ISTTOK Tokamak Engineering and Operation

This repository contains the source code for the tomography session at the ISTTOK training program on Tokamak Engineering and Operation.

The tomography session takes place on Friday, July 26, 2019 at 9:00 in room C13.

The code is based on a two-camera setup described in the MSc thesis of F. L. Burnay (in Portuguese). The vertical camera is on the top and the horizontal camera is on the low-field side of the vessel, as shown in the figure below. (A third camera placed at the bottom is currently not operational.)

A more detailed discussion of tomography methods can be found in the PhD thesis of P. J. Carvalho, which also describes a previous setup of the tomography diagnostic at ISTTOK.

For different setups, feel free to adapt this code according to the license terms.

Cameras

• Run cameras.py to find the lines of sight for each camera.

  • The two cameras are referred to as vertical (v) and horizontal (h).

  • The pinhole and detector positions are provided in the code, together with étendue values obtained by calibration.

  • The vessel has a circular cross section that is assumed to be centered at (0,0) in the xy-plane.

  • An output file cameras.csv will be created with the start and end positions for each line of sight, together with the corresponding étendue.

Projections

• Run projections.py to find the projections from pixel values to detector measurements.

  • The pixel resolution for the x- and y-axis are defined in the code.

  • When a line of sight crosses a pixel, the contribution of that pixel is assumed to be proportionate to the length of the intersection between the line and the pixel.

  • When a line of sight does not cross a pixel, the contribution of that pixel is zero, since there is no intersection.

  • Each line of sight is weighted according to the corresponding étendue, which was obtained experimentally.

  • The projections will be saved to the output file projections.npy.

Signals

• Run signals.py to get the signals from each camera, for a given shot number.

  • The code uses the SDAS API that can be downloaded and installed from here.

  • The shot number is indicated in the code. (The present setup is valid from shot number 47216 to 47262.)

  • The data acquisition channels that correspond to each camera detector are indicated in the code.

  • The signal offset is removed based on the signal average for t < 0 s.

  • The signals are subsampled from 10 kHz to 1 kHz.

  • The signals are clipped to zero to remove any negative values.

  • The signals data and time are stored separately in signals_data.npy and signals_time.npy.

Reconstructions

• Run reconstructions.py to reconstruct the plasma profile at specific points in time.

  • The regularization is based on the differences between adjacent pixels.

  • The pixels outside the vessel have an additional regularization imposed on them.

  • The regularization weights are indicated in the code.

  • The time points for the reconstructions are indicated in the code.