I lead the MFE Integrated Modeling group at the MIT Plasma Science and Fusion Center. Our work combines first-principles turbulence simulations, integrated modeling and machine-learning-based optimization to predict and design the performance of magnetic-confinement fusion devices. These are the main projects I work on, and some I worked on before; the complete list of papers is in Publications.
Current research
PORTALS: burning-plasma profiles from nonlinear gyrokinetics
Predicting the temperature and density profiles of a fusion plasma with the most accurate turbulence simulations (nonlinear gyrokinetics) used to be too expensive. PORTALS treats the transport equations as an optimization problem and solves them with surrogate models and Bayesian optimization, reaching the steady-state profiles with a minimal number of nonlinear gyrokinetic simulations and no loss of accuracy.
PORTALS made possible multi-channel predictions of burning-plasma profiles with nonlinear CGYRO simulations, for SPARC and for the ITER baseline scenario, including alpha heating, radiation and energy exchange. It has also reproduced experimental profiles, fluxes and turbulence measurements in DIII-D and in a negative-triangularity plasma in ASDEX Upgrade. PORTALS is open source, as part of the MITIM-fusion framework.
Main papers
- Nonlinear gyrokinetic predictions of SPARC burning plasma profiles enabled by surrogate modelingP. Rodriguez-Fernandez et al., Nucl. Fusion 62, 076036 (2022)
- Enhancing predictive capabilities in fusion burning plasmas through surrogate-based optimization in core transport solversP. Rodriguez-Fernandez et al., Nucl. Fusion 64, 076034 (2024)
- Prediction of performance and turbulence in ITER burning plasmas via nonlinear gyrokinetic profile predictionN. T. Howard et al., Nucl. Fusion 65, 016002 (2025)
- Simultaneous reproduction of experimental profiles, fluxes, transport coefficients, and turbulence characteristics via nonlinear gyrokinetic profile predictions in a DIII-D ITER similar shape plasmaN. T. Howard et al., Phys. Plasmas 31, 032501 (2024)
- Gyrokinetic profile prediction and validation of a negative triangularity plasma in ASDEX UpgradeR. Bielajew et al., Nucl. Fusion 65, 086042 (2025)
Links
MAESTRO: integrated modeling for reactor design
A reactor design needs the whole plasma at once: core profiles, pedestal, equilibrium, heating and divertor limits all depend on each other. MAESTRO couples PORTALS with solvers for each of these, so that steady-state plasmas are predicted self-consistently with full physics models, fast enough to explore and optimize the design of fusion power plants.
The surrogate-based solver is well suited to coupling with external codes, and new surrogate techniques handle the discontinuities in the transport fluxes that come from numerical issues or from very stiff physical instabilities. MAESTRO is part of the open-source MITIM-fusion framework.
Main papers
- Accelerating integrated modeling with surrogate-based optimization: the MAESTRO workflowP. Rodriguez-Fernandez et al., Plasma Phys. Control. Fusion, published online (2026)
Talks
- Accelerating core plasma design and optimization with surrogate-assisted transport workflowsWorkshop IV: Multi-Fidelity Methods to Enable Robust Optimization and Real-Time Control of Fusion Processes, May 2026
▶ Click the image to watch the talk
- Designing tokamak core plasmas: from scaling laws to first-principles simulationsWorkshop III: Fusion Device Design and Engineering, May 2026
▶ Click the image to watch the talk
Links
Core performance of the SPARC tokamak
SPARC is the compact, high-field tokamak designed by Commonwealth Fusion Systems and MIT to produce net fusion energy (Q>1) for the first time in magnetic confinement fusion. I lead the predictions of its core plasma performance, which informed its design and are part of the published SPARC physics basis.
Integrated modeling simulations let us look into the physics of SPARC plasmas before they exist: what regimes to expect, what performance SPARC can achieve, and the kinetic profiles that other physics studies build on. The first predictions were published in 2020 in a special issue of the Journal of Plasma Physics on the SPARC physics basis, featured in The New York Times. Since then they have been refined with nonlinear gyrokinetic simulations of burning-plasma profiles and of the plasmas projected for SPARC’s first campaign.
