Proceedings of Chamonix 2009 workshop on LHC Performance

BUSBAR AND JOINTS STABILITY AND PROTECTION

A.P. Verweij, CERN, Geneva, Switzerland

Abstract and two U-shaped G10 pieces with a length of 190 mm

A defective bus connection between two dipole and a thickness of 1 mm. More details on the joint can be
magnets was the primary cause of the incident in sector 3- found in [3].
4 on September 19th. I will show how this could have
happened, i.e. how a highly resistive joint has caused a
thermal runaway and burned (or opened) before the QPS Copper Wedge

threshold was reached. Furthermore I will present the new

detection limits for the QPS upgrade of the RB circuits, SnAg soldering layer
required to avoid similar thermal runaways in the future.

INTRODUCTION

Triggered by the incident in sector 3-4, the stability of

the RB bus including the soldered joints has been re-
assessed. In the first section the geometry of the bus will
be described, with special attention to the joint. Then the
U profile
protection of the bus will be briefly explained, and three
types of thermal runaway will be differentiated, namely
localised slow, non-localised slow, and fast thermal
Figure 1: Layout of a RB bus and joint.
runaways. Differences between these three types will be
illustrated by means of a simplified bus in adiabatic
In a good joint the resistance between the two SC
conditions. In the next section the ‘old’ QPS threshold is
cables should be less than 0.6 nΩ, so that, even at ultimate
presented, and it is shown in which situations the
current of 13 kA, the heating is less than 0.1 W.
protection would be effective or non-effective.
Furthermore, the joint stabilizer and the bus stabilizer (on
In the following section three scenarios of faulty joints
either side of the joint) should work as a continuous
are introduced that could all have lead to the incident. The
electrical shunt to the cables. This is achieved when the
electro-dynamic and thermal behaviour of the bus during
solder fills all the voids in and around the joint (see
the current ramp leading to the incident are then simulated
Figure 2), as well as the thin slots between the bus
for the most plausible scenario. Based on further
stabilizer and the joint stabilizer. Finally, a good joint
calculations the ‘new’ threshold for the QPS upgrade is
should be mechanically strong enough so that it does not
introduced, and the improvement with respect to the ‘old’
open due to the mechanical and electromagnetic forces
protection is discussed. Finally, it is stressed that the
acting on it.
‘new’ QPS threshold cannot protect the bus under all
conditions, due to reasons intrinsic to the design of the
bus.

BUS AND JOINT LAYOUT

A schematic view of the RB bus is shown in Figure 1.

The bus-bar has dimensions of 20 mm x 16 mm, and
contains a single Nb-Ti Rutherford type 02 cable. The
cross-section of the copper ‘bus stabilizer’ (ACu) is about
Figure 2: Cross-section of a properly solder-filled RB
282 mm2 with RRR>120 according to the specification.
joint.
Actual measurements show a RRR of about 210-220 [1]
and 230-300 [2].
The joint between the two superconducting (SC) cables
is 120 mm long, and soldered while being compressed
between a 120 mm long copper wedge and a 155 mm
BUS PROTECTION
long copper U-profile. In the following the wedge and U-
profile together are referred to as ‘joint stabilizer’. The During operation several disturbances can release heat
joint is insulated by means of two U-shaped kapton pieces in the bus. Disturbances are either transient, such as
with a length of 240 mm and a thickness of 0.125 mm, mechanical movement, or beam losses, possibly leading

113
 Proceedings of Chamonix 2009 workshop on LHC Performance

to a local quench, or non-transient, such as resistive [4]. However, a qualitative feeling of the three types of
heating due to a joint resistance or due to a locally non- runaway can be obtained by assuming a certain length of
SC cable (see also Figure 3). bus in the normal state with uniform temperature and no
cooling to the helium. The results of such a simplified
Disturbances approach are given in the following three sub-sections.
Non-transient Transient Transient

