Research

The MFE-IM group is interested in developing and utilizing workflows for tokamak scenario modeling, with an emphasis on core turbulence and transport physics. We are building expertise in the following workflows to predict core plasmas:

• Empirical modeling (low-fidelity): We use Plasma OPeration Contour (POPCON) models to scope the operational space of fusion devices. We work on refining POPCON models by accounting for uncertainties in input parameters (via robust optimization) and assumptions (via uncertainty quantification), refined physics models and engineering considerations.
  E.g. P. Rodriguez-Fernandez et al. Nucl. Fusion 62, 042003 (2022)
• Reduced-transport modeling (medium-fidelity): We use quasilinear transport models such as the Trapped Gyro-Landau Fluid (TGLF) model for the quick evaluation of linear micro-instabilities and the estimation of transport fluxes. We also use such models within time-independent flux-matching frameworks such as TGYRO and PORTALS and time-dependent scenario modeling toolsets such as TRANSP and ASTRA.
  E.g. P. Rodriguez-Fernandez et al. Journal of Plasma Physics, 86(5), 865860503 (2020), C. Holland et al. J. Plasma Phys., vol. 89, 905890418 (2023)
• First principles transport modeling (high-fidelity): We use gyrokinetic simulations with the CGYRO code to study the nonlinear physics of core turbulence in tokamaks. We have also developed workflows that allow us to use nonlinear CGYRO simulations to directly predict core gradients and fusion performance (PORTALS framework).
  E.g. P. Rodriguez-Fernandez et al. Nucl. Fusion 62, 076036 (2022), N. T. Howard et al. Physics of Plasmas 28, 072502 (2021)

We have contributed to the physics basis of the SPARC tokamak and we continue fruitful collaborations with Commonwealth Fusion Systems as they develop a robust experimental plan to achieve net plasma gain and burning plasmas in the SPARC tokamak. Members of MFE-IM also lead the SPARC Performance & Transport (P&T) group whose mission is to provide physics-based simulations of SPARC and ARC scenarios.

As a group, we are interested in the use of medium and high fidelity transport modeling directly during the design of burning plasma experiments and fusion pilot plants. We believe that to achieve fusion device solutions that are both feasible and economically attractive, we must utilize core transport physics models from the beginning of engineering design.

The confidence in the predictions of future fusion devices relies upon our capability to reliably simulate current experiments. In MFE-IM, we are interested in using high-fidelity transport simulations to simulate core plasmas in existing and past tokamaks (DIII-D, ASDEX Upgrade, Alcator C-Mod and JET) and compare them to diagnostic measurements of kinetic profiles, turbulence and inferred transport coefficients. Over the years we have built tools, workflows and metrics to help understand where and when transport models fail in describing the physics of the tokamak core.

While we are involved with numerous core transport validation exercises, the areas of special interest for MFE-IM are:

• Multi-scale turbulence and isotope effect
  Even though small scale fluctuations are usually expected to have a small impact on transport in realistic tokamak core plasmas, in some special situations they (and their coupling to larger scales) might be necessary to reproduce experiment results. This is known as multi-scale or cross-scale turbulence effects and the simulations were pioneered by MFE-IM scientist N.T. Howard in the 2010s. Understanding in which situations multi-scale simulations are required remains an important area of research and of interest to the MFE-IM group, including whether or not they can contribute significantly to the so-called isotope effect observed experimentally in current day devices.

  

• Impurity transport
  While most validation exercises of the core of tokamaks are performed with respect to energy fluxes, impurity transport is gaining attention in the fusion community, as future burning plasmas and pilot plants will suffer from some level of impurity accumulation. This is especially damaging in the burning-plasma regime and with high-Z wall materials (as most devices are scoped). Understanding how heavy, non-hydrogenic (impurity) ions are transported and distributed in the plasma core is an important focus of the MFE-IM group.

  

• Perturbative transport
  Static, power-balance transport validation studies can only consider single-point, standalone experimental data. Understanding how transport responds to perturbations around stationary conditions is key in revealing whether a transport simulation properly simulates the underlying physics mechanisms or if it matches transport levels fortuitously. In MFE-IM, we are interested in finding out ways to utilize perturbative transport tools (particularly, cold and heat pulses) to validate transport models.

  

• Multi-field, multi-channel validation
  Well-diagnosed tokamak plasmas, particularly those with turbulence measurements, provide an important opportunity to validate transport simulations. Cutting-edge measurements of core fluctuations such as those provided by the Correlation Electron Cyclotron Emission (CECE) diagnostic, can be compared directly to the behavior of nonlinear turbulence simulations. In our group, we collaborate with experimental teams that develop these diagnostics to carry out multi-field (e.g. kinetic profiles, fluctuations), multi-channel (e.g. electrons, ions, impurities) validation of core models.

  

In MFE-IM we are interested in utilizing Bayesian Optimization (BO) techniques to solve complex problems in fusion energy and plasma physics. We developed the PORTALS tool to enable the interaction of standard BO workflows (particularly the BoTorch package) with fusion and plasma physics tools such as TGLF, TGYRO, CGYRO, FREEGS and others.

So far, we have applied BO to the following problems:

• PORTALS: Flux-matching solver of expensive transport models. Using nonlinear CGYRO.
  e.g. P. Rodriguez-Fernandez et al. Nucl. Fusion 62, 076036 (2022)
• FreeGS-U: Grad-Shafranov geometry finder with user-specified constraints. Using FreeGS.
  e.g. P. Rodriguez-Fernandez, lecture at the 2022 MIT Computational Physics School for Fusion Research.
• VITALS: Multi-channel, multi-field, uncertain-input validation of local transport models. Using TGLF.
  e.g. P. Rodriguez-Fernandez et al. Fusion Technol. 74:1-2, 65-76 (2018)
• ESPRESSO: Shape-optimization of high-temperature superconductor solenoids. Using COMSOL.
  In collaboration with N. Riva and S. Packman at the MIT Plasma Science and Fusion Center
  e.g. S. Packman et al. EUCAS 2023

At MFE-IM we are especially interested in finding ways to incorporate physics understanding to aid in the optimization process and in the creation of reliable surrogate models that can accelerate the convergence of Bayesian and surrogate-based optimization workflows.

In MFE-IM we acknowledge that the direct use of full-physics models in scenario modeling tools is cumbersome and can be very computationally expensive. Consequently, an area of active research in MFE-IM is the development of machine-learning enhanced models (such as Neural Networks and Gaussian Processes) to reproduce the output of black-box type simulations of core transport and scenario modeling tools. We are currently working in developing models for impurity transport and energy fluxes specifically for the parameter space of the SPARC tokamak to be implemented in flight-simulators for scenario and control purposes.