Main papers
- Predictions of core plasma performance for the SPARC tokamakP. Rodriguez-Fernandez et al., J. Plasma Phys. 86, 865860503 (2020)
- Overview of the SPARC physics basis towards the exploration of burning-plasma regimes in high-field, compact tokamaksP. Rodriguez-Fernandez et al., Nucl. Fusion 62, 042003 (2022)
- Core performance predictions in projected SPARC first-campaign plasmas with nonlinear CGYROP. Rodriguez-Fernandez et al., Phys. Plasmas 31, 062501 (2024)
- Gyrokinetic simulation of turbulence and transport in the SPARC tokamakN. T. Howard et al., Phys. Plasmas 28, 072502 (2021)
- Overview of the SPARC tokamakA. J. Creely et al., J. Plasma Phys. 86, 865860502 (2020)
- Projections of H-mode access and edge pedestal in the SPARC tokamakJ. W. Hughes et al., J. Plasma Phys. 86, 865860504 (2020)
Links
Physics basis of the ARC fusion power plant
ARC is the fusion power plant that Commonwealth Fusion Systems plans to build in Virginia, designed for about 1.1 GW of fusion power and at least 400 MW of net electricity. Our group predicts its performance and transport with integrated modeling and nonlinear gyrokinetics, as part of the ARC physics basis published in 2026.
The work also studies the effect of: pedestal conditions, tungsten and other impurities, the plasma composition and shape, and the effect of fusion-born alpha particles on turbulence.
Main papers
- Overview of the physics basis for the ARC fusion power plantJ. C. Hillesheim et al., J. Plasma Phys. 92, E69 (2026)
- Performance and transport in the ARC tokamakN. T. Howard et al., J. Plasma Phys. 92, E67 (2026)
- Core-edge integrated modeling of ARC: on the effect of impurity transport and detachment conditionsM. Muraca et al., Nucl. Fusion, published online (2026)
- Impact of energetic alpha particles on core turbulence in an ARC-class fusion power plantJ. Hall et al., arXiv preprint (2026), submitted to Nucl. Fusion
- Steady state, core, operational optimization of an ARC-like tokamak via plasma composition and shapeA. Saltzman et al., arXiv preprint (2026)
Alpha particles and turbulence in burning plasmas
In a burning plasma, the alpha particles born from fusion reactions become the main source of heat, but their effect on confinement is still under study. With collaborators, we study it with simulations that evolve turbulence, alpha heating and the plasma profiles together, and find that fast particles can suppress ion-scale turbulence and improve confinement.
Alpha particles weakly destabilize Alfvén eigenmodes, which drive zonal flows that shear the turbulence apart. Less heat is lost, the profiles peak, alpha heating grows, and the loop reinforces itself: a feedback with no analogue in present-day devices, where external heating dominates. Similar effects from heating-generated fast ions could raise the fusion gain of SPARC plasmas.
Main papers
- How Fusion-Born Alpha Particles Suppress Microturbulence in Burning PlasmasA. Di Siena et al., arXiv preprint (2026)
- Impact of energetic alpha particles on core turbulence in an ARC-class fusion power plantJ. Hall et al., arXiv preprint (2026), submitted to Nucl. Fusion
- Predictions of improved confinement in SPARC via energetic particle turbulence stabilizationA. Di Siena et al., Nucl. Fusion 63, 036003 (2023)
Testing turbulence models against experiments
Predictions are only as good as the models behind them. Together with experimental teams, we compare gyrokinetic and reduced turbulence models with measurements of turbulent fluctuations and with how transport and rotation change with the hydrogen isotope (hydrogen, deuterium, tritium) in ASDEX Upgrade, JET and DIII-D.
At ASDEX Upgrade, Correlation Electron Cyclotron Emission measures electron temperature fluctuations that are compared with synthetic diagnostics applied to the simulations, across L-mode, I-mode, H-mode and negative-triangularity plasmas. At JET, Ohmic and L-mode plasmas in hydrogen, deuterium and tritium were compared for the first time, showing how intrinsic rotation depends on the isotope mass, and modeled with JETTO-TGLF.
Main papers
- Gyrokinetic profile prediction and validation of a negative triangularity plasma in ASDEX UpgradeR. Bielajew et al., Nucl. Fusion 65, 086042 (2025)
- Investigation of collisionality as a mechanism for confinement saturation in L-mode, I-mode, and H-mode plasmas with strong electron heating at ASDEX UpgradeC. Yoo et al., Phys. Plasmas 33, 032508 (2026)
- Integrated modelling simulations of Ohmic and L-mode JET discharges in H, D and T using JETTO-TGLFH. G. Dudding et al., Plasma Phys. Control. Fusion 68, 075030 (2026)
- Fusion plasma turbulence research beyond the burning plasma era: perspectives on transport model validation in fusion and fissionA. E. White et al., Front. Nucl. Eng. 3, 1380108 (2024)
- Isotope effects on energy transport in the core of ASDEX-Upgrade tokamak plasmas: Turbulence measurements and model validationP. A. Molina Cabrera et al., Phys. Plasmas 30, 082304 (2023)
- Isotope effects on intrinsic rotation in hydrogen, deuterium and tritium plasmasM. F. F. Nave et al., Nucl. Fusion 63, 044002 (2023)
- Edge turbulence measurements in L-mode and I-mode at ASDEX UpgradeR. Bielajew et al., Phys. Plasmas 29, 052504 (2022)
Earlier work
Perturbative transport in tokamaks
Why does the temperature at the center of a tokamak increase when the edge is cooled down? During my Ph.D. at MIT, supervised by Prof. Anne White, we explained this 20-year-old puzzle with local turbulent transport, with no need for nonlocal effects.
