What are the dangers of slow current removal from superconducting magnets?

[jreelawg]

If you took an insulated fully charged superconducting capacitor, and suddenly warmed it above the critical temperature, what would you observe happen? For example, if the superconductor discharged in space. What form would the energy released take?

 

[Bob S]

In superconducting systems, the energy is actually stored as inductance (½LI²), not capacitance (½CV²), but the discharge of a superconducting magnet "going normal" for whatever reason is a very serious problem.

In some superconducting magets, like the ones in MRI "machines", the magnets are charged up to full current, and they are switched to persistent mode, where they can run for months without discharging. If the liquid helium runs out, the magnet will warm up and go normal, the coil becomes resistive, and the magnet discharges into the coil with a L/R (L over R) time constant, so the coil absorbs the stored magnetic energy and gets warm (but it doesn't melt).

In some high-performance superconducting magnets like the ones used in superconducting particle accelerators, the stored magnetic energy density is very high, and when the magnet goes normal (which can be triggered by beam loss or by tiny resistances) the energy has to be removed very fast using active "quench protection" circuits to prevent the magnet from being destroyed. About 10 days after the CERN LHC (Large Hadron Collider) was first turned on in September 2008, a small resistance (perhaps 200 nano-ohms) in a superconducting cable junction carrying roughly 7000 amps created enough heat (I²R = 10 watts) to cause a section of superconducing cable to go normal and melt. The cable and roughly a kilometer of superconducting magnets were destroyed before the quench protection circuit was triggered. The LHC was off for over a year for repairs.

See http://cdsweb.cern.ch/record/1150661

So in both cases, the stored energy is converted to heat (I²R).

Bob S

 

[berkeman]

Nice post, Bob.

 

[mheslep]

... About 10 days after the CERN LHC (Large Hadron Collider) was first turned on in September 2008, a small resistance (perhaps 200 nano-ohms) in a superconducting cable junction carrying roughly 7000 amps created enough heat (I²R = 10 watts) to cause a section of superconducing cable to go normal and melt. The cable and roughly a kilometer of superconducting magnets were destroyed before the quench protection circuit was triggered. The LHC was off for over a year for repairs.

See http://cdsweb.cern.ch/record/1150661

So in both cases, the stored energy is converted to heat (I²R).

Bob S

Imagine the like happening to http://www.iter.org/mach/magnets" If it all went in a second that's ~4X the power of shuttle launch. Good stuff for a Bond film, if not a practical power source:
Hugo Drax: Jaws, Mr. Bond must be cold after his swim. Place him [next to the failing superconductor] where he can be assured of warmth.

 

[Mike_In_Plano]

As I recall reading, energy storage devices may have the superconductor wrapped about a copper cylinder or tube. When the coil quinches, it's resistance exceeds that of the copper, and the field dumps it's energy into the copper where it becomes heat. The thermal mass of the copper absorbs the energy.

 

[Bob S]

Superconducting wire is often made by drawing ingots of niobium rods in a matrix of copper rods down to strands of wire less than 0.5 mm diameter, with imbedded filaments of superconducting wire several microns in diameter. The ratio of copper to superconductor determines whether the thermal mass of the copper matrix can absorb the energy when the wire quenches. For high current cables, the wires are wound into rectangular or keystoned shapes for magnet coils.

In the case of LHC (the 26-km circumference superconducting accelerator ring at CERN) the superconducting magnet cable needed external quench protection if a section of cable went normal. Cable splices between magnets were sandwiched and brazed inside copper blocks to provide thermal mass to limit damage while the quench protection system rerouted the current (up to 12,000 amps). Some brazed cable joints (inside liquid helium cryostats) were faulty, and one failed at about 7,000 amps shortly after commissioning began, leading to about 1 km of magnets being damaged. The LHC was off for more than one year to repair the damaged magnets. Before LHC raises the current to the design value of 12,000 amps, all roughly 10,000 brazed cable joints will be inspected and /or rebrazed. This will require another year of downtime.

Bob S

 

[mheslep]

Superconducting wire is often made by drawing ingots of niobium rods in a matrix of copper rods down to strands of wire less than 0.5 mm diameter, with imbedded filaments of superconducting wire several microns in diameter. The ratio of copper to superconductor determines whether the thermal mass of the copper matrix can absorb the energy when the wire quenches. For high current cables, the wires are wound into rectangular or keystoned shapes for magnet coils.

