Validation of a full-plasma integrated modeling approach on ASDEX Upgrade

Abstract

In this work we present the extensive validation of a refined version of the integrated model based on engineering parameters (IMEP) introduced in reference (Luda et al 2020 Nucl. Fusion 60 036023). The modeling workflow is now fully automated, computationally faster thanks to the reduced radial resolution of the TGLF calculation, and it includes the modeling of the toroidal rotation, which was still taken from experimental measurements in our previous work. The updated model maintains the same accuracy as its previous version when tested on the cases presented in the initial publication. The confined plasma, from the magnetic axis to the separatrix, is simulated without using any experimental information from profiles measurements, and the inputs of IMEP are the same engineering parameters used when programming a plasma discharge. The model validation database consists of 50 ASDEX Upgrade (AUG) stationary (over a few energy confinement time) H-mode phases, which largely cover the entire AUG operational domain. The prediction of IMEP is compared with experimental measurements and with scaling laws, such as the IPB98(y,2), the ITPA20-IL, and AUG specific regressions. This modeling framework has proven to be very accurate over the entire set of 50 cases, with a significantly lower mean relative error with respect to each of the scaling laws considered, accurately reproducing the change in pedestal and core confinement caused by a change in plasma current, heating power, fueling rate, triangularity, magnetic field, NBI voltage (i.e. the effect of a change in the core particle source), and heating mix (e.g. correctly predicting the effect on confinement caused by a change in T e/T i). Plasma confinement is correctly described by IMEP also for two particular operating regimes, such as the ITER baseline scenario, and the QCE regime (quasi continuous exhaust, also referred as type-II and small ELMs). This work clearly demonstrates the power of this approach in pulling out physics mechanisms to interpret subtle interdependencies and that a 1D integrated model can reproduce experimental results over very large parameter variations with a higher accuracy than any statistical regression. This approach has therefore the potential to improve the prediction of the fusion performance in future tokamak reactors.

Full-radius integrated modelling of ASDEX Upgrade L-modes including impurity transport and radiation

“…A clear objective for future work is to extend the applicability of the framework to full-radius simulations of seeded H-mode plasmas. This could be done by including a description of impurities in IMEP [12,13]. Three key elements to be addressed, moving from L-mode to H-mode simulations, are the estimation of turbulent impurity transport coefficients in the pedestal, the impact of impurities on the SOL model that provides boundary conditions at the separatrix in IMEP, and effects of high impurity content and radiation on the pedestal stability.…”

Towards fully-predictive transport modelling in ASDEX Upgrade H-modes

“…The IMEP workflow [3,4] has been applied for full-radius modelling of ASDEX Upgrade discharges, not making use of any experimental input and limiting empirical device-related information to a minimum. The boundary condition of the IMEP workflow in ASTRA is set to the separatrix, with the formulas in [14], equations ( 5) and (8), respectively for T e,sep and n e,sep .…”

“…For ions it is assumed T i,sep = 2 * T e,sep , a robust observation in ASDEX Upgrade, due to the lower ion parallel heat conductivity in the SOL. This applies in case of attached divertor, which is the case in the discharges in [3,4]. The effects of momentum losses, power losses and divertor heat flux broadening are described by a coefficient c pres , derived by regression analysis of ASDEX Upgrade measurements.…”

“…The IMEP workflow prescribes the selection of the simulation with the highest stable pedestal pressure profile according to the MISHKA code [17]. With the original IMEP workflow as in reference [3], typically 20 ASTRA-MISHKA simulations were necessary to evaluate the target pedestal; in the refined workflow discussed in [4], each database points requires only 12 simulations. A comparison with the widely known EPED model [18] is discussed in detail in reference [3].…”

“…In this paper we show recent progress both in integrated modelling [3,4] and in the detailed validation of quasi-linear theory-based transport models in the core, for both electronheated and ion-heated ASDEX Upgrade H-mode plasmas. Two quasi-linear transport model are considered: TGLF [5] and QuaLiKiZ [6], the latter also in its Neural-Network version [7], which is extremely fast and already validated on a few tokamaks.…”