DIII-D RESEARCH OPPORTUNITIES FORUM FOR THE 2008 EXPERIMENTAL CAMPAIGN Login | Review | Submit | Logout | Help

Questions about this website? Contact Andrew LeBlanc
Questions about ROF? Contact Chuck Greenfield

Print this page
Title 97: Operational implementation of model-based shape control (MIMO)
Name:Michael Walker () Affiliation:General Atomics
Research Area:Model based Control Presentation time: Requested
Co-Author(s): D.A.Humphreys, J.A.Leuer
Description: The purpose of this experiment is to test revised versions of multivariable shape controllers on DIII-D, test algorithms that support the use of such controllers, and study the resulting effects of such controllers on shape and stability control. This work is a continuation of the testing of implementation of components of the combined inner loop, outer loop, and feedforward control approach which was begun following the first implementation of a MIMO (multiple-input-multiple-output, but here used to also refer to model-based design) controller in 1999. The long term objectives of this work are to provide integrated control capability for AT operation, to ensure robust stability (including reducing plasma oscillations), obtain more precise shape control, deal effectively and systematically with control limitation problems such as coil current sharing/fighting, coil current limits in high performance discharges, and VFI bus voltage limitations on control, and to provide methods for reducing shape development time.
Experimental Approach/Plan: We propose to use a small number of dedicated 2-hour experiments, plus several piggyback experiments to complete the implementation and test the multivariable controllers that support all patch panels for LSN, USN, and DND plasmas. A final half-day experiment would be used to perform a systematic evaluation of the capabilities of such controllers.
Background: Design of multivariable linear controllers requires a sufficiently good model of the plant response. Over the last several years, detailed models of DIII-D coils, vessel, power supplies, and plasma response have been developed and validated. These models have been used both in designing linear controllers and constructing simulation tools for testing of the controllers. A controller developed using these tools was first successfully tested experimentally in 1999. A more sophisticated version, which dealt with nonlinear limitations such as coil current limits and the conflict between vertical and shape control algorithms, was successfully tested experimentally in 2005 and 2006 on LSN and USN plasmas respectively. The extension of this algorithm to DND plasmas and to one particular DND configuration (nearly complete), which was known to be difficult to control, was performed during 2007. The two-hour blocks used to test the multiple required algorithm upgrades were essential to debug hardware and software problems; many problems appeared only during operational use, not in simulation. A new PCS test bed, which duplicates a portion of the PCS hardware for testing purposes, is expected to substantially reduce the number of problems discovered only during operation, which should significantly reduce the requirement for dedicated experimental testing time and remove much of the risk of hosting a piggyback experimental test.
Resource Requirements:
Diagnostic Requirements: Fast magnetics data acquisition desirable for piggybacks, necessary for dedicated exp.

Essential diagnostics 2 kHz magnetics desirable during piggyback test intervals, 5 kHz magnetics required during dedicated experimental time; voltage (fault system) diagnostics;
Power supply fault data, PCS data acquired at maximum rate for a portion of discharge.

Highly desirable diagnostics Thomson (multipulse core, divertor), tile current array, SPRED, MDS, MSE, fast SXR, vessel motion diagnostic, Vert./Horiz. ECE, CER, Zeff, Bolometers, interferometry.
Analysis Requirements: --
Other Requirements: The PCS test bed system needs to remain working and representative of the DIII-D PCS during all of operations.