CER

N-A

CC

-201

4-00

4325

/04/

2014

EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH CERN – ACCELERATORS AND TECHNOLOGY SECTOR

CERN-ACC-2014-0043

STUDIES AND IMPLEMENTATION OF THE PS DUMMY SEPTUM TO MITIGATE IRRADIATION OF MAGNETIC SEPTUM IN STRAIGHT

SECTION 16

C. Bertone, J. Borburgh, D. Bodart, R. Brown, S. Burger, S. Damjanovic, P. Demarest, R. Fernandez Ortega, J. A Ferreira Somoza, D. Gerard, S. Gibson, S. Gilardoni, M. Giovannozzi (ed.), G. Le Godec, C. Hernalsteens, M. Hourican, N. Jurado, J.-M. Lacroix, S. Mataguez, G. Métral, C. Pasquino, E. Perez-Duenas, S. Persichelli, B. Salvant, R. Steerenberg, P. van Trappen

Abstract

During the commissioning phase of the PS Multi-Turn Extraction (MTE) and the following operational period a number of limitations were observed, namely activation of the magnetic septum in straight section 16, fluctuation of the fraction of particles trapped in the islands, and instabilities of the injection trajectories in the SPS. These issues have been addressed in many ways, by either dedicated measurement campaigns or paper studies to determine mitigation measures. In this document the proposed solution of installing a dummy septum in straight section 15 of the PS ring to reduce the activation of the extraction magnetic septum is described in details.

Geneva, Switzerland, April 29 2014

1. Introduction

The PS Multi-turn extraction (MTE) [1] was proposed to mitigate the extraction losses due to the slicing process that is at the heart of the Continuous Transfer (CT) technique. Such a beam manipulation has been proposed in the seventies to transfer a beam from the PS to the SPS (see Ref. [2] for an overview on the subject and references therein). The implementation, described in detail in [1], was completed in 2008 when the hardware commissioning and the first beam commissioning started (see Refs. [3, 4] for detailed accounts on these activities). In 2010 the PS and SPS start-up were planned to be performed only with the MTE beams, with the CT kept as fall back solution [5, 6]. Reasonable intensities, in excess of 2×1013 p/PS batch, were achieved. However, in May 2010 a very large activation of the magnetic extraction septum 16 was measured and the IEFC decided to revert to CT for delivering beam to the SPS [7]. This was based on the need to minimise the time required for an intervention in the septum area in case of failure after several months of operations. It is worth emphasising here that a) it was not possible to predict the activation of the septum before having MTE in operation as to estimate the septum activation via numerical simulations was not considered realistic or accurate enough by the experts; b) the increased activation of the septum was not considered to have an impact on the reliability of the device, but only on the cool down time required to access the area in case of an intervention. It is also important to stress that such an increased cool down time would have had a negative impact on the LHC physics run in case of break down in the PS in the extraction region. The septum activation is due to the longitudinal structure of the beam (completely debunched) and the long rise time of the kickers compared to the revolution period. In the MTE Design Report [1] two strategies were considered and proposed to reduce the impact of the kicker rise time: firstly, a bunched version of the beam (either using h=8, or h=16 in the PS); secondly, a thinner magnetic septum. The first option could not be evaluated on paper and only beam tests could provide the final answer. This has been obtained during the beam commissioning period and the bunched version of the MTE beam had to be discarded due to the too large losses in the SPS [8-10]. The transfer of bunched beam could be revised in case that the 200 MHz system of the SPS would be upgraded as proposed in the framework of the SPS-LIU project [11-13]. The second option was considered to be too expensive for the actual benefit that it could bring and therefore it was not approved. Parenthetically, radioprotection measurements have clearly shown that the rest of the PS ring benefited from the MTE, as global beam losses were much reduced with respect to the corresponding situation with CT in the entire ring, except in the extraction septum region.

At the same IEFC meeting it was also decided to pursue the studies on the performance of the MTE beam using one or two cycles extracted towards the SPS as this had been evaluated not to be dangerous for the septum 16. Since that moment, intense efforts were devoted to performing systematic measurement campaigns as well as paper studies to find mitigation measures to the observed phenomena that were, and still are, limiting the MTE performance. These phenomena are: beam losses on the magnetic septum due to the beam lost during the rise time of the kickers; fluctuations in the fraction of beam trapped in the islands; fluctuations in the injection trajectories in the SPS.

These issues have been studied extensively and two dedicated mini-workshops were organised in 2010 and 2011 and the progress constantly monitored by the IEFC (see Refs. [14-36] for more detail). A number of mitigation measures to reduce the activation of the extraction septum have been found, like the installation of a dummy septum [37] or a hybrid MTE extraction scheme. In the next sections the issues are briefly reviewed, together with the proposed solutions.

1

As a last point, we think it is necessary to underline that MTE is certainly a challenging beam manipulation that requires tight tolerances on the operation of all machine systems (magnets, beam instrumentation, controls, RF) such that any deviation from that is enhanced by MTE. Nevertheless, some of the perturbing effects might also be affecting operational beams, however, to a much lower extent. It is worth mentioning that in the case of the LHC beams, an empty cycle is added in front of the operational ones in order to provide reproducible magnetic conditions.

2. Issues

2.1 Magnetic septum activation

The activation of the magnetic septum is the result of particles lost during the rise time of the extraction kickers. These losses are unavoidable due to the longitudinal structure of the beam that is required by the SPS, based on arguments of beam loading in the cavities and quality of the spill provided to the fixed target experiments in the North Area. It is worth mentioning two points: firstly, the possibility of leaving a hole in the longitudinal beam structure has been considered, but, independently on any consideration about its feasibility in the PS, discarded once more based on arguments related with either SPS beam dynamics or experiments optimal running conditions. Indeed, the gap in the beam structure would be repeated four times in the SPS after injection and this has been considered a serious performance limitation. Furthermore, the local beam intensity should have been increased to compensate for the missing beam in the four gaps. On the other hand the fixed target experiments are already at the limit of their capabilities and any intensity increase should be coupled with an increase in the spill length in order to be fully exploited by them.

As a second point, one could also argue that the initial stage of the beam commissioning has induced higher losses on the septum with respect to a possible future routine operation (in [1] typical losses were estimated at the level of 2 %). Furthermore, the various fluctuations affecting the trapping and the extraction trajectories should also have an adverse impact on the septum activation (see next section). Nonetheless, it is clear that on the long term a robust and sound solution is mandatory.

Lastly, the obvious solution of reducing the rise time of the extraction kickers has been considered and looked at in detail1. However, it turned out that the beneficial impact on extraction losses was rather limited (only 80 % reduction of losses for a 100 % reduction of the rise time) and this had to be compared against a rather high implementation cost and an increased complexity of the whole system. Therefore, this option was eventually dropped.

2.2 Trapping fluctuations and trajectory instabilities

Another issue was the observed fluctuation in the fraction of particles trapped in the islands. The optimal performance of the SPS relies on an approximately equal sharing between the particles in the islands and the beam left in the centre of phase space. Such a fraction, however, changes between different cycles with a pattern that is not completely random and hence seems to exclude a beam dynamics source, such as beam instability, but rather points towards a hardware issue. Lengthy measurements campaigns have been organised to understand the phenomenon and find a solution. In 2010

1 Such activity was a follow up from the 2011 IEFC Workshop [27].

2

the studies were performed by using the nominal scheme, i.e., the split beam, and the fraction of trapped particle was studied as a function of key beam dynamical quantities, like tune, chromaticity, octupolar strength, properties of the transverse damper that is used to excite the beam during the resonance crossing, without finding a good correlation with any measurable quantity. In 2011 it was decided to simplify the situation and, instead of using split beams, a single bunch was kicked inside empty islands and the orbit system was used to measure the reproducibility of the islands’ position under the influence of several external parameters. These measurements seemed to indicate not only a contribution of the cycle preceding the one that is used for our tests, but also of the overall super-cycle composition. In addition, variations from one day to another of the beam physical parameters, for an equal set of hardware settings, were observed in many occasions. There were also a number of independent observations made on different beam types, such as LHC beams, indicating that the PS main magnets might have a role in some of the observed behaviours.

Finally, it was also observed a trajectory fluctuation from cycle to cycle at injection in the SPS. This effect could be partially mitigated with the transverse damper, but it was clearly something that could not be really accepted for routine operation. The key point, however, is that off-line analysis of the measured data seems to indicate that the observed fluctuations at the SPS could be compatible with the observed fluctuations of the islands’ position in the PS from the measurements performed in 2011. This would mean that the solution of the trapping fluctuations would possibly imply also a full or at least a partial solution for the injection trajectories fluctuations in the SPS.

3. Proposed solutions

3.1 Reduction of Magnetic septum activation

Two solutions that will be pursued in parallel are the installation of a dummy septum and the implementation of a hybrid MTE extraction.

By dummy septum, we mean a device provided by a blade intercepting the beam during the rise time of the kickers, but that is not generating any deflection [37]. The idea is to absorb almost all particles that would be otherwise intercepted by the magnetic septum blade during the five extraction turns and without interfering with the circulating beam during injection and acceleration. This last point has to be valid for any kind of beam produced by the PS, but also for extractions other than MTE. The new device would be enclosed in a concrete shielding, similar to what is done for the internal dumps in straight section 47 and 48 of the PS ring, in order to minimise the level of radiation in the area, but also to protect from the remnant dose the personnel in case of an intervention on the device. Therefore, this approach aims at creating a well-shielded area in which all extraction losses are concentrated. Such a strategy has been verified by radioprotection experts and approved [38]. The dummy septum will be designed as a passive device, requiring only a minimum maintenance and a very low risk of failure. Moreover, as the amount of material potentially exposed to beam is much less than in the case of a standard septum, the dummy septum activation will be naturally reduced. The solution envisages the installation of the new device in straight section 15, with knock on consequences on the elements currently installed there (the whole straight section should be emptied to leave only the dummy septum), and on the extraction trajectories of all the fast-extracted beams. So far, the attempts to change the extraction trajectories of the beams were rather successful and it should be possible to generate a region of about 10 mm in the horizontal plane

3

that is never swept by any of the extracted beams: the blade of the dummy septum will be located inside this area. Some complementary studies and beam measurements have been carried out during the physics run in early 2013, just before LS1, to provide the final confirmation of the feasibility of this solution.

By hybrid MTE extraction, we mean a new scheme in which the extraction of the split beam is performed using also the electrostatic septum that is at the heart of the CT scheme, plus eventually some of the CT kickers. Of course, while for the CT the electrostatic septum is used to deflect particles and also to slice the beam, in the case of the MTE only its first function will be used, as the beam will be split transversally by the resonance crossing mechanism. The constraint of this approach is that it should not impact on the availability of the CT during the transition phase. Indeed, such an approach would imply keeping in operation some CT hardware that was originally assumed to be de-commissioned with the advent of MTE. It is also clear that in principle either of the two proposed solutions would be enough for mitigating the activation of the magnetic septum. Nevertheless, we think worth it to pursue the two approaches in parallel and even have both in operation to provide the best suppression of the losses in the PS ring. The constraint of having always available the CT imposes not to alter the layout of the CT extraction elements, which, in turns, implies applying all the optics manipulations that are performed during the CT extraction also to MTE. This point is certainly not optimal as, e.g., the fast and large tune variation over the last 5-6 ms during the raise of the extraction bumpers and the quadrupole kick enhancement system used for the CT would affect the split beam as it would move islands in phase space and an appropriate compensation should be envisaged. This and other points have been considered during beam-based measurements in 2011 and continued in 2012 to assess the actual feasibility. Another delicate point is the fast bump that extends over a larger fraction of the PS ring than the nominal MTE fast bump. For the CT-like fast bump a turn-by-turn correction strategy for adequate closure should be devised and is currently under study on paper.

3.2 Fluctuations and trajectory instabilities

This issue has been tackled on several fronts, but it has to be highlighted that the rather complex structure of the PS main magnet, including several additional coils, i.e., the so-called pole-face windings and figure-of-eight loop, is not yet very well modelled, which makes it rather difficult to pursue theoretical studies on the possible sources of the observed fluctuations, in spite of interesting recent advances [39, 40]. The current working hypothesis is that the magnetic state of the PS main magnet depends not only on the previous cycle, but also on the global super-cycle composition. While this hypothesis is somewhat confirmed for the main field and the quadrupolar component, the situation of higher order field components, relevant for MTE, is not clear. It is worth stressing that some improvement in the stability of the trajectories of the islands was observed when the preceding cycle was a 26 GeV/c one. Such a cycle should have the effect of resetting the magnetic state of the PS main magnets to a better defined state with respect to the case in which the preceding cycle is randomly chosen among those available at the PS. For these reasons, it has been decided to push for the installation of magnetic probes in the reference dipole unit 101 to allow measuring directly, and not only via beam based measurements, the higher order field components.

4

4. Dummy septum to mitigate the irradiation of the magnetic septum

4.1 Hardware design

The dummy septum (TPS15) shall reduce the activation of the magnetic extraction septum SMH16 by using a copper blade to absorb the particles which sweep across the septum blade during the rise time of the extraction kickers. A sketch of the layout of the extraction region towards the SPS is shown in Figure 1.

Figure 1: Dummy septum location in straight section (SS) 15 is just upstream of the magnetic extraction septum in SS16.

The position of the blade can be adjusted by means of a remote displacement system which allows for the precise positioning of the blade with an accuracy of 0.1 mm. The range of movement is from 80 mm to 100 mm from the PS orbiting beam for operation. The angle of the blade can also be adjusted by ±10 mrad to allow for an accurate alignment of the copper blade with respect to the extraction beam. In the case when the dummy septum is not used, the dummy septum blade can be moved to the park position at 130 mm from and parallel to the orbiting beam.

A cross section of the dummy septum including also the main internal components can be seen in Figure 2.

The dummy septum blade is mounted on a solid copper baseplate which transfers the heat deposited by the beam in the blade via a copper conductor to a water cooled circuit. This avoids having a water cooling circuit under vacuum, and reduces significantly the risk of a vacuum failure of the cooling system.

SMH16

TPS15

Extracted Beam

Circulating Beam

5

Figure 2: View of the dummy septum from the upstream side.

The vacuum vessel (see Figure 3), is equipped with standard UHV conflat type flanges with copper joints. The downstream side of the vessel is equipped with an end cover which uses a wheeler type flange. The end cover can be removed to allow access to the inside of the vessel to facilitate assembly of the under vacuum components. The end cover also has a racetrack section vacuum tube which allows the passage of the beam and maintains continuity of the impedance. Following simulations of the outgassing of the various components and the available pumping capacity in the vicinity (see Section 4.3), it has been decided that no additional pumping is necessary. Nevertheless, the vacuum tank will however have a flange where a NEG cartridge can be fitted to improve the vacuum if required.

Figure 3: The vacuum vessel is made of the main tank body (left) and cover with beam pipe (right).

The blade support is mounted on a copper base which is located on precision recirculating guides running on hardened shafts. The base plate shall be displaced using trapezoidal leadscrews driven through

Motor

Dummy septum blade Beam impedance screen

Circulating Beam

Actuator rods/bellows Ejected Beam

6

bevel type gearboxes by 48 V, 3 phase AC motors upstream and downstream. Each motor can be controlled independently either from a local station or from the control room. Position measurement on each side is provided by rotary potentiometers as well as linear potentiometres for redundancy and calibration local readout can be made, using dial guages with a resolution of 10 µm. End switches control the maximum in and out positions. The detail of the movement system is shown in Figure 4.

Figure 4: The up and downstream remote displacement systems are located outside of the vacuum vessel and on the inside of the PS ring, to limit the dose.

