MODES OF THE COMPONENTS OF THE ESS LINAC
M. Eshraqi, R. De Prisco, R. Miyamoto, Y. I. Levinsen European
Spallation Source, Lund, Sweden.
May 27, 2015
1 Introduction
In the following we report on the study of the failure of a single
component, either RF cavity or quadrupole. For now it is assumed
that the field in the element drops to zero instantly and makes it
completely un-functional, even destructive to the beam, while the
other elements are unchanged. This assumption is not necessarily
very realistic since field of either a cavity or a quadrupole has a
finite decay time determined respectively by their Q-value or
inductance. Additionally, the machine protection system (MPS)
should stop the beam at the front-end with a short delay in the
order of few microseconds. Therefore, the losses shown in the
following represent unrealistically bad cases even ignoring the
decay of the field, and so only their patterns have meaning and
their absolute scale should not be taken literally. The purpose of
this study is to understand the worst case scenario in the absence
of a functional MPS.
These simulations cover the failures of the quadrupoles and
cavities of the MEBT, field in the DTL tanks, quadrupoles and
cavities in the superconducting linac. For the case of cavities in
the superconducting linac, a few representative cavities have been
considered, 3 cavities in the spoke section, 2 cavities in the
medium beta section and 2 cavities in the high beta section.
One should emphasize that no other errors are considered in these
studies and inclusion of dynamic and static errors will affect the
conclusions.
2 MEBT
2.1 Overview
Due to the lack of a periodicity, a pattern of losses caused by a
failure of an element in the MEBT could differ for each element.
Hence, we simulate losses caused by a failure of one element of the
MEBT for all the buncher cavities as well as all the quadrupoles.
Exception for the forth quadrupole surrounding the chopper since,
as of February 2015, the layout around the chopper and its dump is
likely to be changed with respect to the lattice used in this
study.
A detailed study of beam losses was conducted and presented at the
HB’14 Workshop [1]. The same lattice was used in the following
study. The lattice version or name of each section is
• MEBT: 2014.v1
• DTL: v86
• HEBT: raster27
1 × 105 macro particles in a Gaussian distribution at the entrance
of RFQ is transported through the RFQ (the version of 2013
baseline), and used as the input to the MEBT. Please note that
this
1
lattice is almost identical to one so-called “2014 Baseline”, the
snapshot at the end of 2014, but some details may be slightly
different. As presented in [1] and seen below, the MEBT scrapers
have a significant impact on the beam losses cause by issues on the
transverse planes. Hence, we always repeat a case of a simulation
with and without the scrapers.
Figure 1 shows the total particle losses in the linac caused by a
complete failure of each MEBT element. The schematic on the top
represents the MEBT lattice and the bars are located at the
location of the failed element. The complete failures of the
quadrupole magnets have immediate effects and a large fraction of
the beam is lost in the MEBT and the following DTL but almost no
loss occurred in the SC linac or HEBT. For the failure of a buncher
cavity however, even a complete failure does not create such a fast
effect. A large fraction of the beam reaches the target, but
failure of a buncher cavity causes the losses throughout the whole
linac.
DTL Tank 1
Q1 Q2 Q3 Q5 Q6 Q7 Q8 Q9 Q10Q11 0
20
40
60
80
100
With scrapers
Without scrapers
Figure 1: The total losses in the whole linac in number of
particles due to a complete failure of each MEBT lattice element,
comparing the cases with and without the three MEBT scrapers. The
bars are shown at the corresponding location of each element. The
schematic on the top represents the MEBT lattice, where the blue
boxes above (below) the line are the focusing (defocusing)
quadrupoles, green boxes are the buncher cavities, and the red
lines and triangles are the chopper and its dump. (Losses into the
three MEBT scrapers are not shown.)
