Web Newsletter
Front Page 
Livingston
Caltech
MIT | The LIGO Web Newsletter |
Front Page |
Livingston |
Caltech |
MIT |
LIGO Hanford Observatory News
Go Back Two Steps for Earthquake Repairs!
Interferometers do not like to be shaken. The 6.8 Nisqually
Earthquake on February 28, 2001, was the largest temblor to
strike Western Washington in over 50 years. Even while engineers
were assessing the damage done to the Capitol Dome in Olympia,
LIGO staff on the opposite side of the Cascade Mountain range
in Eastern Washington were investigating to determine what
damage may have been caused to the two Hanford Observatory
interferometers. The 2-km interferometer was scheduled to
participate in a
weekend-long engineering run, named E3,
starting in 10 days. This was to be the first coincidence
run between the two LIGO observatories. The full-up Hanford
interferometer (WA2K) was scheduled to take data in coincidence
with one of the 4-km arms of the LIGO Livingston detector.
People were rushing to implement a feed-forward tidal compensation
system for WA2k, hoping to be ready for the E3 run. (The feed-forward
system would prevent the Earth Tides from stretching the interferometer
out of range for its control system by anticipating the stretching of
Earth's crust due to lunar and solar gravity, and gently pushing the
optic within the buildings to compensate. This would cancel the large
tidal drifts--typically measured in tenths of a millimeters--and extend
locked intervals to several days at a time.) Meanwhile, another crew was
actively installing mirrors into the chambers for the 4-km long Hanford
interferometer (WA4K). And finally, a visiting
National Science Foundation
review panel, composed of scientists with extensive experience with major
science facilities, was on site to review our operations and the proposed
funding of operations for the next five years. Then the ground shook,
startling some but escaping the notice of others who had been focused on
their tasks. But very quickly it dawned on people that the interferometer
had been shaken up. We knew this was not good.
One of the first questions often asked of serious accident victims
is, "Can you wiggle your toes?" Stan Whitcomb, the leader of the
commissioning effort at Hanford, went to the control room shortly
after the shaker passed and tried wiggling mirrors, one by one.
It quickly became obvious that some of the mirrors had sustained
damage and repairs would be needed. Immediately we began to mobilize
efforts to enter the vacuum chambers and start inspections of the
mirrors. Since much of the development of LIGO's suspended mirrors
was done at Caltech--in the midst of California's earthquake
country--we were not unfamiliar to the damage earthquakes can do
to suspended mirrors. The LIGO prototype 40-meter interferometer
at Caltech was typically jolted about once a year by shakers during
the 90's, and we learned to incorporate earthquake safety into our
designs. In the photo at left above, those shiny screws aimed at the
periphery of the purplish optic are called "Earthquake Stops" and
their job is to catch the mirror as gently as possible when a jolt
comes. Earthquake stops are also incorporated into the optical tables
and vibration isolation systems to limit their motions. We have enough
confidence in these stops to know that the mirror substrates, surfaces
and coatings would do just fine, even in a fairly large local earthquake.
Still, the "Achilles heel" of the design is the set of attachments glued
onto the mirrors. By design, the mirrors are supposed to float as freely
in space as possible, so that they are free to follow the currents of
space induced by gravitational waves. That means the attachments between
the mirror and the rest of the world have been made as tenuous as possible.
There are two sets of attachments: a suspension fiber and guide rod assembly;
and the magnet/standoff assembly. These are shown in the photo on the right.
The 25-pound mirror is balanced on a single loop of thin (0.01 inch) steel wire to an accuracy of about 1/100th of a degree. This is done with the guide rod that has a small groove to locate the wire. Balancing requires getting the guide rod in the right place and the wire properly seated in the groove to an accuracy thousands of times smaller than the wire diameter. Not surprisingly, a good shaking changes the balance and this needs to be trimmed up by hand. This alignment trimming is not a difficult procedure, but it does require a person to enter the vacuum chamber to tweak it up.
Much more worrisome is the magnet/standoff assembly (visible below the guide rod in the photo at right). The standoff is used to isolate the magnetic material--with its high internal friction--from the pristine fused silica of the mirror. This keeps the thermal vibrations of the atoms in the magnet from driving large (meaning more than 1/1000th of a proton diameter) wave-like distortions of the mirror's surface. Unfortunately, this assembly has a long lever arm and a small footprint by design, features that make the structure fragile if knocked. Topping that off, clearances between the magnet and the surrounding coils and shadow sensors are restricted to about 1-2 millimeters. In our post-quake check up, we found that these attachments had come unglued on about a dozen mirrors. Repairing the damage requires removing the mirror assemblies from vacuum chambers, removing the attachments and recleaning the mirrors, and then gluing new attachments onto the mirrors, following all the requisite vacuum baking and balancing steps.
How do we do better with our earthquake safety procedures in future? A clue is that the damaged attachments were part of the earliest optics installed. Typically the gaps between the installed optics and the stops were set to approximately one millimeter. We will now set the gaps closer to one-half millimeter and tighten the tolerance on that spacing. Hopefully this will improve survivability in subsequent quakes.
We are taking the occasion of opening the chambers to make detector improvements as well as repairs, by installing new shadow sensors throughout WA2K. These will be less sensitive to disruption by scattered light and the replacement will make it possible to inject more laser light into the interferometer to increase our sensitivity.