Figure 5: Energy deposition as predicted by FLUKA for the blade

Due to the energy deposition in the blade by the beam-matter interaction, there will be a corresponding rise in temperature of the components under vacuum. Finite element analysis has been performed to assess the temperature rise and distribution. The results of these simulations are shown in the Figure 5 and Figure 6. The heat deposited in the blade has been estimated using the FLUKA code [41, 42]. The resulting energy deposited in the copper blade per unit volume and time as a function of the longitudinal coordinate is shown in Figure 5, featuring the peak very close to the entry point of the blade with a

7

following almost exponential decrease. In all the simulations the standard intensity of 1011 p/s has been used2.

The outcome of the FLUKA simulations has been used as input for the heat transfer simulations. The heat is evacuated by means of a cooling rod which is connected to the PS demineralised water supply. This ensures that the temperature of the cool end of the rod remains at approx. 18-20º C and the subsequent maximum temperature of the blade turns out to be 35º C. A temperature probe (PT100) shall be fitted to the blade to monitor the temperature. The overall layout of the cooling system is shown in Figure 6.

A number of components have been added to the layout of the dummy septum in order to cope with requirements imposed by impedance considerations. The details of this analysis will be given in Section 4.4. Nonetheless, here we anticipate the hardware choices based on that analysis. Indeed, it turned out that a beam impedance screen should eliminate any dangerous resonances. For this reason a stainless steel beam screen has been integrated in the vacuum vessel. The screen is connected to upstream and downstream ends of the vessel using multi-contacts which permit relative movement between the blade support and the vacuum vessel. The impedance beam screen is visible in Figure 7. Furthermore, an RF cover plate has also been integrated to reduce the effect of internal geometry which may create low frequency trapped modes.

Figure 6: Heat transfer of the energy deposited in the dummy septum blade. This image shows the results of the simulation of the temperature distribution of the blade, support and thermal conductor.

It is worth stressing that the main design principle of the dummy septum is that it should require essentially no maintenance and, in the event of damage, a spare should be available for replacing the installed device. In fact, the activation of the device prevents any possibility of repairing action, a part from very minor fixes. This also implies that special care should be taken in the design of the various sub components such that they are radiation hard and in order to minimise the intervention time. These considerations have been incorporated in the design of the supporting system, which allows for a rapid exchange due to the plug and play concept (see Figure 8). Adjustable supports have been fitted to each vacuum vessel and each of the vacuum vessels shall be pre-aligned to the main support in the accelerator.

2 This value corresponds to a rounding up of the official figure of 0.8×1011 p/s for typical super-cycles as reported in L. Bruno et al. “PS Radiation Working Group – Final Report”, CERN-ATS-2011-007, 2011.

Cooling rod

8

The main support comprises a shielding block which incorporates the sockets which receive the tank supports. This allows for rapid exchange of the system in the tunnel, while at the same time it guarantees the alignment of the newly installed dummy septum. The single downstream foot determines the longitudinal position and the upstream feet position the vessel in radial, angular and height.

Figure 7: Main tank body showing the installed beam screen (left) and the sliding contacts (right).

Figure 8: TPS support system, showing the plug and play sockets for the tank, eliminating the need for realignment of the TPS in case of an exchange.

As far as the controls are concerned, the copper blade position will be controlled via the two motors, which allow both controlling the position and the angle relative to the incident beam. The position and the angle of the blade will be accessible in the control room, and will be expressed in the reference system of the circulating beam. Also the information on the blade temperature will be made available in the control room via the standard controls interfaces. It is also worth recalling that it is planned to have one complete spare of TPS15, which will be prepared after the operational device will be completed and installed in the PS ring.

9

In the event of a failure of the TPS or adjacent magnets, an intervention may be required. In order to quantify the time of the intervention and the estimate of the dose for the specialist personnel, a study has been undertaken to assess the consequences of several possible failure modes. It is worthwhile stressing that the dummy septum is an element essential for the MTE Hence any failure might affect the beams delivered to the SPS for Fixed Target physics. On the other hand, the TPS15 is not a device required for all other fast extractions towards the SPS. In particular, it is of no use for any flavour of LHC beams (protons or ions). Therefore, the underlying concept in the design of the TPS is that it should not create any stop for the LHC physics. This means, that each failure event can be classified based on its impact on the LHC or SPS fixed target physics programme. This distinction will be made clear in the following discussion.

The worst case scenarios involve replacement of the complete TPS and require a bake out of the sector. Due to the presence of the laminated septum, SMH16, a period of 5 days is required to perform the complete bake out cycle. The actual time required for the replacement of the TPS depends mainly on the levels of radiation in the immediate vicinity and may involve a period of cooling down time before any intervention may be performed. In Table 1 some of the possible types of failure and the corresponding action required to restore the PS to operational mode are listed.

Table 1: List of possible failure scenarios for the TPS15.

Failure type Classification Action Comments

Failure of magnet 14 or 15

Serious Remove TPS and shielding before dismantling the magnet.

Affects all beams. Loss of beam time, minimum 5 days intervention (bake out).

Vacuum Leak on TPS

Serious Repair or replace TPS Depends on results of diagnostics. Affects all beams.

Bellows Failure Serious Replace TPS Affects all beams. Loss of beam time, 5 days (bake out).

Blade drive system Depends on results of diagnostics

Repair or manually displace.

Depends on results of diagnostics. Possibly affects MTE beam.

Beam instrumentation

Depends on results of diagnostics

Repair or replace TPS Depends on results of diagnostics.

Camera Low Replace Camera Rapid intervention (5-10 min).

Beam screen insertion system

Blocked IN: serious

Blocked OUT: acceptable

Displace manually

Replace at next TS

Screen normally used during PS commissioning. Failure very unlikely due to failsafe design of BTV.

10

Some of the failure scenarios reported in the previous table deserves some further comment. In case of problems with the blade drive system with the blade in the operational position no real issue is expected. Indeed the repairing action could be scheduled during a planned stop of the physics programme. Of course, in case of troubles during setting up, this failure will have a direct impact on the fixed target physics at SPS, but no impact whatsoever on the LHC physics.

These considerations also apply to the failure of beam instrumentation components. In fact, unless a vacuum leak is generated, only the MTE beam might be affected by beam instrumentation issues. Moreover, in case of routine operation period, the failure of instrumentation might be completely irrelevant also for the MTE operation.

4.2 Instrumentation

A beam observation system has been foreseen in the TPS15 design in order to measure the position of the extracted beam so to adjust precisely the extraction bump and the blade position. It consists, as shown in Figure 9, of an insertion device holding a screen made of Al2O3:Cr and a camera setup whose details are shown in Figure 10. The following constraints have been considered in the design phase:

• Limited, easy, and fast maintenance due to the very high radiation dose accumulated with the operation of TPS15.

• The IN position of the screen will be at 2 mm from the blade, no matter the actual blade position3.

• The park position for the screen should ensure that it will never be in the beam extraction aperture independently on the copper blade position.

• The whole system should be adjustable such to avoid the acquisition of saturated images, e.g., due to optical density filter, in the case of future beam intensities up to 3-4×1013 p.

3 The blade position can be adjusted by ±10 mm around the nominal position of 90 mm.

11

Figure 9: Beam observation system installed on the TPS15.

The choice of the camera is rather limited in this environment where the radiation level is expected to be rather high. The Vidicon tube based homemade camera will be used since it is by far the most radiation resistant camera. Indeed, the design of its head is free of active electronic components. The filter wheel and the illumination system have been designed in such a way that they can be replaced very rapidly with no need of any readjustment.

Figure 10: Exploded view of the camera setup installed of the fuse silicate viewport.

Also, in view of minimising the maintenance needs and to reduce the radioactive dose delivered to personnel in case of intervention, a special design of the screen insertion device has been developed. It consists of a pneumatic actuator without sealing, which reduces the number of parts that could be damaged by radiation. It is worth stressing that the design is considered to be failsafe i.e. the parking position is the default one in case of failure of electrical power cut, compressed air missing, etc.

To overcome the constraint of the positioning of the screen with respect to the blade of the septum, the two movements are mechanically linked together. The screen has two defined positions: OUT (corresponding to park position) and IN (corresponding to beam measurement position). The pneumatic actuator of the screen is mounted on the base of the copper blade movement as shown in Figure 11. This has the advantage to ensure that the screen will follow any displacement of the blade. The IN position of the screen will always be 2 mm from the blade. Using the same logic the OUT screen position will always be 62 mm from the blade. This ensures that even when the blade is in the position closest to the circulating beam, i.e., at 80 mm, the screen will be at 142 mm, i.e., beyond the blade park position of 130 mm.

The screen is a 1 mm thick aluminium oxide wafer doped with chromium (Al2O3:Cr) with dimensions of 43 mm×82 mm. The observation area is 40×56.6 mm2 positioned at 45º with respect to the beam trajectory which makes the real, projected observation area 40×40mm2. Marks are engraved for calibration purposes. The support of the screen is equipped with fine adjustment screws for precise alignment of the screen (angular and position), as shown in Figure 12.

12

Figure 11: Blade and screen movements are mechanically linked together to ensure that any movement of the blade will keep the distance blade/screen at 2 mm when screen is inserted IN and at 62 mm when it is in park position.

Figure 12: Dimensions of the Al2O3:Cr screen (left) and 3D view on its 45º adjustable support (right).

The control and acquisition electronics will use the standard VME card used for any video camera beam observation system at CERN (called BTV system). This VME card can drive the screen actuator, the filter wheel, the lights (two independently) and triggers the synchronous beam capture with a resolution of 400×300 pixels. The optics of the system is a single specific camera lens of 50 mm focal length. A non-browning treatment guarantees the efficiency of the optical aperture in this high radiation level area. The overall resolution of the system is about 250 µm/pixel.

Successful tests have been performed, showing that the screen would not be saturated when interacting with high beam intensity of up to 4×1013 p.

As a final remark, it should be mentioned that the experience gained with the design and construction of the BTV described in this section has led to the decision of changing the current design in order to simplify a number of items, and in particular by removing the bellow installed inside the TPS15 vacuum vessel. A possible design, still under study, is show in Figure 13. The new version of the BTV will be installed in the spare TPS15. Hence, strictly speaking, the TPS15 installed and its spare will not be exactly the same, but this does not have any impact on performance.

13

Figure 13: Possible design of the BTV that should be mounted on the spare TSP15.

4.3 Vacuum aspects

The TPS15 is located in vacuum sector 20. The gas load has been evaluated using the existing literature of outgassing data for the materials involved in this component [43]. Considering the actual vacuum system; after modelling the entire vacuum sector, the gas load can be evacuated without any modification in the total pumping speed, keeping the vacuum requirements on the range 0.1-1×10-8 mbar. To cope with possible modifications, e.g., the installation of ferrites in order to mitigate possible impedance issues, an extra pumping port DN150 has been installed such that adding a new pump, e.g., CapaciTorr 2000 NEG pump, can be envisaged, should additional gas loads not considered in this study be observed during operation.

The current layout of the PS vacuum sector 20 is shown in Figure 14 (upper part) while in the lower part the corresponding PS ring layout is reported. Sixteen 16 Sputter Ion Pumps (SIP) of 220 l/s N2 equivalent nominal speed keep the pressure profile within the operational limits, while 5 Penning gauges (VGP13, VGP13a, VGP16, VGP17, and VGP18) are used to monitor the pressure profile in the sector.

Figure 14: PVSS vacuum layout of PS sector 20 (upper) and general layout of the sector 20 of the PS ring (lower).

14

To simulate the full vacuum sector the analogy between vacuum and electrical networks was used [44]. Computational tools developed for electrical network analysis (ENA) such as LTSPice were used to produce a model of the studied vacuum sector. The analogy uses the equivalency shown in Table 2

Table 2. Analogy between electrical and vacuum networks.

Electrical network Equivalent vacuum network Capacity C [F] Volume V [l] Electrical current I [A] Flow Q [mbar l/s] Inverse of resistance 1/R [1/Ω] Conductance C [l/s] Voltage V [V] Pressure p [mbar]

𝐼 = 𝐶𝑑𝑉𝑑𝑡

𝑄 = 𝑉𝑑𝑝𝑑𝑡

𝐼 =𝑉𝑅

𝑄 = 𝐶 ∙ ∆𝑝

To evaluate the conductance of each element of the vacuum system, a 3D Monte Carlo simulation was carried out for each component of the TPS15. The data calculated using the code MOLFLOW [45] were used as inputs for the ENA software. This technique allows the evaluation of time-dependent pressure profiles. The pressure is then computed in specific locations of the network, simulating the position of the existing gauges in the vacuum sector. In SS15, there are no dedicated pressure gauges. Nonetheless, the pressure was computed in the middle of the vacuum chamber to evaluate if a local burst of pressure could affect the beam. The layout of the sector 20 of the PS ring is shown in Figure 14 (lower part), while the electrical network is shown in Figure 15.

Figure 15: Equivalent electrical network for the vacuum line of the PS sector 20

Three types of gas sources have been identified:

• Water vapour outgassing from the vacuum chamber (unbaked stainless steel). • H2 outgassing, as contribution from the bulk of the metallic components. • Ion-induced desorption, caused by the proton beam impinging on the dummy septum copper

blade.

15

From experimental data, the water vapour outgassing from a clean metallic surface is described by the following law [43]:

𝑄𝐻2𝑂 =3 ∙ 10−9

𝑡[ℎ] �𝑚𝑏𝑎𝑟 ∙ 𝑙𝑠 ∙ 𝑐𝑚2 �,

while the hydrogen outgassing from a stainless steel bulk produces a constant gas load equal to [43]:

𝑄𝐻2 = 3 ∙ 10−12 �𝑚𝑏𝑎𝑟 ∙ 𝑙𝑠 ∙ 𝑐𝑚2 �.

The water vapour gas load is dominant at the beginning of the pump down phase. The hydrogen gas load will be dominant for pump down times longer than 1000 h (≈ 40 days). In [46] a collection of ion-induced desorption experimental data can be found and

Figure 16 shows the desorption yield (the number of molecules desorbed per ion perpendicularly impinging on a surface) as a function of the ion species and its energy per nucleon. The spread of data is big, and can be as high as 107 for gold ions. In addition, no data were found for protons. Some measurements on protons were carried out and reported in Ref. [47], showing a yield less than 1. A safety factor of 10 was used per each of the main gases usually desorbed: H2, CO, CO2 and CH4.

Figure 16: Ion desorption yield for different ions as function of the ion energy per nucleon.

In order to benchmark the electrical network model, the pressure profiles along the whole sector were simulated considering the same boundary conditions which occurred during the intervention on the septum 16 that took place at the end of January 2012. In that case, after 12 days of pump down the pressure attained along the sector was in the low 10-9 mbar range. In Figure 17 the comparison between the simulated pressure profile along the sector and the experimental one is shown and a good agreement is found.

16

Figure 17: Comparison between the experimental and the simulated total pressure profile (N2 equivalent) referring to the pressure profile for the PS vacuum sector 20 after the intervention on septum 16 on January 2012. No beam losses are assumed for this simulation.

Once the model has been benchmarked, the current layout SS15 was replaced by the TPS15. Figure 18 (left) shows the comparison between the pressure profile in the actual sector and the pressure profile in the future sector with the dummy septum installed (N2 equivalent). Considering a proton beam loss of 1×1011 p/s; the contribution of H2, CO, CO2 and CH4 to the total pressure is negligible, as visible in Figure 18 (right).

Figure 18: Total Pressure profile (N2 equivalent) for PS vacuum sector 20 including TPS15 without any beam loss taken into account (left). Partial pressure profiles due to ion induce desorption (right).

As already mentioned above, a scenario where ferrites would be installed in the dummy septum was also studied. The boundary conditions considered for this case were 1000 cm2 of exposed surface and an experimental value of water outgassing after 10 hours of 2 × 10−9 �𝑚𝑏𝑎𝑟∙𝑙

𝑠∙𝑐𝑚2 � , which is a value that has been observed in past acceptance tests. This experimental value can vary by several orders of magnitude depending on the grade of the ferrite, its shape, and thermal treatments. Hence, before any installation, the ferrite must pass a number of vacuum acceptance tests.

