fast rundown 21.12.1998

A fast rundown from 3000 A to 0 A was initiated by a faulty water-flow meter in the power supply of the magnet. The fast rundown takes about 20 min. However, the Labview controlsystem displayed zero current immediately after the current rundown started. This triggered a chain of reactions which caused a quench about 3-4 min later.

coil current

Labview displays 0 current immediately after the fast rundown started.

coil temperature 1

About 3-4 min after the fast rundown started the quench detector fired at a current of still about 2300 A. It triggered coil heaters and disconnected the power supply from the coils. This action should avoid that all energy is dumped in a single coil. Most of the energy is dumped in a quench resistor located in the power supply. Only coil B (TC1208) got normal conducting immediately.

coil temperature 2

The coil temperatures increased by a few K only. The coils were cooled down again until the liquid in the heat exchanger vessel connected to the coils was used up. Without liquid in the heat exchanger the coil temperatures increased a second time. After the He compressor started again (about 13:48), the liquid in the 150 l storage vessel was used to cool the system again. The second warm-up is most pronounced for the last coils along the He cooling pipe. The heat exchangers in between the coils were inefficient due to the lack of liquid in the heat exchanger vessel. The small change of coil temperatures displayed by the temperature sensors in the lower figure after the current went up to 3000A (12:30-13:30) is an artefact caused by strong forces on the coils slightly displacing some coils with respect to the sensors. This geometrical effect depends on the coil position (upwards, sideways or downwards).

JT temperature

The temperature increase at the JT valve behind the last coil is most pronounced since its heat capacitance is small.

pressure

Warming up the single phase He by a few degree only results in a strong expansion of the He. The density changes from that of a liquid to the density of a 'normal' gas. An increase of the pressure was the result and the He compressor was unable to pump all the He released within a few min back into the buffer. It stopped due to overpressure after about 1-2 min at about 13:38 and was restarted around 13:48.

liquid He level

The increase of the pressure in the 150 l storage container to about 2.2 b converted the liquid into single phase He. Single phase He has about the same density as the liquid but is a gas. The pressure dropped after the compressor started again and the liquid phase returned.

current lead temperature

curent lead flow

enlarged time scale:

coil current

Coil temperature 1

coil temperature 2

vacuum

The vacuum detoriated twice. The reason is most likely due to an increase of the temperature of the current leads at their coldest point (frozen gas evaporates). It is caused by the lack of flow of cold He gas through the leads (see below). The flow through the current leads is controlled by a PID controller which should keep an intermediate lead temperature at its set point. Labview reduced this set point after the fast rundown started by about 10K since the current reading showed zero current. This caused an overreaction of the PID controller.

current lead flow

After Labview got zero current, it reduced the set point for the intermediate current lead temperature. As a result, the flow through the leads first increased and then was reduced after the lead temperature went below the set point (76K, see below).

current lead temperature

Once the flow through the current leads goes to zero for more than 30 s an emergency cooling is activated, driving the lead temperature distinctly below the set point (76K).

pressure

Summary:

The fast rundown of the current was caused by a faulty water-flow meter in the power supply of the magnet. The quench appearing 3-4 min later was most likely caused by an overreaction of a PID controller which should stabilize the flow through the current leads. The sudden change of the current read by Labview (not related to the real change of the current) caused Labview to change the set point of the lead temperature suddenly by nearly 10K. This confused the PID controller.

Solution:

Never allow the PID contoller to close completely the valve controlling the cold gas flow through the current leads. Tests at zero current indicate that zero flow for only a few seconds result in an immediate warm-up of the coldest point of the leads. Evidence for this behaviour is the drastic detoriation of the vacuum at zero lead flow. Already a small flow avoids this problem.

Last update 25.12.98
by Wolfgang Koenig