Characterization and core renovation of beam stoppers for personnel safety

To minimize the risk of accident on personnel accessing the PS complex at CERN transfer lines are equipped with beam stoppers. They immediately shut the beam aperture whenever members of personnel enter an accelerator or test facility. Most of them consist of cores out of stainless steel designed for a proton beam-pulse energy at 9.0 kJ during the 70's. Further data on the design history of the beam stoppers is scarce. With the current pulse energy at 35.0 kJ steel reaches its structural limits. Based on the current function of the core requirements on the structural and beam-interaction performance of beam stoppers for personnel safety are drawn. To meet those requirements a compact core compatible to all PS complex is developed, replacing five existing design types. The new core introduces high-density sliced materials which withstand upgraded beam-pulse energies up to 92.5 kJ.


Introduction
Particles sent to the Large Hadron Collider (LHC) at CERN are first accelerated in the Proton Synchrotron (PS). It is fed by the Proton Synchrotron Booster (PSB) with four stacked synchrotron rings that receive protons from linear accelerators. Not only does PS extract LHC-dedicated beam, but it also provides the neutron Time-of-Flight facility (nTOF) [1], the Antiproton Decelerator (AD) [2] and the East Area (EA) [3] with protons up to 26 GeV/c.
The risk of beam accident of technical personnel intervening in the different facilities and acceleration stages has to be the lowest. In order to mitigate this probability Elements Important for Safety type beam (EIS-b) are installed in the beam transfer lines [4,5]. For redundancy each transfer line has two independent EIS-b of two different operating modes: a magnetic bender and a beam-intercepting material block [6]. In normal beam operation both EIS-b do not interact with the -1 -beam, however they are actuated during intervention campaigns or personnel intrusion to a facility. Secondary particles produced by the beam impact in the block are absorbed by external shielding.
A beam stopper consists of that block, which is also referred to as core, in a vacuum chamber. The design criterion met in the 70's was to use common, reliable and cheap materials which could sustain one beam impact of 2 × 10 12 particles per pulse at 28 GeV [7]. Stainless steel was chosen as the core material because the beam power was considered negligible [7]. Since then no further design record has been available and the thermal and structural performance of the PS beam stoppers are yet to be analyzed. Documentation on the minimum core size, structural criterion for the core, actuation time and maximum number of allowed pulses is scarce or non-existent.
The highest beam intensity has been increased to 8.7×10 12 ppp (particles per pulse) at 26 GeV/c for nominal LHC beam and to 20 × 10 12 ppp at 14 GeV/c for machine development campaigns [8].
The goal of this paper is to first analyze the thermo-mechanical and beam-attenuation performance of the beam stoppers currently active in the PS complex. Using their configuration and best practices we specify guidelines for a beam stopper design and devise beam stopper cores capable of receiving high-intensity beams.

Current beam stoppers
In this section we present the operation and describe the different design types of the beam stoppers existent in the PS complex.

Operation
The position of the stopper core is managed by the personnel access system [6,11] which is dedicated to personnel access to beam facilities and independent from the control system of the accelerator machine [11].
During beam operation the stopper core rests out of the beam line (OUT-BEAM position). The core is linked to a one-way pneumatic actuator at 4 bar while the solenoid of the pneumatic valve is powered at 25 V (figure 1a). The air pressure force balances the weight of the core. By cutting the power supply in the solenoid the air pressure in the pneumatic actuator is released, inverting the resultant force in the actuator. The core is left to fall in beam line (IN-BEAM position), until reaching an end stop (figure 1b). The IN-BEAM and OUT-BEAM positions are read by end-switches. Any intermediate position is not detected [12].
The actuation of the beam stopper is fail-safe because its core moves always in beam position whenever power cut or air-pressure failure in the stopper occur. The core also moves in beam position when [6]: • a beam operator grants access to the downstream facility. The beam is already dumped upstream and does not hit the stopper core.
• someone forces an access door of a downstream accelerator. The access system automatically moves the upstream stopper core in beam position before the beam is dumped. This is an accidental case as the beam hits the material core.

