Search

Distributed systems have been widely adopted for deep neural networks model training. However, the scalability of distributed training systems is largely bounded by the communication cost. We design a highly efficient collective communication library, namely Alibaba Collective Communication Library (ACCL), to build distributed training systems with linear scalability. ACCL provides optimized algorithms to fully make use of heterogeneous interconnects simultaneously. And the experimental results show significant performance improvement.

The future high luminosity (Hi-Lumi) upgrade of the Large Hadron Collider (LHC) at CERN will include eight (plus two spares) 10.2 m-long Cryo-assemblies which will be components of the triplets for two LHC insertion regions. Each cold mass in the Cryo-assemblies will consist of two 4.2 m-long Nb3Sn high gradient quadrupole magnets, designated MQXFA, with aperture 150 mm and operating gradient 132.2 T/m, for a total of twenty magnets. Before assembling and testing the final cold masses at Fermilab, the component quadrupoles are being tested first at the vertical superconducting magnet test facility of the Superconducting Magnet Division (SMD) at Brookhaven National Laboratory (BNL), in superfluid He at 1.9 K and up to 18.0 kA, in accordance with operational requirements of the LHC. The tests of the first two full-length prototype quadrupole magnets MQXFAP1 and MQXFAP2 at BNL have been reported previously. The first two pre-series magnets, the first two that will be used in the LHC, have also now been tested. This paper reports on the quench test and training results of these two magnets. The test results of these magnets will be important for validating the final MQXFA design for operational magnets.

MQXF is the Nb 3 Sn Low-β quadrupole magnet that the HL-LHC project is planning to install in the LHC interaction regions in 2026 to increase the LHC integrated luminosity. The magnet will be fabricated in two different lengths: 4.2 m for MQXFA, built in the US by the Accelerator Upgrade Project (AUP), and 7.15 m for MQXFB, fabricated by CERN. In order to qualify the magnet design and characterize its performance with different conductors, cable geometries and pre-load configurations, five short model magnets, called MQXFS, were fabricated, assembled and tested. We compare the mechanical behavior of short model magnets using experimental data and new numerical models that take into account the measured coil sizes as a function of position. -MQXF is the Nb3Sn Low-β quadrupole magnet that the HL-LHC project is planning to install in the LHC interaction regions in 2026 to increase the LHC integrated luminosity. The magnet will be fabricated in two different lengths: 4.2 m for MQXFA, built in the US by the Accelerator Upgrade Project (AUP), and 7.15 m for MQXFB, fabricated by CERN. In order to qualify the magnet design and characterize its performance with different conductors, cable geometries and pre-load configurations, five short model magnets, called MQXFS, were fabricated, assembled and tested. We compare the mechanical behavior of short model magnets using experimental data and new numerical models that take into account the measured coil sizes as a function of position.

MQXF is the Nb$_3$Sn Low-β Quadrupole magnet that the HL-LHC project is planning to install in the LHC interaction regions in 2026 as part of an upgrade to increase the LHC inte-grated luminosity by about a factor of ten. The magnet will be fab-ricated in two different lengths: 4.2 m for MQXFA, built in the US by the Accelerator Upgrade Project (AUP), and 7.15 m for MQXFB, fabricated by CERN. In order to qualify the magnet de-sign and characterize its performance with different conductors, cable geometries and pre-load configuration, five short model magnets, called MQXFS, were fabricated, assembled and tested. The latest model, MQXFS6, uses a new powder-in-tube (PIT) su-perconducting wire, featuring a bundle barrier surrounding the filaments. The coil and the support structure were equipped with strain gauges and optical fibres to monitor strain during assembly, cool-down and excitation. In this paper we further develop the conventional azimuthal preload analysis and introduce a new set of tools for MQXF coil pack characterization which we use to an-alyse the behaviour of MQXFS6 room temperature preload and to reanalyse all the short models tested at CERN. A comparison is made between all the studied magnets revealing new characteriz-ing preload parameters. MQXF is the Nb$_3$Sn Low-β Quadrupole magnet that the HL-LHC project is planning to install in the LHC interaction regions in 2026 as part of an upgrade to increase the LHC integrated luminosity by about a factor of ten. The magnet will be fabricated in two different lengths: 4.2 m for MQXFA, built in the US by the Accelerator Upgrade Project (AUP), and 7.15 m for MQXFB, fabricated by CERN. In order to qualify the magnet design and characterize its performance with different conductors, cable geometries and pre-load configuration, five short model magnets, called MQXFS, were fabricated, assembled and tested. The latest model, MQXFS6, uses a new powder-in-tube (PIT) superconducting wire, featuring a bundle barrier surrounding the filaments. The coil and the support structure were equipped with strain gauges and optical fibres to monitor strain during assembly, cool-down and excitation. In this paper we further develop the conventional azimuthal preload analysis and introduce a new set of tools for MQXF coil pack characterization which we use to analyse the behaviour of MQXFS6 room temperature preload and to reanalyse all the short models tested at CERN. A comparison is made between all the studied magnets revealing new characterizing preload parameters.

