Team

The digitalization of the process industry requires components that monitor themselves and detect and communicate their condition and any faults.

Design and operation of modular process plants in the pharmaceutical and specialty chemicals industry enables more flexibility and energy savings. However, increased wear in such plants is to be detected and distinguished from sensor errors. Unfortunately, there are often only few measured variables available. Therefore, knowledge about the behavior of the modules in the form of models for pumps, motors and valves will be used to calculate additional variables at the module level.

These quantities serve as symptoms for wear detection at the system level. Here, a decision on the location and severity of wear or sensor faults is made from the individual symptoms and knowledge of the overall system topology. The aim here is to map the empirical knowledge of plant experts in a logic-based diagnostic system.

The high flexibility of a modular plant also requires flexible wear detection. This is achieved by ensuring that the modules, their measured variables and the derived symptoms remain the same in all plant configurations. Only their interconnection is adapted. This means that only the diagnostic system has to be adapted on the basis of the derived logic.

Simulation models and a demonstrator system are available for the investigations. The demonstrator is a modular mixing plant consisting of five dosing modules and two mixers and is equipped with extensive measurement technology.