17

Figure 19 (left) shows the effect of the ferrite in the dummy septum pressure profiles after 1000 hours of pumping. Water desorbed from this porous material is the main gas source. The installation of a NEG pumping system could mitigate this effect, as it is seen in Figure 19 (right).

Figure 19: Effect on the pressure profile of the installation of a block of ferrite (left) and of the NEG pumping system (right).

To avoid the impact of this scenario, the design of the vacuum tank has been implemented with an extra pumping port of DN150. In case of installation of ferrites, a NEG cartridge, e.g., CapaciTorr 2000, could be installed to cope with the extra load as shown in Figure 20. Such a solution would also comply with the limited space available.

Figure 20: Integration of the CapaciTorr 2000 NEG pump at the bottom of the TPS15 vacuum tank (left) and detail of the NEG cartridge (right).

Figure 21 (left) shows the effect of an extra pump (CapaciTorr 2000), locally connected to the bottom part of the vacuum tank, which helps lowering the contributions of the different gases proportionally to the delivered pumping speed. For the sake of comparison, in the right part, the impact of the beam screen

18

on the efficiency of the additional pumping has been tested. When the impedance beam screen is removed, an increase of the pumping efficiency is clearly observed.

Figure 21: Effect of the additional NEG pump on the partial pressure profiles including beam losses (left) and impact of the beam screen, which has been removed for this special simulation (right).

It is very important to underline that a getter pump has to be activated in situ using a dedicated power supply each time the sector is vented to air. That regeneration might be also needed when CO saturation of the getter surface reaches the maximum capacity of the pump. To evaluate the time between regenerations, a time-dependent simulation was carried out. The saturation effect has been taken into account considering all the previously mentioned gas loads as CO, and therefore contributing in the same amount to the saturation of the NEG pump. Cabling of this pump would be required to pilot its activation or regeneration remotely. A summary plot is shown in Figure 22.

Figure 22: Total pressure profiles (N2 equivalent) for the three case studies, namely: nominal TPS15 configuration (including ferrites); nominal TPS15 configuration (including ferrites) and additional pumping; nominal TPS15 configuration (including ferrites), additional pumping, but without the impedance beam screen.

19

A number of configurations of the TPS15 have been considered, namely with and without impedance beam screen and with and without additional pumping capacity. In conclusion, if the final impedance measurements will show that there is no need for additional ferrites, then the existing pumping capacity is enough to ensure the appropriate vacuum quality in the PS sector 20. In case ferrites would be mandatory, then an additional NEG cartridge could be easily installed thanks to the flexible design of the vacuum vessel of the TPS15, but prior to installation in the tunnel of the dummy septum.

Finally, it is worth recalling that all the standard procedures for vacuum acceptance for the final installation of the TPS15 are summarised in [48].

4.4 Impedance aspects

Impedance studies have been performed for the dummy septum to evaluate longitudinal and transverse coupling impedances taking into account the beam parameter of current and future beams, i.e., after the planned injectors upgrade. These studies have been complemented by bench measurements and the final outcome of these studies is the basis for the acceptance for the installation of TPS15 in the PS ring

The analysis has been performed on simplified 3D geometries, i.e., neglecting the BTV, the holes in the beam impedance screen, the screws inside the tanks, increasing the length of the vacuum pipes at the extremities of the TPS15, and representing the sliding contacts as small bricks of perfectly conducting material. Such geometries have been imported in the impedance simulation code from mechanical CATIA drawings. For time domain simulations, CST Particle Studio Wakefield Solver [49] has been used to obtain the wake potential generated by a Gaussian bunch circulating inside the TPS15. In CST the beam is always defined as a pencil beam with no transverse size, and in this context only ultra-relativistic beams have been considered. The longitudinal wake potential can be calculated by CST Particle Studio locating the beam and the integration path at the geometrical centre of the structure; instead, the transverse dipolar wake potential can be obtained by displacing the beam in the transverse direction, while the integration path remains fixed at the centre. The beam coupling impedance of the structure is then evaluated by Particle Studio from the Fourier transform of the wake potential.

To crosscheck results obtained from time domain, CST Microwave Studio frequency domain simulations have been performed. The evaluation of the frequencies of eigenmodes resonating in the TPS15 is done by the Eigenmode solver, while Q factor, shunt impedance Rs, and R/Q are obtained from the post-processing. The correct evaluation of the resonance parameters is fundamental to obtain good accuracy in the estimation of the impact on coupled bunch instability.

The analysis made shows that a beam passing in the dummy septum generates sharp resonances (trapped modes) in the coupling impedance; trapped modes’ frequencies also correspond to the eigen-frequencies of the closed structure. Since low-frequency trapped modes are a potential source of coupled bunch instability, two different solutions for reducing their impact on the stability of the beam have been considered, namely the use of sliding contacts or the insertion of a brick of ferrite. Both methods have the effect of damping, or in some circumstances cancelling, low frequency resonances. Furthermore, an estimation of the impact of the longitudinal impedance of the TPS15 in comparison to the total impedance budget of the PS will also be given.

To perform longitudinal and transverse impedance simulations, several aspects of the beam operation with the TPS15 have to be considered. During operation, the beam circulates in a position displaced by 27 mm from the geometrical centre of the septum. During extraction the beam moves from the circulating

20

position to few millimetres from the copper blade in about 6 ms. The beam then stays close to the blade for only few turns prior to extraction. The design of the septum considers the blade positioned between 80 and 100 mm from the nominal circulating beam position. The nominal blade position considered in these has been at 90 mm from the nominal circulating beam.

To perform longitudinal impedance simulations in CST Particle Studio, both the beam and the integration path have to be at the same position. Then, the longitudinal impedance can be evaluated at different distances from the location of the blade. From simulations, excitation of trapped modes in the longitudinal and transverse impedance spectra due to the passage of a beam of milliseconds bunch length of 26 cm has been observed. The frequencies of the trapped modes, which also correspond to the eigenvalue of the closed structure, do not depend on the beam position. On the contrary, amplitudes of impedance peaks are increasing while the relative distance between the beam and the blade is decreasing. Before extraction, while the beam covers 90 mm in about 6 ms to approach the blade, a significant increase of shunt impedance for each trapped mode has been observed. During extraction, when the beam is close to the blade at a minimum distance of 5 mm, the maximum of the impedance peaks’ amplitude is reached (see Figure 23). This effect is due to the strong electromagnetic field trapped at the edges of the blade after the passage of the beam. Since the inner geometry of the dummy septum is strongly asymmetric, all trapped modes excited by the passage of the beam, have both longitudinal and transverse components.

The dipolar impedance can be evaluated with CST Particle Studio by shifting the beam in the transverse direction and by performing the integration of the field along the central axis. Similarly, the quadrupolar impedance can be obtained by shifting the integration path while keeping the beam in the centre. The dipolar (resp. quadrupolar) component is then obtained by subtracting from the simulated dipolar (resp. quadrupolar) transverse impedance the same term evaluated in the centre and then dividing by the displacement. For the transverse impedance, the same increase of amplitude of the peaks while the beam is approaching the blade has been observed. For this reason, only numerical examples for the longitudinal component of the impedance are shown, since the transverse one shows a similar behaviour.

Figure 23 shows the real part of the longitudinal impedance excited by a bunch of rms length 26 cm and charge q = 1 nC, circulating 5 mm away from the axis of the copper blade. The bunch length has been chosen to obtain a good resolution in the desired frequency range, according to the relationship valid for CST 2012, namely fmax = 𝑐

1.69 σb, where c is the speed of light in vacuum and σb the bunch length. The

wake potential has been evaluated through the Direct Integration Method using a wake length of 100 m. As far as the boundary conditions are concerned, a perfect electric conductor has been defined on all the outer surfaces, except for the beam entrance and exit planes that have been defined as open boundaries (perfect matching layer) due to the beam pipe aperture. No symmetry planes have been used.

Therefore, beams circulating inside the dummy septum excite several low frequency sharp resonances on the real part of the longitudinal and transverse impedance. The first trapped mode resonates at 118 MHz with a Q factor of 2616 and, since Q depends only from the geometry, it will be constant for the same mode while the beam is moving from the nominal position towards the blade prior to extraction. For the PS, it is assumed that only resonant modes with a frequency lower than 300 MHz represent potential issues for coupled bunch instability: the instability rise time decreases significantly while the frequency of the mode is diminishing. For this reason, the mode at 118 MHz is studied in more detail, while the other trapped modes at higher frequencies are not anticipated to enhance the coupled bunch instability.

21

Figure 23: Real part of longitudinal impedance excited by a bunch of milliseconds length of 26 cm circulating at 5 mm from the axis of the copper blade.

In Table 3, frequencies, Q factors and shunt impedances are summarized for the first 10 eigenmodes of the TPS15. These parameters show a very good agreement with the same parameters extracted from Wakefield simulations for the first mode. The shunt impedance has been evaluated with the two solvers both with the beam placed in nominal position and with the beam at 5 mm from the blade: between the two positions there is an increase of shunt impedance for the mode at 118 MHz by a factor 600.

Table 3: Eigen modes evaluated with CST Microwave Studio.

Frequency [MHz] Q Rs [Ω]

(Circulating) Rs [Ω]

(Extraction) 1 118 2655 62 36176 2 295 3975 76 74899 3 331 3947 7 5153 4 362 4727 2 2909 5 420 4987 13 10109 6 441 4885 22 18914 7 495 5777 3 2097 8 533 7597 9 8852 9 616 3585 3 2145

10 656 5805 18 13988

For the mode at 118 MHz, the shunt impedance evaluated at 5 mm from the blade axis is about 36 kΩ, which corresponds to about 10 times the shunt impedance for a single 200 MHz cavity loaded with three pin diodes lines. Therefore, even if the beam is going to circulate in this position only for few turns before extraction, this mode should be considered as a potential source of coupled bunch instability.

To investigate more deeply the impact of this mode on the beam stability, the coupled bunch instability growth rate can be calculated with the following formula, which is valid for a mode fully coupled with the multi-bunch spectrum:

22

α =c2ηcqNb

2 L2E0ωsωrRe{Z(ωr)},

where ηc is the slippage factor, q is the bunch charge, Nb is the number of bunches, L the length of the machine, E0 the beam energy, ωs the synchrotron frequency, and ωr the resonance frequency. Since the TPS15 is going to be present in all operational conditions and with several types of beam, growth rates should be evaluated in several conditions, also assuming worst case scenarios, and different beam energies, i.e., intermediate energy corresponding to the case of MTE beams and high energy corresponding to the case of LHC beams. To this extent, beam parameters after LIU implementation have been considered for the estimation and are summarized in Table 4.

Table 4: LIU parameters for 25 ns bunch spacing considered for coupled bunch calculations.

E0 [GeV] 13 26

RF voltage VRF [kV] 165 100 Harmonic number 21 84 Number of bunches 18 72 Charge per bunch [C] 1.28×10-7 3.2×10-8 Slippage factor 0.0163 0.0215 rms bunch length σb [ns] 3 3

Furthermore, to be closer to the real conditions, the actual Gaussian shape of the bunch has to be considered. Therefore, the shunt impedance Rs evaluated with CST Microwave Studio has to be corrected with the following form factor to account for the beam power spectrum

R′s = Rse−(ωrσb)2 .

When such a correction is taken into account, the amplitude R′s of the shunt impedance is drastically reduced. In Table 5, the values of growth rate for the 118 MHz mode, evaluated both at intermediate and high energy, for different beam positions are summarised.

Table 5: Coupled bunch instability growth rates evaluated for different beam position inside the TPS15.

Displacement [mm] Rs [Ω] R′s [Ω] α [s-1]

13 GeV α [s-1]

26 GeV 0 640 10 0.15 0.08

20 3385 53 0.82 0.43 40 14762 231 3.59 1.87 60 49215 770 11.97 6.25

These growth rates, evaluated with the beam at several circulating positions, are very small compared to the typical growth rates measured in the PS [50]. Nevertheless, they become non-negligible when the beam is approaching the blade. Indeed, in the final position the rise time of 170 ms evaluated at 13 GeV could sound alarming, but since the beam is extracted in 6 ms, corresponding to 3 synchrotron periods for

23

the PS, the instability does not have the time to develop. Therefore, these considerations allow to state that the coupled bunch instabilities should not be enhanced by this mode.

Nevertheless, two modifications of the inner design have been studied with the aim of further reducing the impact of this mode. The first solution consists of inserting sliding contacts between the impedance screen and the blade displacement system to cancel the mode at 118 MHz and to avoid the generation of trapped modes below 300 MHz. Figure 24 (left) shows that all the energy released by the beam is trapped in the small volume between the impedance screen and the blade displacement system. Similarly, a small amount of energy is also trapped at the edge of the blade, in the small space between the impedance screen and the blade itself. Hence, to cancel the mode, it is necessary to create contacts between the impedance screen and the blade displacement system. Such a solution cannot be easily implemented in the space between the impedance screen and the blade itself. In Figure 24 (right) the 118 MHz mode electric field calculated on a section between the screen and the displacement system is shown.

Figure 24: Left: field lines of the 118 MHz electric field showing that all the power is trapped between the impedance screen and the blade displacement system, and the impedance screen and the blade itself. Right: 118 MHz mode electric field calculated on a section between the impedance screen and the blade displacement system.

The second solution envisages the insertion of a block of ferrite TT2-111R, which will not have the effect of cancelling the mode, but will reduce Rs and Q of the mode itself and, as a consequence, the impact on coupled bunch instability. Several simulations have been performed with CST Particle Studio, varying dimension and position of the ferrite. As a general rule, the brick of ferrite should be placed inside the tank where the magnetic field is more intense; therefore, the suggested position is the one shown in Figure 25, where a brick of 24×7×395 mm3 has been placed between the displacement system and the impedance screen. The reduction of Rs of the 118 MHz mode has been estimated to be about a factor 600, as shown in Figure 26. A further reduction of shunt impedance can be achieved by increasing the length of the ferrite brick (along the horizontal axis in Figure 25). As far as the power loss is concerned, it has been estimated that the heating would be about 1.8 W for such geometry if the ferrite is included as the mode falls inside the PS bunch spectrum and at 118 MHz the power is about -20 dB. All in all, the power deposited should not pose any problem and the foreseen cooling system should easily cope with it. As an outcome of these studies, the decision has been taken to install the sliding contacts. The option of installing a block of ferrite is left as a fall back solution to be implemented only on the basis

24

of the results of the impedance bench measurements. Therefore, the blade displacement system has been equipped with a support that could be used to house the ferrite block.

Figure 25: Magnetic field of the 118 MHz mode with the insertion of a brick of ferrite

Figure 26: Real part of longitudinal impedance excited by a bunch of rms length of 26 cm circulating at 5 mm from the axis of the blade with and without the ferrite brick.

The imaginary part of the longitudinal impedance of the PS has been evaluated with measurement campaigns [51] to be

Z(p)p

= (18.4 ± 2.2) Ω.

In this simulation, a long bunch (σb=1.5 m) and a wake of length 80 m have been used to obtain the impedance in a low frequency range. When a bunch of such length is at the nominal circulating position, it excites an imaginary part of the longitudinal impedance that is purely inductive, and the effective impedance has been evaluated to be Z(p)

p= 0.001 Ω . In comparison with the measured value, the

contribution of the TPS15 to the PS longitudinal impedance budget is expected to be negligible. When the bunch circulates at 5 mm from the blade before extraction the effective impedance has been evaluated to

Ferrite block

25

beZ(p)p

= 0.12 Ω . Albeit much larger than in the previous condition, the contribution of the dummy

septum to the PS longitudinal impedance budget is less than 1 %. For the sake of comparison, the 200 MHz cavities provide a contribution of about the 4% of the total longitudinal impedance. Hence, no issue is expected under any of the operational conditions foreseen.