The future high luminosity (Hi-Lumi) upgrade of the Large Hadron Collider (LHC) at CERN will include eight (plus two spares) 8.4 m-long cryostatted cold masses which will be components of the triplets for two LHC insertion regions. Each cold mass will consist of two 4.2 m long Nb3Sn high gradient quadrupole magnets, designated MQXFA, with aperture 150 mm and operating gradient 132.6 T/m, for a total of twenty magnets. Before assembling and testing the final cold masses at Fermilab, the twenty component quadrupoles will be tested first at the ver-tical superconducting magnet test facility of the Superconducting Magnet Division at Brookhaven National Laboratory (BNL), in superfluid He at 1.9 K and to 18.0 kA, to meet LHC operational requirements. The first two full-length prototype quadrupole magnets, MQXFAP1 and MQXFAP2, have been tested at BNL. This paper reports on the quench test and training results of these magnets, and also the retest of the first prototype, rebuilt and designated as MQXFAP1b. The test results of these magnets will be important for validating the MQXFA design. The future high luminosity (Hi-Lumi) upgrade of the Large Hadron Collider (LHC) at CERN will include eight (plus two spares) 8.4 m-long cryostatted cold masses which will be components of the triplets for two LHC insertion regions. Each cold mass will consist of two 4.2 m long Nb3Sn high gradient quadrupole magnets, designated MQXFA, with aperture 150 mm and operating gradient 132.6 T/m, for a total of twenty magnets. Before assembling and testing the final cold masses at Fermilab, the twenty component quadrupoles will be tested first at the vertical superconducting magnet test facility of the Superconducting Magnet Division at Brookhaven National Laboratory (BNL), in superfluid He at 1.9 K and to 18.0 kA, to meet LHC operational requirements. The first two full-length prototype quadrupole magnets, MQXFAP1 and MQXFAP2, have been tested at BNL. This paper reports on the quench test and training results of these magnets, and also the retest of the first prototype, rebuilt and designated as MQXFAP1b. The test results of these magnets will be important for validating the MQXFA design.

The US Accelerator Upgrade for the HiLumi-LHC (US-HL-LHC AUP) project and CERN are designing and fabricating superconducting quadrupole magnets for the interaction regions of the High Luminosity Large Hadron Collider (HL-LHC). The triplet is made of three optical elements: Q1, Q2, and Q3. The Nb3Sn quadrupole magnets operate in superfluid He at 1.9 K with a nominal field gradient of 132.6 T/m. The three inner triplet elements are connected together with superconducting buses. The design and fabrication of the through and local buses is carried out at Applied Physics and Superconducting Technology Division (APS-TD) at Fermilab (FNAL). This paper reports the characterization of the bus-bar thermo-electric properties. The bus was tested with a short Nb3Sn magnet (MQXFS1e) in the vertical test facility of the APS-TD at FNAL. The test demonstrated that the bus design is adequate since no spontaneous quench took place up to 17.89 kA current value. Quench propagation velocities were investigated over a range of currents that is typical for accelerator superconducting magnets. Temperature margins were found above that required for the Hi-Lumi triplet bus. The design guarantees the protection of the bus at operational current value according to the quench detection voltage threshold (100 mV) established for the Hi-Lumi LHC interaction region.

