Computation of Normal Conducting and Superconducting Linear Accelerator (LINAC) Availabilities
A brief study was conducted to roughly estimate the availability of a superconducting (SC) linear accelerator (LINAC) as compared to a normal conducting (NC) one. Potentially, SC radio frequency cavities have substantial reserve capability, which allows them to compensate for failed cavities, thus increasing the availability of the overall LINAC. In the initial SC design, there is a klystron and associated equipment (e.g., power supply) for every cavity of an SC LINAC. On the other hand, a single klystron may service eight cavities in the NC LINAC. This study modeled that portion of the Spallation Neutron Source LINAC (between 200 and 1,000 MeV) that is initially proposed for conversion from NC to SC technology. Equipment common to both designs was not evaluated. Tabular fault-tree calculations and computer-event-driven simulation (EDS) computer computations were performed. The estimated gain in availability when using the SC option ranges from 3 to 13% under certain equipment and conditions and spatial separation requirements. The availability of an NC LINAC is estimated to be 83%. Tabular fault-tree calculations and computer EDS modeling gave the same 83% answer to within one-tenth of a percent for the NC case. Tabular fault-tree calculations of the availability of the SC LINAC (where a klystron and associated equipment drive a single cavity) give 97%, whereas EDS computer calculations give 96%, a disagreement of only 1%. This result may be somewhat fortuitous because of limitations of tabular fault-tree calculations. For example, tabular fault-tree calculations can not handle spatial effects (separation distance between failures), equipment network configurations, and some failure combinations. EDS computer modeling of various equipment configurations were examined. When there is a klystron and associated equipment for every cavity and adjacent cavity, failure can be tolerated and the SC availability was estimated to be 96%. SC availability decreased as increased separation distance between failures is required for operation. Another configuration is a power supply servicing two or more klystrons and associated equipment. It makes a tremendous difference in availability if a power supply services adjacent klystrons, every other klystron, every third klystron, etc. The SC availability plummets to -82% if the required separation distance between failures is 1 or 2 and the high voltage power supply (HVPS) services adjacent klystrons. However, if the HVPS drives alternate klystrons, a required separation of one cavity between failures can be tolerated. One proposed SC LINAC design configuration has one power supply driving six cavities. In this configuration, any component failure will likely cause LINAC shutdown. The estimated availability for these conditions is 86.3%. This gain over 83% NC availability results from the higher availability of the SC tuner system. Some conclusions of this study are: (1) An SC LINAC availability can be substantially ({approx}13%) higher than an NC LINAC if the capability of the SC LINAC to tolerate cavity failure is harnessed by configurations that allow operation with failure. About 6% of this increase in availability is caused by lower SC tuner system failure rates and repair times. (2) Quadrupole magnets and power supply are primary contributor to SC LINAC unavailability. (3) Power supply to klystrons and the tuner resonance control system are primary contributor to NC LINAC unavailability. In summary, an SC LINAC availability is not inherently higher than an NC LINAC. SC availabilities can be made substantially higher than an NC LINAC availability if the design capitalizes on advantages offered by certain equipment configurations. Even though input failure rate and repair data to these calculations may be uncertain, this analysis indicates that it is unlikely that the LINAC can achieve its high availability requirements of 98% without astute use of equipment configuration to allow fault-tolerant operation.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- DE-AC05-00OR22725
- OSTI ID:
- 885853
- Report Number(s):
- ORNL/TM-2000/93; TRN: US0604161
- Country of Publication:
- United States
- Language:
- English
Similar Records
Simulated performance of the superconducting section of the APT linac under various fault and error conditions
Design, construction, system integration, and test results of the 1 MW CW RF system for the e-gun cavity in the energy recovery LINAC at Brookhaven National Laboratory