The transverse impedance aspects will be covered in a future study as the transverse PS impedance budget needs to be reviewed in detail.

As already mentioned, impedance measurements have been performed to confirm the results of numerical simulations, to test the effectiveness of the sliding contacts, and to assess the need of ferrite. A detailed account of the measurement results can be found in Ref. [52]. The impedance has been measured stretching a metallic wire of 0.5 mm radius inside the TPS15. The wire is then connected to two matching resistors, and finally welded to a device called suco-box. This setup allows measuring the transmission coefficient S12 by means of a vector network analyser. Using special flanges, the impedance for three situations has been measured, corresponding to the wire in nominal position for the circulating beam, wire at 30 mm from the circulating beam position, and wire at 60 mm from the circulating beam position. During all the measurements the copper blade was positioned at 90 mm from the circulating beam position. The measurement setup is shown in Figure 27.

Figure 27: Experimental set-up used for the impedance measurements of the TPS15.

In Figure 28 (upper) the comparison of the impedance measurement for the three wire positions is shown. No mode at 118 MHz is visible, thus indicating that the sliding fingers are working as expected. Indeed, the first excited mode has a frequency of 270 MHz, which is too high to be a source of coupled bunch instability in the PS. Furthermore, its amplitude increases when the wire is set closer to the blade. The 270 MHz mode is generated by resonances due to the gap between impedance screen and copper blade. Measurements have also been compared with simulations in time and frequency domain and the comparison in shown in Figure 28 (lower). Both measurements and numerical simulations agree on the frequency of the first trapped mode resonating in the TPS15. The source of discrepancy for the higher frequency modes is not clear, and further analysis is required. Nonetheless, the outcome of the

26

measurements confirms that the dummy septum can be safely installed in the PS ring and that the installation of ferrite is not needed.

Figure 28: Upper: The S21 transmission coefficient measured for three positions of the wire. Lower: Comparison between the simulated and measured S21 transmission coefficient with the wire in extraction position.

4.5 Shielding layout

Particular efforts have been devoted to the design and the optimisation of the local shielding for the TPS15. After several studies on the best suited material, concrete turned out to be the most effective choice. The need for a high level of optimisation suggested abandoning the choice of using standard CERN concrete blocks. Therefore, a series of concrete blocks has been specially designed and produced. The optimisation has been performed firstly at the level of the minimisation of the radiation dose outside the local shielding. Then, simplicity in mounting and dismounting the structure has been a second criterion. A third criterion has been to provide an easy access to existing systems, such as bus bars of the PS main magnets, without requiring any dismounting of the local shielding. Finally, the dimensions of the blocks have been determined on the need to protect the workers in the area in case of partial dismounting

27

of the local shielding. The detail of the shielding is shown in Figure 29, where the total set of concrete block (upper), the overall dimensions (middle), and the 3D integration situation (lower) are reported.

Figure 29: Final layout of the local shielding around TPS15. The total set of concrete block (upper), the overall dimensions (middle), and the 3D integration situation (lower) are shown here.

28

5. Performance analysis of dummy septum

5.1 Shielding efficiency

To quantify the effectiveness of the shielded TPS15 for reducing the activation in the region of SS16, comparative simulations of the radiation field in this region induced by beam losses in TPS15 and in the magnetic septum 16 have been performed using the Monte Carlo code FLUKA [41, 42]. The complete geometry of the PS extraction region as implemented in the code, including the two septa and the proper materials throughout, is shown in Figure 30. A more detailed description of the beam extraction region can be found in Ref. [38].

Figure 30: PS extraction region with the dummy septum surrounded by its local shielding in SS15, and with the magnetic extraction septum SMH16 surrounded by a vacuum tank in SS16. The tracks show a single interaction event in the TPS15 copper blade.

A detailed survey of the residual ambient dose equivalent rates along the SS16 was performed in December 2009 during the annual winter shutdown, 40 days after the end of operation with the MTE scheme at a beam intensity of 1013 p/s, and measured beam losses of ~1 % in this region. A two-dimensional color contour plot of the residual ambient dose equivalent rates in the horizontal plane centred on the PS ring axis and calculated with FLUKA is shown on the middle plot in Figure 31 for the same beam losses and cooling time conditions as for the detailed survey in the area. To facilitate the comparison, the survey data are superimposed to a dose equivalent rates profile extracted from the two-dimensional color contour plot and, as seen on the right plot in Figure 31 looking at the red curve and black stars, a very good agreement in absolute terms is found between the survey data and the results from the FLUKA simulations. While the results are shown for 40 days of cooling time for the sake of comparison with the survey data, several other cooling times were also considered. After benchmarking the FLUKA calculation for the existing configuration, the second step of the analysis is performed in an analogous way, assuming now that the beam losses of 1 % occur in the shielded TPS15. The left panel of Figure 31 shows the results in the form of a two-dimensional horizontal map of the residual ambient dose equivalent rates for beam losses in TPS15. As done for the losses in the SS16, a one-dimensional

29

projection of these results along the magnets (external to the PS ring) for the beam losses in SS15 is shown in the right panel of Figure 31.

As visible from Figure 31, the comparison with the results for the dummy septum is very encouraging. By relocating the beam losses to the shielded dummy septum, the production of secondaries and resulting activation are shifted upstream of the septum 16 where no sensitive equipment is present, strongly reducing the radiation level around SS16. While the total beam losses are not reduced, but only shifted, a local shielding around TPS15, which could not be integrated in SS16, strongly reduces the residual dose rate levels around the beam loss region in SS15, by about a factor of 8. For instance, after a cooling time of 40 days, the dose rate of ~2 mSv/h in the SS15 area is reduced to an average of 250 μSv/h.

Figure 31: Left: comparative results on the residual ambient dose equivalent rates for beam losses induced in TPS15 (left) and SMH16 (right) for cooling times of 40 days. Right: one dimensional projections of H*(10) along magnets (external to the PS ring) for beam losses in TPS15 (blue), in SMH16 (red), and measured data (black).

The importance of the possibility to implement shielding around TPS15 was identified early in the project. While preliminary studies were conducted using a bulk and conceptual shielding configuration (as shown in Figure 30) around the dummy septum, the space limitations necessary to guarantee the accessibility of some PS ring components, as well as weight constraints for handling of shielding blocks, and for the tunnel structure, were considered early in the analysis to assess the impact of a realistic shielding configuration. After several iterations, once the final shielding configuration was specified, a comparative FLUKA simulation was then performed to verify that the residual dose rates outside the local proposed design [53]. While several material combinations had been studied for the shielding, it was demonstrated that concrete, considering the radiological parameters of interest and the cost, was the best material choice. The final geometry as implemented in FLUKA is shown in Figure 32.

In addition to the activation and associated residual dose rate, the impact on the stray radiation levels at the surface of the dummy septum with the beam losses concentrated in the SS15 area was also taken into account in the analysis. The earth thickness on top of the extraction region is thinner than in other

30

areas of the PS ring while the beam losses are locally higher. A detailed shielding analysis was therefore performed to specify the additional shielding thickness necessary to extract the beam using the MTE scheme while respecting CERN dose objectives for personnel and the public exposure. The effect of shifting the beam losses to the dummy septum can be seen in Ref. [54] in Figures 27 and 28. Due to the locally confined single source and to the fact that the total radiation flux is conserved, the maximum dose rate associated with the future SS15 layout without local shielding would be a factor of 2 higher than for the present situation. Therefore, in addition to the lateral parts to protect personnel during accesses, the roof part of the local shielding is mandatory during operation, with a thickness optimized to reduce the radiation field at the ground level by a factor of 2. This is then exactly the same maximum value (simulated and measured) during the MTE operation for 1013 p/s with the main source in SS16. The recommendation also holds for the new additional shielding on top of the PS extraction region, which is presently being installed [54]. It is worth noting that all the optimization criteria for the additional shielding on top of the PS ejection region, including the safety margin by a factor of 3, will remain valid if the beam losses would be shifted upstream to the planned locally shielded TPS15.

Figure 32: Geometry of the final local shielding around TPS15 as implemented in the FLUKA simulation.

5.2 Shadowing effect of magnetic septum

The FLUKA studies described before have assessed the performance of the dummy septum by considering that a pencil beam would either hit the dummy septum or the actual septum blade. The success of the shadowing method strongly depends on the geometrical masking of the magnetic septum blade by the dummy septum blade. In particular, it is crucial to assess the level of shadowing under

31

realistic simulation conditions, including, e.g., the sensitivity to TPS15 blade positioning and angle setting, which, by the way, will also provide useful information for the beam commissioning strategy. To this aim, a massive campaign of combined FLUKA and MAD-X/PTC tracking simulations has been launched.

These simulations are based on a realistic beam distribution obtained with measured beam parameters (beamlet emittance and rms momentum deviation) and on the PS nonlinear model. The strategy used to perform these simulations is to split the distribution into two parts: one for the losses created by the extraction of the islands and one for the losses created by the extraction of the core. The situation just prior to extraction features a slow bump at its maximum and the external island is close to the dummy septum blade in SS15 and close to the magnetic septum blade in SS16. Then the fast closed bump is switched on, the outermost island moves towards the external sides of the blades and some fraction of the beam (depending on the ratio between the rise time of the kickers and the period of a full machine turn) interacts with the blades. The same process is repeated when the core is extracted, although the losses are then lower due to the much faster rise time of the extraction kickers (KFA71 and KFA4) compared to the fast closed bump kickers (KFA13 and KFA21) (see Fig. 33).

To generate the realistic beam conditions corresponding to the islands and the core we create a distribution representing the time-cumulated beam distribution as it evolves during the rise time of the fast closed bump, or extraction kick in the case of the core. The time-cumulated distribution is the integral of the beam distribution at the entrance of SS15 during the rise time of the kickers. The distribution of initial conditions for FLUKA is created by summing the distributions of 50 time-slices representing 50 steps of the rise time of the kickers (both for the island and for the core).

The strength of the kickers as a function of time is modelled as in Fig. 33. The rise time (10 %-90 %) is 360 ns for KFA13 and KFA21 (islands fast closed bump) and 70 ns for the extraction kick of the core (KFA4 and KFA71). Based on that model, the time-cumulated distribution for the island is the integral over 609.0 ns and the one for the core is integrated over 118.5 ns.

Figure 33: Strength vs time of the island kickers KFA13 and KFA21 (rise time 10-90 % of 360 ns) and of the core kickers KFA71 and KFA4 (rise time 10-90 % of 70 ns).

32

It follows that the intensity represented by the realistic time-cumulated beam distributions is given by

(1)

where τrev is 2100 ns at extraction. The total intensity considered in these studies is 2.6×1013 protons. These kickers’ strengths were used in MAD-X/PTC simulations together with the full PS model to obtain the islands’ fixed points as a function of the bump evolution and similarly to obtain the core centroid. A linear interpolation is used in between the data points simulated from the model.

In order to improve the statistical significance of the results we used weighted distributions to construct the time-cumulated data. Indeed the distribution corresponding to the initial time step will hardly interact with the dummy septum as only the particles in the tail of the distribution will have a chance to be affected by the dummy septum blade. On the other hand, a large fraction of particles in the distribution corresponding to the time step at which the beam centre is crossing the blade will interact. Therefore, we introduced a parabolic multiplication factor function of the slice time to increase the number of particles and hence the interaction rate of the beam distributions far from the blade. The time-cumulated distribution thus features an artificially increased number of particles close to the positions corresponding to the initial and final fixed points. This effect is then weighted appropriately in all the subsequent processing and taken into account in the error estimates. This procedure allows increasing the number of equivalent events, and thus their statistical significance, by a factor 2.8. Figure 34 shows the factors used to increase the number of events in the distributions as a function of the slice time.

Figure 34: Number of initial conditions for FLUKA as a function of the time-slice for the island (blue) and the core (red). The solid curves represent the multiplication factor applied to the time-slices (dots).

These steps allowed obtaining the realistic beam distributions required as input for the FLUKA tracking. Figure 35 displays the weighted distributions of initial conditions for FLUKA for the island (upper) and the core (lower) during the rise time of the extraction kickers.

33

Figure 35: Input distribution for FLUKA for the island case (upper) and the core (lower). The blade of the dummy septum is represented by the grey band.

The distributions are normalized using Eq. (1) for a total intensity of 2.6×1013 protons, corresponding to 1013 p/s used in the rest of this report. These input distributions have been tracked using the complete FLUKA model of SS15. In total 1.1×106 initial conditions are tracked with FLUKA, for the island and for the core. The tracking is done from the upstream part of SS15 to the downstream one. The position of the blade is varied from its nominal position, at 85.3 mm from the beam axis, towards the inside or the outside of the machine in steps of 1 mm. A scan over the angle is also performed, in steps of 2 mrad. The blade creates a hole in the beam distribution as its length is about 3 nuclear interaction lengths for protons at 14 GeV/c. Figure 36 represents the beam distributions at the downstream end of SS15 (left part), after the FLUKA tracking for the nominal settings of the blade (islands – upper part – core – lower part).

The effect of the dummy septum blade is clearly visible, both for the island and for the core: it creates a hole in the distribution where the density goes to zero. The small phase advance along the drift is also visible, as it tends to shift the distribution towards higher amplitudes. Indeed, the fixed point has a non-

34

zero angle at that location, both for the island and the core. As the blade is long compared to the nuclear interaction length, the hole is already present after a short drift; we thus observe the tendency of the distribution shifting towards the outside of the machine (positive values of X). That effect is more visible for the core, as the angle of the fixed point is larger.

Figure 36: Beam distributions (island – upper – core - lower) downstream of SS15 (left) and upstream of SS16 (right). The hole created by the dummy septum is clearly visible. The blade of the dummy septum is represented by the grey band, while that of the magnetic septum is represented by the red band.

This tracking allows computing the losses induced by the interaction with the dummy septum blade. The associated errors represent a 3-sigma confidence interval taking into account the weighting of the distribution. The losses are expressed in a percentage of the total beam intensity of 2.6×1013 protons.

Figure 37 displays the cumulated losses in SS15 (upper) for the different configurations where the TPS15 blade is shifted in position. One observes that the losses induced by the island (around 0.5 %) are larger than those induced by the core (around 0.06 %) as the rise time of the kickers is shorter for the core by a factor 5. However the losses do not scale linearly with the rise time, because other parameters are of influence, such as the distribution shape, which is different between the core and the island.

The distributions downstream of SS15 have to be propagated through MU15 towards the upstream end of SS16. The FLUKA model does not include the dipolar magnetic field of the main magnet and the associated curvature and higher-order multipolar fields. Therefore, that part of the tracking had been performed using the detailed PS model in MAD-X/PTC. The tracking simulations take into account the limited aperture in MU15 and one obtains additional losses at the end of MU15. These losses come from the scattered particles that reached the downstream end of SS15 but with angles too large to allow them to survive up to the upstream end of SS16. These losses are shown in Fig. 37 (lower). For the nominal

35

position of the dummy septum blade one obtains total losses of around 0.7 % of the total beam intensity in the fraction of PS circumference between SS15 and MU15. These values are compatible with the design assumptions of the dummy septum including also some safety margin.

Figure 37: Losses in SS15 (upper) and MU15 (lower) as a function of the relative dummy septum blade position. The losses induced by the island (blue) are summed with the losses induced by the core (red). These are the total losses (green).