2.2 Failure of a buncher cavity
In this section, we look at the losses caused by a failure of each
buncher cavity. Figure 2 shows loss densities in J/m in MEBT (left
column), DTL (middle column), and the rest of the linac (SC linac
and HEBT) for a complete failure of the first buncher (top row),
second (middle row), and the third (bottom). As discussed above,
the effect of a buncher failure is not immediate and only the
failure of the first buncher causes losses within the MEBT itself,
but all the three cases cause losses
2
in the DTL, beginning of the spoke section, high-β section, and
HEBT up to the dogleg. Since the problem is in the longitudinal
plane, the scrapers have almost no impact for these cases.
Buncher1
1
10
102
103
104
0 10 20 30 40 0.1
1
10
102
103
0 100 200 300 400 500 600 0.1
1
10
102
103
104
Buncher2
1
10
102
103
104
1
10
102
103
104
1
10
102
103
104
Buncher3
1
10
102
103
104
1
10
102
103
104
Position in DTL [m] 0 100 200 300 400 500 600
0.1
1
10
102
103
104
Position in SC linac [m]
Figure 2: Losses in MEBT (left column), DTL (middle column), and SC
linac and HEBT (right column) due to a complete failure of MEBT
Buncher1 (top row), Buncher2 (middle row), and Buncher3 (bottom
row). (Losses into the three MEBT scrapers are not shown.)
2.3 Failure of a quadrupole
In this section, we look at the losses caused by a failure of each
quadrupole. For all cases of quadrupole failures, no loss is
observed beyond DTL except for the ones in the very beginning of
the spoke section. We can also conclude that the scrapers prevent
losses in DTL. Exception for the losses caused by the quadrupoles
towards the end of the MEBT, for which the scrapers cannot act
effectively. This is good for DTL but such cases may have to be
taken into account for the design of the scrapers.
Figure 3 is structured in the same way as Figure 2, but now for the
first seven quadrupoles of the MEBT. Please note that the failure
of the second quadrupole could cause losses all over the first half
of the MEBT, and that the third quadrupole could cause losses into
the chopper dump. As for the scrapers discussed in the previous
paragraph, this might need to be considered for the design of the
chopper dump. What should be noted is that, even with the scrapers,
we could have losses in the third buncher cavity. The losses in the
DTL are larger for the middle triplet (Q4-Q6).
Figure 4 shows the cases for the last four quadrupoles of the MEBT.
Because the last and third scraper is located between the ninth and
tenth quadrupoles, the scrapers cannot act on failures of these
quadrupoles very well (or at all for the failure of the tenth and
eleventh).
2.4 MEBT Conclusions
The failures of the buncher cavities do not have immediate effects
and a large fraction of the beam actually reaches the target, but
cause losses throughout the linac, including the SC sections,
raising some concern. The failures of the quadrupoles have
immediate effects and the losses are either in MEBT or DTL. The
losses caused by the quadrupole failures could be improved by a lot
with the MEBT scrapers for some cases. Some cases of the quadrupole
failures may cause considerable
3
Quad1
1
10
102
103
104
105
106
0 10 20 30 40 0.1
1
10
102
103
104
105
0 100 200 300 400 500 600 0.1
1
10
102
103
104
105
106
Quad2
1
10
102
103
104
105
106
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Quad3
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Position in DTL [m] 0 100 200 300 400 500 600
0.1
1
10
102
103
104
105
106
Quad5
1
10
102
103
104
105
106
0 10 20 30 40 0.1
1
10
102
103
104
105
0 100 200 300 400 500 600 0.1
1
10
102
103
104
105
106
Quad6
1
10
102
103
104
105
106
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Quad7
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Position in DTL [m] 0 100 200 300 400 500 600
0.1
1
10
102
103
104
105
106
Position in SC linac [m]
Figure 3: Losses in MEBT (left column), DTL (middle column), and SC
linac and HEBT (right column) due to a complete failure of a single
quadrupole in the MEBT. From top to bottom are Quadrupole 1 to
Quadrupole 7. Losses into the three MEBT scrapers are not
shown.