Author(s): Vallone, G; Ambrosio, G; Anderssen, E; Bourcey, N; Cheng, DW; Ferracin, P; Grosclaude, P; Guinchard, M; Bermudez, SI; Juchno, M; Lackner, F; Pan, H; Perez, JC; Prestemon, S; Semeraro, M; Triquet, S | Abstract: The Nb 3 Sn low-β quadrupole MQXF is being developed as a part of the High-Luminosity large hadron collider (LHC) upgrade project. The magnet will be produced in two different configurations, sharing the same cross section but with different lengths. A 7.2-m mechanical model of MQXFB was recently assembled at european organization for nuclear research (CERN) with one copper coil, two low-grade coils, and one rejected coil. Coil dimensions were measured with a portable coordinate measurement machine. The coil pack shimming was designed in order to optimize the field quality and the contacts between the coils and the collars. The azimuthal preload target was defined using the short models experience. The mechanical behavior during loading was monitored by means of strain gauges. The results demonstrated that the structure can provide the required prestress to the coils.

The High-Luminosity Large Hadron Collider (HLLHC) upgrade requires new quadrupoles, MQXF, to replace the present LHC inner triplets. The MQXFA magnet is the first prototype that has a 150 mm aperture and uses Nb3Sn superconducting technology in a 4.2 m magnetic length structure. The support structure design of the MQXFA magnet is based on the bladder-and-key technology, where a relatively low pre-stress at room temperature is increased to the final preload targets during the cool-down by the differential thermal contraction of the various components. The magnet support structure components experience different load levels from pre-load to cool-down and excitation. Consequently, a few parts experience high stresses that may cause localized plastic deformations or internal fracture development. The concept presented in this paper for the failure assessment of support structures integrates nonlinear finite element analysis with detailed sub-models and fracture mechanics into an advanced engineering tool. The nonlinear FE solutions enable estimations of the structural response to the given loads, and the advanced fracture analysis with failure assessment diagram (FAD) assesses the structure safety index of results obtained from the FE model. The paper describes how the MQXFA shell end segments are being optimized based on the failure analyses. The high luminosity large hadron collider (LHC) upgrade requires new quadrupoles, MQXF, to replace the present LHC inner triplet magnets. The MQXFA magnet is the first prototype that has a 150-mm aperture and uses Nb3Sn superconducting technology in a 4.2-m magnetic length structure. The support structure design of the MQXFA magnet is based on the bladder-and-key technology, where a relatively low pre-stress at room temperature is increased to the final preload targets during the cool-down by the differential thermal contraction of the various components. The magnet support structure components experience different load levels from pre-load to cool-down and excitation. Consequently, a few parts experience high stresses that may cause localized plastic deformations or internal fracture development. The concept presented in this paper for the failure assessment of support structures integrates nonlinear finite-element (FE) analysis with detailed sub-models and fracture mechanics into an advanced engineering tool. The nonlinear FE solutions enable estimations of the structural response to the given loads, and the advanced fracture analysis with failure assessment diagram assesses the structure safety index of results obtained from the FE model. The paper describes how the MQXFA end-shell segments are being optimized based on the failure analyses.

Author(s): Vallone, G; Ambrosio, G; Anderssen, EC; Bajas, H; Bourcey, N; Cheng, DW; Chlachidze, G; Ferracin, P; Grosclaude, P; Guinchard, M; Bermudez, SI; Juchno, M; Pan, H; Perez, JC; Prestemon, S; Strauss, T | Abstract: The Nb3Sn quadrupole MQXF is being developed as a part of the large hadron collide (LHC) High Luminosity upgrade. The magnet design was tested on 1.5-m-long short models, sharing the same cross section with the full-length magnets. Various azimuthal and longitudinal preloads were applied, studying the impact on the magnet training and on its mechanical performances. The experiments demonstrated the possibility to control the magnet prestress. However, various factors, coil size among the others, may affect the stress variation between and within each winding. This variation could prevent the magnets from reaching the magnet performances, as for example as a result of the critical current reduction of the Nb3Sn strands. This paper analyzes the mechanical performances of the short models, studying in particular the stress variation on different coils. The measured coil size was used as input in the numerical simulations, and the results were then compared with the strain gauge measurements. Finally, the short model experience was used to evaluate the feasibility of a loading operation that does not rely on the strain measurements.