Figure 36 (right part) displays the beam distribution at the upstream end of SS16 for the island distribution (upper) and the core (lower). The blade of the magnetic septum is shown as a red band. It can be observed that, the dynamics of the islands and the core being different, the dip in the beam distribution at the beginning of SS16 is not located at the same position with respect to the magnetic septum blade. Nevertheless, as the contribution of the core losses to the total losses is smaller than the islands’ one, an important gain can still be obtained from the optimised position. Additionally one also observes that, although the dummy septum creates a perfect hole in the distribution, that hole is repopulated as it arrives

36

in SS16. This is due to the non-negligible phase advance from SS15 to SS16. As a consequence, it is not possible to suppress completely the losses at the magnetic septum.

The last step consists in computing the losses at the magnetic septum 16. To do so, a simple model of a 3 mm thick fully absorbing blade is considered. The integral of the properly weighted distribution is computed as a function of the blade position. This is done for the islands and for the core, separately, as well as for the different dummy septum configurations under consideration.

Figure 38 displays the losses at the magnetic septum (percentage of the total beam intensity) as a function of the position of the blade of SMH16. The colours correspond to the different settings of the dummy septum blade. A minimum can clearly be found in the different cases. That minimum is close to 0.1 % of the total beam intensity. This is to be compared with the losses that would happen at the magnetic septum in the absence of the dummy septum. Those losses are shown in purple. Also shown are the cumulated losses at TPS15 and MU15. The minimum is found for the nominal blade position for which the losses in SS16 are 0.13 % of the total beam intensity. Compared to the losses in the absence of the dummy septum (0.47 %), this provides a reduction by a factor 3.6. The losses corresponding to the nominal setting of the dummy septum blade (85.3 mm) shown in light blue in Fig. 38 have their minimum for a magnetic septum position of 57.75 mm. That blade position, very close to the nominal value for the magnetic septum, is an acceptable value for the extraction of the MTE beam, but also for the modified extraction schemes of the other fast extracted beams.

Figure 38: Losses at the magnetic septum 16 as a function of the septum position. Total losses at TPS15 and MU15 are also shown for different dummy septum blade positions. Vertical dashed lines indicate the minima of each configuration. Configurations were the blade is displaced by ±3 mm and ±4 mm are not shown as they have their minima for unachievable positions of the magnetic septum (the values are not lower than the shown configurations).

In order to reduce the losses by a larger factor, additional configurations were considered in which the TPS15 blade was rotated. This has been done for the nominal position of both blades. The TPS15 blade is rotated in the horizontal plane around an axis located at its geometrical centre. A positive angle corresponds to the downstream part of the blade moving towards the outside of the machine. A rotation of

37

the blade allows increasing its apparent thickness, thus creating a larger hole in the distribution. That hole will eventually lower the losses at the magnetic septum.

Figure 39 (left part) displays the beam distribution of the island downstream of SS15 for a blade rotated by +4 mrad. The geometry of the hole can be compared with Fig. 4. As a consequence the density of particles interacting with the magnetic septum is reduced, as is also shown in Fig. 7 (right part).

Figure 39: Beam distributions (island – upper – core - lower) downstream of SS15 (left) and upstream of SS16 (right) for a blade rotated by +4 mrad. The hole created by the dummy septum is clearly visible. The blade of the dummy septum is represented by the grey band, while that of the magnetic septum (in its nominal position) is represented by the red band.

The losses created at the dummy septum for the different configurations of the rotated blade are summarized in Fig. 40 (upper part). For the +4 mrad configuration the losses are increased by 20 %. Figure 40 (lower part) displays the losses arising in MU15 for the different configurations of the rotated blade, compared to the reference case. For the blade rotated by +4mrad this is an increase of 35%. These additional losses can potentially be beneficial if they lead to a decrease of the losses at the magnetic septum.

Figure 41 displays the total losses in SS15+MU15 for the different configurations, as well as the losses at the magnetic septum as a function of the magnetic septum blade position.

At this point of this analysis it is worth showing the expected dose rates due to primary beam interaction with the dummy septum or SMH16 blade. The outcome of these numerical simulations is shown in Fig. 42, separately for the island (left) and the core (right). The selected example demonstrates

38

the two-dimensional projections of the residual dose rates in the z-y plane, at the beam level, after a cooling time of 40 days (z- beam direction, x-vertical and y-lateral direction) due to the interaction with the TPS15 blade (upper part). The radiation field resulting from the secondaries produced was then determined by FLUKA transport through SS16. The detailed geometry of the magnetic septum 16 including a uniform field B = 0.6 T over the magnetic length (~2.2 m seen by the beam) has been included in the FLUKA simulation [38]. The expected dose rates due to primary beam interactions with the blades of the magnetic septum are shown in the figure (lower part), separately for the island (left) and the core (right). This type of analysis is reported in more detail in the next section.

Figure 40: Losses in SS15 (upper) and MU15 (lower) as a function of the dummy septum blade angle.

The interpretation of the efficiency of the dummy septum has to take into account the actual residual ambient dose reduction factor in SS16. Results concerning the percentage of losses in SS15 and SS16 for the different optimized dummy septum settings are used to obtain the ambient dose along the SS16.

39

FLUKA simulations have been performed, using the realistic beam distributions, to evaluate residual ambient dose pattern in SS16 based on four types of beam losses: losses in SS15 due to either the island or the core distributions impacting onto the TPS15 dummy septum, and losses in SS16 for either of the two distributions. The relative losses obtained from the MAD-X/PTC tracking studies discussed before have then been used to weight these four contributions, allowing reconstructing the dose map in SS16.

Figure 41: Losses at the magnetic septum 16 as a function of its position for several values of the TPS15 blade angle. Total losses at SS15 and MU15 are also shown, for the different dummy septum blade positions. Vertical dashed lines indicate the minima of each configuration.

From the dose maps in the whole 3D volume around SS16 the dose has been projected onto a line parallel to the beam pipe, at the beam level, running 10 cm from it, either on the inside, or on the outside of the machine. The results are shown in Fig. 43 (right - internal side – left - external side). In the upper part the residual ambient dose in the absence of the dummy septum is shown, and it serves as a reference for the computation of the reduction factor. It should be noted that the simulated dose pattern does not reproduce the measured one. No attempt to re-scale the simulation results in order to match the measured data has been made. Such a discrepancy has been explained assuming that the measurements were performed in a commissioning period leading to a much less favourable situation in term of losses due to non-perfect machine setting. Already at this level the improvement brought by TPS15 is clearly visible. Nonetheless, the ratio between the doses with and without dummy septum is the final outcome of these studies. The doses and the reduction factors are shown in the lower part of Fig. 43 (right - internal side – left - external side) as a function of the z-position in SS16. Table 6 summarizes the key values of the reduction factors for the different cases.

The best blade configuration (+4 mrad) shows an average reduction factor of 7.0 on the external side of the ring, and of 9.1 on the internal side. It is worth stressing that the reduction factor is computed with respect to the simulated level of losses without the TPS15. Therefore, the achieved results reported in Table 6 seem rather encouraging.

The analysis performed so far allows quantifying the improvement of the beam losses and radiation doses with and without TPS15 for the MTE beams. Of course, it is also interesting to compare the future

40

situation with respect to the old and current situation of PS operating with CT beams. As anticipated in Ref. [1], the MTE should be superior in terms of losses around the PS ring. This has been confirmed by measurements of beam losses taken during typical PS run with CT and MTE beams. The outcome of such measurements in shown in Fig. 44 (left part) where the beam losses over the straight section of the PS ring are compared between CT and MTE: the improvement is clearly seen, with losses well-confined to the extraction region for the MTE beams, unlike the CT ones, where losses are rather uniformly spread around the whole machine circumference.

Figure 42: Expected residual ambient dose equivalent rates after a cooling time of 40 days due to a fraction of the primary beam (island – left - core - right) interacting with the dummy septum blade (upper) and with the blade of the magnetic septum (lower).

41

Figure 43: Upper plots: Residual ambient dose rate in SS16 (10 cm from the beam pipe at the beam level) after 40 days of cooling time. The simulated dose in the absence of the dummy septum (red markers) is shown together with the dose for the two optimized settings of the dummy septum blade (green and blue markers). Lower plots: Corresponding reduction factors (ratio between the dose without and with dummy septum) are shown for the two configurations. The region inside and outside the PS ring is shown in the left and right plots, respectively.

Table 6: Reduction factors for two configurations of the dummy septum blade and for the external and internal side of the PS ring. Minimum, maximum, and average values over the z-coordinate are given.

Internal side External side Blade 0 mrad Blade +4 mrad Blade 0 mrad Blade +4 mrad

Mean 6.6 9.1 5.6 7.0 Max 7.0 9.8 6.1 7.8 Min 5.8 7.4 4.4 4.9

In the left part of Fig. 44, a zoom around SS16 is shown, where the residual doses are reported, based on actual measurements for CT, and on numerical simulations for MTE. The cooling time in this case is only two weeks and the distance from the beam line in the horizontal plane of the measurement points is 40 cm (in previous plots it was 10 cm). Also in this case, the improvement for MTE beams, with or without TPS15, with respect to CT beams is clearly visible.

42

Figure 44: Comparison of losses along the PS ring for MTE and CT. In the left part, the beam loss monitors readings are shown for CT and MTE, while in the right part, the radiation dose in SS16 for CT and MTE are compared. The improvement is clearly seen.

In summary, the solution based on TPS15 has enabled reaching the original goal of these studies. With the optimized configuration of the dummy septum blade the radiation field and resulting activation in the whole environment of the magnetic septum 16 associated with the future MTE operation will be below the present level associated with the CT operation

At the end of this section it is worth stressing that the simulations with realistic beam distributions are also useful for the commissioning with beam as they indicate an optimization strategy. Indeed a linear relationship with a coefficient close to 1 is found between the position of the dummy septum blade and the optimized position of the magnetic septum. This is shown in Figure 45. Additionally it is found that the values of the minimum losses at SMH16 have very little dependence on the TPS15 position. This allows removing one free parameter in the beam-based optimization, while focusing on angular scans of the TPS15 blade.

Figure 45: Optimised position of the magnetic septum blade as a function of the position of the dummy septum blade. A linear fit is also shown.

43

5.3 Dose delivered to personnel in case of interventions

Although the dummy septum has been designed with reliability as paramount criterion, the possibility of failure and the need for a human intervention for repair cannot be excluded and have been taken into account in the design phase. The possible intervention scenarios are listed in Table 1 together with the consequences on operation. Simulations of the residual dose rates in the immediate vicinity of the dummy septum have been used to estimate the doses for all possible failure scenarios as well as for annual preventive maintenance and vacuum testing. The shielding has been optimized to facilitate the access to some areas of the dummy septum such as the cooling and beam observation system. The possibility to replace sensitive equipment as the camera without the removal of shielding blocks has been considered in the design and integration study. For the other actions some parts of the shielding blocks have to be removed to allow access to the concerned equipment. However, the shielding design has been carefully optimized (see Figure 29) such that the remaining parts would still provide efficient shielding during these interventions. A comprehensive Work Dose Planning has been compiled and is shortly summarized in this chapter. It shall be used in conjunction with the established procedure for any intervention which would require a Work and Dose Planning to be established and approved by the concerned safety officer considering details of the intervention and the actual radiological conditions.

Radiation protection is based on three general principles: justification, optimization, and limitation of potential human exposure to ionising radiation [55]. The principle of optimization consists of finding a right balance to minimize the individual doses and the number of persons exposed, having taken into account economic and social factors. To be sure that significant effort will be invested on optimization, a CERN design criterion of 2 mSv per person per intervention is introduced, and optimization of interventions must be taken into account in the design phase to reduce personnel exposure.

FLUKA simulations of three-dimensional residual effective dose rate maps for different cooling times allow estimating and optimizing the individual and collective doses for possible intervention processes. The two-dimensional projections of residual effective dose rates in the x-y plane averaged over a length of Δ z=110 cm along SS15, in the form of colour contour plots, are shown in Fig. 46.

The coordinate system used is such that z represents the beam direction, x the vertical and y the radial direction. To cover all possible cases, spanning from a short intervention occurring during the operation period to a long one as part of the annual shutdown, different cooling periods corresponding to radioactive decays between 1 hour and 4 months were considered in this analysis. For instance, Fig. 46 shows the residual dose rates for cooling times of 1 hour (top plot), 3 days (middle plot) and 1 month (bottom plot), respectively. A conservative approach has been used assuming 100 % shadowing efficiency for TPS15, i.e. 1 % of the primary beam intensity of 1013 p/s lost within SS15. All the results in this section are based on tracking of the realistic beam.

Figure 47 shows the one-dimensional projections of the residual effective dose rates along the lateral y-direction, at the beam level, averaged over a length of Δ z=110 cm along the SS15. The vertical dashed lines indicate the lateral positions of the shielding walls and the septum tank. It is clearly seen that outside the local shielding the residual effective dose rates after a cooling period of 1 hour are on average 0.7 mSv/h and by factors of about 5 and 10 smaller for the cooling periods of 3 days and 1 month, respectively.

44

Figure 46: Two-dimensional projections of residual effective dose rates in the x-y plane averaged over a length of Δ z=110 cm along SS15. The selected examples refer to cooling times of 1 hour, 3 days and 1 month.

Figure 47: One-dimensional projections of the residual effective dose rates shown in Fig. 46 along the radial y-direction at the beam level, averaged over a length of Δ z=110 cm along SS15.

45

Table 7: Detailed information concerning the activities required to perform a replacement or removal of the complete TPS15.

In order to make an estimate of the collective and individual doses associated with the different interventions, in addition to the calculated effective dose rate, a detailed description of the work steps to be accomplished with the associated duration, the location, the number of persons involved as well as the frequency is necessary. An example of such a detailed list for the first three possible interventions indicated in Table 1, classified as serious and for which the replacement or removal of the complete TPS15 is required, is shown in Table 7. Part of the local shielding blocks, the top and the two side blocks, have to be dismantled (the local shielding without these blocks is shown in the two figures of Table 7). Quite some effort has been spent to design the shielding according to the ALARA principle. For example, the final shielding layout has been optimized such that the two flanges (position indicated in the lower figure of Table 7) could be removed with the shielding in place. This will reduce the dose delivered to personnel executing this work by a factor of two. For the other activities, for which dismantling of part of the shielding is required, the remaining blocks have a minimum height of 1.4 m, providing effective protection from exposure during interventions. The TPS15 has also been designed according to the ALARA principle, with a possibility of rapid exchange due to the plug and play support system.

As already mentioned, one way to assess the values of accumulated individual and collective doses for each work step during the replacement of the complete TPS15 system is to use the three-dimensional

46

residual effective dose rate maps calculated with FLUKA. However, in this particular case quite some computation is needed, since these matrixes should be calculated for each cooling time considered and in the following geometrical configurations: i) local shielding in place; ii) top part of the shielding removed; iii) top and two side shielding blocks removed; iv) dummy septum removed. For each of the four situations, examples after a cooling time of 3 days of the two-dimensional projections of residual effective dose rates in the x-y plane, averaged over the length of Δ z=110 cm along SS15, are shown in Fig. 48.

Figure 48: Two-dimensional projections of residual effective dose rates in the x-y plane averaged over the length of Δ z=110 cm along SS15 for 4 different options for a cooling period of 3 days, namely: i) local shielding in place; ii) top part of the shielding removed; iii) top and two side blocks removed; iv) dummy septum removed.

Based on all these dose rate maps, information on the work procedure, the respective location and duration, the individual and collective doses accumulated during the exchange of the complete TPS15 system, can now be computed. Table 8 shows the results for each work step and for a cooling period of 1 hour. Similar tables are computed for nine different cooling periods between 1 hour and 4 months and can be found in [53]. Final results on individual and collective doses as accumulated during the replacement of the complete TPS15 system are summarized in Table 9. All results are compared to the design constraint of 2 mSv/person/intervention, with doses exceeding this value marked in red. A minimum waiting time of 8 hours is required to assure that no person would reach the intervention limit of 2 mSv. In order to calculate the collective dose for the entire intervention, all individual doses have to be summed up. The collective doses for the replacement of the complete TPS15 system (marked in yellow in Table 9) are expected to be within 1.5 and 6.5 mSv for a cooling time between 1 hour and 3 days. The content of Table 9 is also shown in graphical form in Fig. 49.

