Thermonuclear fusion: Artificial intelligence from the frontiers of science to customer segmentation

Cheap, infinite, safe and clean energy

Artificial Intelligence from Thermonuclear Fusion research to sales generation or risk control.

Today, energy and the environment are two of the major concerns of states and people. Apart from environmental considerations, energy prices have reached unprecedented levels, causing major economic and geostrategic problems.

In this context, to speak of the thermonuclear fusion: an infinite, cheap, safe and clean energy sourcemay seem opportunistic. For those who think this way, let them review, for example, the film Aliens: The Return of 1986, where the atmosphere of the planet where the colony (and the aliens) was located had been made breathable using large industrial infrastructures and thanks to the energy of a fusion reactor. Imagine if in those years it had already reached the cinema, since when were these processes being investigated?

If you allow me, let's look at several topics in order: what is fusion, how can artificial intelligence (AI) help us to obtain this new energy source? And our experience bringing developments from the field of thermonuclear fusion to customer insight and risk prediction and management.

The above path does not seem the most logical: starting from one of the most exciting fields of physics today (the physics of magma generated in a thermonuclear fusion experiment in a tokamak-type device), to the knowledge of my clients, their business development and risk management. This journey has a common thread: the use of artificial intelligence capabilities and the enormous flexibility that allows us to attack different problems, in different sectors, obtaining really important results and returns.

What is thermonuclear fusion?

Let's start with the first question: what is fusion? And for that let's turn to the experts, according to the National Security Council "Nuclear fusion is a nuclear reaction in which two nuclei of light atoms, generally hydrogen and its isotopes (deuterium and tritium), join to form another heavier nucleus, usually releasing particles in the process."

If the mass of the nucleus is lighter than the original masses, energy is generated. This released energy is given by the very famous equation of E=mc² by Albert Einstein. Fusion is not the same as nuclear fission; in the latter, the atom is broken, generating a chain reaction and releasing energy. Fission is the basis of our current nuclear power plants.

And after this paragraph, someone may ask, what is this for?

• For life to exist, fusion is the process that occurs in the sun; in which hydrogen nuclei are fused to form helium.
• If we were able to industrialize fusion power generation, we would have clean, highly efficient and practically inexhaustible energy. 

A fusion reaction can release approximately ten million times the energy released by a chemical process. such as oil burning.

One of the isotopes of hydrogen used in fusion is deuterium, well, the energy content in one liter of seawater is equivalent to the energy content provided by 250 liters of oil.

We can already begin to glimpse the difficulties we have in industrially generating energy from the fusion process: joining two atoms by overcoming the enormous electromagnetic forces requires tens of millions of degrees Celsius. At those temperatures, matter forms what is called magma, which is a different state from the three we were taught at school: solid, liquid and gas. This is how we can understand what the Nobel Prize in Physics Pierre Gilles de Gennes: "We say we will put the sun in a box. The idea is attractive, but the problem is that we don't know how or what to make the box out of. ". 

One of the ways to make such a "box" is in the form of a toroid (like a hollow donut) where the magma is confined, without touching the walls, using large magnetic fields. This device is called Tokamak and is the basis of the design of the device. ITERwhich is perhaps the most ambitious energy project in the world.

Another Tokamak device is the JETWe have worked with him and he will help us to specify how AI can help physicists to understand the laws of fusion and help us to create algorithms and technologies that will allow us to move forward providing services to customers in other sectors, such as security, risk management and even commercial development of channels in mass consumption. These examples will be discussed briefly at the end.

How artificial intelligence can help physicists to obtain energy through thermonuclear fusion

Let us define in the context of the problem:

• The fundamental laws of physics of the plasma that is generated and confined in a device such as JET are not known.
• There is a huge amount of data, coming from the readings of the device's sensors, in each discharge (experiment where a fusion reaction is achieved) and that these data usually present a lot of noise (or what is the same: poor data quality).

For example, the JET database was over 100 TB 5 years ago, and was doubling annually. ITER, will collect between 500,000 and 1,000,000 signals per download; so the expected data rate is Pbytes/year.

• The process is extremely complex, where it would be necessary to optimize three very difficult parameters to bring them at the same time to the values needed to produce energy at industrial level:
  • Plasma density.
  • Highest possible temperature. The highest temperatures in the solar system have been achieved at JET, 10 times larger than at the center of the Sun.
  • Longest possible confinement time.

Given the above context, it is easy to understand that the modeling and analysis capabilities of large amounts of data provided by AI can be very useful in helping to understand fusion processes, to foresee unwanted phenomena and ultimately to control thermonuclear fusion devices in order to produce energy in an industrial and very profitable way. In short, to achieve the so-called "Breakeaven Fusion", to produce much more energy than that needed to start and maintain the fusion process.

Fusion work can be thought of as something to which a lot of resources are devoted but with little impact on people's day-to-day lives. This is not the case, some examples of developments that have been made possible by this research:

• Medicine: Superconducting magnets, which are used in magnetic resonance imaging.
• Environment: Membranes of a palladium alloy, designed for melt wastes, are effective for treating wastes from the chemical and automotive industries.
• Remote control: Remote control techniques that are employed in EUROfusion's Tokamak JET are being applied in high-energy physics, space science, nuclear material dismantling, and current surgical practices.

Also in GAMCO we have worked to help create models that segment the large amounts of data available, identifying the most relevant variables, out of the thousands available, to explain certain phenomena. 

These developments have not only subscribed to the field of physics but also to other more day-to-day developments:

• Automatic segmentation: applied to obtain segments with common customer characteristics in the development of commercial channels. This customer segmentation is implemented in our SAIL solution (Sales Artificial Intelligence Launch).
• Identification of the most relevant variables: these developments have enabled us to implement the explainability of the outputs of our predictive models. For example, in advanced risk management (ARM - Advanced Risk Management), being able to define why non-payment alerts are triggered on customers who have so far paid their credit obligations.