A layer-based approach to build up diagnosis applications in process industries

We present a generic approach in the form of a layer-based framework for building up applications that perform diagnosis or asset management functions. This framework is designed to run fully automated, gaining and managing all required information. Thereby, only little engineering effort is required and the framework is able to scale with the complexity of the plant structure. This is realized by mapping different plant situations to simplified and unified situations in real time. The modular layers decouple semantically differing tasks and the corresponding information from each other, which makes the framework modular and thus open to plug in new plant elements to be considered, new or changing plant properties or new diagnosis or asset management functionality. The framework presented here is already implemented in a German BP refinery, realizing a completely automated pump monitoring application.