Digitizing a brownfield rig or plant? Let’s move the mountain – with help of AI!
Beyond the overly stretched “Digital Twin” in colorful 3D animation, one may wonder how to derive actual value from it, knowing that a rig or plant consists of hundred thousands of pieces of hardware, designed, assembled, and turned around a few times before the advent of digitally charged CAD and CAE. Hence, is it possible indeed to transform existing P&IDs of a brownfield plant into a fully integrated digital twin, with each element logically and digitally connected to all other relevant elements?
The simple answer is yes. However, the immediately raised question is “at what effort” and is it worth the investment of capital, human resources and frustration making sense of symbols, attributes and missing information? Prior to answering that question, let us look at why a digitally connected plant is more than a 3D-animated model of the asset.
First, a digitized plant offers optimization potentials from engineering to operations and maintenance. Management of P&IDs is centralized and efficient. There is only a single truth, always the right information available, no search time, easier collaboration with subcontractors, and foundation to merge with Computerized Maintenance Management System (CMMS) and other systems.
Optimize maintenance strategy using machine learning
Second, it allows a deep graph analysis for engineering and maintenance optimization. Based on the conversion of P&IDs into graphs, it is possible to identify the most important equipment and therefore optimize maintenance strategy. In addition, one can semi-automatically define “communities” in the graphs, which serve as foundation for equipment modules. Module-based engineering has gained significant importance for engineers but now it is finally possible to use existing P&IDS for a first robust definition of these modules using machine learning. An additional side benefit is the automated completeness check of P&ID, when receiving from 3rd parties, respectively automated risk analysis of old configurations vs new ones.
And third, even process-based analytics for actionable insights are possible. Identify causal relationships between equipment is one of the prerequisites when it comes to predictive maintenance. Here, the graph serves as foundational layer connecting the relevant tags and their values with each other.
Moving from P&IDs with individual symbol standards to commonly used CAE-systems
As becomes evident by now, digitized P&IDs are the foundation of any meaningful Digital Twin. Coming back to the effort required to reap those benefits. The digital subsidiary of global industrial services provider Bilfinger has collaborated with the most advanced experts on AI-based engineering document processing, such as the German Research Center for Artificial Intelligence (DFKI), and developed a solution (PIDGraph), which recognizes symbols, connections, text and attributes on a P&ID, and merges this information into a graph-based file structure. This in turn is exported via a specialized API into commonly used CAE-systems or data analytics solutions for further utilization. This process is seamless, accurate, fast and cost-efficient.