Jackub Urban (IPP Prague), the developer of the WOrkfloW framework in Python (WOWP), visited the GA theory group for two weeks to embed the WOWP software into OMFIT. WOWP is a workflow engine that adopts a data-driven workflow execution. The various components of the workflow, called actors, are organized into a oriented graph. The resulting graph allows the workflow engine to automatically execute the actors based on their data-dependencies. Furthermore, the framework can naturally take advantage of the graph knowledge to execute components in parallel, when that is possible, thus greatly simplifying the task of leveraging high performance computing platforms for the execution of complex workflows. The OMFIT and WOWP frameworks are complementary to one another, and their integration greatly enhances each other’s capabilities. With Dr. Urban, OMFIT was also installed at IPP Prague and was further extended to work with the COMPASS experimental database (CDB). Profile reconstruction of the experimental COMPASS tokamak data was achieved using the OMFITprofiles module, thus providing the first step towards kinetic equilibrium reconstructions for COMPASS.
Further analysis of the NIMROD simulations of massive gas injection (MGI) into non-rotating plasmas with pre-existing large islands (Feb 19 highlight) has found that the appearance of a 4/2 harmonic of the imposed 2/1 mode plays a crucial role in altering the rate of parallel impurity spreading and the consequent distribution of radiated power. The impurities spread most rapidly at q=2 due to the short parallel connection length leading to fast temperature equilibration along field lines. The presence of a large 2/1 island somewhat alters this picture and slows the parallel impurity spreading. When the 2/1 island is broken up into smaller island chains by the appearance of the 4/2 harmonic, the impurity spreading rate increases. Although the 4/2 appears only for some 2/1 island phases (relative to the gas jet), an additional simulation showed that the deliberate imposition of a 4/2 harmonic could reproduce the rapid spreading that occurs when the 4/2 appears spontaneously.
The kinetic energetic particle (EP) transport code EPtran has been applied to analyze a DIII-D tilted Neutral Beam Injection discharge that was used in a recent validation study in collaboration with a Peking University graduate student Sheng He. The more accurate EPtran code recovers results from the ALPHA EP density transport code used in the previous validation study, with the exception that ALPHA overestimates the effect of the weak ITG/TEM very low energy EP transport at the discharge center. The central EP pressure becomes more peaked than the experimental pressure. This disagreement is addressed by including EP drift orbit effects in EPtran. This actually resolves a long-standing weakness of the Alfven eigenmode critical gradient model, which typically acts only at mid-core r/a ~ 0.5, making it difficult to account for the central EP transport.
A previously developed refractory pellet ablation model has been applied to analyze Lithium, Beryllium, and Boron ablation rates and incorporated into the PELLET source code. An “ablation coefficient”, Ca— analogous to a dimensionless drag coefficient in the aerodynamic formula - depends on pellet specific thermo-physical quantities and other parameters, and can only be determined numerically by solving the gas dynamic equations describing the ablation flow for each parameter set. This is a time consuming process because for each run an iteration procedure has to be performed to match boundary conditions of the ablation flow variables at the pellet surface with certain phase transition relations. For efficient use in the PELLET code, we derived a fitting function for the numerically generated Ca data, which provides startlingly accurate results over a relatively wide range of inputs. Its dependence on the pellet radius and plasma density also satisfies known functional requirements in two asymptotic limits. In the limiting case and/or where pellet shielding becomes vanishingly small, the correct scaling should be that of a “bare pellet”. In the extreme limit of large pellet radius and/or high plasma density, the ablation coefficient becomes a numerical constant.
Teobaldo Luda, Orso Meneghini, Phil Snyder, Gary Staebler, and Ron Waltz attended the US Transport Task Force (TTF) Workshop in Denver, Colorado this week. A new working group led by Gary Staebler was initiated at the Workshop to promote the use of predictive modeling in the planning of experiments. Phil Snyder and Anne White gave presentations to the group showing examples of the successful use of predictive modeling to plan experiments. The first meeting was well attended and the group had a good discussion of the process and benefits of predictive experimental design.
These highlights are reports of research work in progress and are accordingly subject to change or modification