Generating information from numbers: Interview with software experts Thomas Voggenberger and Josef Scheuer about the GRS analysis simulator

26.11.2018

The GRS analysis simulator has undergone continuous development at GRS Garching since the 1980s. The software, which is funded by the Federal Ministry for the Environment and the Federal Ministry of Economics, is one of the most important tools in the field of nuclear safety in Germany. Thomas Voggenberger has been project manager for the development of the analysis simulator for ten years. Next year, he will hand this role over to his colleague Josef Scheuer. In an interview, the two software experts explain how the analysis simulator works.

What can a layman imagine an analysis simulator to be like?

Voggenberger: There are simulators in many areas. The best known is the Microsoft flight simulator. The person sitting at the flight simulator selects an airplane, adjusts certain weather conditions or perhaps also turbulences, and then sees how he copes with it.

The analysis simulator we use here at GRS is all about a nuclear power plant. The user sits in front of a computer simulating a nuclear power plant or a research reactor. Via the screen, the user can act in a real control room, similar to a reactor operator. In process diagrams, he can see what is happening in the plant, e.g. how high the pressure or the temperatures are, which is represented by changing colours and numbers. By clicking on the corresponding graphic symbol, he can intervene in the process and, for example, switch off pumps or close valves.

The analysis simulator is similar to the training simulators in which personnel from nuclear power plants are trained. In contrast to the analysis simulator, the training simulator also displays a large part of the control room in terms of hardware. In order to carry out realistic training, training simulators must describe the processes at least in real time, which leads to limitations in the modelling of the physical processes.

Scheuer: In the analysis simulator, this is similar, but the purpose and the user group are different. Analysis simulators have a much wider user base, including research, universities, expert organisations and industry. And the user has a different focus. He is trying to investigate physical processes more closely in order to find out whether limit values are exceeded, whether the materials can withstand them, whether the temperature drifts into critical ranges.

With the analysis simulator it is also possible to simulate very complex scenarios. In the analysis simulator, any failure assumptions and combinations can be assumed. This is only possible to a limited extent in the training simulator. There, you are meant to learn how to operate the system safely. In the analysis simulator, one concentrates on the analysis of the plant behaviour and receives information partly in very high temporal and spatial resolution and not only at the measuring points of the plant.

In summary, it can be said that the analysis simulator is used to examine cases that have occurred or are hypothetical. Various plant conditions and events can be investigated: from normal operation, to a malfunction, an accident, or a severe accident with core damage.

Thomas Voggenberger and Josef Scheuer working with the analysis simulator ATLASHow exactly does the GRS analysis simulator work?

Voggenberger: The analysis simulator essentially consists of three components: the simulation program, the data sets, and the graphical user interface. The starting point is always the simulation program, which calculates these physical processes and can simulate the components of the plant. The computer program developed and used by GRS, which is used very frequently, is called ATHLET. At the same time, there are various other codes with which the program can be extended or applied, depending on which processes are to be analysed in which area of the plant.

These simulations are often very complex in terms of computing time. Depending on the object of the study, this can range from a few minutes to several weeks or even months. In the case of time-intensive analyses, an evaluation carried out parallel to the calculation is not very practical. In these cases, the calculations are carried out in advance. The results of the calculations are then subsequently evaluated.

Scheuer: You can't afford that with a training simulator. Here, the results are presented in real time. For this purpose, other calculation methods are also used, which, for example, make compromises when it comes to accuracy. The analysis simulator, on the other hand, is all about being accurate.

Voggenberger: In order to be able to simulate plants with the computer program, a corresponding plant model must be created in the form of a data set. Since the program only describes physical processes, a wide variety of plant types can be simulated, such as different types of pressurised and boiling water reactors. The plant type is determined via the data set, which contains extensive data on the reactor core, the cooling systems, and the instrumentation and control technology.

How long does it take to create the data sets?

Voggenberger: To give you a rough idea: When you create a new dataset for a plant, it can easily take several thousand working hours, including the performance of test calculations. The data sets must also be updated if changes are made to the nuclear power plant. This is why we are constantly collecting such data. Other recurring developments in the data sets concern maintenance work, which goes hand in hand with the updating of the computer programs, and model refinements in order to be able to simulate the real behaviour better and better.

The third component of the analysis simulator is responsible for the visualisation. By visualising the data, interactive control is possible and you can intervene in the simulation, e.g. by opening or closing valves. The visualisation program "talks" to the simulation program in the visual sense and translates its results into images.

In which situations do you start the simulator?

Voggenberger: We use the simulator to study real events in nuclear power plants in depth. Detailed statements on complex facts can really only be made on the basis of a simulation. We look at whether the events could also have occurred in other nuclear power plants. Also, we are interested in how they could have evolved if certain safety systems had not intervened. The simulator is also used for training members of staff of GRS or of authorities.

Scheuer: In the past, the simulator was also used to develop a screen display design for digital control rooms. It was investigated how displays and control elements are best arranged in order to be able to react quickly and error-free in the event of an accident.

Would you be able to study or recalculate the reactor accidents at Chernobyl and Fukushima with the simulator and if so, have you done so?

Voggenberger: In the Chernobyl accident, reactor power increased by a factor of about 100 within of seconds. This gigantic increase led to the well-known steam explosions that destroyed the reactor. GRS has recalculated the power increase in the reactor core in the months following the accident. We were not able to calculate the behaviour of the entire plant before the power increase with our analysis simulator because we simply lacked the data.

Irrespective of the recalculation of the accident sequence at Chernobyl, we have compiled data sets for this Russian reactor type in cooperation with Russian expert organisations. This type of reactor is completely different from the concept, design and operation of Western reactors. With these data sets, safety analyses are possible. They have been used to study the backfitting measures needed to prevent such an accident and to demonstrate their effectiveness.

The reactor accident at Fukushima was extensively investigated analytically. Two things were done. On the one hand, analyses were carried out to better understand the accident at Fukushima. On the other hand, the event was applied to German plant designs to see what would happen here. Ultimately, such real accidents are also recalculated in order to test and demonstrate the usability of the computer programs and models.

In your opinion, what makes the analysis simulator a special, important tool?

Scheuer: The analysis simulator makes it possible to develop a deep understanding of the behaviour of a plant. This makes it a valuable tool for me. The visualisation helps immensely to understand the figures. It makes an important contribution to avoid possible misinterpretations and to recognise correlations faster.

Voggenberger: There are very complicated events, with enormous amounts of result data and figures. Even for our experts this is a lot. Without the graphical user interface, the result would be a list of figures. The analysis simulator facilitates the evaluation of the results. This allows even people who do not know the actual code to understand the results.

Operation via a user interface is also much easier. For example, many commands or controls of components such as "open" or "close" can be executed via defined buttons by means of mouse clicks.

Do you have any plans and wishes as to how the analysis simulator should develop over the next few years?

Voggenberger: We want to reprogram the graphical part of the analysis simulator and bring it up to date. The programs must also be adaptable to the new operating systems and must be renewed and further developed against this background.

Scheuer: Since the calculation codes are constantly evolving, the analysis simulator must also be able to visualise the new developments. In the future, online simulations are to be controlled more flexibly and the results of several running calculations are to be compared with each other. In addition, we want to present thermohydraulic effects in more detail resolution. The increased use of 3D representations and virtual reality are also topics that are often requested by users.

Thank you very much for the interview!

Find out more about the ATLAS analysis simulator