Resistive joint, Sudden increase Cable

non-SC cable in R_joint quench Non-localised slow thermal runaway
The typical characteristic time tTR of the thermal
Stable resistive Recovery runaway can be defined as the time it takes to warm up
heating Cooling>Heating the bus from the current sharing temperature (equal to
Cooling>Heating
about 9 K) to 500 K at constant current. τTR depends
basically on the amount of stabilising copper (which is in
good contact with the superconductor), the cooling
Localised slow Non-localised slow Fast thermal
conditions, and the operating current. Figure 4 shows tTR
thermal runaway thermal runaway runaway vs. I for a properly soldered RB bus with ACu=282 mm2,
Good thermal and
electrical contacts.
Good thermal and
electrical contacts.
Bad thermal and
electrical contacts assuming RRR=200, and adiabatic conditions. τTR is
No propagation to bus. Propagation to bus.
typically 90 s at 13 kA and 105 s at 11.85 kA.
Old QPS acts/protects here

1000
Figure 3: Schematic view of bus/joint protection and
different types of thermal runaway.
100
t_TR [s] τRB/2

Below a certain current, the resistive losses (that are the 10

consequence of the initial disturbance) will be equal or
smaller than the cooling, leading to a stable state, or
possibly to recovery of the SC state. However, above a 1 A_Cu=282 mm^2
A_Cu,eff=180 mm^2
certain current, the resistive losses exceed the cooling, A_Cu,eff=100 mm^2

causing a temperature rise of the bus, hence increasing the 0.1

A_Cu,cable=13 mm^2

resistivity, which in turn increases the resistive losses and 0 2000 4000 6000 8000 10000 12000

so on. A so-called thermal runaway occurs that can only I [A]

be stopped by reducing the current in the circuit. This is Figure 4: Typical characteristic time of a thermal runaway
accomplished by putting a dump resistance in series as for a standard bus (ACu=282 mm2), a single SC cable
soon as the QPS system detects a voltage over the bus (ACu=13 mm2), and two cases with reduced Cu stabilizer
(Vbus) larger than a certain threshold Vthr. The external (ACu=100 and 180 mm2).
dump causes an almost exponential current decay with a
time constant τRB=104 s. A properly functioning
protection should limit the maximum bus temperature A non-localised slow thermal runaway can be defined
Tmax to below 500 K, i.e. the temperature at which the as a runaway without overheating of the bus, which
kapton insulation and solder inside the bus/joint could means (in first approximation):
start to melt.
As a first approximation the voltage over the bus can be
(tTR − tthr ) > τ RB (2)
2
written as: with tthr the duration between the start of the thermal
⎛ ρ (T ( z )) ⎞ runaway and the detection of the quench (i.e. Vbus>Vthr)
Vbus = I ⎜⎜ R jo int + ∫ dz ⎟⎟ (1)
As an example, consider a normal zone of 2 m length
⎝ ACu ⎠ with I=10 kA, RRR=200, Vthr=200 mV, and adiabatic
with Rjoint the effective resistance of one or more joints conditions, so tTR=150 s. Figure 5 shows the start of the
and ρ(T(z) the temperature dependent resistivity of the thermal runaway (at t≈0 s) resulting in a slowly increasing
copper stabilizer. The integral has to be taken over the voltage that reaches the threshold after about 70 s, when
length of the normal zone. the bus has a temperature Tthr=95 K. The small resistive
heating, combined with the high enthalpy of the bus,
As shown in Figure 3, three types of thermal runaway limits the maximum temperature (Tmax) to about 260 K at
can be distinguished, namely ‘slow non-localised’, ‘slow the end of the exponential current decay. Condition (2)
localised’, and ’fast localised’. The entire electro-dynamic confirms that no overheating occurs because (150-
and thermal process of a thermal runaway is very 70)>104/2.
complex due to the strong dependency of various
parameters on temperature, field, and current, and is
therefore solved numerically with the computer code QP3
114
 Proceedings of Chamonix 2009 workshop on LHC Performance

stops growing (for example because it enters a part of the

Slow (non-localised)
500 thermal runaway 10000 bus with better cooling) the thermal runaway remains
localised. In this case, the local resistivity of the normal
Peak temp. [K], Voltage [mV]