Author(s): Wang, X; Ambrosio, GF; Cheng, DW; Chlachidze, G; Dimarco, J; Ghiorso, W; Hernikl, C; Lipton, TM; Myers, S; Pan, H; Prestemon, SO; Sabbi, GL | Abstract: The U.S. High-Luminosity LHC Accelerator Upgrade Project, in collaboration with CERN, is developing Nb 3 Sn quadrupole magnets (MQXFA) to be installed at the interaction region of the LHC. The project will deliver 20 MQXFA magnets in 10 cold masses. These magnets need to meet the stringent requirements on field quality at the nominal operating current. Compared to the mature NbTi accelerator magnet technology, achieving excellent field quality can be challenging for Nb 3 Sn magnets. To help track, understand, and allow effective correction of geometric field errors, field quality measurements at room temperature during the MQXFA assembly stage was planned for the project. The measurements also intend to evaluate the magnetic axis and twist angle along the magnet aperture. We report the first measurement on a prototype MQXFA magnet using a recently developed measurement system. The magnetic axis and twist angle met the acceptance criteria. Further development needs for the room-temperature measurements were discussed. We expect that statistics obtained from such measurements throughout the project will provide insight into future applications of high-performance Nb 3 Sn accelerator magnets.

During the development of MQXF, the new Nb3Sn quadrupole to be used in the large hadron collider (LHC) inner triplets for the High Luminosity upgrade, three short models were tested: MQXFS1, MQXFS3, and MQXFS5. These models differ in the use of thin or thick laminations for the iron components, in the coil design, and in the superconductive strands, rod restack process (RRP) or powder in tube (PIT). In the MQXF design, the azimuthal prestress is provided at room temperature by means of the bladder-key technology, and it is further increased during the cooldown by the differential thermal contraction of the various components. Four aluminum rods provide the longitudinal prestress. Both systems allow for a flexible control of the amount of prestress applied. As a consequence, it was possible to test the models exploring different azimuthal and longitudinal prestress conditions, in an attempt to understand their impact on the magnet performances. This paper studies the mechanical behavior of these short models, also providing the strain and stresses measured by means of strain gauges installed on the aluminum shell, on the winding poles and on the rods. Finally, the paper compares the measures with the results from finite element (FE) models.

In the framework of the High-Luminosity Large Hadron Collider, the installation of a new generation of quadrupole magnets is foreseen on each side of ATLAS and CMS experiments. The new magnets are based on Nb3Sn technology and shall be able to reach an ultimate current of 17.9 kA with a peak field of 12.3 T in the coil. In 2016 and 2017, the first two short models, called MQXFS3 and MQXFS5, have been tested at 4.2 and 1.9 K in the two new test benches at the European Organization for Nuclear Research. This paper presents the result of the quench performance of the two models; the first magnet reached nominal but failed to reach ultimate, showing detraining in one coil. MQXFS5 reached ultimate performance without any detraining phenomena, validating the PIT conductor used for the first time in this magnet program.

Software-defined networking envisions the support of multiple applications collaboratively operating on the same traffic. Policies of applications therefore require composition into a rule list that represents the union of application intents. In this context, ensuring the correctness and efficiency of composition for match fields as well as the associated actions is the fundamental requirement. Prior work however focuses only on the composition of match fields and assumes simple concatenation for action composition. We show in this paper that simple concatenation can result in incorrect behavior and inefficiency of packet processing. To address this issue, we formalize the action composition problem and propose two graph-based computation models to facilitate efficient composition of action lists. Our proposed approach has been integrated into the CoVisor code base and the evaluation results show its fitness for purpose.

