs
Small foils
Connections

Soldering
 Silver-loaded epoxy
 Finger-stock

Quench forces
Stephen Haywood
End-cap Closeout Review
42
7 Dec 2007
From Patrick
Using the Segmented Plate as our grounding ring was a very bad idea.
because also others use these plates to mount parts this has caused a lot of
problems. If we would get a short now it can be a million things and we
cannot systematically disconnect parts any more to find the short, which is a
big concern. So lets hope this does not happen!
Stephen Haywood
End-cap Closeout Review
43
7 Dec 2007
Integration
Stephen Haywood
End-cap Closeout Review
44
7 Dec 2007
Fit Rear Support Panel (Wing)
 Transfer to Cantilever Stand (CLS)
 Transfer loads to CLS with Rear Support Yoke and PPF1 & TPP
Support Yoke
 Insert ITE , using ITE Carriage
 Fit Link Plate
 Add OTE, using OTE Mounting Frame
 Insert into TRT
 Add Front Support Panel (Wing) with Bridge Support System
 Add Mechanisms and support from Rails
Also
 PPF1 & TPP to TRT Arms
 Shim Adjustment Stands
Not needed?
 Transportation Lock System
 Services Support Ring

Stephen Haywood
End-cap Closeout Review
45
7 Dec 2007
Issues


Did we anticipate all requirements?
(Eg. Split Front Support was anticipated right from beginning)
Was Tooling optimal and fit-for-purpose?
Stephen Haywood
End-cap Closeout Review
46
7 Dec 2007
From Patrick
Orange cantilever stand: Wrong paint was used. When using alcohol to clean
the paint was removed, so we were not really able to properly clean.
The "front transfer plates" were not very flat ~ 5mm. I still believe that this is why EC-C. For EC-C this
plate was mounted and stressed several parts and when removed, changed something on the EC.
I still believe that this is one of the causes why we have bigger leaks in the thermal enclosure than
on EC-A. On Side A the "front transfer plate" was re-worked and made flat to be well within
0.5mm.
Integration: I have never liked the way the EC was supported on the cantilever
stand. Also the transfer from the assembly frame to the cantilever stand and
from the cantilever stand to the EC trolley we should do differently next
time. The EC should have been supported in a kinematic way.
With the tooling provided we did not have any precise control over what we
were doing. Moving the EC onto the EC trolley we ended up 2mm away
from what we aimed and had a twist. At the end this was accepted as more
or less a good position. For Side A we had to make special to lift the EC in
the frame and reposition the EC at the front. It was too difficult and time
consuming to make tooling to reposition the EC at the rear.
At the end everything turned out to be ok, but it wasn't easy.
Stephen Haywood
End-cap Closeout Review
47
7 Dec 2007
From Patrick
A private comment on building detectors. You may not want to use this or word
it differently:
A lot of problems with building the EC's was on the manpower involved and
specifically on EC-C. The detector should be built by technicians and not by
physicists or engineers. I think most of the time people not "qualified" were
working on the EC. When doing these difficult projects people with the right
skills are absolutely needed. I also see this general change in institutes
where mechanical departments are reduced or even cancelled and I think
this is very bad. We need the best technicians to make the best detectors:
i.e. fix all the problems that we cause.
Also engineers should be much more involved in building detectors, especially
the parts that are designed. If this does not happen the engineer will not
learn from his mistakes.
One annoying thing I had with EC-C is that people were always too much in a
hurry. At the end this resulted in much more problems and delays. I also
believe that EC-C was shipped too early too CERN. People travelling to
CERN are not very efficient in working hours and EC-C had enough
problems which could have been solved in Liverpool. This would have
saved a lot of time.
Stephen Haywood
End-cap Closeout Review
48
7 Dec 2007
From Patrick
The trap we always fall into is accepting parts that we know are wrong, but due
to time pressure we accept them. This causes many more problems, than
solving the original problem.
Stephen Haywood
End-cap Closeout Review
49
7 Dec 2007
General
Interfaces & Communication
 CERN
 Nikhef
 Valencia
 Ned for G&S
RAL Organisation
 Proj Man: Debbie  Stephen
 Dept: Debbie  Eddie  Mike
 Senior Engineer to replace Debbie?
 Additional Engineer to support Jason?
Stephen Haywood
End-cap Closeout Review
50
7 Dec 2007
Scheduling
Most things took longer than expected – never enough time/people
Use of MS Project
 Did we get level of detail right?
 Need to schedule time for