400
Voltage
8000 zone, and therefore its temperature, will be much higher
by the time the threshold is reached (see eq. 1). An

Current [A]
300
Temperature
6000
example is shown in Figure 7, with the same conditions as
before (see Figure 5), except that the normal zone is now
Vthr
200 4000
only 0.4 m long in stead of 2 m.
Current

100 Tthr 2000

1100
Slow (localised)
1000 thermal runaway 10000
0 0
900
0 tthr 80 160 240 320 400

Peak temp. [K], Voltage [mV]

Time [s] 800 8000
Temperature
Figure 5: Non-localised slow thermal runaway, with 700

Current [A]
limited maximum temperature because the threshold is 600 6000

reached in an early stage of the runaway. 500

400 4000
Current
300
Tthr
200 2000
Localised slow thermal runaway 100
Vthr Voltage

0 0
Figure 4 shows that for a properly soldered bus (so 0 80 tthr 160 240 320 400

ACu=282 mm2) condition (2) is met for any current

Time [s]

assuming that tthr is small. In order to reach the threshold Figure 7: Localised slow thermal runaway, with
as fast as possible it is important to have a fast growing Tmax>500 K because the threshold is reached in a late
(or propagating) normal zone. The propagation speed of a stage of the runaway.
normal zone depends on the current, the bus
characteristics, and the cooling conditions, as shown in The figure shows that the temperature of the bus is
Figure 6. The current I* denotes the current below which already 250 K by the time the quench is detected (tthr=120
the propagation speed becomes negative, in other words, s) and the current discharge starts. The final temperature
the normal zone recovers. The typical I* for the RB bus- is about 740 K which is well above 500 K, in line with
bar inside a HeII bath is 16 kA, which means that those condition (2) since (150-120)<104/2.
parts of the bus are cryostable under all LHC operating The only solution to avoid overheating is to set the
conditions. However, due to the thicker insulation, the threshold at a lower voltage. For this specific case tthr
typical I* for the RB joint is 8 kA. Other parts of the bus, should be smaller than (150-104/2)=98 s, which would
such as the lyre, and the bus-bar inside the key of the cold have required a threshold smaller than about 120 mV. At
mass will again have different I*. higher currents, tTR is smaller and therefore also tthr and
hence Vthr have to be smaller.
2.5 RRR=240, 0.3 mm It is important to note that shorter localised normal
RRR=240, 1.7 mm
RRR=240, adiab
RB bus, T=1.9 K zones will require even smaller Vthr. Potentially dangerous
2
RRR=120, 0.3 mm
RRR=120, 1.7 mm
areas in the circuit are therefore badly cooled short parts
RRR=120, adiab of the bus such as the joints and plugs. Very short
RRR=70, 0.3 mm, CEA exp. 1.9 K
Propagation speed [m/s]

1.5
RRR=70, 0.3 mm, CEA simulation 1.9 K insulated parts are however safe again, due to the
RRR=70, 0.3 mm, CEA exp. 4.2 K
RRR=70, 0.3 mm, CEA simulation 4.2 K longitudinal cooling through the stabilizer.
1

Localised fast thermal runaway

0.5

I*bus
Rewriting eq. (2) shows that thermal runaway to above
0
500 K is unavoidable for tTR<0.5τRB+tthr. This can occur
0 4000

I*joint
8000 12000 16000 20000 24000
when the stabilising copper is in bad thermal contact with
Current [A]
the cable, and at the same time has a high longitudinal
resistivity (or even electrical discontinuity). This means
Figure 6: Normal zone propagation speed vs. current for that the effective amount of stabilizer cross-section ACu,eff
various RRR and insulation thicknesses (calculations is reduced, hence increasing the resistive losses, and
using QP3). reducing the thermal enthalpy. Two examples are shown
in Figure 4 for ACu,eff=100 and 180 mm2, demonstrating
As long as a normal zone is propagating, the thermal clearly the safe current becomes smaller for reduced
runaway is often non-localised, and the bus voltage will ACu,eff. An example is shown in Figure 8, with the same
usually reach the threshold fast enough to limit Tthr and conditions as before (see Figure 7), except that
Tmax. However, when an initially propagating normal zone ACu,eff=100 mm2 in stead of 282 mm2, and Vthr=10 mV.
115
 Proceedings of Chamonix 2009 workshop on LHC Performance

