Compressing the Timeline to a Fusion Future

General Fusion worked with Veryst Engineering to simulate and better understand the internal behavior of a fusion demonstration machine and predict its performance.


By Joseph Carew
July 2025

Magnetized Target Fusion (MTF) power plants have the potential to produce significant amounts of energy with comparatively inexpensive technology and without releasing carbon emissions. In a General Fusion MTF machine, hydrogen plasma (such as deuterium–tritium, or D-T) is injected into a liquid metal vessel formed inside the fusion machine. From there, an array of pistons compress and reshape the liquid metal vessel around the plasma, increasing the density and temperature of the plasma to fusion conditions — making fusion happen! It is a pulsed approach that repeats once per second in a commercial plant. The liquid metal wall of the vessel captures the energy of the neutrons, converts it to heat, and then carries it to a heat exchanger to create steam and ultimately produce electricity.

The engineers at General Fusion are using MTF to bring fusion power to the commercial power grid by the early to mid-2030s. This journey has reached a major milestone with the construction of Lawson Machine 26 (LM26), General Fusion's large-scale fusion demonstration machine. Through the use of multiphysics simulation, General Fusion has successfully designed and optimized LM26 and made a major step toward the team's fusion-powered vision for the future.

LM26 Relies on Electromagnetic Compression of a Lithium Liner to Reach Fusion Conditions

In late 2024, LM26 (Figure 1) was built to de-risk General Fusion's eventual commercial MTF machine. The team's project goals for LM26 are to achieve 1 keV, then 10 keV, and finally, to achieve the equivalent of scientific breakeven.

Figure 1. The team at General Fusion assembling LM26.

LM26 is designed to inject a spherical tokamak plasma into the chamber and compress it using a solid lithium liner to reach fusion-relevant plasma temperatures. During the compression, a toroidal magnetic field is generated by an axial current flowing in the hourglass structure that keeps the plasma confined and stable. This process creates the plasma temperatures and densities that lead to the fusion of plasma ions and the release of energy in the form of neutrons.

Figure 2. A rendering of General Fusion's commercial MTF machine design.

In General Fusion’s MTF power plant concept (Figure 2), pistons trigger the compression process instead of magnetic coils like in LM26. Pistons are unique to General Fusion’s approach, as other fusion methods rely on superconducting coils, lasers, or other expensive equipment.

"For MTF, the larger the initial plasma volume, the more time it can stay hot, which gives us more time to compress the plasma to fusion conditions," said Jean-Sebastien Dick, an engineering analysis manager at General Fusion. "At General Fusion, we have been iterating on the process and developing a power plant operating point that is not only commercially viable but also very competitive against the other type of energies on the market."

Modeling Magnetomechanical Compression of a Solid Lithium Liner

LM26 is targeting the key temperature thresholds of 1 keV, 10 keV, and the equivalent of scientific breakeven by compression of a solid lithium liner. With the COMSOL Multiphysics® software, the team was able to model and measure the internal effects to predict the performance of the LM26 design (Figure 3).

"When I joined General Fusion, looking at the type of challenges we were facing, I thought that COMSOL Multiphysics would be a great tool to add to our list of software," Dick said. "This is what got us to interact with Veryst Engineering, who are well-known experts in the field of multiphysics. They know COMSOL very well."

Figure 3. The LM26 geometry.

Calibrating a Lithium Material Model with Help From Veryst Engineering

General Fusion’s partnership with Veryst Engineering, a COMSOL Certified Consultant and engineering consulting company that specializes in highly nonlinear simulation and material modeling, was essential in the development of LM26. Sean Teller, a principal engineer at Veryst, worked alongside Dick to develop material models that enabled the team to accurately simulate the response of the lithium liner. This information was critical for accurate, predictive modeling of the LM26 liner trajectories, which enabled General Fusion to create and assemble LM26.

As Teller explained, "We used COMSOL Multiphysics simulation with integrated experimental plans and validation to enable the team at General Fusion to quickly iterate on designs of LM26. The predictive models are critical for achieving fusion conditions on the road to viable and abundant clean fusion power."

One of these experimental tensile tests included measuring the material response of solid lithium. Using a high-speed camera and impact load cells, Veryst and General Fusion heated the lithium with a pair of ceramic heaters and pulled the sample to failure in order to measure the stress versus strain response (Figure 4). The results of these experiments were then used to calibrate a Johnson–Cook model (Figure 5).

Figure 4. This experimental setup features a tensile test of the lithium (center, black with silver dots) between two black ceramic heaters.

"The full model is quite complex," Teller said. "It utilizes a moving mesh for the lithium and the compressed plasma as well as nonlinear solid mechanics and the Johnson–Cook material model, and EM forces from modeled circuits drive the compression of the lithium. The lithium liner impacts the hourglass device, so capturing the nonlinear contact is crucial to perform accurate predictions." To add to the complexity, heat transfer occurs between all of these model components.

Different LM26 designs were able to be weighed simultaneously and in the same space in the COMSOL® software. Veryst and General Fusion used a time-dependent, fully coupled solver and automatic remeshing to capture the large deformation and pressures inside LM26.