4
Quad8
1
10
102
103
104
105
106
0 10 20 30 40 0.1
1
10
102
103
104
105
0 100 200 300 400 500 600 0.1
1
10
102
103
104
105
106
Quad9
1
10
102
103
104
105
106
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Quad10
1
10
102
103
104
105
106
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Quad11
1
10
102
103
104
105
106
1
10
102
103
104
105
106
Position in DTL [m] 0 100 200 300 400 500 600
0.1
1
10
102
103
104
105
106
Position in SC linac [m]
Figure 4: Losses in MEBT (left column), DTL (middle column), and SC
linac and HEBT (right column) due to a complete failure of MEBT
Quad8 (first row), Quad9 (second row), Quad10 (third row), and
Quad11 (fourth row). (Losses into the three MEBT scrapers are not
shown.)
5
losses into the scrapers and the chopper dump so further studies,
including the thermomechanical calculations of these devices
themselves, may be needed. Finally, a more detailed study taking
into account the decay of the field should be conducted.
3 Drift Tube Linac
The drift tube linac of the ESS is composed of five independently
powered tanks. Each of these tanks is powered by a 2.8 MW klystrons
that are fed by independent modulators. The failure of each
modulator or klystron will result in a field decay in the DTL tank
with a time constant that is inversely proportional to the quality
factor, Q, of the DTL tank. In this study the fields are instantly
dropped to zero and the transients are not included. This is worse
than the real life, since the machine protection system should stop
the beam within few micro-seconds. In this study 1×106
macro-particles are used for simulations that start at the
beginning of the DTL and are tracked to the target. The power loss
in case of an individual cavity failure, or an individual permanent
magnet quadrupole (PMQ) failure are presented in the
following.
3.1 DTL RF failure
All the five tanks of the DTL have been turned off one by one. Tank
one failure results in a complete beam loss before the end of DTL.
In other words, with no power to tank 1 the beam will not reach the
SC linac. Tank 2 failure results in significant losses within the
DTL and also the rest of the linac with complete beam loss before
the target. Tanks 3 through 5 failures cause no losses in the DTL
tanks, but all the beam will be lost in the linac. These losses are
presented in Fig. 5 and the transmission is shown in Fig. 6.
0 50 100 150 200 250 300 350 400 450 500 550 600
10−2
10−1
Figure 5: Losses due to individual DTL tank failure
Without power in tank 1 the loss pattern will not be affected by
the field value in the other tanks since all the beam is lost
before leaving tank 1. However, having tank 1 fully powered beam is
almost fully transported through the DTL even if all the downstream
cavities are turned off. This is explained by the smooth phase
advance variation along the DTL.
3.2 Demagnetized PMQ in the DTL
Increased temperature of the PMQs above their Curie temperature
demagnetizes them. Although this does not happen instantly in this
study we assume an instant demagnetization of few of the PMQs along
the DTL to look at the resulted loss. The losses are presented in
Fig. 7.
6
0 50 100 150 200 250 300 350 400 450 500 550 600 0
20
40
60
80
100
Tank 5
Tank 4
Tank 3
Tank 2
Tank 1
Figure 6: Transmission as a result of individual DTL tank
failure
0 5 10 15 20 25 30 35 40 45 50 10−2
10−1
(W )
Figure 7: Losses due to individual PMQ failure. Each failed
quadrupole is marked with vertical dashed lines and the
corresponding losses have the same color. The vertical dotted line
at ∼43 m indicated the end of DTL.
Failure of each PMQ results in losses that happen very locally
within few meters of the failed quadrupole, either in the same tank
or within the DTL.