CERN introduced a formalized approach to ALARA, where an intervention is classified according to three ALARA levels. Level 3 intervention requires formal approval from the ALARA Committee. The

47

classification thresholds for level 3 are 1 mSv for the individual and 5 mSv for the collective dose [56]. The individual and collective dose values above level 3 threshold (and below the design limit of 2 mSv per intervention) are indicated in darker green and orange in Table 9. As visible from the table and from Fig. 49, even though the individual doses stay below the design constraint of 2 mSv per person and intervention after a cooling time of 8 hours, a minimum waiting time of at least 1-2 days will have to be considered to avoid the ALARA committee.

Table 8: Accumulated dose per person for each work step required to replace TPS15 in the case of cooling time of 1 hour.

Possible interventions in connection with beam instrumentation are linked with less severe failures and might concern diagnostic of faults, repair with medium risk classification, or maintenance procedure, e.g., for camera, light or filter wheel exchange, all with low risk classification. The beam instrumentation equipment is placed in the less shielded side of the TPS15 assembly, facing towards the external side of PS ring, i.e., the side closer to the dummy septum blade. Nevertheless, access to equipment is relatively easy as shown in Fig. 50. If repair in-situ is chosen, then dismantling of the shielding is not necessary. The camera can also be exchanged without removal of the local shielding blocks. The procedure to evaluate the individual and collective doses for these interventions is exactly the same as described previously. Only the final results are reported here, but all individual steps of the analysis, including the detailed information on the work procedure, the location, and the duration, can be found in [53].

Final results on individual and collective doses as accumulated during the beam instrumentation activities are summarized in Table 10. The same colour code as for Table 9 is used. Two beam instrumentation specialists are involved, one to establish the diagnostics and the other one to perform the repair. The results show that a minimum waiting time of 1 day is required to assure that no person would

48

reach the intervention limit of 2 mSv. The collective dose is expected to be within 2.3 and 5.3 mSv for a cooling time between 1 hour and 3 days. However, to lower the personnel exposure to the one corresponding to level 2 interventions, a longer waiting time between 1 and 2 weeks has to be envisaged. A possible solution would be to add one more person to the team working on the beam instrumentation. This would reduce the minimum waiting time to 3 days and the intervention could be classified as level 2 based on the ALARA criteria. Of course, the collective dose would not significantly change as it would remain of the order of 2.3 mSv.

Table 9: Dose per person and intervention as accumulated during the replacement of the complete TPS15 system for a cooling period between 1 hour and 4 months.

Figure 49: Individual (left) and collective (right) doses as accumulated during the replacement of the complete TPS15 system for cooling time between 1 hour and 4 months.

49

The individual and collective doses as accumulated during the maintenance procedure for the camera, light and filter wheel exchange are summarized in Table 11. The results are evaluated for the three different options of camera exchange defined in [53]. For each of the options a rapid exchange is possible.

Figure 50: Position of a worker performing interventions on beam instrumentation equipment. Part of the magnetic unit 14, the TPS15 tank assembly, and the BTV are visible. The local shielding is in place.

6. Changes to the extraction schemes from the PS ring

6.1 Fast extractions

New extraction schemes for the fast extracted PS beams were designed and successfully tested, aiming at reducing the horizontal displacement in SS15 while providing a clean extraction process [57-60]. This is necessary as the usual PS fast extractions for the LHC, AD and nTOF beams would lead to extraction trajectories incompatible with the presence of the dummy septum blade. Additional constraints were also imposed in terms of extraction losses and extraction conditions (horizontal position/angle) for the LHC-type beams.

The new schemes exploit all the existing hardware, excluding the elements used for the CT extraction, i.e. the BFA21, the BFA9-staircase and the electrostatic septum. Figure 51 displays the elements used for the fast extractions. The unmodified scheme is the same for all beams and consists of the classical combination of a slow quasi-closed bump (generated by 4 bumper magnets located in SS12, 14, 20 and 22 powered with the same currents) and a fast kicker, located in SS71/794. The new schemes differ from the original one by two key ingredients: they exploit the independent powering of the 4 bumpers and they use 3 additional kickers located in SS4, 21 and 13 and originally installed only for MTE. Unbalancing the kicks of the bumpers 12/20 and 14/22 allows to increase the angle of the closed bump in SS15 while maintaining a constant position in SS16. New slow closed bumps have thus been implemented for each scheme with the additional advantage of a better bump closure. Using additional fast kickers allows benefiting from the phase advance relationships between these kickers and SS15 and 16 as summarised in Table 12.

4 Kicker 71/79 consists of two groups of magnets. The phase advance between them is 180° and they can be considered as a single kicker. The fine delay between the two kicks is always taken into account.

50

Table 10: Dose per person and intervention accumulated during intervention on beam instrumentation for which a repair in-situ would be possible.

Table 11: Dose per person and intervention accumulated during the maintenance of the camera, light and filter wheel exchange.

As seen from Table 11, a minimum waiting time of 8 hours is required to assure that the beam instrumentation specialist performs this maintenance without reaching the intervention limit of 2 mSv. Therefore, to reduce the personnel exposure to the one compatible with level 2 interventions, a waiting time of 1 or 3 days would be required depending on the option selected for exchanging the camera. Therefore, in case of interventions not requiring the replacement of the TPS15, a cooling time between 1-3 days seems appropriate and compatible with a level 2 ALARA interventions.

One can see that a positive kick with KFA71 leads to a larger displacement in SS15 than in SS16. On the other hand KFA13 and KFA4 will have their kick translated into a larger displacement in SS16 than in SS15. Therefore reducing the kick from KFA71 and using additional kicks from KFA4/13 will help reducing the displacement in SS15 while keeping it constant in SS16. KFA71 is still needed, as the two other kickers do not have enough strength. The use of KFA21 follows the same idea, although, as its polarity is inverted it will actually reduce the excursion in both sections, more in SS15 than in SS16. The new schemes thus use KFA21 in combination with increased kicks from the other 3 kickers.

51

Although the use of additional kickers can pose problems in term of reliability, it should be noted that the kickers KFA4, 13 and 21 are mandatory for the extraction of the MTE beams. Therefore, a failure of one of these kickers would stop the MTE beam operation and would allow moving the TPS15 blade to its park position, thus reverting to the usual extraction scheme.

Figure 51: PS extraction elements and schematics of the new nTOF extraction scheme. The slow bump magnets are shown in red, the fast kickers in dark blue, while the MTE kickers (five-turn long pulses) are depicted in light blue.

Table 12: Horizontal phase advance relationships

KFA21 KFA71 KFA4 KFA13

SS15 313° 270° 248° 45°

SS16 331° 293° 271° 68°

The installation of the TPS15 imposes the removal of a dipole magnet used for high-energy orbit distortion correction. The old scheme was using two dipole magnets, namely one in SS15, which had to be removed in order to make room to the TPS15, and another one in SS60. An extensive simulation campaign has been performed and a new set of orbit correctors was defined. It consists of three dipoles in SS5, 18 and 60, the first one being already installed in the PS ring. The implementation of the new extraction scheme has been tested with that orbit correction in place.

The results of the extensive measurement campaign performed for the various beam types are reported and discussed in the following sections.

6.1.1 LHC-type beams

The variants of the LHC type beams (various longitudinal structures, transverse emittances and intensities) are accelerated to 26 GeV/c and fast extracted. The extraction beam line which connects the

52

ring to TT2 is located at about the middle of the MU16, and the impact of stray fields on beam quality is non-negligible. The reason is that the optics parameter at the entrance of TT2 strongly depends on the beam trajectory through the stray field itself. For this reason, we imposed the new extraction scheme for LHC to recover the same extraction conditions as the old one. The modifications to the extraction scheme of the LHC beams are twofold: a modification of the high energy orbit correction due to the change of corrector magnets and the change in the extraction itself. The new orbit correctors proved to be very effective. The rms closed orbit is reduced by a factor of 2, performing better than the old one (see Fig. 52).

Figure 52: Measured closed orbit at 26 GeV/c for LHC beams. The uncorrected orbit is shown in green; the orbit resulting from the old correction scheme is in blue; the one resulting from the proposed correction scheme is displayed in red. Both the rms and the peak-to-peak value have been improved.

The new scheme uses 4 bumper magnets with independent current and keeps the usual kicker KFA71 for the fast extraction. The use of the additional kickers is not needed for the LHC beam as the available aperture in SS15 in the presence of the dummy septum blade is large enough. Figure 53 shows the closed extraction bumps when the closed orbit is subtracted. The two bumps have different positions in SS15 and SS17 but the same position and angle in SS16. The new bump also features a closure 60±13 % better. The results show a slight increase of the transverse position of the extraction trajectory in SS15, from 71.4±0.7 mm to 72±1 mm.

The beam has been successfully injected in the SPS and scraping measurements were performed to assess the transverse beam quality. The measurements did not reveal significant differences between the two beams and it was concluded that the new scheme does not induce emittance growth or tails, as depicted in Fig. 54, where the comparison between scraping measurements done for the old and new schemes in the SPS is shown. Similar results have been obtained for the LHC Pb54+ ion beam. Additionally a first version of the new scheme has been used for the physics fills 3377 and 3378 of the LHC. The bump was not fully optimized in terms of the horizontal position of the trajectory in SS15 but the same procedure was applied to recover the same extraction conditions as for the old scheme. The new bump was transparent to the performance of the LHC beam.

53

Figure 53: New and old slow extraction bumps for the LHC beams at 26 GeV/c.

Figure 54: Scraping measurement in the SPS.

6.1.2 nTOF beams

The beam sent to the nTOF experiment is a single high intensity (around 750×1010 p) bunch undergoing a bunch rotation a few milliseconds prior to extraction. Two main issues complicate the extraction of that beam: a large transverse emittance and a large ∆p/p value (induced by the longitudinal bunch rotation) which, due to the non-zero dispersion in the extraction region, also leads to a larger beam size. Moreover the dispersion function is about 40% larger in SS15 than in SS16. This makes the beam even larger and reduces the aperture margin in SS15. In addition the nTOF beam exists also in the form of a so-called parasitic beam: a two-bunch beam is accelerated to 20 GeV/c, one bunch is fast extracted while the other one is kept in the machine and further accelerated to 24 GeV/c and slow extracted in towards the EAST area in the SS61. The intensity of the parasitic nTOF bunch is limited to 400×1010 p, implying a smaller transverse size.

54

The new hardware was used for the orbit correction at 20 GeV/c (nominal beam) and led to a reduction of the RMS orbit distortion of 20%. Secondly a new closed bump was obtained, allowing a reduction of the closed bump in SS15 by 10±1 mm while being acceptable in SS16. A reduction of the horizontal position of the extraction trajectory was first obtained with the additional kickers KFA4 and KFA13 but it was soon realized that this was not enough to overcome completely the aperture restriction.

A final scheme using the kickers 71, 4, 13 and 21 was designed. It is depicted in Fig. 51. On top of the slow closed bump the bunch is kicked 5 times by the different kickers. The strength of these kicks is compatible with the aperture in the rest of the machine and it leads to a drastic improvement of the situation in SS15, as shown in Table 13.

The situation for the parasitic nTOF beam is more complex. Indeed, the kickers KFA13-21 generate a pulse corresponding to 5 machine turns thus also kicking the non-extracted bunch for 5 turns. It is possible to obtain a fast closed bump between SS13 and 21. In the new scheme, the parasitic nTOF bunch experiences the same extraction trajectory as the nominal nTOF bunch, while the other bunch sees a fast closed bump only. We successfully assessed that the additional manipulation of the second bunch does not induce any emittance growth. The results are summarized in Table 13.

Table 13: Improvements in SS15 of the new nTOF schemes

KFA71 KFA71, 04, 13 KFA71, 04, 13, 21

Nominal 90.3±0.7 mm 73.2±0.8 mm 65.1±0.9 mm

Parasitic 70.6±0.4 mm NA 64.7±0.1 mm

The new schemes for the nTOF beams allow overcoming the aperture restriction with a gain in horizontal displacement of the extraction trajectory for the nominal beam (parasitic) of 25±1 mm (5.9±0.4 mm). It also induced a reduction of the beam losses at extraction as measured by the Beam Loss Monitors (BLMs). This is clearly seen in Fig. 55 where the signals provided by the BLMs at PS extraction for the old and new scheme are shown. In the ring, we observed a reduction of a factor 1.5 in SS9, which is one of the less shielded sections of the machine, and a reduction in four sections downstream of the extraction septum, indicating a cleaner extraction. No significant difference could be found in the other parts of the ring. The losses at the beginning of the transfer line could also be optimized. The new extraction scheme was successfully used in operation for physics from the December 12 until December 16 of the 2012 run.

6.1.3 AD beam

The AD beam consists of 4 bunches of total intensity around 1500×1010 p extracted at 26 GeV/c. A bunch rotation is performed prior to extraction, leading to an increase of ∆p/p and thus of the transverse beam size due to the non-zero dispersion in the extraction region as for the nTOF beam. The original extraction scheme is not compatible with the aperture restriction of the dummy septum. Results with beam showed that the optimum scheme comprises the kickers KFA71, KFA13 and KFA21. A new closed orbit correction and slow closed bumps have been put in place, prior to the setting up of the extraction

55

kickers. The results show a reduction of the transverse position of the extraction trajectory in SS15 of 11.5±1.5 mm, for a position of the new scheme of 68.7±0.7 mm. The beam with the new settings was successfully sent to the target for antiprotons production on December 15 and 16 during the 2012 run.

Figure 55: BLMs readings for the nominal nTOF beam. A reduction of the losses is visible in SS9 and in the SS’s downstream of the magnetic septum 16, indicating a cleaner extraction.

6.1.4 Dummy septum aperture in SS15

The horizontal position of the TPS15 blade is imposed by the optics between SS15 and 16. The maximum shadowing efficiency has been found for a blade located at 85.3±0.5 mm. The dummy septum aperture is shown in Fig. 56 with the extraction beam envelopes, including the extracted MTE islands. The extraction trajectories are compatible with the position of the blade. The beam sizes assume Gaussian distributions and assume 3 σ (betatronic) and 2 σ (synchrotron) standard deviations. Table 14 displays the beam sizes along with the distance between the edge of the envelope and the blade for the different beams. The latter is a measure of the available aperture margin.

Figure 56: Dummy septum aperture and beam envelopes. Vertical positions of the AD and nTOF beams are artificially shifted by ±15mm for clarity.

56

6.2 Continuous Transfer

Although the MTE extraction is foreseen to replace the old CT transfer, the existing CT-specific hardware is still in place in the PS ring and is kept operational at least for the few years to come. The two extraction methods have been simultaneously used, i.e., in the same super-cycle, prior to the installation of the dummy septum. The design of the dummy septum maintains this mode of simultaneous operation available, which explains the need to verify in detail the compatibility between the two types of operations.

Table 14: Horizontal beam size and distance from the blade for the main fast extracted beams from the PS ring.