JINST 14 T01011
Upon permission of the downstream areas [6] the personnel access system puts the beam stopper out of the beam line by energizing the solenoid and pressurizing the cylinder.

Beam-stopper types
The PS complex (figure 2) has six types of beam stoppers (figure 3).  hits the shells radially interacting with a total of 14.6-mm material depth. Except for BI-FA all PS stopper cores are massive cylinders out of steel for axial beam impacts.
Downstream in the BI line a stopper type 1c out of stainless steel provides a core length of 600 mm. The core is pivoted and has a diameter of 200 mm.
In the transfer lines in the PS injection (BTP), extraction (F16), AD (FTA) and nTOF (FTN) a combination of stoppers type 1c and 2c are used to increase the total length for beam attenuation. The stopper type 2c has two mechanically independent cores in the same vacuum chamber. However their control systems are linked and both cores can only be moved simultaneously in beam line.

JINST 14 T01011
Each core has a diameter of 200 mm and a length of 550 mm spaced by 50 mm. Both 1c and 2c cores travel 175 mm in less than 17 seconds.
The beam stopper in the transfer line to ISOLDE (BTY-11 o ) has a core tilt of 11 o and a vertical pneumatic mechanism. The core is made of stainless steel 304 and has a diameter of 140 mm with a length of 600 mm. The beam aperture of 140 mm is closed in 4 seconds.
The extraction line to the East Area has two beam stoppers type EA-1.5m. Each core is made of CK45 steel with 1.5-m length and a diameter of 300 mm. It is screwed to a driving mechanism consisting of forks and chains. The core falls 260 mm in 17 seconds.
Each experimental line inside the East area -T8, T9, T10, T11 -is equipped with a beam stopper type EA-TX. The core is made of iron coated with copper and has a diameter of 300 mm and a length of 600 mm. The core falls 260 mm in less than 30 seconds actuated by a vertical pneumatic cylinder.

Methods and core requirements
In accidental scenarios beam hits the stopper core. The core must be able to attenuate a great amount of beam particles while resisting high thermal stresses caused by accidental impacts of particle beam pulses. In this section we introduce methods for modeling the thermo-mechanical and beam-attenuation performance of the current stopper cores.

Thermo-mechanical modeling
The interaction between beam and matter induces focused heat deposition in the core material. We use the Monte Carlo particle transport and interaction code FLUKA [13,14] to simulate the energy deposited in the core material normalized to one particle. The energy density per beam particle (Q bp ) is converted to power density per pulse ( Q pulse ) through eq. 3.1.
Here n p is the number of particles per pulse and µ the beam pulse length. For bunched beams µ corresponds to the number of bunches multiplied by the bunch length.
The energy deposited in the stopper core is imported as internal heat generation in a thermal model in ANSYS 17.1. In our models we calculate the difference in the imported energy between ANSYS 17.1 and FLUKA to be less than 0.1%. The difference in the peak temperature in both models is smaller than 1.2%. The thermal model in ANSYS 17.1 determines the temperature distribution through eq. 3.2 [15].
Where T is the temperature, t the time, ρ the mass density, c p the specific heat capacity and k the thermal conductivity.
The temperature distribution at each time step is imported in an ANSYS 17.1 structural model [15] which is solved through implicit methods such as [16] and [17]. Our maximum time step length (∆t) for structural simulations follows the Courant-Friedrichs-Lewy condition (eq. 3.3) [18].

JINST 14 T01011
Here c is the speed of sound in the material (eq. 3.4), E the Young's modulus and ν the Poisson's ratio.
The initial temperature for all simulations is 22 • C. Because of vacuum there is no convective heat exchange in the core. For stainless steel the emissivity factor for the radiative transfer between the core and the vacuum chamber is assumed to be 0.5. In the structural model the core is fixed through weak springs [15].
For the thin graphite BI-FA core the thermo-mechanical performance is computed via layered shell elements [19]. To quickly assess the material integrity we use the Coulomb-Mohr criterion [20]. Here a safety factor F < 1 indicates the material is safe.
The metallic cores are modeled with solid elements and evaluated through the von-Mises yield criterion [21,22]. Considering σ e the equivalent von-Mises stress and σ y (T) the temperaturedependent tensile yield strength a value σ e /σ y (T) < 1 indicates the deformation induced by thermal gradients is reversible (elastic), otherwise it is permanent (plastic).