The work combined experiments in three devices (Alcator C-Mod, DIII-D and ASDEX Upgrade) and simulations with two leading transport solvers (TRANSP and ASTRA), in collaboration with scientists at MIT, General Atomics, Princeton and the Max Planck Institute for Plasma Physics. It led to an article in Physical Review Letters, my Ph.D. thesis (Del Favero Doctoral Thesis Prize) and a review in Reviews of Modern Plasma Physics.
▶ Click the image to watch
Main papers
- Explaining Cold-Pulse Dynamics in Tokamak Plasmas Using Local Turbulent Transport ModelsP. Rodriguez-Fernandez et al., Phys. Rev. Lett. 120, 075001 (2018)
- Perturbative transport modeling of cold-pulse dynamics in Alcator C-Mod Ohmic plasmasP. Rodriguez-Fernandez et al., Nucl. Fusion 59, 066017 (2019)
- Predict-first experiments and modeling of perturbative cold pulses in the DIII-D tokamakP. Rodriguez-Fernandez et al., Phys. Plasmas 26, 062503 (2019)
- On the correlation between ‘non-local’ effects and intrinsic rotation reversals in Alcator C-ModP. Rodriguez-Fernandez et al., Nucl. Fusion 57, 074001 (2017)
- The local nature of the plasma response to cold pulses with electron and ion heating at ASDEX UpgradeC. Angioni et al., Nucl. Fusion 59, 106007 (2019)
- Local transport dynamics of cold pulses in tokamak plasmasP. Rodriguez-Fernandez et al., Rev. Mod. Plasma Phys. 6, 10 (2022)
- Perturbative transport experiments and time-dependent modeling in Alcator C-Mod and DIII-DP. Rodriguez-Fernandez, Ph.D. Thesis, Massachusetts Institute of Technology (2019)
In the press
- Integrated simulations answer 20-year-old question in fusion researchMIT News, February 2018
- Pablo Rodriguez-Fernandez resolves a fusion paradox to receive Del Favero PrizeMIT News, November 2019
Surrogate techniques for model validation
VITALS (Validation via Iterative Training of Active Learning Surrogates) explores efficiently the large parameter space needed to validate plasma transport models against experiments. It uses the same genetic algorithms aided by surrogate models that I developed for turbine design.
▶ Click the image to watch
Main papers
- VITALS: A Surrogate-Based Optimization Framework for the Accelerated Validation of Plasma Transport CodesP. Rodriguez-Fernandez et al., Fusion Sci. Technol. 74, 65–76 (2018)
- Criteria for the importance of multi-scale interactions in turbulent transport simulationsA. J. Creely et al., Plasma Phys. Control. Fusion 61, 085022 (2019)
Links
Turbine optimization with evolutionary algorithms
With Prof. Giacomo Persico at the Laboratorio di Fluidodinamica delle Macchine of Politecnico di Milano, I developed shape-optimization tools for turbomachinery blades of Organic Rankine Cycles: genetic algorithms for constrained, multi-objective problems, with Gaussian process regression to cut the number of CFD simulations.
On an ORC benchmark case, the workflow halved the pressure losses of a single blade with respect to the baseline design, a 20% improvement over earlier gradient-based methods.
Main papers
- High-Fidelity Shape Optimization of Non-Conventional Turbomachinery by Surrogate Evolutionary StrategiesG. Persico et al., J. Turbomach. 141, 081010 (2019)
- Automatic design of ORC turbine profiles using evolutionary algorithmsP. Rodriguez-Fernandez et al., Proceedings of the 3rd International Seminar on ORC Power Systems (ASME ORC 2015) (2015)
- Development of shape-optimization tools for the aerodynamic design of turbomachinery bladesP. Rodriguez-Fernandez, M.Sc. Thesis (Proyecto Fin de Carrera), Universidad Politécnica de Madrid (2015)