Journal of the Korean Physical Society, Vol. 54, No. 5, May 2009, pp. 1925 1930
Operational Experiences of the Spallation Neutron Source Superconducting Linac and Power Ramp Up Sang-Ho
Kim
Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, U.S.A.
for all SNS teams
(Received 4 September 2008) The spallation neutron source (SNS) is a second generation pulsed neutron source and designed to provide a 1-GeV, 1.44-MW proton beam to a mercury target for neutron production. Since the commissioning of the accelerator complex in 2006, the SNS has started its operation for neutron production and beam power ramp-up has been in progress toward the design goal. All subsystems of the SNS were designed and developed for substantial improvements compared to existing accelerators because the design beam power is almost an order of magnitude higher compared to existing neutron facilities and the achievable neutron scattering performance will exceed present sources by more than a factor of 20 to 100. In this paper, the operational experiences with the SNS Superconducting Linac (SCL), Power Ramp-up Plan to reach the design goal and the Power Upgrade Plan (PUP) will be presented including machine, subsystem and beam related issues. PACS numbers: 29.17.+w, 74.25.Op, 74.25.Bt, 74.25.Fy Keywords: Spallation, Superconducting linear accelerator, Accumulator ring, Neutron source I. INTRODUCTION
The spallation neutron source (SNS) accelerator complex consists of a negative hydrogen (H ) RF volume source, a low-energy beam transport (LEBT) line with a rst-stage beam chopper, a 4-vane radio-frequency quadrupole (RFQ) for acceleration up to 2.5 MeV, a medium-energy beam transport (MEBT) line with a second-stage chopper, six drift-tube linac (DTL) tanks up to 87 MeV, four coupled-cavity linac (CCL) modules up to 186 MeV, a superconducting linac (SCL) with 11 medium-beta cryomodules (up to 379 MeV) and 12 highbeta cryomodules (up to 1000 MeV), a high energy beam transport (HEBT) line, an accumulator ring with associated beam transport line, a ring-to-target beam transport (RTBT) line and a mercury target. At full duty, the linac will produce a 38-mA peak, chopped H beam for 1 ms at 60 Hz. The H beam loose electrons at the injection to the ring where 700-ns long midi-pulse beam is accumulated over 1060 turns through multi-turn chargeexchange injection, reaching an intensity of 1.5 1014 protons per pulse. After beam accumulation in the ring, the beam is extracted by using the extraction kickers during a 300-ns long midi-pulse gap in a single turn and is transported to the mercury target through the RTBR line. Figure 1 shows the layout of the SNS accelerator
E-mail:
[email protected]; Fax: +1+865-574-6617
complex. A series of the beam commissioning tests, initiated in 2002 and completed in May 2006, was performed in seven commissioning runs for Front-End, Drift Tube Linac Tank 1, Drift Tube Linac Tanks 1-3, Coupled Cavity Linac, Superconducting Linac, Accumulator Ring and beam on. Ocial SNS operations for scheduled neutron scattering experiments was started in October 2006. The initial instrument suite was commissioned and the user program began in 2007. The SNS is now nearly two years into the initial operational phase. It was envisioned to ramp-up the beam power to 1.4 MW, the beam availability to 90 % and the accelerator operating hours to 5000 in the rst three years following construction [1,2]. The SNS superconducting linac is the largest application of RF superconductivity to come on-line in the last decade and has been operating with a beam for almost two years. As the rst operational pulsed superconducting linac, many of the aspects of its performance were unknown and unpredictable. Much experience and many data have been gathered on the pulsed behavior of cavities and cryomodules at various repetition rates and at various temperatures during the commissioning of the components and during beam operations. A careful balance has been achieved between safe operational limits and the study of the conditions, parameters and components that create physical limits [3{5]. The SCL is running at a 0.9-GeV output energy and is providing reliable operations. The power ramp up plan to reach
-1925-
-1926-
Journal of the Korean Physical Society, Vol. 54, No. 5, May 2009
Fig. 1. Layout of the SNS accelerator complex. Table 1. High-level beam parameters achieved to date. Parameters Design Beam Energy (GeV) 1.0 Peak Beam Current (mA) 38 Average Beam Current (mA) 26 Beam Pulse Length (s) 1000 Repetition Rate (Hz) 60 Beam Power on Target (kW) 1440 Linac Beam Duty Factor (%) 6 Beam Intensity on Target (protons per pulse) 1.5 1014 SCL Cavities in Service 81
