Practical Application of Digital Twin of a Process Plant

The digitalization of a brownfield process plant and the generation of a digital twin respectively is a time-consuming and expensive task. The as-is state needs to be captured and engineers need to remodel the plant manually within the CAD system. By applying object recognition from point cloud this task can be automated. However, various factors influence the results. Object recognition from a point cloud highly depends on the quality of a point cloud. Environmental influences such as dust, dirt, vapors, and more are captured as well and need to be filtered out. Additionally, according to the specific domain also constructive influences appear. Depending on the size of the plants the extent of used steel construction or additional platforms and stairs or ladders can be found. These elements hurdle the scanning activities and lead to scanning shadows. In consequence, fragmentations appear, which reduce the quality of the results. Here also the density regarding the packaging of the piping system and supporting constructive structures play a role. Domain-specific attributes of piping systems need to be considered as well. For some domains, the use of highly polished materials is required while others are made for more rough environments. Additional aspects such as insulation or painting change the appearance to some extent. All these factors affect the object recognition process as well. In addition, components’ specific properties come into place. While some components are simple connections between two components, others lead to a change of direction within the course of a piping system or even change attributes such as diameter. All that makes post-processing and optimization of parameters necessary. The practical application is presented in the context of examples from different domains. Various point clouds are analyzed according to the above-mentioned factors. Finally, the automatically generated piping systems are presented and approved in terms of integration with the original point cloud.