LHC nTOF

(nominal) AD

Beam radius [mm] 6.8±0.3 14.7±0.6 11.4±0.3

Distance [mm] 8±1 20±1 16.6±0.9

The CT extraction proceeds in three successive phases: first a slow closed bump is established around SS16 and SS31. Then a five-turn long fast closed bump is turned on around the SS31. That bump is closed between SS21 and 9. An additional kick provided by the staircase kickers, varying on a turn-by-turn basis, displaces a fraction of the beam on the other side of the foil of the electrostatic septum 31 (SEH31) thus producing the beam slicing. After having received a kick from SEH31, the slice oscillates around the deformed closed orbit and is extracted in SS16 thanks to the magnetic septum SMH16. For the operation with the dummy septum that scheme is unchanged. The beam stays on the inner side of the TPS15 blade during the slow and fast bump while the extracted slice is pushed to the outer side of the blade by the kick imparted by the SEH31.

The normalised emittance of the circulating beam is 10 µm and the 1-σ relative energy spread is 3×10-

4. Figure 57 (upper left) displays the beam envelope (3 σ betatronic and 2 σ longitudinal) around the closed orbit at the maximum of the slow closed bump (blue) and at the maximum of the fast closed bump as generated by the pedestal kickers (red). One can observe that the bump is rather properly closed between the SS9 and 21. The trajectory and the envelope of the extracted slices are shown in green in Fig. 57 (upper right), while in the lower part of the same figure a zoom around the extraction region is reported. The dummy septum blade is represented by a black rectangle. The extracted slice has a trajectory that passes on the outer side of the TPS15 blade.

To analyse in more detail the available aperture in SS15 one has to consider different scenarios of the slicing process. Indeed, the beam can be sliced such that the results parts are equal in terms of either intensity or emittance5. When the emittances are equalized each slice has an emittance corresponding to 30 % of that of the circulating beam. On the other hand, if the intensities are made equal, then the slice

5 Of course, in practice, only the first approach can be applied as by no means the emittances of individual slices can be measured.

57

with the largest emittance has 40 % of the initial emittance. Additionally the horizontal beta-functions of the slices are not equal to those of the initial beam [1]. The ratio between the beta-function of the slice and that of the circulating beam can be as larger as a factor of 2.2. For our considerations, an approach based on the worst case scenario has been considered. Indeed, the fast bump is not perfectly closed around the ring. Furthermore, the average position and angle of the slices at the location of the electrostatic septum is not the same [1]. This means that the position and size of the slices at the location of TPS15 is changing on a turn-by-turn basis. Therefore, the set of different conditions has been used to define a resulting envelope of beam sizes at the location of TPS15

Figure 57: Upper left plot: Available aperture for the CT beam for the maximum amplitude of the slow closed bump (blue) and fast closed bump (red). Upper right plot: the extracted slice is shown in green. Lower plot: Zoom of the aperture situation around the extraction region. The dummy septum blade is represented by a black rectangle. The detailed aperture restriction of SEH31 is not represented.

Figure 58 displays the aperture at the entrance flange of TPS15. It considers the most pessimistic slice in terms of beam envelope, the case of equal emittances, where the emittance of the slice is 30 % of the initial beam emittance and where the beta function ratio is 2.2. The beam stays about 7 mm clear of the blade, which represent 39 % of the full width of the slice.

Therefore, operation of the CT extraction with the dummy septum blade in its nominal position is possible without any modification of the CT scheme. The circulating beam and the extracted slices are not affected by the aperture restriction and their trajectories avoid the blade with a comfortable margin.

58

Figure 58: Aperture at the entrance flange of TPS15. The beam envelopes corresponding to the maximum of the fast closed bump (red) and to the extracted slice with the largest envelope (green) are shown.

6.3 Multi-turn extraction

The MTE operation with the dummy septum requires modifications to the extraction conditions. Indeed, the aperture restriction induced by the dummy septum imposes tighter conditions on the extractions bumps, slow and fast. By means of a closed slow bump the external island is first brought close to the internal side of the dummy septum and to the magnetic septum blades. Then, using a fast closed bump, the island is made to jump on the external side of both blades. Finally, for the fifth extracted turn, the core is extracted using additional deflections imparted by two more kicker magnets.

In order to optimize these bumps, the horizontal tune was chosen to provide an appropriate separation between the external island and the beam core. The strengths of the non-linear magnets (sextupoles and octupoles) are kept at their nominal values. However, additional sextupole magnets, to be installed in SS60 and SS94, are used to optimize the phase of the islands at SMH16. The optimized current has been found to be -100 A. With this configuration the chromaticity of the core and of the islands will change with respect to the values corresponding to the nominal settings for the beam splitting. However, this is considered to be going in the right direction as, e.g., for the islands the chromaticity is moving towards zero and for the beam core it remains small. In any case, numerical simulations will be performed to check this point. Figure 59 displays the corresponding islands positions and envelopes (3 σ betatronic, 2 σ longitudinal), along with the machine aperture.

The closed slow bump can use up to five different dedicated bumper magnets controlled with independent currents. An additional dipole corrector magnet had to be used in order to fulfill these conditions. The dipole installed in SS5 is customarily used for correcting the distortion of the closed-orbit at high-energy, but in this framework it is used as part of the extraction bump, which is therefore closed between SS5 and SS22. The constrains imposed on the bump are the overall closure in SS22 and its position and angle at TPS15. Of course, as there is no bumper magnet between TPS15 and SMH16, the position and angle at SMH16 are a direct consequence of the position and angle at TPS15. Therefore the angle at TPS15 has been optimized with a trade-off between the position at TPS15 and SMH16 and the angle at SMH16. Figure 60 displays the slow closed bump for the islands, along with their horizontal

59

envelopes (3 σ betatronic, 2 σ longitudinal). The dummy septum blade is shown as a black rectangle. The slow bump is optimized in a way that the external island is close to the dummy septum blade as well as close to the SMH16 blade, on the inner side of the machine.

Figure 59: MTE islands trajectories and envelopes prior to extraction. The upper plot displays the whole ring circumference. The lower plot shows a zoom around the extraction region.

Table 15 summarizes the values of the bumper magnets used to create the slow bump.

Just prior to extraction, a five-turn long closed bump is created around SS9 and SS21, on top of the slow bump. That bump is generated by three kicker magnets: two are the MTE-dedicated kickers KFA13 and KFA21 and a third one, originally used for the CT extraction, BFA9, is used to help closing the bump properly. It should be noted that this bump is closed for the core of the beam, only, while it is not perfectly closed for the islands. Over the four first turns the islands will not have the exact same position in SS15 or SS16. Nonetheless this effect has been taken into account using tracking simulations (see later). Figure 61 displays the beam trajectories and envelopes at the maximum of the fast-closed bump. One can observe that the external island jumped from the internal side of the dummy septum blade to the external side.

60

Table 15: Magnet strengths used for the generation of the slow bump.

Magnet Current [A]

PR.DHZ05 119.9

PE.BSP12 367.4

PE.BSP14 0

PR.DHZ18 101.5

PE.BSP20 107.7

PE.BSP22 45.8

Figure 60: MTE islands trajectories and envelopes at the maximum of the slow closed bump. The upper plot displays the whole ring circumference. The lower plot shows a zoom around the extraction region.

61

The core of the beam is extracted during the fifth turn using additional kicks imparted by the kickers KFA71 and KFA4. Their effect on the kicked beam core is shown in Fig. 62. It is worth noting that the KFA4, initially installed to provide simply more flexibility in the extraction of the core, has a crucial role now as it allows optimising the position of the kicked beam in SS15 and SS16 for both aperture and extraction efficiency.

Figure 61: MTE islands trajectories and envelopes at the maximum of the fast closed bump. The upper plot displays the whole machine circumference. The lower plot shows a zoom around the extraction region.

The beam envelopes and the available aperture at the different stages of the beam extraction, i.e., with the slow bump, with the fast bump, are analysed in more detail for SS15 and SS16. Figure 63 displays the cross section of the vacuum chambers at SS15 (left column) and SS16 (right column), including also the beam positions and envelopes before and during the extraction process. The situation for the first four extracted turns, corresponding to the four beamlets, is shown in the upper row, while the beam core is represented in the lower row. It is worth noting that the vertical shift introduced in the upper row is artificial and should simply improve the readability of the plot. The vacuum chambers drawn in these figures are those of the upstream magnet, not the ones of the vacuum tank of TPS15 (resp. SMH16). In

62

any case it is apparent that the vertical aperture is not a concern here. The slow bump is optimized in the sense that its maximum corresponds to the external island being very close to the inner side of the blades of TPS15 and SMH16. The fast bump allows jumping towards the external side of both blades, with additional margin at the level of SMH16.

Figure 62: trajectory and envelope of the core at extraction during the fifth turn.

Figure 63: Aperture and beam envelopes in SS15 (left column) and SS16 (right column). The circulating islands are shown in blue including the effect of the slow bump, while the extracted islands are shown in red and are visible in the upper row (the vertical shift of the beam envelops is artificially introduced to improve the readability of the plot). The extracted core is shown in green and is visible in the lower row.

63

Table 16 summarizes the strength of the different elements for the magnets used to create the bumps for the first four turns as well as for the fifth one.

Table 16: Kicker strengths used for generating the fast bumps.

Magnet Voltage [kV]

PE.BFA9 6.0

PE.KFA13 64.8

PE.KFA21 68.6

PE.KFA71 579.8

PE.KFA4 78.0

Following the detailed optimisation of the fifth extracted turn, the estimate about the actual beam losses as given earlier should be reviewed in the direction of an additional reduction. In fact, it has been checked that the further optimisation should imply a reduction by 7 % of the estimated losses for the beam core on the SMH16. This is a small gain, but nevertheless a gain.

6.4 Summary of PS ring hardware changes

For the sake of clarity, the complete list of hardware changes to the PS ring entailed by the implementation of the dummy septum solution will be given in the following. Indeed, as it is customary in the PS ring, any change in the machine layout has a far-reaching knock on effect, which, sometimes, makes it difficult to perform the required changes.

• The SS15 has been emptied in order to make room to the TPS15. The quadrupole for the gamma jump transition has been moved to SS99, while the dipole used in the slow bump of the fast extracted beams and for the correction of the orbit distortion at high energy should be removed.

• The correction of the closed orbit distortion at high energy will be performed using three dipoles in SS5, 18, 60. The first one has been installed for the beam tests performed during the 2012-2013 physics run.

• The dipole in SS18 will have a double role, namely the correction of the closed orbit distortion and the generation of the slow bump for the MTE beam. This implies the need of a programmable power converter capable of generating both a trapezoidal function (distortion of high energy closed orbit) and a sine-like function (slow bump).

• In terms of power converters these are the plans: o The programmable power converter connected to the dipole in SS15 will be used to

power the dipole in SS5 during the 2014 physics run. In parallel, a power converter of type S250 will be built to be installed in Bldg. 355 and finally connected to the dipole in SS5 for early 2015. The programmable power converter will then be moved to Bldg. 365 and it will become a spare for the other power converters already located there.

64

o The spare of the programmable power converters installed in Bldg. 365 will be used to power the dipole in SS18. To note that the programmable power converters in Bldg. 365 are used for MTE (sextupoles and octupoles) and slow extraction (quadrupoles, sextupoles, and slow bumps). In case of problems with one of the operational converters, one used for the MTE sextupoles could be used, without any severe impact on MTE operation. When the programmable power converter from Bldg. 355 will be moved to Bldg. 365 the situation of spares will be restored.

o The capacitor discharge power converter connected to the dipole in SS18 till 2013 will be disconnected and used as a spare for the power converters of the slow bump.

It is also worth recalling that in the framework of LIU studies aimed at studying the compensation of the space charge resonances using octupoles new sextupole and octupoles will be installed in the PS ring. These new elements will also be beneficial for an improved control of the PS generalised working point, i.e., linear and higher order chromaticities, for the MTE beam [61].

7. Commissioning strategy for the dummy septum

The commissioning of the new trajectories for the fast extracted beams, also including the dummy septum, requires the availability of the main instrumentation, in particular the ring BCT, the orbit system, the BLMs (including the fast signals for PE.BLM15 and PE.BLM16 or equivalently of the PE.FBLM15 and PE.FBLM16 connected to OASIS), the MTV in the magnetic extraction septum SMH16 (PE.MTV16) and the BTV of the TPS15 (PE.MTV15). Of primary importance is the evaluation of the trajectories in TT2, which will be achieved with the SPS-type BPMs and the new F16.UDC106, and of the extraction efficiency, estimated via the several BCTs available.

The beam profiles will be measured by both the TT2 OTRs (F16.MTV201, F16.MTV218, F16.MTV229, and F16.MTV241) and the SEM grids or wires depending on the beam size (F16.MSG257/258, F16.MSG267/268, F16.MSG277/278). While the logging of the TT2 pickups will be available in TIMBER, it is not clear yet whether the logging of the OTR images will be possible.

Concerning the extraction elements, all the kickers should be available since the first day of the run, i.e., PE.KFA13, PE.KFA21, PE.KFA4, PE.BFA9 (pedestal and staircase), PE.BFA21 (pedestal and staircase) and PE.KFA71. The SPS requires, as part of the beam commissioning, a 1-turn CT extracted beam for setting-up. For this reason, also the electrostatic septum PE.SEH31 should be operational.

The operation of the fast extracted beams will be resumed after LS1 with the new bump configurations tested during the end of the 2012-2013 run, as described earlier.

7.1 Software tools needed

A LabView-based proto-type software needed for the correct setting up of the extraction was developed and tested already during the 2012-2013 run, and a final Java-based version should be available since early stages of 2014 start-up. In particular:

• The high energy orbit correction for the 26 GeV/c beams will be computed by YASP, taking into account the new correctors PE.DHZ05, PE.BWS16-18 as well as the PE.DHZ60 already used in the past.

65

• The slow and fast bump closure, and the evaluation of the RF correction for the fine synchronization during the rise of the slow bump required for the LHC-type beams, will be performed by a dedicated application developed in JAVA by specialists in the Operations group. A first version of the application written in LABVIEW was already in use during the 2012-2013 run.

• The trajectory steering in the TT2 line will be performed by YASP. The newly installed pickup F16.UDC106 will be included in the list of the available monitors.

7.2 Cold checkout tests

A number of tests should be done during the cold checkout period to ensure the correct functioning of the dummy septum mechanics, with special care to the blade movement and the correct positioning of the BTV screen. These items will be checked also prior to the installation of TPS15 in the tunnel and the proposed list of tunnel tests includes:

• Check of blade movements with remote control including also putting the blade in park position.

• Check the blade movement with manual system. • Check the movement of the screens and the cameras of PE.MTV15 and PE.MTV16. • Check of movement of the screen of PE.MTV15 when the blade is displaced. • Check the correctness of the blade position given by the control system with respect to the

position red by the motor controls in the tunnel. • Check stability of the blade position vs time to exclude the presence of unwanted movements.

This needs logging of the acquired blade position in TIMBER that will be made available before start-up.

The remote control of the different devices will be possible as of February 2014, including the full integration of the different knobs and acquisitions in the working sets and the general PS controls infrastructure.

7.3 Schedule of the beam tests

Tentatively, the new extraction schemes including the correct position of the TPS15 blade should be put in operation before the start of the physics run for the AD and nTOF facilities in order to avoid lengthy iterations in the setting up process. According to the 2014 preliminary schedule of the injectors (see Fig. 64) the first beam should be available in the PS by the end of week 25. It is reasonable to suppose that the beam setting up, including the re-commissioning of the beam instrumentation, would take about one week. During that week, the beams will be extracted with the same slow and fast bumps used during the last period of the 2012-2013 run, but it is unlikely that systematic measurements could be possible in view of performing a proper setting up of the dummy septum. For this reason, the studies of the positioning of the dummy septum blade could take place in week 27, still before the start of the physics run. It is important to notice that the blade cannot be moved in PPM mode, thus implying that the identification of its best location should be done during a period of dedicated beam time.

66

Figure 64: Preliminary 2014 injector accelerator schedule (version 05, October 18, 2013).