Individually achieved 1.01 40 24 1000 60 550 3.5 1.3 1014 78
Highest production beam 0.89 32 18 580 60 550 3.5 0.53 1014 76
the design goal is set and various e orts are in progress. II. SNS STATUS
Table 1 compares the high level beam parameters achieved individually and altogether for neutron production with the design ones. The column of `Individually achieved' in Table 1 correspond to maximum values with other parameters at a lower condition. For example, the beam current was about 10 mA at 1 Hz for a 1.01-GeV linac output energy. Figure 2 shows the history of the SNS beam power on target during neutron production runs from the start of the ocial operation of the SNS in October 2006. The beam power on target is presently about 0.55 MW and will be ramped up to 0.7 MW at the end of this year. III. SCL OPERATIONAL EXPERIENCES
Extensive testing has been conducted on the SNS superconducting RF modules starting in late 2004. Behavior peculiar to the SNS cavities and the overall design,
Fig. 2. Beam power ramp-up since the ocial operation has started in October 2006. as well as due to the pulse nature of the operation, have been observed. A number of components have shown peculiar behaviors. For instance, the cold cathode gauges, which provide interlock protection to the fundamental power coupler window, have behaved erratically, with a response that is often not correlated at all with other independent measurements of gas pressure at the window. Also, the higher order mode (HOM) notch RF lters, which are designed to extract RF power at fre-
Operational Experiences of the Spallation Neutron Source { Sang-Ho Kim and for all SNS teams
-1927-
quencies potentially harmful to the stability of the beam, have shown signs of multipacting, discharges and transient detuning, which can lead to critical components damage. These and other limitations have required extensive studies and close attention in setting gradients, which are both necessary for adequate energy gain and for reliable operation. 1. Operational Flexibility
One of the main characteristics of the superconducting linac is its exibility and adaptability to adding or removing cavities. On many occasions, cavity elds had to be temporarily or permanently changed, or in some cases, cavities had to be removed from service. The linac
exibility has demonstrated itself in allowing complete retuning and rephrasing of the linac over times on the order of a few minutes. This characteristic is important as not just cavities, but other systems that may be removed temporarily from service and operations, will interrupt operation only for a modest amount of time. 2.
Limits and Limiting Factors of SRF Cavities
The SCL is operating at a 0.89-GeV output energy with 76 cavities in service out of 81 cavities, which is much lower than the design energy. Most of the cavities exhibit heavy eld emission and/or multipacting, which directly or indirectly (through heating of end groups) limits the gradients achievable in normal operation with the beam. The overall phenomena are complex and the nal operational cavity gradients need to be determined individually for each cavity based on the equilibrium between electromagnetic, electron emission and thermal phenomena, each a ecting the overall stability of the system on a pulse-by-pulse basis. In addition to individual cavity eld emission limitations, collective e ects have been observed, which a ect neighboring and second neighbor cavities. Heating of cavity elements are driven not only by the amplitude, but also by the relative phase of neighboring cavities. Since in the SNS linac, neighboring cavities' amplitudes and phases are correlated, operation in a heavy eld emission region is prevented by stability concerns, thus limiting the nal available energy. Figure 3 plots the operating gradients of the SRF cavities in the SNS SCL. Due to the lack of the nal linac output energy, the gradient of each cavity is set to maximize the gradient based on the collective limiting gradients achieved through a series of SRF cavity/cryomodule performance test at SNS, rather than setting uniform gradients as designed. As seen in Figure 3 cavities in the medium-beta section of the SCL are operating above the design gradient of 10.2 MV/m whereas those in the high-beta sections of the SCL are operating
Fig. 3. Accelerating gradients of the SNS SCL. below the design one of 15.8 MV/m, mainly due to radiation and related heating e ects. 3.
