Development of a parallelized open-source python library for synthetic diagnostics and inversions for fusion devices

Virtually all magnetic fusion devices resort to tomography diagnostics for a variety of plasma emissions. Reconstructing the signal from a simulated emissivity requires modeling the geometry and is used for code validation or diagnostic design. Solving the inverse problem is useful for data interpretation and requires geometry modeling and inversion-regularization routines. An open-source parallelized python library was developed to provide a common and reliable tool for solving the direct and inverse problems for synthetic diagnostics.

Transcript

1. Development of a parallelized open-source python library for synthetic diagnostics and inversions for fusion devices Laura S. Mendoza1, Didiver Vezinet2 1INRIA Grand-Est, TONUS Team, Strasbourg, France 2CEA, Cadarache, France ICIAM 2019, Valencia, Espa˜ na
2. Context: Energy generation Current solutions present some drawbacks: Limited resources Production of carbon dioxide Radioactive waste Not too efficient Harmful to surrounding environment ⇒ Fusion reactor: cleaner, more reliable, more powerful energy source ? ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 2
3. Context: Controlled fusion and magnetic confinement D-T Fusion reaction n n n n n n Deuterium Tritium Helium Temperature > 100 Million◦K. ⇒ Gas composed of positive ions and negative electrons: plasma ⇒ Plasma responds strongly to electromagnetic fields ⇒ Energy breakeven point still not obtained: Q = Eoutput Einput = 0.67 ⇒ Current reactors: differnt shapes, sizes, heating methods, confinement techniques, etc. ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 3
4. Toakmak diagnostics Diagnostics: Set of instruments to measure for the uderstanding, control and optimization of the plasma performance. Magnetic diagnostics: currents, plasma stored energy, plasma shape and position; Neutron diagnostics (ie. cameras, spectrometers, etc.): fusion power; Optical systems (interferometers): temperature and density profiles; Bolometric systems (tomography): spatial distribution of radiated power; Spectroscopic: X-ray wavelength range, impurity species and density, input particle flux, ion temperature, helium density, fuelling ratio, plasma rotation, and current density. Microwave diagnostics probe the main plasma and the plasma in the divertor region in order to measure plasma position. ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 4
5. Tomography diagnostics Mi(t) = ˚ Vi # » ε(x, t) · #» n Ωi dV Direct problem (synthetic diagnostic): Simulated emissivity −→ integrated measurements Spatial integration Inverse problem (tomography): Integrated measurements −→ Reconstructed emissivity Mesh and basis functions construction, spatial integration, data filtering, inversion routines, etc. Tomography very sensitive to errors, noise and bias −→ Reputation for low reproducibility / reliability ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 5
6. A code for Tomography for Fusion Develop a common tool: Accessible to everyone (open-source) Generic (geometry independent) Portable (developped in Python) Optimized (reliability and performance) Documented online Standardization of diagnostics Long-term costs saving to the community For tomography diagnostics: The Tomography for Fusion code (ToFuab) arepository: https://github.com/ToFuProject/tofu bdocumentation: https://tofuproject.github.io/tofu/index.html ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 6
7. Tofu’s structure Geometry Data Simulated emissivity Spatial Integration Reconstructed emissivity Inversion routines Basis functions generation Geometry matrix Data Treatment ToFu Exp. Data ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 7
8. Geometry reconstruction: ray-tracing techniques To reconstruct emissivity we need to take account: Geometry defined with minimal data polygon (R, Z) extruded along ϕ Symmetry of vessel along ϕ Upto hundreds of structural elements in vessel Scale of the vessel: 104 bigger than smaller structural detail ICIAM 2019 – Laura S. Mendoza (laura.mendoza@inria.fr) Wednesday 17th July, 2019 8