Beam attenuation
At low energies a particle beam is stopped mainly due to ionization losses in the intercepting material. Due to the increasing cross section interaction and the decreasing particle's energy through matter the primary particles come to a complete stop at a characteristic material depththe projected range [23].
At high beam energies inelastic nuclear scattering [24] with target nuclei becomes more relevant than ionization losses. The inelastic nuclear scattering length λ inel (eq. 3.5 [25]) describes the length of target material in which the intensity of the beam is reduced to 1/e (≈ 0.368) of the initial intensity due to inelastic nuclear scattering interactions in the target. It quantifies the beam attenuation produced via inelastic nuclear scattering.
Here I 0 is the initial intensity of a particle beam impinging on the target, I the beam intensity that passes through the target without undergoing inelastic nuclear scattering, λ inel the inelastic nuclear scattering length and L the length of a target. Table 1 shows the values of λ inel for materials used in today's stoppers.
Ionization losses are dominant for proton energies lower than 160 MeV incident on today's core materials. There the projected range is shorter than the nuclear interaction length. On the contrary, at beam energies higher than 1.4 GeV, nuclear interaction becomes dominant. The length of the stopper core depends on the projected range at low beam energies and the nuclear interaction length at high beam energies.
We assess the lateral containment of electromagnetic shower of particles produced by the beam-matter interaction via the Molière radius R m [26]. A cylinder with the radius of 3.5 R m is -6 - able to absorb 99% of the energy of the electromagnetic shower produced after the interaction of the beam with the target [27]. For steel R m = 17.6 mm [28]. Most of the stopper cores in the PS complex are made of steel and have a radius of 100 mm which corresponds to 5.7 R m .

Results of current stopper cores
This section discusses the performance of the current beam stoppers on the perspective of thermal and structural response. The beam parameters for the analysis are summarized in table 2, where σ x and σ y are the beam sizes at 1 σ following a two-dimensional Gaussian profile.

BI line
For the BI-FA stopper core the two inner shells are modeled with two layers, the two outer shells with 30 layers using layered shell elements [19].
-7 - Figure 4 shows the Bragg peak, the point at maximum beam energy deposition due to ionization, lies between the last shells hit by a 50-MeV proton beam. We calculate more than 99% of the energy of the incoming primary particles is absorbed by this core, while the energy absorbed by BI-SW is nearly zero. The material is safe as the maximum F = 0.19 in the core material.  Increasing the proton energy to 160 MeV causes most of the beam particles to pass through the graphite core. Only 10.3% of the incoming beam energy is absorbed by the BI-FA and 94% by the BI-SW core. The peak temperature in all shells is nearly the same at 199 o C with a maximum safety factor F = 0.12. Due to the beam dilution caused by BI-FA the peak temperature rise per pulse is only 7.2 o C in the BI-SW core.
In case the actuation of the BI-FA core fails a single beam impact causes concentrated plastic deformation in the BI-SW core due to short beam projected ranges. The peak equivalent stress is evaluated at 210 MPa for the 50-MeV beam and 397 MPa for the 160-MeV beam, exceeding the elastic limit of stainless steel [33]. In addition the BI-SW core is exposed to excessive peak temperature rise of T max = 385 o C (50 MeV) and 750 o C (160 MeV).
During the next long shutdown at CERN both BI-FA and BI-SW stoppers will be dismantled. The personnel protection system of the BI line will be integrated to Linac4 [34] where a beam stopper core out of a graphite R4550 rod ( 80 mm x 120 mm) shrink-fit in a stainless steel clad withstands few 160-MeV proton beam pulses [35].