SCL Tune-up
Compared with the design gradients, the actual operating gradients have large uctuations. It is necessary to smooth the longitudinal focusing by adjusting the synchronous phase of several cavities, particularly around the unpowered ones, to preserve the beam emittance in the linac. It is also important to model the linac phase oscillation and damping curves to provide helpful information about the longitudinal lattice. It is equally important to optimize transverse focusing in the linac, which is done by applying a code developed in the XAL infrastructure. Tune-up of the cavity phase is based on the phase-scan signature-matching method. During the linac tune-up and beam energy ramp-up, the focusing quads in the SCL and the downstream line are adjusted for several intermediate energies to reduce beam loss. It is time consuming to scan all cavities in the linac, so a fast cavity fault recovery method, based on RF cavity phase scaling has been developed. Phase scaling can adjust the entire linac in a few seconds, as it does not need any beam phase measurement [6]. 4. Beam Loss
Low beam loss was anticipated in the SCL because of the beam pipe being large compared with the normal
-1928-
Journal of the Korean Physical Society, Vol. 54, No. 5, May 2009
Fig. 4. Beam loss monitor (BLM) and neutron detector (ND) readings in the SNS SCL at 550 kW operations. The unit is rad/pulse. The charge per pulse is about 10 C. conducting linac and because of the large beam acceptances in both the transverse and the longitudinal directions. As beam power goes up, an unexpected large beam loss and activation are observed in the SCL. It appears that the SCL beam loss is mainly a longitudinal issue, as the shifting phase of upstream cavities may signi cantly a ect the beam loss pattern. Both simulation and measurement show that the hottest activation spot, located between cryomodule 2 and cryomodule 3, is caused by a beam phase halo at the DTL-CCL transition. The origin of the beam longitudinal halo is not fully understood yet. The total amount of beam in the halo is larger than 1 10 4 and mitigating beam loss and activation in the SCL will be critical for the SNS power ramp-up. In computer models, a perfectly-tuned normal conducting linac provides a much better solution, as growth of halos is controlled and most halo particles are cleaned up in the upstream linac, but this may not be the case in a real machine. Figures 4 show the measured SCL beam loss for a 550-kW neutron production in August 2008. The SCL activation measured 20 hours after the 500-kW production run in July 2008 is shown in Figure 5. From this measurement, we could linearly scale the SNS beam power up to approximately 1 MW and have a tolerable residual activation for hand-on maintenance. More study is, however, necessary to understand the origin of the beam halo and to further reduce the SCL beam loss to protect hardware in the tunnel.
IV. POWER RAMP-UP
Most of the equipment in the SNS requires substantially higher operational ratings compared to existing accelerators, because the design beam power is almost an order of magnitude higher compared to existing neutron facilities. As the beam power was increased at higher duty factor during previous runs, down-times of some equipment, such as the LEBT and the high-voltage convertor modulator, led to lower machine availability than expected. Some systems like the SNS SCL are the rst attempt for pulsed operation. Many of the aspects
Fig. 5. SCL activation measurements 20 hours after a 500kW production run at a 30-cm distance.
Fig. 6. The SNS beam power ramp-up status and plans. of its performance are unknown and unpredictable, for which it takes time to understand the systems as a whole and/or need additional performance improvements. The power ramp-up plan has been revised based on the operational experiences and on understandings of the limits and the limiting conditions through extensive studies, with more emphasis on machine availability (Figure 6). The plan covers the main driving factors for beam power, such as the chopping eciency, ion source improvement, high-voltage convertor modulator (HVCM) improvement and the SCL output energy. The followings are short descriptions of the issues and the plans for the SCL power ramp-up.