Admin
 Holidays
 Contingency


Difficulty of individuals undertaking multiple parallel tasks
Time for Prototyping
Should we have been more involved at CERN, especially post
Integration?
Note: Jeff is still helping with wiring.
Stephen Haywood
End-cap Closeout Review
51
7 Dec 2007
Manpower





Never enough
Would have benefitted from separate Manager & Engineer
Eg. Difficult to manage contractors and do work
Importance of getting good people
Avoid loosing key people until work really finished (Eg. Ian)
If Manpower “envelope” is constant, front-load effort to allow for
problems
Not ideal for Dept planning; cost of keeping a large team in light of
delays and changes
Stephen Haywood
End-cap Closeout Review
52
7 Dec 2007
RAL Support
Workshops
Metrology
SSTD Workshop (John Spencer)
Contractors:
 Designers (John, Steve, Alan, Tommy)
 Technicians (Cooling Lab)
Lab space:
 Clean Rooms
Needed much more space than anticpated
 Cooling Lab
 R12
 Metrology
 Stuart’s area
Stephen Haywood
End-cap Closeout Review
53
7 Dec 2007
Cost





Consistently underestimated by factors of 2-3*
Initial inquiries gave false expectations or expected savings did not
materialise
“Cheap” materials turned out to be less cheap, eg. Airex
Bits and bobs can add up
Cost of effort in labour-intensive work adds to total cost (Cooling
pipes, Stiffeners, Membranes)
* The exception was the Support Structures which cost 0.6 M$, while
we feared they could be >>1 M$
Stephen Haywood
End-cap Closeout Review
54
7 Dec 2007
Financial Control









I struggled to follow costs in Famis
Identification of expenditure in Famis often obscure; lots of junk
Would have helped to subdivide FK36000 – I could not understand
how much we had spent on Discs, TE etc
Engineers spent money without communicating (did try to control
this several times)
Bits and bobs add up
I failed to understand budget and contingencies (from Janet)
Janet seemed to understand the bottom line and have significant
contingencies
Despite my lack of control, it seems miraculous there was not a net
overspend on the ATLAS Eng
In future, prepare very carefully before embarking on project and
consider parallel system (personal spreadsheets) to track
Stephen Haywood
End-cap Closeout Review
55
7 Dec 2007
Expenditure
Discs
Structures [1]
TE [2]
Services
Ass/Integration [3]
Total
Expenditure
416
341
41
185
50
1033
Budget
353
86
57
50
[1] UK paid 20% ie £61k of £317k PCI Contract
[2] We only supplied parts and paid for some tooling, we did not build OTEs or ITEs or
Rear TE Pads
[3] Estimates – not updated, but probably in budget
Uncertainty due to difficulty of assessing and I have not revisited …
ultimately bottom line from Famis
Stephen Haywood
End-cap Closeout Review
56
7 Dec 2007
From Disk Review:
One needs far greater resources (money, time, effort) than is often
guessed (take first estimate and increase by factor of 2 to 3 (p?).
What is expected to be trivial at an early stage in the design often can
prove more complicated when considered towards then end of the
design phase, when there may be many constraints.
From Patrick:
For time, money and manpower you should next time take into account
at least a factor 2.
Stephen Haywood
End-cap Closeout Review
57
7 Dec 2007
Other Things
Documentation
 Pro-E & Intralink
 EDMS
 CDD
Reviews
 Internal
 CERN (FDR, PRR, regular meetings)
Stephen Haywood
End-cap Closeout Review
58
7 Dec 2007
From Patrick
At the end I am satisfied with what we have achieved, but we have certainly not
chosen the easiest way. The result could have been better if more qualified
people (technicians) would have been based at CERN for the last two
years.
The last two years have been difficult for people travelling a lot to CERN,
especially for their personal lives. Next time an attempt should be made to
get more people based at CERN.
Stephen Haywood
End-cap Closeout Review
59
7 Dec 2007
ATLAS SCT EC Eng Team @ RAL
Workshop
TD
PPD
Alan Austin
Paul Barclay
Richard Apsimon
Geoff Burton
Steve Butterworth
Lewis Batchelor
Eamonn Capocci
Mike Curtis-Rous
Jeff Bizzell
Colin Dabinett
Peter Ford
Stephen Haywood
Mike Harris
Debbie Greenfield
Richard Holt
Cyril Locket
John Matheson
Mike Tyndel
Graham Rolfe
Craig McWaters
Chris Nelson
John Noviss
Brian Smith
Jason Tarrant
Steve Temple
Ian Wilmut
Stephen Haywood
End-cap Closeout Review
60
7 Dec 2007
Well done
&
Thank you
… the drinks
are on me
Stephen Haywood
End-cap Closeout Review
61
7 Dec 2007
Download