Exhibitors & Products
Events & Speakers

How can digitalization with the help of "digital twins" support the mass production of hydrogen by electrolyzers to meet the growing demand for green hydrogen?

Large-scale hydrogen electrolyzers are a relatively new technology that is not yet present in the grid on a large scale. Electrolyzers therefore need to be monitored from the outset to gain new knowledge for efficient operation and to be able to produce large quantities of hydrogen cost-effectively.

In addition, the integration of large capacities of electrolyzers into our power grid requires that they be used in a way that serves the grid, for which they must be made more flexible - i.e., digitalized. Digital twins are therefore required in parallel with the technical plants in order to be able to use electrolyzers "en masse".

What role does Digital Twin software architecture play in optimizing hydrogen production and improving electrolysis plant operations?

"Digital twin" is a colorful term used in the industry to describe many concepts of varying power - from digital plant configuration to simulation models to real-time imaging of individual plants in the field with live information on current status. It is therefore imperative to create a common understanding of the different manifestations/variants of Digital Twins and to talk about minimum standards. Only in this way can digital twins from different manufacturers be combined and integrated into system landscapes for higher-level control and optimization. The easiest way to achieve this is to agree on software architectures and standards - preferably open source.

How can digital twins of electrolyzers be used to perform both long-term and near-real-time analyses to enable rapid response to unforeseen events?

Machine learning methods are essential for analyses using digital twins. Measured data collected over long periods of time can be used to create data-driven models that can be used to make predictions about the future behavior of electrolyzers. Behind this, however, are computationally intensive processes that - assuming modular software architectures - can be relocated to the cloud, for example, if necessary.

To detect anomalies in real-time, we use data stream technologies - so-called stream processing - to be able to react immediately when events occur. Technically, our projects at OFFIS combine data-driven models, which have been trained on long-term data but can then be used directly on data streams for classification, for example, with physico-chemical models for this purpose.

Data-driven models are typically faster in application, but not in training. In return, the accuracy of physicochemical models is usually higher, provided the processes have been properly understood and modeled. This is not yet necessarily the case for new electrolyzer technologies.

To what extent can the integration of renewable energies into the power grid be improved by producing green hydrogen via hydrogen electrolysis?

The mass production of hydrogen, e.g. when there is surplus energy from wind power, is a very significant contribution to sector coupling - relieving the burden on our electricity grids by producing hydrogen on site and feeding it into our gas grid. Electrolysis at energy-rich sites means that electricity does not have to be transported as far and our power grids are relieved.

Therefore, a major goal of our research at OFFIS is the use of digital twins for system-serving operation. Here, electrolyzers are to adapt to the state of the grid and thus offer system services - such as flexibility - for the grid.

What challenges must be overcome in order to successfully implement digital twin technology in hydrogen electrolysis and optimize the efficiency of hydrogen production?

For the technology behind digital twins, data availability must be increased. This means that sensor technology must be available to collect data over the entire life cycle. The concept of Digital Twins helps organize the amount of data that accumulates over the lifetime of electrolyzers so that it can be used to draw conclusions about aging behavior, which is not well enough understood today. Knowledge of aging effects is important for predicting the effects of different operating modes of electrolyzers on lifetime and, if necessary, for improving them through gentler operating modes and timely maintenance measures. All this can help to develop and mass produce.