Note that the maximum bus temperature will exceed interrupted coinciding with a bad thermal contact between
500 K during the discharge even if the detection voltage cables and stabilizer.
(and hence tthr) are very small. No protection scheme is Summarised, the old QPS was mainly designed to
possible as soon as a fast thermal runaway starts. protect the bus in case of a non-localised thermal runaway
(see also Figure 3), a situation that can only occur in large
parts of the bus with non-zero propagation speed, i.e. with
poor heat transfer to the helium (see Figure 6).
Localised fast thermal
1000 runaway 10000 Calculations with QP3 show that only the 15 m long part
of the bus located in the key of the main dipoles was
Peak temp. [K], Voltage [mV]

800
Current
8000 effectively protected, and probably only for currents
Voltage
below about 10-12 kA.

Current [A]
600 6000
Temperature

400 4000

SIMULATION OF THE 19/9/08 INCIDENT

200 Tthr 2000

0 Vthr 0
Analysis of the incident in sector 3-4 occurring on
0
tthr
5 10 15 20 25 30 35 40 19/9/2008 resulted in the following findings:
• The QPS triggered at the maximum current of
Time [s]

8715 A.
Figure 8: Localised fast thermal runaway, with
• The bus threshold of 1 V was reached before a
Tmax>500 K because local resistivity is too high and local
voltage increase was observed in any of the dipoles.
enthalpy too low.
• The voltage increase on the bus was extremely fast:
from about 10 mV to 1 V in about 1 s.
As already mentioned before, Figure 4 and equation (2) • The resistive voltage on the bus increased very likely
are based on a simplified model without taking into with about 10 mV during the last minute before the
account heat transfer along the cable and towards the incident.
helium. Including both types of heat transfer will make
the thermal runaway slower (and hence Tmax smaller), Post analysis of calorimetric data, performed during a
especially if the length of bus with reduced ACu,eff is current plateau at 7 kA, revealed an additional local
small. Note also that in the worst case, all current passes power of 10.7±2.1 W at 7 kA [8], corresponding to an
through the cable (ACu,cable=13 mm2), and the bus will additional resistance of 220±40 nΩ.
behave like a fuse with tTR<1 s at high currents, see These findings clearly indicate that a phase of initially
Figure 4. stable resistive heating moved into a fast thermal
runaway, eventually leading to local melting of the bus
and arcing, see also Figure 9. In order to have a very fast
THE ‘OLD’ QPS THRESHOLD thermal runaway, the heating should occur in the
superconducting cable, and the bus should be electrically
interrupted and in bad thermal contact with the cable.
The ‘old’ QPS threshold on the bus was set at 1 V as
recommended in 2000 [5], [6]. Additional calculations in Disturbances
2006 confirmed that this threshold was sufficiently low Non-transient Transient Transient

even in the case of local resistive heating due to faulty Resistive joint, Sudden increase
in R_joint
Cable
quench
non-SC cable
splices [7]. The latter however assumed that the bus
around the faulty splice is continuous, implying perfect
electrical contact between the joint stabilizer and the bus
Stable resistive Recovery
stabilizer. New calculations with QP3 clearly show that heating (cooling>
the threshold of 1 V was far too high because possible (cooling>heating) heating)

coincidence of different longitudinal variations in the bus

was not properly taken into account in the past. As shown
before, these longitudinal variations could lead to
Localised slow Non-localised slow Fast thermal
localised slow or fast thermal runaways resulting in thermal runaway thermal runaway runaway
temperatures above 500 K. Localised slow runaways Good thermal and Good thermal and
Bad thermal and
electrical contacts. electrical contacts.
would result from variations in the propagation speed, No propagation to bus. Propagation to bus.
electrical
contacts

eventually becoming very low or even 0, which could The incident

then result in a very localised normal zone reaching
already a high temperature by the time the threshold was Figure 9: Schematic view of sequence of phases leading
reached. Localised fast runaways could occur in parts of to the incident in sector 3-4.
the bus were the copper stabilizer was (partially)
116
 Proceedings of Chamonix 2009 workshop on LHC Performance

