Throughout the lifecycle of an industrial installation, information about that installation is shared through various diagrams. Even in today’s digital world, that installation knowledge is locked away in analog formats such as legacy, paper documents. To realize the value of that information, it’s essential to bridge the gap between the analog and digital as seamlessly as possible.
Analog documents are still a concern
Many facets of life and industry today are digitized. However, analog documents (paper documents, image files, etc.) still play a role in various ways. The most obvious as we transition into digital systems is that many older installations were constructed even before the advent of CAD systems. Information for these installations is only held in legacy, paper documents.
Analog documents are still introduced in processes for new installations as well. Bid packages will often only include PDF documents that indicate basic design. Additionally, project deliverables might be provided in non-CAD or similar formats.
Inefficiencies created by analog documents
The increased productivity of the digital age is centered around the management and interchange of digital information. Companies today are fully reliant on information being accessible and understandable through these digital systems. When dealing with analog documents, a digitization process (often labor-intensive) must take place before the efficiencies of digital systems can be realized.
This decrease in productivity can be expensive when analog documents aren’t received on a predictable basis. For teams that only deal with this issue infrequently, extracting information from the documents is often handled by engineers who might be more focused on design, compounding the loss in productivity.
The concept of "digitization" might take on different meanings depending on the person or context. If we take it to mean the removal of inefficiencies caused by handling analog information, then the digitization should result in data that’s ready for use in the digital systems we use today.
Given this definition, there are a few different components of the digitization that must be addressed. These steps are as follows:
- Scanning a paper diagram to make it available for use on a computer
- Using optical character recognition (OCR) and similar technologies to extract text, tables and drawn elements from the diagram. However, OCR isn’t sufficient as it doesn’t establish an understanding of the information and therefore doesn’t result in immediately-usable information.
- Establishing engineering design and relationships ("semantic processing") so that the information is truly useful.
The promise of artificial intelligence (AI)
Existing technologies have proven effective at handling the first few steps into digitalization. However, the final step of extracting engineering meaning and context from data has often been handled by engineers or other skilled individuals. Given the complexity of the subject, as well as the variability in presentation from different sources, it’s exceedingly difficult to pull meaning from documents in a deterministic manner.
As deep learning and AI solutions become more practical, it’s increasingly possible to mitigate or even obviate the need for engineers to handle this laborious task. By incorporating these advanced technologies, we can move closer to a fully automated digitalization process.
Removal of inefficiency is the goal, not the removal of analog documents
Analog documents exist in both legacy assets and day-to-day processes. Without a means to translate and use their information in digital systems, the productivity gains of these systems are reduced. Understanding that we’ll continue to work with analog documents for the foreseeable future, we must find a way to incorporate these documents into our digital world.
New AI technologies are making it increasingly possible to incorporate analog information seamlessly into digital processes, removing the inefficiencies and difficulties seen in the past. This allows us to implement new digital innovations confidently, without creating a divide between the old data and the new.
You can read more about machine learning in industrial drawings in this blog article or more on digital transformation of the EPC industry in this blog article.