Current stoppers with steel cores
We evaluate the ratio between the equivalent stress and the temperature-dependent yield strength in the geometric model of the steel cores (table 3). For the core material to be safe max σ e /σ y (T) < 1.
For 20 repeated pulses on the BTP stopper T max = 90.1 o C and σ e /σ y (T) = 0.64. In case a BTY beam hits the BTP stopper, e.g. due to a failure on the beam transfer, the BTP core presents some plastic deformation after three repeated pulses.
A cycle of six spills can hit the EA stopper core after the slow beam extraction from the Proton Synchroton to the East area [32]. Each spill is 0.4-second long repeating each 2.4 seconds. The -8 - Table 3. Peak temperature, maximum σ e /σ y (T) from one proton beam pulse and minimum number of repeated pulses to plastic deformation in steel cores. Critical beam scenarios are those of F16 and the upgraded BTY in which one single pulse causes plastic deformation. Although the stopper core is still operational the accumulation of plastic deformation under frequently repeated pulses lead to the start of fracturing (figure 5a). Once the yield strength is reached adding more thermal strain increases the plastic strain ε p , conducting to the start of fracture at ε tensile . For stainless steel ε tensile is about 0.2 [36]. Leaving the stopper core to be cooled down over long periods between beam impacts dampens the increase of plastic strain in the F16 stopper core (figure 5b).

Design of a new PS stopper core
This section is about a new stopper core designed to be structurally safe under upgraded beam scenarios and conform to beam-attenuation requirements. For standardization we aim for a common -9 -

JINST 14 T01011
stopper and core design for all PS complex.
For design simplification the core has no active cooling system. In our analysis the core is passively cooled only through radiation to the vacuum chamber.

Beam parameters
The core takes several repeated beam pulses while moving from OUT-BEAM to IN-BEAM positions. The slowest core nowadays takes about 17 seconds between both positions. Knowing the smallest pulse repetition rate of all beam scenarios is 1.2 s a beam stopper is impacted by 15 repeated pulses. We use this number of repeated pulses as a design requirement. Although the beam could hit different spots on the core we assume in our analyses the repeated pulses hit only the core center.
The most damaging beam is the upgraded beam for the F16 line (2.3 × 10 13 protons per pulse, 26 GeV/c, beam sizes σ x = 1.5 mm and σ y = 1.39 mm). Our beam parameters for validating the core integrity has the same intensity, energy and beam size as the most damaging beam with a repetition rate of 1.2 s following the shortest repetition rate in the PS complex [8].

Beam attenuation
In order to maintain the current beam-attenuation performance we keep the same accumulated multiples of the inelastic nuclear scattering length (λ inel ) in each beam line (table 4). The East Area is the most protected facility because of the high number of spills in a supercycle of the extracted beam. The nuclear interaction length is calculated for steel at different proton energies: λ 2.0GeV = 14.97 cm and λ 20−26GeV/c = 15.56 cm. A core protection length of 3.75 λ inel (table 4) is a fitting compromise between the EA, FTA and FTN lines while keeping the core length short. The other lines (BTP, BTY and F16) benefits from an increase in the total protection length.
Following the current electromagnetic containment of stainless steel stoppers with a radius of 100 mm we design our core with at least 5.7 R m .

Single-block configuration
The new material must have a high yield strength in order to prevent permanent deformation. In addition high thermal-shock resistance (R t , eq. 5.1) and low temperature rise per beam pulse reduce -10 -thermal stress peaks. Stainless steel (SS304L) is again proven to be unsuitable compared to other materials (table 5).
Here α is the coefficient of thermal expansion. Low-density materials such as aluminum (Antico 100) and titanium (TiAl6V4) alloys have low peak temperature rise and can withstand several repeated pulses. Nevertheless stopper cores with those materials have to be very long for personnel protection due to their long nuclear interaction lengths.
The copper alloy CuCr1Zr fails the safety criterion on one single pulse impact despite its high thermal-shock resistance. The nickel alloy Inconel 718 is the only high-density materials able to withstand repeated pulses thanks to its high yield strength. Although the shock resistance of the molybdenum alloy TZM is the highest the peak temperature rise per pulse in this material is also high, making it less suitable as a core material than Inconel 718.