In the case of LHC (the 26-km circumference superconducting accelerator ring at CERN) the superconducting magnet cable needed external quench protection if a section of cable went normal. Cable splices between magnets were sandwiched and brazed inside copper blocks to provide thermal mass to limit damage while the quench protection system rerouted the current (up to 12,000 amps). Some brazed cable joints (inside liquid helium cryostats) were faulty, and one failed at about 7,000 amps shortly after commissioning began, leading to about 1 km of magnets being damaged. The LHC was off for more than one year to repair the damaged magnets. Before LHC raises the current to the design value of 12,000 amps, all roughly 10,000 brazed cable joints will be inspected and /or rebrazed. This will require another year of downtime.

Bob S

That seems to be an untenable design, when a failure in one part of a system of 10,000 can not just disable, but damage or even destroy a sizable portion of the entire system.

 

[Bob S]

That seems to be an untenable design, when a failure in one part of a system of 10,000 can not just disable, but damage or even destroy a sizable portion of the entire system.

Hopefully, the LHC quench detection circuit has been improved. There are about 10,000 external 12,000-amp superconducting magnet interconnection splices, and 14,000 internal magnet 12,000-amp splices. Including the 600-amp splices, there are over 100,000. See Table 1 in

https://espace.cern.ch/acc-tec-sector/Chamonix/Chamx2010/papers/NCL_2_01.pdf

These are all inside the liquid helium cryostats.

One of the problems in detecting a resistive splice (say 10 nano-ohms) in the magnet circuit is that if the magnet current is ramping up (dI/dt > 0), there is a large V = L dI/dt voltage, which makes detecting the resistive IR drop difficult.

I should point out that the 6,280-meter circumference superconducting Tevatron at Fermilab has been running since 1983 with no major mishaps.

Bob S

 

[mheslep]

Hopefully, the LHC quench detection circuit has been improved. There are about 10,000 external 12,000-amp superconducting magnet interconnection splices, and 14,000 internal magnet 12,000-amp splices. Including the 600-amp splices, there are over 100,000. See Table 1 in

https://espace.cern.ch/acc-tec-sector/Chamonix/Chamx2010/papers/NCL_2_01.pdf

These are all inside the liquid helium cryostats.

One of the problems in detecting a resistive splice (say 10 nano-ohms) in the magnet circuit is that if the magnet current is ramping up (dI/dt > 0), there is a large V = L dI/dt voltage, which makes detecting the resistive IR drop difficult.

Why not simply open the circuit if the SC goes normal and suffer the high L di/dt voltage, which would result undoubtedly in some severe local arcing, but avoid explosive IR heating?

should point out that the 6,280-meter circumference superconducting Tevatron at Fermilab has been running since 1983 with no major mishaps.

Bob S

Both LHC and Fermi using old school, low temperature, super conductors I see?

 

[Bob S]

Why not simply open the circuit if the SC goes normal and suffer the high L di/dt voltage, which would result undoubtedly in some severe local arcing, but avoid explosive IR heating?

If the current is removed from a string of superconducting magnets too slowly when a small section goes normal (= quench), the local ∫I²R·dt heating is excessive and the magnet coil is destroyed. If the magnet current circuit is suddenly opened and there is arcing, there is a very high V = L dI/dt voltage in every magnet in the string, there is local turn-to-turn and coil-to-ground arcing in every magnet, and the magnet coil insulation is destroyed. There is only a small window of the magnet current L/R (i.e., I(t) = I₀·e^-Rt/L) decay rate, where dI/dt is neither too slow nor too high, and the magnets survive.

Read about the LHC magnet quench protection systems in Section 9.4 of

https://edms.cern.ch/file/445850/5/Vol_1_Chapter_9.pdf

Bob S

 

[mheslep]

If the current is removed from a string of superconducting magnets too slowly when a small section goes normal (= quench), the local ∫I²R·dt heating is excessive and the magnet coil is destroyed. If the magnet current circuit is suddenly opened and there is arcing, there is a very high V = L dI/dt voltage in every magnet in the string, there is local turn-to-turn and coil-to-ground arcing in every magnet, and the magnet coil insulation is destroyed. There is only a small window of the magnet current L/R (i.e., I(t) = I₀·e^-Rt/L) decay rate, where dI/dt is neither too slow nor too high, and the magnets survive.

Read about the LHC magnet quench protection systems in Section 9.4 of

https://edms.cern.ch/file/445850/5/Vol_1_Chapter_9.pdf

Bob S

Interesting. Thank you.