7.4 Beams required for setting up and setting up steps

The beams needed for setting up can be sorted according to when they are needed, in view of preparing the beams for physics. In the following the list of beams is given, highlighting the underlying physics user as, in the initial steps of the beam commissioning, not all the beam parameters might be nominal:

• nTOF– like beam, also at intensity lower than nominal and with no bunch rotation before the extraction. The beam can be used for the setting up of the nominal nTOF beam extraction at

67

20 GeV/c. It is worth stressing that an optimization of the transverse emittances is required at the production in the PSB, in particular for the horizontal plane, to increase the available aperture at the location of TPS15.

• AD – like beam to set up the extraction at 26 GeV/c. • EAST – like beam with the parasitic nTOF bunch. The extraction of the nTOF bunch is done

using also the PE.KFA13, which has a fixed pulse length of 5 turns. The EAST bunch trajectory will also be perturbed and, as described earlier, a fast bump will be implemented by closing the perturbation with the PE.KFA21, and PE.KFA4.

• Small-emittance, low-intensity, single-bunch beam at h=8 on a 26 GeV/c and 20 GeV/c cycles without bunch rotation at the end of the cycle. This beam will be used to measure trajectories and available aperture during the fast extraction. Very likely an EAST-type beam would be sufficient.

• Pencil-like beam to kick in the islands and set up MTE extraction on a 14 GeV/c cycle. • MTE nominal beam. • Setting up of LHC-type beams with the new extraction (this depends on when the LHC beams

will be available).

A setting up procedure could be thought as follows:

a) The extraction septum PE.SMH16 blade should be put at the nominal position (i.e. a position already optimized during the beam setting up).

b) The setting up of new slow and fast bumps should be done for a nTOF-type beam at 20 GeV/c, starting with a low intensity beam. This should be followed by a comparison of the extraction trajectories with those expected from the MAD-X-model. Then, the transfer lines TT2 and FTN should be steered with special care to the beam size and position at the location of the neutron production target and the proton transmission efficiencies. If possible, the optics of TT2 and FTN should be re-measured for the sake of comparison with their reference.

c) The setting up of the EAST cycle with parasitic nTOF beam should be done at 24 GeV/c. In this case particular attention should be paid to the closure of the fast bump for the EAST bunch.

d) The setting up of new slow and fast bumps should be done for an AD type beam at 26 GeV/c following the same procedure mentioned for the nTOF beam. In this case the optimization of the trajectories should be repeated once the AD machine will be available to optimize the antiproton production rate. The quantity to maximize in this case is the antiproton intensity measured in the AD ring after the first cooling at high energy. The optics of TT2 and FTA should be re-measured for the sake of comparison with their reference.

e) The setting up of bump and extraction for 1-turn CT beam at 14 GeV/c should be performed. In this case there is no need of measuring the optics in the extraction lines since this beam will be used for a limited amount of time during the initial stage of the SPS start-up [61].

f) The setting up of new slow and fast bumps for a LHC-type single bunch beam at 26 GeV/c as for nTOF and the AD beam. In this case, the quality of the extracted beam should be determined by the profiles at the OTRs (F16.MTV201, F16.MTV218, F16.MTV229 and F16.MTV241) and the SEMFILs (F16.MSG257, F16.MSG267, F16.MSG277). The optics of TT2 should be re-measured. Once the SPS will be available, probably in September 2014, a second campaign of optics measurements will be done including TT10 and the SPS to evaluate any optical mismatch.

68

The measurements in TT10, in particular of the dispersion function, could be advanced depending on the availability of the transfer line magnets and instruments. The control of the fine synchronization during the bump rise should be done in view of the setting up of multi-bunch beams.

g) The dummy septum should be put in nominal position6 and then losses should be optimized by moving the blade. As shown earlier, for the fast extracted beams there should be no losses at the location of the PE.TPS15, thanks to the optimization of the extraction trajectories. For this reason, no losses should be observed on the BLM PE.BLM15 or the fast one BLM PR.FBLM15. The beam trajectory and shape should be controlled at location of the PE.MTV15 and PE.MTV16 and the TT2 SEMGRIDs, in particular if the profiles present sharp edges in one of the two planes (or both). Being the PE.TPS15 blade movement non-PPM, a unique position will be declared operational for all the beams. The setting up mentioned in this step will be also be repeated during dedicated beam time, since it is not possible to move the blade between the park and the operational position between two consecutive cycles.

h) The setting of the MTE beam extraction at 14 GeV/c should take place in two steps. At first, a pencil beam with one single bunch should be kicked in the islands and then extracted. The last turn trajectories will be adjusted to jump the septum blade. Then the trajectory will be optimized considering that an increase of losses at the BLMs in SS15 should correspond to a decrease of them at the PE.BLM16, without affecting the extraction efficiencies. The procedure should be repeated first with a bunched, and then with a debunched beam, possibly with the beam occupying the entire ring circumference. The advantage of the use of a bunched beam is the possibility to measure the trajectory at the last turn. The disadvantage is due to the fact that the loss pattern will not represent the real operational case. The losses, in fact, are produced by the combined effect of the long rise time of the extraction kickers and the longitudinal beam structure required to transfer the beam to the SPS. In a second stage, a nominal beam should be used and the extraction trajectories optimized as for the first step. Any further optimization of the blade position and angle should be followed by a cross check of the losses during the extraction of the non-MTE beams and possibly an iteration on the settings of their extraction elements.

i) The setting up of the multi-bunch LHC type beam extraction will be performed according to the procedure sketched at point f), considering that the blade position at that moment should be already optimized.

8. Conclusions

In this report, the analysis of the main issues encountered during the MTE beam commissioning and first period of operational beam have been discussed in detail. The dummy septum to be installed in the PS ring in SS15 is aimed at addressing the irradiation of the magnetic septum in SS16. The design of the new design and ancillary systems has been reviewed in detail, together with the several studies carried out to optimise the overall performance of the dummy septum solution. According to the outcome of the detailed studies reported here, the implementation of the dummy septum will allow reducing drastically the beam losses and doses for MTE beams with respect to the situation without dummy septum. More quantitatively, the reduction of activation at the location of the magnetic septum 16 should be between a factor of 6 to 9 for the internal side of the straight section and 4 to 7 for its external side. The range of the

6 The value of the local closed orbit should be added to the nominal position from MAD-X simulations.

69

improvement factor depends on whether an optimised dummy septum blade orientation is considered. Furthermore, the final radiation doses in SS16 should be lower with MTE and dummy septum than with the CT beams.

The proposed solution has imposed a number of modifications to the PS ring layout, which have been also discussed in this report. In particular, the relocation of some active beam elements, and the need for additional aperture in SS15 required an in-depth review of the fast extraction schemes for the PS ring towards the TT2 transfer line and the other systems further downstream.

A number of beam measurements have been performed already during the 2012-2013 physics run in order to anticipate on possible issues of the proposed extraction schemes and the impact on the operation performance of the various beams.

Based on this analysis the planning and the strategy of the beam commissioning of the dummy septum has been described. The next steps will be the installation activities, which are supposed to start by the beginning of January 2014, the cold check out, and then the commissioning of the new device with the first beams in the PS ring.

Acknowledgements

Several people have contributed to these studies in various ways. We would like to thank G. Arduini, B. Goddard, V. Mertens, E. Métral for fruitful discussions. D. Aguglia, Y. Bernard, O. Berrig, J. Kuczerowski, P. Lelong, T. Masson, E. Matli, S. Pelletier, B. Pinget, S. Reignier, S. Rösler, J. Vollaire are warmly acknowledged for their support and help.

70

References

[1] M. Giovannozzi (ed.) et al., “The CERN PS multi-turn extraction based on beam splitting in stable islands of transverse phase space: Design Report”, CERN-2006-011.

[2] S. Gilardoni, D. Manglunki (eds.) et al., “Fifty years of the CERN Proton Synchrotron: Volume 1”, CERN-2011-004.

[3] S. Gilardoni et al., “Installation and Hardware commissioning of the Multi-Turn extraction at the CERN Proton Synchrotron”, CERN-ATS-2009-139.

[4] S. Gilardoni et al., “First Results for the Beam Commissioning of the CERN Multi-Turn Extraction”, CERN-ATS-2009-038.

[5] M. Giovannozzi, “Status and plans for PS Multi-turn extraction”, presentation at the IEFC meeting held on 8 of January 2010.

[6] M. Giovannozzi, “Linac3 shielding”, presentation at the IEFC meeting held on 22 of January 2010.

[7] M. Giovannozzi, “Status of MTE”, presentation at the IEFC meeting held on 25 of May 2010.

[8] T. Bohl, “Observations in the SPS of beam with various longitudinal parameters and extracted from the CPS using CT”, CERN-BE-Note-2009-013.

[9] T. Bohl, “Low intensity beam extracted from the CPS via CT or MTE with various longitudinal parameters”, CERN-BE-Note-2009-014.

[10] T. Bohl, “Low intensity MTE beam in the SPS”, CERN-BE-Note-2009-015.

[11] E. Shaposhnikova, E. Ciapala, E. Montesinos, “Upgrade of the 200 MHz RF System in the CERN SPS”, CERN-ATS-2011-042.

[12] H. Damerau, et al., “Upgrade Plans for the LHC injector complex”, CERN-ATS-2012-111.

[13] H. Timko, et al., “Experimental and simulation studies of the PS-to-SPS beam transfer efficiency”, CERN-ATS-Note-2012-059 PERF.

[14] MTE Workshop 2010, http://indico.cern.ch/conferenceDisplay.py?confId=105961.

[15] S. Gilardoni, “Status of commissioning with beam”, presentation at the MTE Workshop 2010.

[16] M. Giovannozzi, “Measurements’ analysis”, presentation at the MTE Workshop 2010.

[17] R. Steerenberg, “Operations point of view: PS”, presentation at the MTE Workshop 2010.

[18] K. Cornelis, “Operations point of view: SPS”, presentation at the MTE Workshop 2010.

[19] M. Widorski, “Radioprotection issues”, presentation at the MTE Workshop 2010.

[20] P. Freyermuth, “Power converters stability”, presentation at the MTE Workshop 2010.

[21] J.-J. Gras, “General status of instrumentation”, presentation at the MTE Workshop 2010.

71

[22] L. Sermeus, “Considerations on performance of extraction elements”, presentation at the MTE Workshop 2010.

[23] M. Giovannozzi, “New MTE extraction scheme to mitigate irradiation of SMH16”, presentation at the MTE Workshop 2010.

[24] M. Giovannozzi, “Summary of the MTE mini-workshop”, presentation at the IEFC meeting held on 1 of October 2010.

[25] M. Giovannozzi, “MTE action plan - what is foreseen in 2011?”, presentation at the IEFC meeting held on January 2011.

[26] M. Giovannozzi, “MTE roadmap – Beam physics aspects”, IEFC Workshop 2011.

[27] MTE workshop 2011, http://indico.cern.ch/conferenceDisplay.py?confId=153628

[28] S. Gilardoni, “Status of commissioning with beam”, presentation at the MTE Workshop 2011.

[29] A. Lachaize, “Measurements’ analysis”, presentation at the MTE Workshop 2011.

[30] S. Bart Pedersen, “Status of instrumentation”, presentation at the MTE Workshop 2011.

[31] M. Giovannozzi, “Mitigation measures for extraction losses”, presentation at the MTE Workshop 2011.

[32] J. Borburgh, “Impact of MTE on PS septa”, presentation at the MTE Workshop 2011.

[33] L. Ducimetiere, “Kicker options for MTE”, presentation at the MTE Workshop 2011.

[34] E. Carlier, “Electronics issues for MTE”, presentation at the MTE Workshop 2011.

[35] M. Giovannozzi, “Conclusions from the MTE workshop”, presentation at the MTE Workshop 2011.

[36] M. Giovannozzi, “Summary and recommendations of the MTE workshop on 3 October”, presentation at the IEFC meeting held on 21 October 2011.

[37] B. Goddard, private communication.

[38] S. Damjanovic, J. Borburgh, S. Gilardoni, M. Widorski and J. Vollaire, “Radiological impact study of a protective device for the septum PE.SMH16”, CERN-DGS-2011-003-RP-TN, 2011.

[39] M. Juchno, “Electromagnetic FEM analysis of the CERN Proton Synchrotron main magnetic unit”, CERN-THESIS-2009-175.

[40] M. Juchno, “Magnetic Model of the CERN Proton Synchrotron Main Magnetic Unit”, CERN-ATS-2011-255.

[41] A. Ferrari, P. R. Sala, A. Fassò and J. Ranft, “FLUKA, a Multi-particle Transport Code”, CERN-2005-010, 2005.

[42] A. Fassò et al., “The physics models of FLUKA: status and recent developments”, MOMT005, in Proceedings of CHEP2003, eConf C0303241 (2003), arXiv :hep-ph/0306267, 2003.

[43] J. M. Lafferty, Foundations of Vacuum Science and Technology, Wiley, New York, 1998.

72

[44] S. R. Wilson, J. Vac. Sci. Technol. Vac. Surfaces Films 5, p. 2479, 1987.

[45] R. Kersevan and J.-L. Pons, American Vacuum Society, 27, p. 1017, 2009.

[46] E. Mahner, I. Efthymiopoulos, J. Hansen, E. Page, H. Vincke, Phys. Rev. ST Accel. Beams 7 103202, 2004.

[47] W. Fischer, M. Bai, J. M. Brennan, M. Blaskiewicz, P. Cameron, H. C. Hseuh, H. Huang, W. MacKay, T. Roser, T. Satogata, L. A. Smart, D. Trbojevic, S. Y. Zhang, “Vacuum pressure rise with intense ion beams in RHIC” ”, WEPRI034, in proceedings of EPAC02, p. 1485, 2002.

[48] C. Pasquino, J. A. Ferreira, P. Demarest, “PS Dummy Septum: Vacuum”, EDMS Document No. 1316314, 2013.

[49] http://www.cst.com/

[50] H. Damerau, et al., “Longitudinal coupled bunch instabilities in the CERN PS”, FRPMN069, in proceedings of PAC07, p. 4180, 2007.

[51] M. Migliorati, et al., Phys. Rev. ST Accel. Beams 16, 031001, 2013.

[52] S. Persichelli, O. Berrig, J. Herbst, J. Kuczerowski, M. Migliorati, B. Salvant “Impedance studies of the dummy septum for CERN PS Multi-Turn Extraction”, in preparation.

[53] http://ab-project-mte.web.cern.ch/AB-Project-MTE/Minutes/Index.html.

[54] S. Damjanovic, S. Roesler and M. Widorski, “Shielding Studies for the Ground Level above Straight Section 16 of the PS”, CERN-DGS-2011-066-RP-TN, EDMS 1300759.

[55] International Commission on Radiological Protection, 1990 Recommendation of the International Commission on Radiological Protection, ICRP Publication 60, Annals of the ICRP 21, 1-3, Pergamon Press (1991).

[56] ALARA Review Working Group (ARWG), EDMS 1244380, 2013.

[57] S. Gilardoni, et al., “Modified Extraction Scheme for the CERN PS Multi-Turn Extraction”, CERN-ATS-2012-164.

[58] H. Bartosik, et al., “Proposal of a Dummy Septum to Mitigate ring irradiation for the CERN PS Multi-Turn Extraction”, CERN-ATS-2012-165.

[59] C. Hernalsteens, et al., “Design and Beam Measurements of Modified Fast Extraction Schemes in the CERN PS for Installing a Dummy Septum to Mitigate Ring Irradiation”, WEPEA056, in proceedings of IPAC13, p. 2633, 2013.

[60] C. Hernalsteens, et al., “Numerical Simulations to Evaluate the Performance of CERN PS Dummy Septum to Reduce Irradiation for the Multi-Turn Extraction”, WEPEA057, in proceedings of IPAC13, p. 2636, 2013.

[61] K. Cornelis, private communication.

73