"All of this required tight integration between the physical tests and the finite element models to gain fundamental insights into this compressor design," Teller said.

Figure 5. The plot shows the experimental data (square data points) as compared to the material model prediction (dashes).

During the validation campaign of the models, which led to the development of LM26, General Fusion compressed 40 lithium liners using electromagnetic compression to validate the COMSOL model. The team conducted physical experiments using a small-scale prototype of the compression system (Figure 6).

Figure 6. Prototype 0, the small-scale version of LM26's compression system.

To measure the deformation of the liners, General Fusion developed a structured light reconstruction (SLR) technique. This involved the use of sheets of laser light to extract the velocity at multiple points in the liner. General Fusion also used photon Doppler velocimetry to measure the velocity of the center point of the lithium liner. This combination enabled General Fusion to recreate the deformation observed in the physical experiments and compare that to the simulation results for validation. They then used that rate- and temperature-dependent material model (Figure 5) in subsequent simulations of the plasma compressor and found good agreement between the test data and the data acquired through previous tests.

"The performance of the compressor would not have been possible without the insights gained from the early simulations and the multiphysics model, in particular, that helped drive this design," Teller said. "These validations increase confidence in future modeling efforts to further drive the devices to achieve the Lawson criterion and clean power."

Working with Compressor Impedance

One of the key components of LM26 is its electromagnetic compressor. This portion of the machine is responsible for the rapid compression of the magnetized plasma. A successful electromagnetic compressor design must be able to match its impedance with the compression time of the machine. When the impedance and compression follow the same time scale, that enables an "efficient compression". An "efficient compression" converts a significant fraction of the initial stored electrical energy into kinetic energy.

Modeling and simulation enabled General Fusion to adjust the impedance of the power supply, see how design alterations would impact the performance, and maximize the compression efficiency.

In order to tune the impedance, Dick used the software to make adjustments to the number of turns in the compressor's coils, altered the initial distance between the liner and the coils (known as the "air gap"), and altered how the liner was compressed over its trajectory. Additionally, the liner shape along the compression needs to be controlled in order to ensure that the plasma stays stable, requiring iteration to the liner thickness and the axial spacing between the coils. Dick solved the model after making different design adjustments and compared the results to see if the machine could achieve a stable plasma compression.

"We have done multiple material characterization campaigns to make sure that this liner behaves as expected under the high strain rates and the high plastic strains we are experiencing in these compressors," Dick said.

Validating the Models

As with the experimental setups for the validation of the material characteristics, General Fusion performed internal validations to ensure that the advection of the magnetic field by the liner correctly matched the analytical predictions.

"We have done a separate test with each of these coils to tune their resistance and inductance in their circuits and make sure that they match as close as possible with what we measured with flux loops in our experiments," Dick said.

Dick relied on a 2D axisymmetric simulation (Figure 7) of the machine's operation in order to decrease the solving time.

Figure 7. A 2D axisymmetric model of the machine, showing the hydrostatic pressure contours in the lithium and poloidal magnetic flux contours in all bodies. The central conducting cones are seen on the bottom, and the coils connected to the power supply are the gray bodies on top.

Gaining Data Points with Simulation

"The framework of COMSOL has allowed us to incrementally build in complexity, build confidence in our design intentions, and avoid having to reiterate the design phases," Dick said. "We have not had to change any major parts of these experiments. They were always behaving as intended."

The team at General Fusion is also able to run its simulations in a much shorter time frame. This is thanks to the team's use of the Cluster Sweep node in COMSOL Multiphysics®, which makes it possible to create one large cluster job that spans a number of nodes. The more nodes that have been added directly relates to the amount of parameter values that are computed in parallel. General Fusion used this to tackle multiple parameters in a quicker fashion.

"In the past, running these simulations would have taken multiple weeks or even months, but now we are doing this in less than 24 hours," Dick said. "We are able to get hundreds of simulations done in that time span on our cluster."

The team was able to use the data from the simulation to develop a safe operating space for the machines with confidence.

The Future of Simulation and Fusion Power

Simulation enabled the General Fusion team to incrementally add in complexity as they developed their LM26 design, and combining real-life experimentation with multiphysics simulation was critical in the development of the model. Multiphysics simulation and fusion research will continue to be inextricably linked as the team pushes fusion power to new heights.

LM26 (Figure 8) achieved "first plasma" in February 2025 and is now forming plasmas regularly as General Fusion’s team optimizes performance in preparation for its next step — compressing plasmas to create fusion and heating from compression.

Figure 8. The fully assembled LM26.

"We are no longer just looking at aerodynamics or fluid dynamics or structural dynamics one sector at a time; we are putting everything together," Dick said. "I really like the way COMSOL is approaching the world of simulation. I like the canonical, incremental approach where you can build in complexity by assembling the physics like pieces of a puzzle. I believe that this is the way the future will go for simulation, because new innovations are very complex and require a lot of different physics to be involved together."