Software‎ > ‎LRDFIT‎ > ‎

LRDFIT Overview

What is LRDFIT?

LRDFIT is an L-R (inductance-resistance) 2D axisymmetric circuit model of a tokamak which can be constrained to fit various combinations of diagnostic data.

In short:  LRDFIT = LR circuit model with Data FITting capabilities

LRDFIT is most commonly used for equilibrium reconstructions, but can also be used to predict time-dependent vacuum field evolution.

(1) Equilibrium Reconstructions - in this mode, the LRDFIT is run in 2 steps:

Step 1 - SVD fit to magnetics data to determine all toroidal currents in system

  • All toroidal currents in the system are determined consistent with the LR circuit model and a "best-fit" to the time-dependent magnetics data. 

  • The toroidal plasma current density is treated as unknown, and is allowed to flow anywhere inside the limiter boundary.

  • Regularization (smoothing) of the plasma current density is utilized to produce an over-constrained system of equations. 

  • The Grad-Shafranov equilibrium force balance constraint is NOT enforced at this stage.

  • The solution of this step provides a good estimate of all toroidal currents in passive conducting structures (vessel, passive plates, etc.)

    • These passive structure currents are commonly treated at "known" in subsequent Grad-Shafranov constrained equilibrium fits

    • Treating these currents as known:  (a) produces similar results to fully time-dependent equilibrium fits, and (b) enables parallelization of the equilibrium fits

  • The solution of this step also provides an initial poloidal flux distribution for the Grad-Shafranov constrained equilibrium fits

Step 2 - Grad-Shafranov equilibrium (GSE) fit to the magnetics and other data

  • The passive conductor currents are treated as known and the coil currents are fit

  • The magnetics data, MSE data, electron temperature iso-surfaces, etc. are used as constraints on the equilibrium fit

  • Basis functions for the p' and FF' profiles are specified, and the "best-fit" weighting of the basis functions is determined

  • The results for individual time-slices are written to files

  • The code can be run interactively, or in parallel/batch mode

  • In batch mode, after multiple time-slices have been computed, time dependent quantities are computed, and the results are archived in MDS+