Samenvatting
The tokamak is the leading reactor design for economically viable clean energy production through thermonuclear fusion. The heat and particle exhaust in such a reactor is guided to a dedicated region called the divertor, where the plasma impacts the divertor targets (i.e. dedicated wall plating). The expected power fluxes impacting the divertor targets during operation in reactor conditions (core temperatures of over 150 million degrees) will exceed present-day materials engineering limits. Therefore, active exhaust mitigation control is required for viable operation, foreseen to be achieved by tailored injection of neutral gas which locally dissipates the fluxes impacting the divertor targets.
However, the development of control-oriented models for these exhaust controllers is still in its infancy. Dynamic investigation of the exhaust problem in fusion, required to benchmark developed models, is still relatively scarce, which we hope to overcome by introducing system identification methods.
In this contribution we show a summary of system identification experiments performed on the research fusion reactor Tokamak à Configuration Variable (TCV) at the EPFL in Lausanne, Switzerland. These experiments were performed to start a database of dynamic measurements for benchmarking control-oriented models, and to demonstrate the use of system identification for systematic design of controllers in fusion. Short and expensive experimental time, in combination with harsh signal-to-noise ratio environments, requires the use of state-of-the-art methods to attain usable data. Due to this conditions
only a few multi-sine (1-5) frequencies can be excited per experiment, and only for three to four periods. We show a range of experimental results which identify the exhaust plasma response to different species of neutral gas injection using special odd and odd-odd multi-sine perturbations, which are analyzed using the local polynomial method. The results have successfully been used for designing a controller and closing the loop.
However, the development of control-oriented models for these exhaust controllers is still in its infancy. Dynamic investigation of the exhaust problem in fusion, required to benchmark developed models, is still relatively scarce, which we hope to overcome by introducing system identification methods.
In this contribution we show a summary of system identification experiments performed on the research fusion reactor Tokamak à Configuration Variable (TCV) at the EPFL in Lausanne, Switzerland. These experiments were performed to start a database of dynamic measurements for benchmarking control-oriented models, and to demonstrate the use of system identification for systematic design of controllers in fusion. Short and expensive experimental time, in combination with harsh signal-to-noise ratio environments, requires the use of state-of-the-art methods to attain usable data. Due to this conditions
only a few multi-sine (1-5) frequencies can be excited per experiment, and only for three to four periods. We show a range of experimental results which identify the exhaust plasma response to different species of neutral gas injection using special odd and odd-odd multi-sine perturbations, which are analyzed using the local polynomial method. The results have successfully been used for designing a controller and closing the loop.
Originele taal-2 | Engels |
---|---|
Aantal pagina's | 1 |
Status | Gepubliceerd - 28 sep. 2022 |
Evenement | 30th Workshop of the European Research Network on System Identification, ERNSI 2022 - Leuven, België Duur: 25 sep. 2022 → 28 sep. 2022 Congresnummer: 30 |
Congres
Congres | 30th Workshop of the European Research Network on System Identification, ERNSI 2022 |
---|---|
Verkorte titel | ERNSI 2022 |
Land/Regio | België |
Stad | Leuven |
Periode | 25/09/22 → 28/09/22 |