TCS Chiller Performance Review

1
TCS Chiller Performance Review

• Phil Willems
• Nov. 17, 2004

2
Design Requirements

1. Laser temperature around room temperature, .5K
2. Flow rates

Device Max power dissipated Flow requirement Source of requirement
Carbon dioxide laser 80W ½ gpm Access Laser Co.
Laser power supply 40W cool to the touch Access Laser Co.
AOM 30W 250 ml/min IntraAction, Inc.
AOM driver TBD TBD IntraAction, Inc.

1. Noise acoustic and EMI must be under ambient at
   nearest sensitive device (optical lever system)
2. Monitoring and control from control room using
   EPICS
3. Leak tight and robust

3
Commissioning Timeline

• Device was installed on 2K Y arm Oct. 27-29
• Device was commissioned from Oct. 29 to Nov. 8.
• Commissioning paused after overheating incident
  Nov. 6-7.
• Forensics from Nov. 8 to today.

4
Installation

• Site chilled water delivered to chiller from over
  100 feet away through uninsulated copper pipe
• This may account for some temperature issues and
  surges during control valve operation
• Site chilled water and TCS table water pipes
  interlaid and touching
• Insulation of TCS table supply pipe required for
  temperature stability
• Prototype control electronics not plug and play
• Control software implemented but not safety
  interlocks

5
Commissioning

• Submerged thermistor failed after 2 days (Nov. 1)
• Thermistor was fine problem later traced to bad
  solder joint outside chiller
• Short term solution was to glue another
  thermistor to pipe leaving chiller
• Laser power stabilized without AOM on (Nov. 3-5)
• Laser power unstable with AOM on (Nov. 5)
• Chiller noise performance verified (Nov. 6-7)
• http//blue.ligo-wa.caltech.edu/ilog/pub/ilog.cgi?
  groupdetectordate_to_view11/07/2004anchor_to_s
  croll_to20041107223232-robert

6
Overheating Incident

• After noise test, chiller was switched on to run
  overnight
• Apparently, one or both pumps failed to switch on
• System ran at very high temperature for about 1-2
  days
• Much damage and corrosion to the chiller
• Thermistors on TCS table showed no overheating,
  no flow to table- TCS table not affected

7
Why a Chiller Is Not a Boiler
8
Overheating Incident Analyzed

• Without submerged thermistor operating, no way to
  monitor chiller reservoir temperature
• Without software interlocks programmed, system
  could not power down safely had the submerged
  thermistor worked
• Had system been fully operational before running
  unattended chiller would have safely shut itself
  down
• Why did pump fail to start?
• LHO reports that all Rio pumps studied failed to
  start 100 of the time

9
Forensics

• Chiller disassembled, analyzed, and rebuilt
• Now it runs well even with the AOM on (for 30
  minutes at least)
• Rio 1100 pumps tested for reliability
• One pump tested by hand- 200 starts without
  failure
• One pump tested by robot- robot itself not 100
  reliable, but pump starts 15,000 times without
  failure
• Rio 2100 pump known to fail at LHO now under test
• Coolant corrosion tests
• Chloramine T not compatible with chiller
  components under accelerated aging test
• Propylene glycol and sodium hydroxide are
  compatible

10
Chiller Performance After Rebuild
11
Final Results Compared to Requirements

• Temperature of chiller and laser held to within
  .5K, as required
• Flow rate is 1.3GPM, as required
• Acoustic and electromagnetic noise is below
  ambient, as required
• Operation fully controlled and monitored from the
  control room, as required
• System is not yet robust to failure modes
• Design not fully implemented, software interlocks
  missing when needed
• New safeties can be added without change to
  conceptual design

12
Are the Design Requirements Good Enough?

• The TCS laser stability is unverified even though
  the TCS chiller meets requirements. Laser power
  and coolant temperature fluctuations are
  frequently correlated.
• Is .5K temperature stability sufficient?
• Is 1.3GPM flow sufficient?
• Recommend that no more TCS chillers be installed
  until reconditioned prototype performance is
  verified.
• New chiller parts will replace boiled parts
• Software interlocks will be fully implemented
• Hardware interlocks under consideration
• Plan new commissioning effort Nov. 29-Dec. 3.