Nuclear energy has tremendous potential as a reliable source of cheap, clean energy. The key impediments to expanded use of nuclear energy are safety (Chernobyl and Fukushima), proliferation (Iran, N. Korea); and disposal of waste which remains dangerously radioactive for thousands of years. Presently nuclear waste is stored and managed by the power plant itself as an interim solution, with the intent of eventually transporting it for long term storage. However political, social, and environmental issues places barrier on selection of any such site, for example rejection of the Yucca Mountain Repository proposal. New generations of reactors are designed to generate less waste, alleviating but not solving the problem.
In our research we discuss a new option: Accelerator Driven Subcritical System (ADSS) or simply Accelerator Driven Systems (ADS), a reactor that consumes rather than produce waste. In this case a proton accelerator is used to produce external neutrons and cause fission of actinide rich conventional radioactive waste. ADS can also use abundant (compared to uranium 235) natural isotopes, uranium 238 and thorium and produce waste with very little contamination of long-lived radioactive isotopes as its fuel. This kind of reactor is subcritical, passively safe, and since the fuel used is non-fissile and unsuitable for weapon grade, it is also free of proliferation risks.
ADS research is ongoing around the world, particular in China, India, and the European Union. One of the major challenges, and the one studies in this work, is building a proton accelerator with unprecedented efficiency, intensity, and reliability. Here in the US, an accelerator called Project X is under development at Fermi National Accelerator Laboratory (Illinois) which is intended primarily for basic research, but it also meets the intensity and efficiency requirements for ADS. The reliability required by ADS is >99% whereas present accelerators have around 85%.
In our research we start with a systematic study of accelerator reliability using Project X as a test bed. We use commercial software called Availability Workbench by Isograph to build the reliability block diagrams of the accelerator systems. Reliability modeling is not trivial because the components interdependencies are pretty complex; further it is a challenge to model such a large system. Here we discuss about the various modeling techniques and validate our methodology by modeling some of the subsystems of the Spallation Neutron Source (SNS), a proton linac at Oak Ridge National Laboratory. SNS is the only linac which has a reliability study performed for the whole system using a spreadsheet. Being a proton linac, it is similar to Project X in many respects.
Reliability analysis and modeling helped us to understand the shortcoming of the design and identify the weak components. To improve system availability, which is of main concern to ADS, automated controls and repair is highly desired. One of the biggest sources of downtime in a linac is beam tuning, which can take several hours and presently done manually. Since that is unacceptable for ADS, we explore automated beam tuning methods. We develop beam models in State Space Method so that traditional control techniques can be applied. Further we discuss the use of χ^2 Minimization Technique for actual beam control. Our tests are done in simulated scale down versions of a linac, and further research is needed to validate the actual implementation of our concept.