Three possible and realistic scenarios fulfilling these and probably also between the SC cable and the bus
criteria can be distinguished (see Figure 10), all of them stabiliser.
requiring a lack of solder between the joint stabilizer and The exact defect causing the incident will never be
the bus stabilizer (i.e. electrical interruption): known, since the joint has evaporated. In the following,
1. A non-soldered joint causing a highly resistive joint, the electromagnetic and thermal behaviour of a ‘Scenario
and at the same time a bad thermal (and electrical) 1’ joint has been calculated using the code QP3. Note
contact between cable joint and joint stabilizer. This however that the other scenario’s would give very similar
scenario could be caused by the absence of the heat results. The following model parameters are taken:
treatment during the joint manufacture, or by a heat • The joint resistance is 220 nΩ which is uniformly
treatment at a temperature below the melting point of distributed over its length.
the solder. • The cable joint has no electrical and thermal contact to
2. A non-superconducting cable and a lack of solder the joint stabilizer.
inside the bus and/or inside the joint (over a few cm). • The Nb-Ti cable in the bus is in perfect electrical and
Extremely large IC-degradation is required so that the thermal contact to the copper of the bus.
cable is non-superconducting at 7 kA, i.e. at a very • The RRR of the copper in the superconducting cable is
low self-field <0.5 T. Following [9] this is possible if 200 and the RRR of the copper stabilizer is 240.
the cable has been subject to high temperatures, • The insulation of the bus is 0.3 mm thick non-porous
typically 550 °C for at least several minutes or 600 °C kapton with negligible heat capacity.
or more for several tens of seconds. Although such a
temperature excursion cannot be excluded, this In order to have a good agreement with the findings
scenario is quite unlikely because it requires the (given at the beginning of this chapter) the transverse
coincidence of two non-conformities (non-uniform cooling (in Watt/m) of the joint is assumed as 6.5*(T-
soldering and overheating). THe), and the heat flow from the insulation to the helium
3. A cable that is heavily damaged at the interface (in Watt/m) is assumed as 3.4(Tins4-THe4)
between joint and bus and a lack of solder inside the
bus and/or inside the joint (over a few cm). A similar The result of the calculation is presented in Figure 11,
scenario as nr 2, but in this case the cause of the cable showing three distinct parts:
degradation is mechanical. In order to see a resistance • 0-7.6 kA: the joint temperature is below the critical
at 7 kA most of the strands should be severely temperature of the SC, so that V=RjointI and P=RjointI2.
damaged. At 7 kA the heating is 11 W and the voltage is 1.5 mV.
• 7.7 kA-8.7 kA: at 7.7 kA the temperature of the cable
1.
wedge inside the joint reaches the critical temperature and
almost instantaneously about 15 cm of cables
bus U-profile bus (corresponding to the length with thick insulation)
becomes normal. The normal zone is expanding
2.
wedge slowly for higher currents, reaching a length of about
30 cm at 8.7 kA, with P=70 W. During the slow
bus U-profile bus expansion of the normal zone the voltage is increasing
3.
from several mV to 10 mV, and the maximum
wedge temperature in the joint from 10 K to about 30 K.
bus U-profile bus

1000
Figure 10: Three possible scenarios that could have been Fast localised thermal runaway at T_max=30 K, P=70 W, V=10 mV, z_norm=0.3 m

the origin of the fast thermal runaway during the incident. Length normal zone [m]
100
The light gray areas denote bad thermal and electrical Power [W]
T_max [K]
P=11 W@7 kA

contact. Green cables are superconducting; red cables are Voltage [mV]
10
normal. The black spot denotes a mechanically damaged
cable.
1

