Sustained nuclear fusion? Not without control engineering

 

Two years ago, ITER brought in Luce to head the Science & Operations Department. A physicist with 37 years experience working with fusion, he heads the department responsible for both the software and the hardware of ITER's plasma control system. "Now, with advances in computer and communications technologies, we have the capacity to apply new principles and techniques in real time and not just in an offline, theoretical sense," says Luce.

 

In the past, most systems had simple input and simple output. But control problems have become more complex, requiring a new generation of algorithms that model much more than just a set of independent sensors and actuators. Changes in one component affect the behaviour of other components. Not only do the new systems have to model the dependencies among the different components, but they also have to model the system as a whole.

 

Because of this interplay among the different inputs and among the different outputs, the matrix models that work so well for control of single-input and single-output (SISO) systems cannot be applied to the control of a fusion reactor, a multiple-input and multiple-output (MIMO) system. Furthermore, to overcome delays in both measurement times and reaction times, the control system has to employ feedforward methods in addition to the more traditional feedback loops.

 

 

A fundamental part of the control, data access and communication (CODAC) system, ITER's plasma control system is responsible for ensuring that each pulse is executed correctly. It does this by taking data from sensors and applying sophisticated algorithms to generate commands that it sends to actuators to control plasma parameters, such as position, shape or stability.

 

"We're starting from a point where we do have experience," says Alberto Loarte, head of the Science Division. "If we compare smaller tokamaks operating at highest power and current to large tokamaks operating at lowest power and current, they run the same. So other than a little tuning, we expect our initial models to be based on the behaviour of known systems such as the JET and JT-60SA tokamaks, which have half the linear dimensions of ITER."

 

"As we move beyond the first stages of operation, we will get into areas where there are more unknowns," says Loarte. "Running our models through simulations, we've found that sometimes when you apply an actuator and expect a certain reaction, the result turns out to be the opposite of what you expected. That's because of the interplay of other factors that are specific to ITER fusion-power-producing plasmas. It's as if you pressed on the accelerator with your foot, and instead of accelerating, the car stops."

 

There is still a lot to learn about system behaviour at later stages of operation, when fusion reactions in the plasma will cause the plasma to self-heat. In these later stages, the control system will apply heat to the plasma, thereby causing fusion—and the fusion will produce heat, thereby raising the temperature of the plasma even more. The relationship between external heat and plasma temperature will no longer be direct, as it is in present tokamak experiments. This makes the job of the plasma control system much more complicated.

 

The stakes are high. Once fusion starts to heat the plasma, if the system fails to adjust exactly as required, the plasma can get too hot for fusion to be efficiently produced. The plasma can even become unstable, causing the plant to stop producing power.

 

Peter de Vries, Scientific Coordinator in the Stability & Control Section led by Joe Snipes, compares the problem to playing mini golf and trying to putt the ball up to a hole on top of a mound. "If you apply too much force you go over the hole and down the other side of the mound. If you apply too little force, the ball doesn't reach the top of the mound and rolls back down. This analogy gives you an idea of how delicate it is for the plasma control system to adjust the external heat."

 

Temperature and density are major concerns—and so is the shape of the plasma, which also determines the behaviour of the reactor. To monitor plasma shape, the tokamak contains hundreds of tiny sensors, which measure the magnetic fields at different locations. All that information is fed into the control system, which then models the plasma shape, and compares that with a reference.

 

"It might find that the plasma wants to bulge out a little here," says de Vries. "So then it calculates what voltages it needs to put on the power supplies to change the currents in all the coils that we have around the machine to push the plasma back into shape. Hundreds of measurements go into a complicated calculation—and out of it go dozens of response requests to make sure everything is kept in the right place, with the right plasma density and temperature."

 

 

Employing some of the latest advances in control engineering, ITER's plasma control system is architected to handle both the basic control functions for early commissioning and the advanced control functions that will be needed for future high performance operation. In preparation for each stage of operation, new functions will be added and existing functions will be adapted to new types of plasmas. All modifications will be integrated into the existing system.

 

In May 2019 the design team reached the halfway point for the final design of the plasma control system for First Plasma operation, the first fully integrated use of all basic tokamak functions. Halfway through scheduled project duration, many of the controllers have been designed and the framework in which the controllers will work is nearly completed.

 

As for the next steps, Peter de Vries says, "After this summer we will begin several months of intensive assessment—testing and correcting—and documentation of the design. If all goes well, we should be finished with system design for First Plasma by this time next year."

 

As holistic and robust control engineering is essential to sustained nuclear fusion, any milestone in the design of the plasma control system is a step forward for ITER.