The German Research Foundation(DFG) is investing a total of 4.5 million euros over three years in five research projects. One of these is the TwinGuide project, which is being carried out by the Fraunhofer Institute for Factory Operation and Automation IFF in Magdeburg and the Hamburg University of Technology (TUHH) together with IPT-Pergande GmbH in Saxony-Anhalt. The aim of the project is to develop an intelligent digital twin for process plants to make process engineering processes safer and more efficient. The development is being tested using a fluidized bed system for spray granulation as an example. With the help of the digital twin, the current status of the physical plant under consideration should not only be displayed in future, but its future behavior should also be predicted and the plant should also be reliably controllable.

Digital twin becomes intelligent

To date, digital assistance systems have enabled access to existing documentation such as CAD, e-plans, protocols and sensor data as status information for the plant. They provide plant operators with relevant information on the current situation, such as documentation, plant parameters or operating states. However, they can also issue warnings about faulty system statuses and at the same time provide predefined recommendations on how possible faults can be rectified. In these cases, however, it is usually already too late; the fault and possibly even a case of damage have already occurred.

The TwinGuide project's intelligent digital twin is being tested on a fluidized bed system for spray granulation at application partner Pergande Group.

© Viktoria Kühne / Fraunhofer IFF

To prevent this from happening, the TwinGuide research project aims to create a digital twin for process plants that will enable predictive control of the future development of the condition of a production plant. To this end, simulations running in parallel to the operation of the plant are to enable a prediction of process stability and product properties on the basis of its digital model.

Increased level of detail through data

The intelligence of the digital twin is defined by the fact that the underlying models lead to decision-making measures much faster than in real time thanks to extensive continuous (offline/online) data acquisition with a very high level of detail and a high degree of processing, so that predictions and operational optimizations are possible. In addition to the actual digital twin, a communication interface will also be developed to enable interaction between the simulation results and the digital process image. By directly linking the current plant data and the status evaluation with the simulation results, the latter are integrated into the plant operation in real time. This allows warnings for unsafe plant conditions to be issued to plant operators with corresponding action scenarios and recommendations.

The implementation in the process control system is carried out by the application partner, the Pergande Group, whereby the transferability to larger plants is to be demonstrated in addition to the use on a laboratory scale. The research partners expect that efficiency in the area of contract production can be increased by around 15 percent through higher plant availability.