End-cap magnet incident

11 December 2007

Left: the intact cryoline on Side A. Right: the crushed cryoline on Side C. The end cap magnet is visible to the right of the cryoline.


On 22 November, during testing of the end cap magnet on Side C, the magnet came in contact with the liquid argon calorimeter (see last week’s report). The experts involved tell the story.

 

Marzio Nessi - ATLAS technical coordinator

During the last three weeks, we have been pursuing current ramping tests on the End-Caps Toroid Magnets, doing them one at a time. After the first initial safety system checks (integrity of the vacuum, power and cryogenics systems), we started ramping up the magnet current. On the 22nd of November, the goal of the test was to ramp up to 75% of nomimal current on ECT-C, then provoke both slow and fast magnet dumps to check the response of the cryogenics and the mechanical behavior of the toroids. This test is not done in a standard ATLAS configuration but in a configuration that brings the end-cap magnet 35 cm closer to the calorimeter end-cap than normal position. Care was taken to ensure there was enough safety margin for this operation. Unexpectedly, at 14.7 kAmp, the EC-Toroid moved toward the calorimeter by about 8 cm. Immediately, all activities were stopped, and the current was released. All means were put in place to visually inspect the Liquid Argon Calorimeter End-Cap services for possible damage. The ATLAS safety management team brought in all experts and the fire brigade with safety equipment just in case liquid argon would spill out.

After assessing for possible damage at 1:00 AM, we put together a crisis unit to discuss with all project leaders involved the best course of action in order to avoid further damage or risk. The decision was taken to empty the liquid argon from the End-Cap calorimeter and to re-position the toroid on air pads. This was done overnight by a crew of about ten people.

In the following days, the toroid was moved back by about one meter and we inspected again for possible damage. No damage was found to the electronic connections or other services, except for the liquid argon exhaust line, the so-called cryoline. This line consist s of a warm (outer) and a cold section (the inner part). Only the warm section was damaged. This week, we are starting repair. This will cause minimal perturbation to the normal schedule for a few days.

After this first incident, End-Cap A was tested at half the current. We added protection to avoid any possible movement. We decided to stop the test not to take additional risk until the reasons for the movement of EC-C are fully understood. By now, the tests of the End-Cap toroids are stopped. They will resume when the entire ATLAS detector is in its final closed configuration around April 2008. At least this test has proven that the EC toroids were working correctly and their infrastructure behaved as expected.

Despite what happened, this incident confirmed that we are very well organized in case of safety problems. We were able to put together an effective team of experts at 2:00, with a phone conference bringing together people from all over the planet to discuss the best course of action. What amazed me was that everybody remained calm, organized and coordinated under the supervision of our Glimos, Olga Beltramello. Nobody argued, nobody got excited but instead everybody acted in an efficient and pro-active manner.

 

Johan Bremer - the view from the cryostats and cryogenics team:

During the magnet tests, one person was stationed in the cavern to watch in case of problems. When he heard some abnormal metallic noise, he called us immediately in the control room. We heard hissing gas. We connected a vacuum pump to the outer line of the cryoline and realised we had a complete vacuum failure - even near the pump the vacuum was only 600 millibar. We immediately emptied the cryoline of its argon after. I also wanted to empty the calorimeter, because in these circumstances having 20 cubic metres of argon in the cavern could be dangerous. We took the decision to empty the calorimeter after proper consultation during a meeting at 02:00 that morning.

Later, in order to assess the damage to the cryoline we had to move the magnet out of contact. The normal way to move the magnet is to put it on airpads. But that moves the magnet up 4-5 mm, and risked causing more damage to the cryoline. So instead, we lowered some of the jacks on which the magnet was sitting, and that allowed us to tilt it away from the cryoline without doing any more damage. When there was a gap of around 1 centimetre between the magnet and cryoline, we put the magnet on air bags and moved it away 80 centimetres.

Then we started to cut the outer line of the cryoline open. The inner line was not badly damaged, which was almost unbelievable. Then we opened up one of the bellows on the outer line just to check it. We didn’t think it was affected, but actually we found it was really badly damaged and will need replacing. However, the bellow on the inner line was intact. The problem now is working out how to replace the outer bellow. New bellows come as closed cylinders, but we can’t install that without opening the inner line. Instead we will probably make the new bellows from a sheet, and wrap it around the inner line before welding it together.

 

Herman Ten Kate - the view from the magnets team:

From our calculations, we determined that the friction coefficient of the magnet against the supporting rails underneath was at least 0.2. The end cap toroid weighs 340 tonnes, which means it should not move at all if the magnetic force of attraction is below 68 tonnes. We were testing at 15 kA, and the magnetic force between the end cap toroid and calorimeter at that current is 20 tonnes. We are absolutely sure of that. So, although the magnet should have been stationary unless there was an attraction of over 68 tonnes, it in fact moved towards the iron at a force of 20 tonnes. Clearly the friction coefficient was much lower than we thought. But the absolute minimum friction coefficient of a metal-to-metal contact is 0.12, according to textbooks. Even at that coefficient, you should need 40 tonnes of attractive force to move the magnet. To this day, we are not sure why the magnet moved at half the absolute minimum friction coefficient. We will make a model setup using the same type of steel to try to understand what happened. But for our magnet tests the calorimeter was in a non-standard configuration. In the final configuration this won’t be a problem.

 

Luis Hervas – the view from the calorimeter team:

The incident will mean quite some work for the cryogenics team, but the calorimeter team won’t be much affected – our schedule is more or less intact. We’ve only lost a week or ten days, which, given te circumstances, isn’t such a big deal. Of course, the calorimeter was emptied of argon as a safety precaution after the incident, so it will need filling again. Until that happens we can’t use the calorimeter in cosmics data taking, but that’s not a major issue. The important thing is that the calorimeter doesn't need to be warmed up, which would have meant a twelve week delay to warm it up and cool it down again. The challenge now for the cryogenics team is to fix the outer pip in situ, without affecting the inner one. Inventive solutions are needed.

Colin Barras

 

Colin Barras

ATLAS eNews

THIS WEEK'S FEATURES

FEATURES ARCHIVE