The High Luminosity (HiLumi) Large Hadron Collider (LHC) project has, as the main objective, to increase the LHC peak luminosity by a factor five and the integrated luminosity by a factor ten. This goal will be achieved mainly with a new interaction region layout, which will allow a stronger focusing of the colliding beams. The target will be to reduce the beam size in the interaction points by a factor of two, which requires doubling the aperture of the low-β (or inner triplet) quadrupole magnets. The use of Nb3Sn superconducting material and, as a result, the possibility of operating at magnetic field levels in the windings higher than 11 T will limit the increase in length of these quadrupoles, called MQXF, to acceptable levels. After the initial design phase, where the key parameters were chosen and the magnet's conceptual design finalized, the MQXF project, a joint effort between the U.S. LHC Accelerator Research Program and the Conseil Europeen pour la Recherche Nucleaire (CERN), has now entered the construction and test phase of the short models. Concurrently, the preparation for the development of the full-length prototypes has been initiated. This paper will provide an overview of the project status, describing and reporting on the performance of the superconducting material, the lessons learnt during the fabrication of superconducting coils and support structure, and the fine tuning of the magnet design in view of the start of the prototyping phase.

The first coupling coil for the Muon Ionization Cooling Experiment (MICE) has been tested in a conduction-cooled environment at the Solenoid Test Facility at Fermilab. An overview of the thermal and mechanical performance of the magnet and the test stand during cool-down and power testing of the magnet is presented.

The first MICE Coupling Coil has been tested in a conduction-cooled environment in the new Solenoid Test Facility at Fermilab. We present an overview of the power and quench protection scheme, and report on the electrical and quench performance results obtained during cold power tests of the magnet.

A solid understanding and accurate analysis of the voltage transients in the quench process of superconducting magnets is the basis for verifying quench protection system design. This paper presents an analysis of the voltage spike that triggered a quench in a serially connected set of superconducting solenoid magnets, which are protected by a stack of cold diodes and dump resistors. The voltage development associated with the normal zone growth is analyzed. The magnets were trained close to the full design current and all the terminal voltages of the coils were captured by a specialized quench detection system. In the analyses described in this paper, the different stages of the voltage transient development will be detailed. Comparisons of simulations and the training results are provided to substantiate the mechanism of the quench voltage behavior in this passive protection scheme.

Quenches were simulated for a long solenoid composed of five separately powered coils. Two coils, at one end of the magnet, are separately powered by 300-A power supplies, so that the uniform field section of the magnet is matched to the rest of a physics experiment. The three coils in the uniform field end are connected in series and are powered by a single 300-A power supply. The two end coils of the three-coil set use separate 60-A power supplies to trim the uniform field. Quench back from the 6061-Al mandrel is an important part of the quench protection for the three-coil section. Quench propagation from one separately powered coil to the next was simulated by using the Opera3D program of the Vector Fields. Low current quench simulations showed that some coils carry currents for a long time before quenching. Since the magnet does not quench all at once, there can be unbalanced forces developed in the coils and the thermal shield.

This paper is the final paper in a series of papers that discusses passive quench protection for high inductance solenoid magnets. This report describes how passive quench protection system may be applied to superconducting magnets that are connected in series but not inductively coupled. Previous papers have discussed the role of magnet sub-division and quench back from a conductive mandrel in reducing the hot-spot temperature and the peak coil voltages to ground. When magnets are connected in series, quench-back from a conductive mandrel can cause other magnets in a string to quench even without inductive coupling between magnets. The magnet mandrels must be well coupled to the magnet circuit that is being quenched. When magnet circuit sub-division is employed to reduce the voltages-to-ground within magnets, the resistance across the subdivision becomes the most important factor in the successful quenching of the magnet string.

