Simserver Simulation of a Model-Based Current Profile Controller in the DIII-D Plasma Control System
Symposium on Fusion Technology
Porto, Portugal, September 27-October 1, 2010
| Abstract |
|
Setting up a suitable current profile, characterized by a weakly
reversed magnetic shear, has been demonstrated to be a key condition
for one possible advanced scenario with improved confinement and
possible steady-state operation. Active feedback control of the
evolution of q(0) and qmin during the initial phase of the discharge
has been already demonstrated at DIII- D changing the plasma
conductivity through electron heating, and therefore modifying the
rate of relaxation of the current profile [1]. The q profile is
obtained in real time from a complete equilibrium reconstruction using
data from the motional Stark effect (MSE) diagnostic. The controller
requests a power level to the actuator (electron cyclotron heating
(ECH) or neutral beam injection (NBI)), which is equal to preprogrammed
feed-forward value plus the error in q times a proportional gain
(non-model-based P controller). Present limitations of this controller
(oscillations and instability) motivate the design of a model- based
controller that takes into account the dynamics of the q profile
response to the different actuators. Experiments at DIII-D focus on
creating the desired current profile during the plasma current ramp-up
and early flattop phases with the aim of maintaining this target
profile during the subsequent phases of the discharge. Since the
actuators that are used to achieve the desired current profile are
constrained by physical limitations, experiments have shown that some
of the desirable target profiles may not be achieved for all arbitrary
initial condition. Therefore, a perfect matching of the desirable
target profile may not be physically possible. In practice, the
objective is to achieve the best possible approximate matching in a
short time window during the early flattop phase of the total plasma
current pulse. Thus, such a matching problem can be treated as a
finite-time optimal control problem for a nonlinear partial differential
equation (PDE) system (the evolution in time of the current profile,
or alternatively q, is related to the evolution of the poloidal flux,
which is modeled in normalized cylindrical coordinates using a PDE
usually referred to as the magnetic flux diffusion equation). A
control-oriented model of the current profile evolution in DIII-D was
recently developed for the plasma current ramp-up and early-flattop
phases [2], and used to synthesize both open-loop [3] and closed-loop
[4] control schemes. In this work, we report on the implementation of
these advanced model-based current profile controllers in the DIII-D
PCS (Plasma Control System) and on the assessment of these controllers
in Simserver simulations. The magnetic diffusion equation is implemented
in Simserver with the ultimate goal of simulating the time evolution
of the plasma current profile in response to the active controllers
running in the DIII-D PCS.
[1] J. Ferron et al., Nuclear Fusion, vol. 46, 2006, p. L13.
[2] Y. Ou et al., Fusion Engineering and Design, vol. 82, 2007, pp. 1153–1160.
[3] Y. Ou et al., Plasma Physics and Controlled Fusion, vol. 50, 2008, p. 115001.
[4] Y. Ou, C. Xu and E. Schuster, IEEE Transactions on Plasma Science, vol. 38, 2010, pp. 375-382.