Tank 1 Tank 2 Tank 3 Tank 4 Tank 5 DTL loss Linac loss Case 1 OFF
ON ON ON ON 100% – Case 2 ON OFF ON ON ON < 0.1 W ∼100% Case 3
ON ON OFF ON ON 0 100% Case 4 ON ON ON OFF ON 0 100% Case 5 ON ON
ON ON OFF 0 100%
Table 1: DTL RF failure
7
3.3 DTL Conclusion
Considerable amount of losses are observed inside the DTL (Tank 1
and 2) only if the Tank 1 is off its nominal voltage. If the Tank 1
is at its nominal voltage and the following tanks are completely
off, only negligible amount of losses occur within the DTL itself
but all the beam is lost in either SC linac or HEBT. These
situations are summarized in Table ??.
In the unrealistically pessimistic cases of completely destroyed
PMQs, the downstream part of the DTL acts as scrapers and a part of
the beam is lost in the DTL. The particles which arrive to the end
of the Tank 5 reach the target without causing any further
losses.
4 Superconducting Linac and HEBT
In the superconducting linac the magnet fields have been turned off
instantaneously from their nominal values to zero. The reason
behind this study is to determine if there are hot spots where
losses will be focused for any failed magnet in the spoke, medium
beta, high beta sections or beam transport system. All the magnets
have been turned off one by one (no two magnets simultaneously) and
losses have been recorded. Since these were extreme failure cases,
which do not define a regular operation of the accelerator the
figures of merit of beam like beam emittance, halo or spatial
distribution are not being analyzed here.
4.1 Cavity failures in the SCL
Upon failure of the RF system the field in the cavity will decay to
zero in the absence of any beam over the time constant of the
cavity which is inversely proportional to the Q-value of the
cavity. In the presence of a high current beam though, the beam
interaction with the cavity will excite the cavity and eventually a
field with 50% of the amplitude in the decelerating phase will
affect the following bunches. In the first part of the study of the
cavity failures, the transients are excluded and the worst case
where the beam is decelerated by the cavity is presented to have an
idea of the maximum beam loss and activation. The transmission of
the linac is shown in Fig. 8 and losses are shown in Fig. 9 for few
cases where individual cavities are failed. As is shown in Fig. 8,
failure of a cavity from the middle of medium β section where beam
is energetic enough does not cause significant losses in the linac
and HEBT, however, since the beam energy is below the acceptance of
the dogleg the remaining beam is entirely lost in the bend area.
Energy gain after the failed cavity is almost zero since beam does
not arrive at the right synchronous phase, Fig. 10.
0 50 100 150 200 250 300 350 400 450 500 550 600 0
20
40
60
80
100
8
0 50 100 150 200 250 300 350 400 450 500 550 600 10−3
10−2
10−1
4.1.1 Correction of the failed cavities
In this part of the study it is assumed that the failed cavities
are de-tuned not to be excited through interaction with the passing
beam. As a result the field in the cavity remains zero, i.e. cavity
act as a drift space. The neighbor cavities (one on each side) are
used to match the beam envelope in longitudinal plane. But the beam
energy will be different as there is no margin reserved in the
cavities to increase the gradients and compensate for missing
cavity. As a result the beam velocity downstream of the failed
cavity is reduced resulting in an increase in the time of flight to
the downstream cavities. Therefore phases of the downstream
cavities should be adjusted to have the same synchronous phase
including delayed arrival of the beam. Transmission, losses and
energy gain after correction of each cavity are shown in Figures
11-13.
4.2 Electromagnet failure in the SCL
The failure of electromagnets in the LWUs will cause significant
mismatch and losses happen rather locally. Losses due to failure of
every 10 quadrupole (one by one) is plotted in Fig. ??. The first
three quadrupoles are located in the spoke section, the second two
in medium β, the third four quadrupoles in high β, followed by
three between the last cryomodule and first dipole, one in the
dogleg and another in the A2T. In the absence of any power supply
the ramp and decay time of field in the quadrupoles in ∼ 4.5 s.