The superconducting coupling solenoid magnet is one of the key equipment in the Muon Ionization Cooling Experiment (MICE). The coil has an inner radius of 750 mm, length of 281 mm and thickness of 104 mm at room temperature. The peak induction in the coil is about 7.3 T with a full current of 210 A. The mechanical disturbances which might cause the instability of the impregnated superconducting magnet involve the frictional motion between conductors and the cracking of impregnated materials. In this paper, the mechanical instability of the superconducting coupling magnet was studied. This paper presents the numerical calculation results of the minimum quench energy (MQE) of the coupling magnet, as well as the dissipated strain energy in the stress concentration region when the epoxy cracks and the frictional energy caused by “stick-slip” of the conductor based on the bending theory of beam happens. Slip planes are used in the coupling coil and the frictional energy due to “slow slip” at the interface of the slip planes was also investigated. The dissipated energy was compared with MQE, and the results show that the cracking of epoxy resin in the region of shear stress concentration is the main factor for premature quench of the coil.

The Muon Ionization Cooling Experiment (MICE) consists of eighteen superconducting solenoid coils in seven modules, which are magnetically hooked together since there is no iron to shield the coils and the return flux. The RF coupling coil (RFCC) module consists of a superconducting coupling solenoid mounted around four conventional conducting 201.25 MHz closed RF cavities. The coupling coil will produce up to a 2.2 T magnetic field on the centerline to keep the beam within the RF cavities. The peak magnetic force on the coupling magnet from other magnets in MICE is up to 500 kN in longitudinal direction, which will be transferred to the base of the RF coupling coil (RFCC) module through a cold mass support system. A self-centered double-band cold mass support system with intermediate thermal interruption is applied to the coupling magnet, and the design is introduced in detail in this paper. The thermal and structural analysis on the cold mass support assembly has been carried out using ANSYS. The present design of the cold mass support can satisfy with the stringent requirements for the magnet center and axis azimuthal angle at 4.2 K and fully charged.

The coupling magnet for the Muon Ionization Cooling Experiment has a self-inductance of 592 H and the magnet stored energy of 13 MJ at a full current of 210 A for the worst operation case of the MICE channel. During a quench, the stored energy and the high conductor current density will cause a large temperature rise and induce considerable impact of stresses. One test coil was built in order to validate the design method and practice the stress and strain situation which occurs in the coupling coil. In this study, the analysis on stress redistribution in the sub-divided winding during a quench was performed. The stress variation may bring about failure of epoxy resin, which is the cause of a new normal zone arising. Spring model for impregnating epoxy and fiber-glass cloth in the coil was used to evaluate the mechanical disturbance by impregnated materials failure. This paper presents the detailed dynamic stress and stability analysis to assess the stress distribution during the quench process and to check whether the transient loads are acceptable for the magnet.

The superconducting coupling coil to be used in the Muon Ionization Cooling Experiment (MICE) with inner radius of 750 mm, length of 285 mm and thickness of 110.4 mm will be cooled by a pair of 1.5 W at 4.2 K cryo-coolers. When the coupling coil is powered to 210 A, it will produce about 7.3 T peak magnetic field at the conductor and it will have a stored energy of 13 MJ. A key issue for safe operation of the coupling coil is the thermal stability of the coil during a charge and discharge. The magnet and its cooling system are designed for a rapid discharge where the magnet is to be discharged in 5400 seconds. The numerical simulation for the thermal stability of the MICE coupling coil has been done using ANSYS. The analysis results show that the superconducting coupling coil has a good stability and can be charged and discharged safely.

This paper describes how a passive quench protection system can be applied to long superconducting solenoid magnets. When a solenoid coil is long compared to its thickness, the magnet quench process will be dominated by the time needed for quench propagation along the magnet length. Quench-back will permit a long magnet to quench more rapidly in a passive way. Quench-back from a conductive (low resistivity) mandrel is essential for spreading the quench along the length of a magnet. The mandrel must be inductively coupled to the magnet circuit that is being quenched. Current induced in the mandrel by di/dt in the magnet produces heat in the mandrel, which in turn causes the superconducting coil wound on the mandrel to quench. Sub-division is often employed to reduce the voltages to ground within the coil. This paper explores when it is possible for quench-back to be employed for passive quench protection. The role of sub-division of the coil is discussed for long magnets.