Diluter-absorber configuration
This configuration consists of a diluter out of a low-density material that absorbs part of the incoming beam energy leaving a smaller amount of energy to a downstream absorber out of a high-density material ( figure 6). Introducing a 300-mm long titanium cylinder reduces the peak temperature rise in our reference Inconel 718 absorber by 52%.
A titanium-alloy diluter absorbs more beam energy for the same material length than an aluminum alloy. In addition the beam attenuation in the titanium alloy is stronger allowing the diluter to be more compact for the same beam attenuation.
The amount of repeated pulses a CuCr1Zr absorber withstands increases to 86 by elongating the titanium diluter to 300 mm. This amount is increased to only 51 in an Inconel 718 absorber.
Using this configuration we select a 250-mm long titanium diluter (figure 7a) followed by a 415-mm long absorber made of CuCr1Zr, profiting from the high thermal conductivity, thermalshock resistance and affordability of this material. This core provides a total of 3.75 λ inel (diluter: 0.95 λ inel ; absorber: 2.8 λ inel ) withstanding 20 repeated pulses of the required beam (figure 7b).    The maximum temperature after 20 repeated pulses in the diluter is T max = 380.0 o C. In the absorber T max = 223.7 o C, max σ eqv /σ y (T) = 0.98 and max σ eqv = 257.6 MPa.
After 15 repeated pulses the beam stopper has to be left to passively cool down in order to avoid the increase of thermal strains. Considering an optical emissivity of titanium = 0.19 and copper = 0.07, a homogeneous temperature around 28.6 o C in the whole diluter is reached after 87 minutes, whereas a homogeneous temperature around 34.0 o C is achieved in 29 minutes in the absorber.

High-density sliced diluter
In order to reduce the core length while keeping the same attenuation factor of 3.75 λ inel we introduce a high-density sliced diluter. Under energy deposition from beam impacts the thin slices allow the material to freely expand in the beam axis, reducing the axial stress component.
-12 - Figure 8 shows a massive Inconel 718 cylinder hit by beam presents permanent deformation while a sliced configuration prevents the material from plastic deformation. In a full-block core out of Inconel 718 max σ eqv /σ y (T) = 1.34 using a fully elastic material model, max σ eqv = 1120 MPa and T max = 707 o C. In a core with 20-mm thick slices out of Inconel 718, max σ eqv /σ y (T) = 0.88, max σ eqv = 776 MPa and T max = 709 o C. A sensitivity analysis for Inconel 718 and TZM slices (figure 8c) indicates each material has an optimum slice thickness for low equivalent stresses. Based on the previous titanium-copper core configuration our new diluter must have a minimum attenuation factor of 0.94 λ inel for thermo-mechanical integrity of the CuCr1Zr absorber. Taking a slice thickness of Inconel 718 at 40 mm we introduce a compact stopper core with four Inconel 718 slices followed by a CuCr1Zr massive absorber. The total length of this core is 564 mm (figure 9) with gaps between parts of 1 mm for allowing thermal expansion of the slices in beam direction. All four slices of Inconel 718 together provide a diluter attenuation factor of 1.03 λ inel . Keeping the diameter of the stopper core at 200 mm the sliced Inconel 718 diluter has 6.7 Rm (R m = 14.9 mm [28]) while the CuCr1Zr absorber block has 6.4 R m (R m = 15.6 mm [28]). Figure 10 presents the thermal and equivalent-stress distribution in terms of σ eqv /σ y (T) in the new sliced core right after 15 repeated pulses. The maximum equivalent stress is 601 MPa in the Inconel slices and 218 MPa in the CuCr1Zr absorber. After repeated pulses the core must be left to cool down for 16 minutes for its maximum temperature to reduce to 40 • C.

