PhD defense Gijs Derks 'Dynamic exhaust modeling for fusion reactors'.

On 2 March 2026 Gijs Derks will defend his thesis called 'Dynamic exhaust modeling for fusion reactors'.

• Promotor: Matthijs van Berkel
• Co-promotors: Egbert Westerhof and Sven Wiesen

Summary
Developing plasma exhaust models for the operation of fusion power plants
The fusion energy that powers the Sun is one of the few virtually inexhaustible and inherently safe energy sources with the potential to meet humanity’s growing energy demand. In an effort to harness fusion energy on Earth,the fusion research community builds experimental reactors that trap fusion fuel and heat it to over 150 million degrees Celsius. Facing extremely hot fuel, it is a major challenge to prevent the reactor wall from melting. Within his research, Gijs Derks developed models for the interaction of the hot fuel with the wall. These models can be used for the operation of power producing reactors in the future.

The heat exhaust challenge
Fortunately, at 150 million degrees Celsius, the fusion fuel transitions from a gas into a plasma: a hot cloud of electrically charged particles that interact with magnetic fields. For example, solar winds (streams of plasma) interact with the Earth magnetic field and are redirected to the north pole, lighting up the atmosphere. This we observe as the northern lights. Similarly, fusion reactors use magnetic fields to prevent extremely hot plasma from direct contact with the wall and instead guide it to an exhaust region with heat resistant target plating. Left untouched, however, the stream of hot plasma exhausted onto these target plates greatly exceeds present-day materials limits and melts them.

To protect the target plates, the stream of hot exhaust plasma must be actively cooled down. This is achieved by injecting gases into the exhaust region to establish a gas atmosphere. Similar to how Earth’s atmosphere protects us from solar winds, the gas atmosphere lights up by the exhaust plasma stream and cools it down significantly such that the wall can be protected. However, injecting too much gas can cool the reactor core and can hamper fusion energy production. This careful balancing act can only be performed by a control algorithm that automatically adapts the gas injection.

Models give direction
The design of these control algorithms requires time-dependent models that tell us something about behavior of the reactor, and (un)expected events. To answer questions as: Can we deal with a change in fusion power? For this case, can we ensure that the wall does not melt? Complex models have been developed in recent decades to describe the interaction between the hot core plasma and the reactor wall, but time-dependence has received little attention. This PhD research developed a simple time-dependent 1D plasma exhaust model called DIV1D.

The DIV1D model assumes that the plasma stream in magnetic fields behaves similar to a river flowing in a riverbed. As a river widens, the flow slows down. One can use the river widening to estimate the flow speed, rather than considering the entire riverbed. Similarly, DIV1D uses the width of the magnetic fields in which plasma exhaust flows. In this way, DIV1D describes the main features of the plasma exhaust, which is demonstrated by comparisons to a 2D model that uses the full magnetic fields.

A reality check
To test whether the time-dependence in DIV1D is good enough to design a control algorithm, its behavior is compared with experimental data from a research reactor in Switzerland. The data was specifically gathered to design an exhaust controller. The comparison shows that DIV1D can align with the measurements if it is coupled to a core plasma and a neutral gas atmosphere outside the plasma. Findings in this thesis provide a step towards the knowledge base for the design of exhaust controllers in future fusion reactors; ones that should produce net energy.