02.08.2013 Views

Research Needs for Magnetic Fusion Energy Sciences - US Burning ...

Research Needs for Magnetic Fusion Energy Sciences - US Burning ...

Research Needs for Magnetic Fusion Energy Sciences - US Burning ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

demonstrably reliable control in proximity to stability boundaries, including excursions expected<br />

during nominal operation, as well as off-normal or fault-triggered excursions. here we define offnormal<br />

events as those transient phenomena that lie outside the envelope of noise and disturbances<br />

that can be robustly stabilized. such events may be recoverable (but are not robustly controlled),<br />

or may be unrecoverable; they must be responded to in an appropriate way to minimize<br />

machine damage and down time. Their probability of occurrence must be reduced to an acceptable<br />

level <strong>for</strong> power plant operation, and this must be accomplished through control design. Reliable<br />

operation under nominally disturbed regimes (e.g., intermittent sawteeth, elms, or transient<br />

magnetic island growth) falls under the stability research category. however, novel algorithmic<br />

control and response solutions beyond nominal control are also needed <strong>for</strong> excursions requiring<br />

transient changes of operating regime (e.g., response to loss of a diagnostic channel or key actuator),<br />

rapid but controlled shutdown (e.g., identification of an impending unrecoverable state), or<br />

emergency uncontrolled shutdown (e.g., identification of immediate unrecoverable state and insufficient<br />

time <strong>for</strong> controlled shutdown, often envisioned as serving to mitigate potential system<br />

damage).<br />

enabling elements of these solutions include a quantifiably reliable systematic approach and corresponding<br />

integrated system <strong>for</strong> response to off-normal events and fault-triggered excursions,<br />

real-time predictors <strong>for</strong> proximity to key operational limits, algorithms and mechanisms <strong>for</strong> mitigating<br />

damage, and recovery and cleanup strategies to rapidly restore plant availability. This will<br />

require development and validation of theory-based predictors <strong>for</strong> proximity to stability boundaries,<br />

including assessment of the potential <strong>for</strong> and development of empirical data-based learning<br />

systems. Provable algorithms <strong>for</strong> response to various off-normal or fault events must be developed<br />

to deal with degradation or loss of diagnostic data or of actuator per<strong>for</strong>mance, impending<br />

or current plasma regime excursion, unexpected disturbances, and impending or current loss<br />

of controllability. control scenarios <strong>for</strong> controlled (“soft”) shutdown, as well as <strong>for</strong> preventing or<br />

minimizing device damage during large-scale off-normal events, must be developed.<br />

RELiabiLity anD CERtiFiCatiOn<br />

key solutions are required to quantify per<strong>for</strong>mance, risk, and reliability of an operating fusion<br />

power plant. The results of this area of research will enable licensing, certification, and economic<br />

attractiveness of a realizable power plant. <strong>Research</strong> requirements identified <strong>for</strong> control reliability<br />

and certification include methods <strong>for</strong> executing per<strong>for</strong>mance assessment (systematically quantifying<br />

per<strong>for</strong>mance reliability), and methods <strong>for</strong> failure modes and effects analysis, a systematic<br />

approach to identifying potential system failure modes, causes, and effects on fusion plant operation.<br />

MODELing anD DESign<br />

most of the requirements <strong>for</strong> comprehensive high reliability control of fusion power plants rest on<br />

the ability to model and design control solutions with specified and/or quantifiable per<strong>for</strong>mance.<br />

This in turn requires computational tools <strong>for</strong> producing control-level models, integrated and sufficiently<br />

comprehensive simulations, and real-time predictive models <strong>for</strong> on-line controller adjustment<br />

or operating regime identification. <strong>Research</strong> requirements identified <strong>for</strong> control model-<br />

97

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!