Description
Vertical displacement events (VDEs) [1] are one of the main causes of major disruption that could damage plasma facing components, especially in reactor-size tokamaks. Elongated plasma, an effective approach to enable high plasma confinement, is inherently vertically unstable. Additionally, perturbations in other parameters, such as poloidal beta βp, internal inductance li and toroidal current density jphi, can also increase the vertical growth rate of VDEs, making it very difficult to precisely control the plasma vertical position. In this work, an integrated workflow is developed to verify and optimize VDE predictions in the HL-3 tokamak. The workflow is divided into two parts, i.e. plasma control and the non-linear plasma model. The plasma model is designed to replace the plasma evolution when there is no discharge commissioning, while the plasma control loop remains the same as the one embedded in the PCS. The objective of feedback (FB) control is R, Z and Ip, with the control strategy based on coil voltages [2]. The reference trajectories for R, Z, Ip and coil voltages (VPF) in feedforward (FF) settings are determined through optimizations or trial and error by pilots. At the core of the workflow is a non-linear plasma mode, which is numerically coupled using the free-boundary equilibrium code FEEQS.M [3] and the fast transport code METIS [4]. The integrated workflow is first used to verify #3293 with a VDE in the HL-3. At the beginning of the simulation, the plasma boundary is forced to match the magnetics-constrained EFIT reconstruction. As the simulation progresses, the plasma boundary deviates from the EFIT reconstruction, particularly during the transition from limiter to divertor. This discrepancy arises because the FB strategy regulates R, Z rather than iso-flux or gap-based approach. However, the boundary remains close to EFIT due to the influence of robust (VPFFF). The overlapping βp and li between FEEQS.M and METIS further demonstrate consistency between equilibrium and transport simulations. Finally, the similarity of Ivv compared to experimental data indicates that a reliable model for vacuum vessel has been employed in FEEQS.M. With the aforementioned verification, the workflow is then applied to optimize the nominal shot to avoid VDE-induced disruptions. It is found that, using the experimental controller gains with optimization for RRef, the dynamic scenario remains a smooth shape transition and jphi diffusion. As an alternate approach, deploying advanced control strategies, such as iso-flux and gap control, to replace simple RZIp loop can also lead to an optimized scenario for VDE avoidance.
Key words：Vertical Displacement Events、Feedback Control、Free-boundary Equilibrium、Fast Transport
References
[1] E. Lazarus et al 1990 Nucl. Fusion. 30 111.
[2] X. Song et al 2021 Nucl. Fusion 61 086010.
[3] J. Blum et al 2019 J. Comp. Phys. 394 594.
[4] J-F. Artaud et al 2018 Nucl. Fusion 58 105001.