Start normal zone

V=0.3 mV

Several non-soldered (dry) joints inside cold masses 0.1

were found in recent months (having joint resistance

0.01
values in the order of 10-100 nΩ) but it should be noted 0 1000 2000 3000 4000 5000 6000 7000 8000 9000

that the inter-dipole joints in the machine have been made Current [A]

using different procedures, equipment, and people. There

is also evidence that solder is frequently lacking in the Figure 11: Simulation of temperature, voltage, power and
gaps between the bus stabilizer and the joint stabilizer, length of the normal zone for the ramp with 10 A/s
leading to the incident in sector 3-4.
117
 Proceedings of Chamonix 2009 workshop on LHC Performance

• Fast thermal runaway, with an increase in temperature

12000
from 30 K to 500 K in a few seconds, see also Figure 25 nOhm
50 Watt
12 showing a zoom of the last 20 s before the incident. 10000

Note that the large noise in the bus voltage makes it

impossible to re-establish the exact voltage shape. 8000

I_run [A]
10 Watt

6000
0.3 mV
0.5 mV

1000 700 4000 1 mV

V_meas (Timber)
V_meas (QPS data)
RRR=240, weak thermal contact (Incident)
800 V_sim [mV] 600 2000
RRR=120, no thermal contact (worst case)
T_sim [K]
Worst case -500 A
600 500 0
10 100 1000
R_joint [nOhm]
Voltage [mV]

400 400

Temp [K]
Figure 13: Thermal runaway current versus joint
200 300
resistance for ‘incident-type’ joint and worst case
0 200
scenario.

-200 100
With the reduced threshold the ‘new’ QPS will protect
-400 0
the circuit for both localised and non-localised slow
-20 -15 -10 -5 0
thermal runaways. At the same time it will trigger a
Time [s]
discharge in case it detects the phase of stable resistive
heating that could eventually cause a slow or fast thermal
Figure 12: Measured and simulated voltage during the last
runaway (see Figure 14).
20 s before the incident. The simulated temperature is
shown as well. Disturbances

Non-transient Transient Transient

Resistive joint, Sudden increase Cable

non-SC cable in R_joint quench
REQUIRED THRESHOLD FOR THE QPS
UPGRADE
Stable resistive Recovery
heating
Many simulations of thermal runaways are performed (cooling>

un
(cooling>heating) heating)
varying the RRR and cooling conditions, in order to find

pro
tec
the ‘worst case scenario’, i.e. the case that gives the New QPS

ted
acts here
lowest thermal runaway current Irun for a given joint
resistance. Figure 13 shows this worst case scenario along Localised slow Non-localised slow Fast thermal
with the scenario leading to the incident, but now thermal runaway thermal runaway runaway
Good thermal and Good thermal and Bad thermal and
calculated for various joint resistances. An additional electrical contacts. electrical contacts. electrical contacts
No propagation to bus. Propagation to bus.
current equal to τRB/2*(dI/dt)=500 A is subtracted from
the worst case in order to assure a safe exponential
discharge of the RB circuit before the fast thermal Figure 14: Schematic view of radius of action of the
runaway starts. The figure shows as well the current at QPS upgrade with the ‘new’ threshold.
which the voltage over the bus reaches 0.3 mV, 0.5 mV
and 1 mV, assuming that the bus and joint are still below
Tcs so that there is no normal zone. Combining the worst
case scenario with the bus voltage it becomes clear that a It is very important to note that the new QPS threshold
threshold of 0.3 mV is needed to assure that the QPS will not protect the circuit in the following two cases:
triggers the discharge before the bus enters into a fast 1. Fast thermal run-away resulting from a sudden
thermal runaway. transient disturbance (without intermediate stable
Note that the fast thermal runaway leading to the sector heating). To avoid such fast thermal runaways one
3-4 incident would have been avoided having needs to assure a good thermal/electrical contact
Vthr=0.3 mV. If the disturbance was non-transient (i.e. the between the SC cables and the joint stabilizer or
resistance of 220 nΩ was already present at low currents), between the bus stabilizer and joint stabilizer. Both are
the new QPS would have triggered at about 2 kA. If the achieved by having a perfect solder filling of the
disturbance was transient (i.e. the resistance of 220 nΩ joints.
occurred suddenly during the ramp), it would have 2. Sudden mechanical opening of the joint. The best way
happened before 7 kA (because additional heating was to avoid this, is by clamping the joint.
observed at 7 kA), so the new QPS would have triggered
somewhere during the ramp between 2 and 7 kA.
118
 Proceedings of Chamonix 2009 workshop on LHC Performance