However, when the power supply is connected (which is the case) the
rise and fall time of the magnet is determined by the architecture
of the power supply and is ∼ 4 ms. This means that by the time the
machine protection systems acts, ∼ 11 µs, the field in the magnets
are only 0.5% below the nominal field.
The losses due to quad failure in, as shown in Fig. 14, happen
mainly very locally within 20 to 40 m of the failed quadrupole,
however, several of the quad failures result in significant loss in
the A2T area, due to local tighter apertures in this area.
5 SCL and HEBT Conclusions
A sudden failure of the cavity causes a mismatch between the
arrival time of the bunch to the following cavities with respect to
the RF wave, which results in null or inefficient acceleration.
These unaccelerated particles will be lost, depending on the energy
at the failed cavity, locally or in the dogleg 10.
9
0 50 100 150 200 250 300 350 400 450 500 550 600 0
500
1,000
1,500
2,000
eV ]
medium β 1
medium β 25
high β 1
high β 41
Figure 10: Energy of the beam when a cavity is failed.
Majority of the single cavity failure cases could be corrected as
represented in Fig. 11. In the SC linac, a few cavities in one
cryomodule or a few cavities linked to one modulator could
potentially fail simultaneously, such cases including their
correction scheme should also be studied. While for the cavities
hosted in the very last cryomodule of the high β the loss of power
would only reduce the final energy, after detuning and matching
without any further studies, the other cases need a detailed
investigation. On the other hand, as the transmission after the
correction of the spoke cavity with highest energy gain is not
recovered completely using only two cavities, failure of a pair of
high energy gain spokes cavities should be studied further by
including larger number of cavities in the correction scheme.
While the risk of a electromagnet failure is much lower than a
failure of a cavity, these cases have been studied and the
conclusion is that while for low energy quads the losses happen
more locally, the quads in the high energy part of the linac cause
losses mainly in the A2T, and the amplitude of these high energy
losses are a function of the phase-advance between the failed quad
and the A2T.
6 Further Studies
The results of the study presented in this note identify the
components critical to the losses and the corresponding patterns of
the losses. As discussed in the Introduction, however, an
estimation of the real integrated losses due to failure of a
component requires to take into account the loaded Q-value of the
cavity or the decay time constant of the quadrupole together with
the response time of the MPS. The total dose level over a long
period of time may be also estimated based on the trip rate of a
similar component used in other operational accelerators.
Loss of several random single cavities is another topic that should
be studied. Possibility of operation with failed components by
readjusting the lattice and lowering the beam current should be
also studied.
10
0 50 100 150 200 250 300 350 400 450 500 550 600
96
98
100
Figure 11: Transmission after the correction
0 50 100 150 200 250 300 350 400 450 500 550 600
10−2
10−1
11
0 50 100 150 200 250 300 350 400 450 500 550 600 0
500
1,000
1,500
2,000
eV ]
Figure 13: Energy after the correction
0 50 100 150 200 250 300 350 400 450 500 550 600 10−4
10−3
10−2
10−1
)
Figure 14: Power lost due to QUAD failure. Vertical dashed lines
indicate the position of failed quadrupoles and similar solid color
bars indicate the losses associated to those. The two vertical
dotted black lines represent the position of dipoles.
12
0 50 100 150 200 250 300 350 400 450 500 550 600 0
20
40
60
80
100
(% )
Figure 15: Power lost before target due to each quad failure in the
SCL and HEBT. Vertical dotted lines indicate the start of Spoke,
Medium-β, High-β, HEBT, DogLeg and A2T sections.
13
References
[1] R. Miyamoto, “An ESS Linac Collimation Study”, HB’14, to be
published.
The latest version of this document could be downloaded from:
https://chess.esss.lu.se/enovia/tvc-action/showObject/dmg_TechnicalReport/ESS-0031413/valid
14
Introduction
MEBT
Overview
Failure of a quadrupole
DTL Conclusion
SCL and HEBT Conclusions