Fusion energy may be the holy grail of clean energy, but its complex processes and machines are a difficult task. Researchers at the Dutch Institute (different) believe that the fusion solution is not only in a deeper understanding of physics, but also in an engineering approach to self-attack systems.
In the final months of 2023, researchers and engineers from across Europe are taking part in a fusion race at the Calham Science Campus near Oxford. Their joint European Torus (JET) reactors have been European flagship experimental fusion equipment for 40 years, resigning and decommissioning in the new year. The final experimental campaign squeezes every bit of scientific insight from a donut-shaped fusion device.
As the final Harley, the Jet Team can even set a new world record for the most fusion energy released in a single shot. This is an important step in harnessing the power of the sun as a clean energy source on Earth. But this experimental campaign is more than a record.
Heat exhaust
Part of Jet’s final experiment is three carefully tuned “shots” (10-second bursts of 100 million degrees Celsius). This has nothing to do with pushing the reactor into maximum power.
Running the three shots is a team of Dutch researchers, including doctoral candidate Thomas Bossman. Bossman, the control engineer at the Dutch Fusion Institute, temporarily manages the humble gas valve at the bottom of the jet. Its important role: to inhale enough gas to cool and diffuse the hot exhaust near the jet’s inner wall, without adding too much gas to sniff out the core of the plasma (high temperature charged fuel particles) where the fusion reaction occurs.
Credit: Courtesy of UKAEA and Eurofusion
Bossman explains: “In jets, plasma is short and relatively benign. Sturdy wall materials like tungsten can directly manage heat.” Future fusion power plants will have stronger orders of magnitude with more stringent conditions. He continues: “If you don’t manage the heat exhaust very carefully, you can damage the inner wall in a few seconds.”
Engineering approach
Physicists call a government where heat is spreading before touching “detachment” on the walls of the reactor. They have been understanding for years the underlying interactions between the outer plasma edge, the plasma core, and the walls of the reactor. Instead, Bossman and his colleagues were able to fine-tune the control algorithm that guaranteed peeling with just 3 seconds of plasma shots, unaware of the underlying physics.
In his PhD work at Differ, Bosman uses the insight that the physics underlying fusion exhaust may be complex, but the entire system may become entangled in the basic model depending on the puff of cold gas. Bossman recalls: “During these shots on the jet, we studied how the exhaust responds to slow and fast changes in the amount of inflated cooling gas. In principle, it was all the information needed to set the rules for the controller.”
Confidence
Bosman and his colleagues from Control Engineering’s Differe team did not just rush to solve Fusion Research’s epic challenge in half a day. To get there, it took years of physics research to understand the heat problem and develop a scenario to manage it. Without that solid foundation, the bossman controller didn’t know where to start or what to achieve.
After the first step of exhaust control was successful in 2021 with a small Swiss fusion experiment TCV, the team was invited to do the same for the larger German fusion experiment Asdex upgrade. Only then was the Fusion community confident enough in new techniques for experiments dedicated to green light in Jet, UK, where experimental time is a rare resource.
Even after this success, Bossman believes separation remains a challenge. He said: “The optimum fusion performance is close to the upper limit where gas bulging began to affect the plasma core. However, once the system is switched, the controller can keep the exhaust accurately in the desired state.”
The complete scenario
Heat exhaust is a rather simple system in the grand scheme of control engineering. The so-called linear equations are sufficient to model the response of exhaust to external stimuli such as gas puffs. However, Fusion knows more complex and nonlinear challenges, such as how to predict and manage instability like sudden lightning-like energy from the edge of a fusion plasma. And what about long-term issues like slow erosion of reactor walls over several months of operation?
One of the leaders of Fusion Research at Differ is Plasma Physist and Computer Modeler Sven Wiesen, an expert in understanding the many interaction processes that take place at the edge of Fusion Reactor.