1. Output Beam Energy
The nal output beam energy mainly depends on the SCL gradients. Presently, the SCL is providing an output energy of 895 MeV without reserve, which is much lower than the design energy due to the facts mentioned in Sec. III. The plan is to achieve a 1-GeV output en-
Operational Experiences of the Spallation Neutron Source { Sang-Ho Kim and for all SNS teams
ergy, with 30- to 40-MeV energy reserve for fast recovery of operation from unexpected long-lead down time of not only cavities but also related systems. Cryomodule repair works are in progress to get all 81 cavities in service. One cavity in cryomodule 11 is not operable due to large fundamental power coupling through the HOM port. One high-beta cryomodule, which had beam line vacuum leak and showed biggest radiation from RF operation, was removed from the linac tunnel to the SNS SRF test facility for repair. It is revealed that the beam line leak resulted from the HOM feedthroughs. Four leaky feedthroughs were removed out of eight HOM coupler feedthroughs. After RF performance tests in the SNS SRF test facility, this cryomodule will be brought in service in the next year. The other high-beta cryomodule (CM19) was removed from the tunnel and has been repaired by removing HOM coupler feedthroughs from one cavity in the SNS SRF test facility. This cryomodule has been back in service since March 2008 and is the best performing cryomodule in the SNS SCL. Very recently, one cavity, SCL-10b, which showed noisy eld probe signals and had been turned o for about a year at 30- and 60-Hz operations, was repaired in the tunnel. The nal output energy will still be about 950 MeV after all cavities are available, which implies that an active e ort is need to improve cavity performances, especially for high-beta cavities, to get a 100-MeV additional output energy, including a certain amount of energy reserve. As mentioned above, high-beta cavities need about 2.5 MV/m additional performance improvements. An e ort is in progress to develop in-situ surface processing, such as helium processing and plasma processing, for the cryomodules in the tunnel without disassembly. 2. Beam
Available RF Power for a High Intensity
Each SCL cavity is fed by an individual klystron rated at a 550-kW RF output at saturation and has independent RF control systems. The eighty-one klystrons for the SCL are powered by HVCMs; four HVCMs running at 69 kV for twelve klystrons each and three HVCMs running at 71 kV for eleven klystrons each. The voltage of the HVCMs needs to be increased up to 75 kV to utilize the rated RF power of the klystrons. To have high-power ramp-up as planned with good machine availability, a decision was made to have one additional HVCM for the SCL installed early next year so that most of the SCL HVCMs will power 10 klystrons at 75 kV with fair reliability. 3. Pulse Width
The beam pulse width is presently a major driving factor for the SNS power ramp-up and depends mainly
-1929-
on the HVCM pulse width, the stability of the chopper and the warm linac conditioning especially the RFQ for a longer pulse. There will be another beam pulse extension which will be obtained by reducing the SCL cavity lling time from 300 s to 250 s with an additional HVCM for the SCL, as mentioned above. V. POWER UPGRADE
Many of the accelerator subsystems are designed to be able to support higher beam intensities and higher beam energies. Upgrades to the SNS accelerator and target systems to increase the beam power to at least 2 MW, with a design goal of 3 MW, are in the planning stages. A beam power upgrade to 3 MW can be achieved by increasing the linac beam energy from 1.0 GeV to 1.3 GeV by adding 9 additional high-beta cryomodules in the already prepared empty slots in the linac tunnel and by increasing beam current from 38 mA to 59 mA. The duty factor will be kept at 6 % and most of the beam transport lines and the ring have the capability for 2-MW, 1.3-GeV operations. The SNS power upgrade project is waiting for CD-1 approval and the planned project period is about 4 years starting in 2012. VI. SUMMARY
The SNS began the ocial operations in October 2006 after completion of the construction project in June 2006. The SNS is in the middle of a three-year power-ramp-up period to reach the design speci cations and high availability for user services. The presently achieved beam power for neutron production is about 0.55 MW. From series of tests and operational experiences, more understandings of systems and their limiting conditions in pulsed mode are being obtained at high duty operation. We are focusing on reliable, well-balanced and stable operations with high machine availability. The beam power is on track with a power ramp-up plan that was generated based on previous operational experiences and studies. The Power Upgrade Project (PUP), which will double the beam power to 3 MW, is in the planning stage and is waiting for CD-1 approval. ACKNOWLEDGMENTS
The authors extend our thanks to all our SNS colleagues who contributed to this work. SNS is managed by UT-Battelle, Limited Liability Company, under contract DE-AC05-00OR22725 for the U.S. Department of Energy. SNS is a partnership of six national laboratories: Argonne, Brookhaven, Je erson, Lawrence Berkeley, Los Alamos and Oak Ridge.
-1930-
Journal of the Korean Physical Society, Vol. 54, No. 5, May 2009 REFERENCES
[1] S. Henderson, [2] S. Henderson
Proceedings of European Particle Accelera-
tor Conference 2008
(Genoa, 2008), p. 2892. (Edingburgh, 2006), p. 345.
et al., Proceedings of European Particle Ac-
celerator Conference 2006
[3] S.-H. Kim, J. Korean Phys. Soc. 52, 714 (2008). [4] S.-H. Kim and I. E. Campsi, PRST-AB 10, 032001 (2007). [5] S.-H. Kim , (Albuquerque, NM, 2007), p. 2511. [6] J. Galambos (Albuquerque, NM, 2007), p. 885. et al.
Proceedings of the 2007 Particle Accel-
erator Conference
et al., Proceedings of the 2007 Particle Ac-
celerator Conference