A superconducting coupling solenoid mounted around four conventional RF cavities, which produces up to 2.6 T central magnetic field to keep the muons within the cavities, is to be used for the Muon Ionization Cooling Experiment (MICE). The coupling coil made from copper matrix NbTi conductors is the largest of three types of magnets in MICE both in terms of 1.5 m inner diameter and about 13 MJ stored magnetic energy at full operation current of 210 A. The stress induced inside the coil assembly during cool down and magnet charging is relatively high. In order to validate the design method and develop the coil winding technique with inside-wound SC splices required for the coupling coil, a prototype coil made from the same conductor and with the same diameter and thickness but only one-fourth long as the coupling coil was designed and fabricated by ICST. The prototype coil was designed to be charged to strain conditions that are equivalent or greater than would be encountered in the coupling coil. This paper presents detailed design of the prototype coil as well as developed coil winding skills. The analyses on stress in the coil assembly and quench process were carried out.

The MICE superconducting coupling solenoid magnet is made from copper matrix Nb-Ti conductors with inner radius of 750 mm, length of 285 mm and thickness of 110.4 mm at room temperature. The coil is to be wound on a mandrel made of aluminum. The peak magnetic field on the conductor is about 7.3 T when fully charged at 210 A. High magnetic field and large size make the stress inside the coupling coil assembly relatively high during cool down and full energizing. The shear stress between coil winding and aluminum casing may cause premature quench. To avoid quench potential induced by stress, slip planes were designed for the coil assembly. In this paper, FE models with and without slip planes for it have been developed to simulate the stresses during the process including winding, cooling down and charging. The stress distribution in the coil assembly with and without slip planes was investigated. The results show that slip planes with low friction coefficients can improve the stress condition in the coil, especially reduce the shear stress largely so that improve the stability.

The MuCool program undertaken by the US Neutrino Factory and Muon Collider Collaboration is to study the behavior of muon ionization cooling channel components. A single superconducting coupling solenoid magnet is necessary to pursue the research and development work on the performance of high gradient, large size RF cavities immersed in magnetic field, which is one of the main challenges in the practical realization of ionization cooling of muons. The MuCool coupling magnet is to be built using commercial copper based niobium titanium conductors and cooled by two cryo-coolers with each cooling capacity of 1.5 W at 4.2 K. The solenoid magnet will be powered by using a single 300 A power supply through a single pair of binary leads that are designed to carry a maximum current of 210 A. The magnet is to be passively protected by cold diodes and resistors across sections of the coil and by quench back from the 6061 Al mandrel in order to lower the quench voltage and the hot spot temperature. The magnet is currently under construction. This paper presents the updated design and fabrication progress on the MuCool coupling magnet.

The superconducting coupling solenoid to be applied in the Muon Ionization Cooling Experiment (MICE) is made from copper matrix Nb-Ti conductors with inner radius of 750 mm, length of 285 mm and thickness of 102.5 mm at room temperature. The magnetic field up to 2.6 T at the magnet centerline is to keep the muons within the MICE RF cavities. Its self inductance is around 592 H and its magnet stored energy is about 13 MJ at a full current of 210 A for the worst operation case of the MICE channel. The stress induced inside the coil during cool down and charging is relatively high. Two test coils are to build and test in order to validate the design method and develop the fabrication technique required for the coupling coil winding, one is 350 mm inner diameter and full length same as the coupling coil, and the other is one-quarter length and 1.5 m diameter. The 1.5 m diameter coil will be charged to strain conditions that are greater than would be encountered in the coupling coil. This paper presents detailed design of the test coils as well as developed winding skills. The analyses on stress in coil assemblies, AC loss, and quench process are carried out.

The Muon Ionization Cooling Experiment (MICE) will demonstrate ionization cooling in a short section of a realistic cooling channel using a muon beam at Rutherford Appleton Laboratory in the UK. The coupling magnet is a superconducting solenoid mounted around four 201 MHz RF cavities, which produces magnetic field up to 2.6 T on the magnet centerline to keep muons within the iris of RF cavities windows. The coupling coil with inner radius of 750 mm, length of 285 mm and thickness of 102.5 mm will be cooled by a pair of 1.5 W at 4.2 K small coolers. This paper will introduce the updated engineering design of the coupling magnet made by ICST in China. The detailed analyses on magnetic fields, stresses induced during the processes of winding, cool down and charging, and cold mass support assembly are presented as well.