For a 100% fail-safe protection all joints should be

Also note that: clamped and the SC cable should be everywhere in good
• All joints will probably see (sooner or later) a transient thermal and electrical contact with the stabilizer. In those
disturbance due to the SC-to-normal transition of the areas where a good contact cannot be guaranteed one has
joint due to a quench in an adjacent magnet. to assure the electrical continuity of the stabilizer.
• Having a relatively low electrical joint resistance does In other words: any part of the bus for which a good
not automatically mean that the joint is perfectly contact cannot be guaranteed and for which (at the same
soldered. place) also the electrical continuity cannot be guaranteed
• Thermal cycling and powering can deteriorate the is susceptible to burn through, resulting in an arc causing
thermal contact between SC cables and joint stabilizer, serious additional damage.
and/ the electrical contact between bus stabilizer and
joint stabilizer.

CONCLUSIONS REFERENCES

Calculations clearly indicate that the origin of the [1] L. Belova et al., “Design and manufacture of the
incident in sector 3-4 is linked to the occurrence of: superconducting bus-bars for the LHC main
- a highly resistive joint between two SC cables being in magnets”, IEEE Trans.Appl. SC, Vol 12, pp. 1305-
bad contact with the copper stabilizer, 1309 (2002).
or: [2] D. Richter, “Unpublished measurement data” (2009).
- a non-superconducting cable, thermally insulated from [3] “Main bus bars interconnections between
the stabiliser coinciding with a longitudinal electrical cryomagnets of the LHC arcs”, LHC Engineering
interruption of the stabiliser (at the end of the joint), Specification, EDMS 108101 (2005).
or: [4] QP3, Computer code for the calculation of Quench
- a heavily damaged superconducting cable, thermally Process, Propagation, and Protection by A.P.
insulated from the stabiliser coinciding with a Verweij (2009).
longitudinal electrical interruption of the stabiliser (at [5] R. Schmidt, F. Sonnemann, “Modelling of the
the end of the joint), quench process for the optimisation of the design and
protection of superconducting busbars for the LHC”,
The current could therefore not (or only for a small Proceedings ICEC-18, also LHC Project Report 389
fraction) bypass the resistive joint or non-superconducting (2000).
cable and due to the small local effective enthalpy, a fuse- [6] F. Sonnemann, “Resistive transition and protection
type thermal runaway occurred, very quickly resulting in of LHC superconducting cables and magnets”, PhD
a burn-through of the bus. thesis, p. 104 (2001).
Reducing the QPS threshold to 0.3 mV will limit [7] M. Calvi, L. Bottura, and F. Rodriguez Mateos,
thermal run-aways caused by resistive joint heating to “Proposed method for the verification of the LHC
well below 500 K. The same threshold can be used for the bus bar splices during commissioning at cryogenic
RQ circuits if powered to 9.5 kA maximum. For higher conditions”, Proceedings ASC-2006, also LHC
currents the threshold has to be slightly reduced due to the Project Report 1004 (2007).
smaller amount of copper stabiliser in the bus. [8] L. Tavian, unpublished data analysis.
The new QPS threshold will however not protect the [9] Th. Schneider, and P. Turowski, “Critical Current
13 kA circuits in case of a ‘fast’ thermal run-away Degradation of a NbTi-Multifilament Conductor due
resulting from a sudden transient disturbance (without to Heat Treatment”, IEEE Trans. Magn., Vol. 30, pp.
intermediate stable heating), and in case of a sudden 2391-2394 (1994).
mechanical opening of the joint. Here ‘fast’ means a
thermal runaway with a characteristic time smaller than
half the discharge time constant of the circuit.

119