Quench Detection and Protection

The millisecond loop that catches a spreading normal zone in the superconducting coils and dumps stored energy safely.

Why quench is special

In a superconducting magnet a quench is the transition of part of the winding to a resistive (normal) state. If the normal zone spreads faster than protection acts, the coil's own stored energy heats that spot and can damage it. For the breeder at 16.84 T peak field this is the highest-severity fast fault; for the burner the stressed 26.49 T plug makes it more acute still. Detection must happen in milliseconds and protection must be independent of the supervisory network.

Detection

The signature is a resistive voltage that cannot be explained by inductance. Co-wound voltage taps and bridge arrangements subtract the inductive component so the resistive residual stands out. The AI contribution is discriminating a true quench from noise, ramp transients, and sensor artifacts fast enough to avoid both false dumps and missed quenches.

def quench_detect(taps, L, dIdt, R_lead, I, thresh_V, persist_ms, dt_ms):
    V_res = taps.total() - (L * dIdt + R_lead * I)
    if V_res > thresh_V:
        quench_detect.count += dt_ms
    else:
        quench_detect.count = 0
    return quench_detect.count >= persist_ms   # require brief persistence
quench_detect.count = 0

Protection

• Fire the dump: open the circuit and steer stored energy into external dump resistors
• Spread the heat: heaters or coupling drive the whole coil normal so energy deposits uniformly, not at one spot
• Confirm safe: verify current decay follows the protected profile

The protection path is hard-wired and supervisor-independent so it acts even if the control plane is down - this is why quench protection anchors the safe-state guarantee. Slow precursors to quench (cooling-margin erosion) are tracked as RUL so many quenches become scheduled interventions instead.