This paper describes the passive quench protection system selected for the muon ionization cooling experiment (MICE) cooling channel coupling magnet. The MICE coupling magnet will employ two methods of quench protection simultaneously. The most important method of quench protection in the coupling magnet is the subdivision of the coil. Cold diodes and resistors are put across the subdivisions to reduce both the voltage to ground and the hot-spot temperature. The second method of quench protection is quench-back from the mandrel, which speeds up the spread of the normal region within the coils. Combining quench back with coil subdivision will reduce the hot spot temperature further. This paper explores the effect on the quench process of the number of coil sub-divisions, the quench propagation velocity within the magnet, and the shunt resistance.

The technique of subdivision protection is extensively applied to protect large superconducting solenoid magnets such as accelerator magnets and MRI magnets. The number of subdivision and the parameters of protection circuit are usually decided by the quench simulation results. The accurate current data during quench process is important to verify the quench simulation code and estimate the magnet performance. But not like voltage signal, the acquisition of which just needs to attach a pair of voltage taps on both side of each protection resistor, current signal is very difficult to be obtained from cryogenic environment. This paper presents an algorithm to calculate the quench current from the quench voltage measurement result. By this algorithm the temperature rise history of protection resistor is obtained simultaneously.

The ability to correct magnetic field errors in a superconducting undulator is critical for the successful application of these devices in future and existing light sources. These field errors, which can emanate from sources such as machining and coil winding imperfections, can lead to reduced light source performance by introducing errors in both the electron trajectory and the relative phase relationship between the oscillating electrons and the emitted photons. In this work, correction schemes are presented, which use a single power supply along with a superconducting switch network to define the path for the current during undulator tuning. The basic switching concept was previously designed and successfully tested at Lawrence Berkeley National Laboratory; the approach presented here is a significant advancement in generalizing and scaling that core concept. A new fabrication method is presented here, which uses lithographic methods to produce current paths and switch heaters on a superconducting film. The effect of an example corrector current path design on the magnetic field is investigated using the Finite Element Method, and the results at various undulator and corrector energization levels are presented. Experimental results from the heater switch concept are also presented. © 2002-2011 IEEE.

The magnet used for the quench protection comparison has an ID of 1.5 m. At a maximum current of ~ 210-A, the stored energy is ~13 MJ. The impregnated magnet coil is 281 mm long and about 105.6 mm thick. The coil is wound on a 6061-aluminum mandrel. The magnet quench protection system is passive. The magnet coil is subdivided with back-to-back diodes and resistors across each of the coil subdivision to reduce the magnet internal voltages. Conservative quench protection criteria were applied when the magnet was designed. These criteria are presented in this paper. Quench protection of the magnet was simulated using three computer codes from three different places. The results calculated using the three codes are compared to the original magnet quench protection criteria used to design the magnet. The three quench simulation codes assumptions are compared. The calculated hot-spot temperature and peak voltages are compared for the three quench simulation codes.

The key to being able to operate the superconducting solenoids in the Muon Ionization Cooling Experiment (MICE) using cryocoolers running at around 4.2 K is the application of high temperature superconducting (HTS) leads. Because the MICE magnets are not shielded, all of them will have a stray magnetic field in the region where the coolers and the HTS leads are located. The behavior of the HTS leads depends strongly on the HTS material used for the leads, the magnetic field and their warm end temperature. A pair of binary leads consisting of copper leads and HTS leads made from oriented multiple strands of BSCCO wires will be used for electrical transfer of the MICE coupling magnet for the purpose of reducing the heat leak through the leads to 4.2 K region. This paper mainly discusses the detailed design of the HTS leads and their cooling. Protection for the HTS leads during a power failure is discussed as well.