Radiation protection aspects
At an intensity of 23 × 10 12 protons per pulse and high energies the new core emits a similar neutron spectrum as the previous steel core (figure 11a).
After the impact of 15 consecutive pulses separated by 1.2 seconds with an intensity of 23×10 12 protons per pulse, the maximum ambient dose equivalent rate reached at 40 cm from the stopper surface is 3.5 mSv/h after 1 hour of cool-down time and 0.2 mSv/h after 1 day of cool-down time ( figure 11b). These ambient dose rates are lower than the ones usually found at the extraction equipment located upstream in the PS machine [38]. The required cool-down times depend on the activity to be performed, being defined together with the CERN Radiation Protection group [38]. All accesses to the beam stoppers, including those for inspection, maintenance or repair, are accompanied by the CERN Radiation Protection group to ensure proper radiation protection of these activities.

Conceptual design of the new beam stopper
Instead of the long 2c-type design we opt for a beam stopper out of a single core in a vacuum chamber for easier maintainability and mechanical integration in the beam line. Figure 12 presents a conceptual beam stopper design for the PS complex [39]. In the new beam stopper the core is held over the beam line by a fork linked to a pneumatic cylinder and connected to a shock plate having shock absorbers and ball bearings. The tank is supported by the top table and the outer bellow. The top table orients the tank in the horizontal plane. The outer bellow allows the core and tank axes to be aligned with the beam line. This vertical alignment is set up by means of shims placed under the tank allowing a maximum tilt of 4 o . In the BTY line the new beam stopper must be moved from the current position with 11 o to a position with a maximum tilt of 6.3 o in order to still meet the electromagnetic containment requirement in the core (5.7 R m ).
During beam operation the core stays in OUT-BEAM position supported by an inner bellow ( figure 13). Whenever members of the personnel wants or forces the access in a downstream area or in fail-safe cases, the pneumatic cylinder is de-pressurized and the core moves to IN-BEAM position. The shock plate slides in the guiding shafts connected to the top table. The fall of the core is damped through shock absorbers on the frame shims.
A dedicated test bench using a pneumatically controlled mechanism demonstrated the core travels from OUT-BEAM to IN-BEAM position in less than 2.0 s [39,40] using a double actuating cylinder. This demonstrator used massive iron blocks equivalent to the total forces the core assembly exerts on the pneumatic actuator and the shock absorbers during the its fall.

Conclusions
We described the performance of the current beam stoppers for personnel safety in the PS complex at CERN. Their steel cores reaches the elastic material limit under a single beam pulse impact with 5.8 × 10 12 particles at 26 GeV/c. We introduced guidelines for the design of a new core in terms of multiples of the nuclear interaction length (λ inel ) and the electromagnetic containment in Moliere radii (R m ). The new core is designed for 15 repeated beam impacts with 23 × 10 12 protons per pulse at 26 GeV/c and beam sizes σ x = 1.5 mm and σ y = 1.39 mm based on the most damaging beam upgrade scenarios.
A comparison of materials with regards to the thermal-shock resistance and peak temperature rise in beam-intercepting metals showed no high-density material directly withstands 15 repeated beam pulses. We proposed a diluter-absorber configuration in which a low-density diluter out of the titanium alloy Ti6Al4V dampens the peak temperature rise in a high-density absorber out of the copper alloy CuCr1Zr.
For even shorter stopper cores for personnel safety this paper brought out high-density sliced diluters. This diluter allows the material to freely expand in beam axis during repeated beam impacts reducing thermal stresses. The new core is 564-mm long, has a diameter of 200 mm and consists of a sliced diluter out of the nickel-alloy Inconel 718 and a massive absorber out of the copper-alloy CuCr1Zr. It provides 3.75 λ inel of nuclear interaction length and at least 6.4 R m of electromagnetic containment.
In the conceptual design of the new beam stopper the core is held by a bellowed fork in vacuum linked to a pneumatic actuator, vertically descending in beam line in less than 2.0 s. The new beam stopper and its compact sliced-diluter core replace five previous designs in the PS complex and will be replicated in each beam line according to beam-attenuation requirements. First prototypes of the conceptual design of the beam stopper will be tested in the beginning of 2019.