Consider the case of Iter, the world’s largest fusion experiment to date. Wiesen said: “We don’t just want to model the very edge of the plasma. What ultimately happens is determined by the upstream process at the center of the plasma.” This includes the effect of the fusion heat generated in the plasma core, the magnetic cage that holds hot masses of charged particles in place, and turbulence that transports the heat outwards to the layer that has been scraped away.
All these processes affect each other in highly nonlinear behavior. Preparation for the ITER design was carried out on computer models such as a scrape-off layer plasma simulator (SOLPS). Wiesen explains: “When ITER designs were completed in the 90s and 2000s, SOLPS was a tool to create a complete exhaust scenario for how to operate it.”
Bells and whistles
From the levels of turbulence and energy explosion to the effects of plasma edges to the walls of the fusion reactor, dissolution can be modelled everything. Wiesen said: “With a complete bell and whistle that includes all the final processes, each scenario took months to calculate on a supercomputer.” Over the years, Wiesen knows that these fidelity models are important for the field. He said, “Simplified code doesn’t cut it. If you do an approximation or skip the process, you won’t get the correct answer.”
High-fidelity models have places to provide a fully detailed working scenario, but the increasing pace of the fusion community requires faster, more agile predictions. Wiesen is part of a movement that uses machine learning and artificial intelligence to train so-called proxy models. They can identify more detailed model output trends and replicate the same overall behavior in minutes rather than months. Or, faster, share physicists. “With the new AI methods, we can now predict future behavior faster than the plasma itself evolves.”
A surrogate model works with confidence if it can interpolate solid data points, but it opens up new possibilities in the world of fusion. Control Engineer Thomas Bosman: “An advanced controller may have a fast model of plasma inside, allowing you to pre-empt what the plasma does. For example, if you’re approaching unnecessary behavior and decide how to fine-tune or avoid conditions, it can look predictive.”
Private party
Asked about applying for a proxy model, Wiesen names stakeholders such as privately funded SPARC projects in the US and public-private project steps in the UK. Interestingly, both have different configurations than existing nuclear reactors such as JET. For example, SPARC operates at much higher magnetic field strength compared to Iter, with the stepped plasma shaped like a cored apple than the doughnut shape found in other reactors. Wiesen: “For all these differences, the underlying physics is basically the same. With proper scaling, if it exists, you can port predictions without any problems.”
Even Iter shares physicists. “To get your first shot from the licensing agency, you need to prove that it will pass in advance.” The Iter is twice the size of its direct predecessor, Jet. However, Wiesen believes that high-speed models could extrapolate the performance of previous devices to ITER parameters in a reliable way. “It would show the confidence they need to approach.”
Network Approach
Do fusion control algorithms always need to rely on physical understanding? Entering the field from a PhD in Systems Theory, researcher Lizan Kivits from Differ Doc has a surprising view on this. Within this topic, network identification is a very new field, with the first scientific publication in 2013. It could also revolutionize the understanding of highly connected networks of components.
Essentially, the field of systems theory is a very mathematical field. Network theory can be used to describe any network, from thousands of components on modern computer chips to millions of components in a fusion reactor. The approach is to focus on interconnected subsystems within a network where both the behavior of the subsystem and their interconnections are considered.
Kivits said: “When my colleagues talk about the physics of plasma, I think it’s difficult to follow them.” However, as soon as she gets a measurement of the physical amount, Kibitt can identify the dynamic interactions between them from the experimental data. She says, “For example, how heat at the core of a plasma affects the number of plasma particles at the edge.” Without using all physical formulas.
Both the engineers and physicists at Differ agree that fusion physics and control theory have a lot to offer each other. Thomas Bossman said: “The academic community is developing very advanced smart control algorithms, but real industrial applications use very simple linear control schemes. Fusion can be one of the few areas where these sophisticated control algorithms need to be implemented in practice.
This article will also be featured in the 23rd edition of Quarterly Publication.
Source link
