Part 2 of the 4 part series: Digital Twins through the full lifecycle
How an operational 3D Digital Twin gets built
In Part 1 of our series, we emphasized the importance of fidelity of a 3D Digital Twin model: that is, that the model remains a true-to-life digital mirror of the world, and that fidelity be maintained throughout the long life of the operation or collection of assets.
Here, we delve into the process for making the twin operational. That is, how do we move beyond the static, one-time output of a design / construction phase into a useful operational tool that benefits a wide number of operational users? To make it operational the 3D Digital Twin must:
Connect to all data sources of record about the asset/process. In fact, it should BECOME the system of visual record associated with the asset records.
Augment and complement (not replicate) the data in those systems.
Establish a fit-for-purpose level of accuracy and precision for the use it will be put to.
Integrate into and participate in the wider data ecosystem – the data and analytics platform where asset, operations, supply chain analytics and machine learning is being done.
Stay in sync with the world – a world that is bound to change over time.
Having considered what a good operational 3D Digital Twin must do, let’s set about describing the process to build it.
Building the Foundation
The starting point for 3D Digital Twin models is to survey and collect all previous digitized inputs that have potential to jump-start the process of building the initial model. From there, two main techniques are employed:
Importing existing 3D CAD engineering models, and then updating those models to reflect real-world gaps, variances, and changes.
When 3D CAD models are not available, reality capture methods are used to create the visual input and “fill in the gaps” where other models lack coverage: laser scanners, LIDAR, drones, and other devices can capture the visual data necessary to create the model foundation.
Almost certainly, building the foundation will require a combination of both. The second, ongoing reality capture will be an important part of the operational program for maintaining the 3D Digital Twin. (See the 3rd Blogpost on the process for maintaining the 3D Digital Twin).
Reality capture device types
Many companies have already dabbled in reality capture efforts over time, without completing the full process needed to build a plant-wide 3D digital twin. Those efforts should not be discarded. This “never throw anything away” approach allows for the incremental building out of the model, rather than requiring one to build out the entire plant in a “big bang.”
Once the previous and current inputs for collecting the digital visual input are completed, a good model build process uses AI, and human-assisted automated processes for stitching the inputs together, create an integrated, complete, and coherent visual foundation for the Digital Twin. Reality capture methods have changed over time, and the cost and level of effort have decreased over time, as have the level of artificial intelligence that can be applied in splicing together the image data to form a coherent whole. As emphasized in its Reality Capture webcast, the Digital Twin Consortium reviewed considerations for selecting the capture method and devices used, level of precision and other factors. Establishing a robust reality capture program is an important part of the digital maintenance program for the 3D Digital Twin.
“Intelligizing” the Twin
Once a foundation is built, AI-powered processes “intelligize” the model: using the information from asset master systems, highly precise attach points called tags are created, and accurate asset, process, and other attribute labeling is added. These tags become the conduit through which relevant data can flow, providing a highly intelligent, spatially aware visual scene that can be manipulated and used in a variety of ways. Tags are also the integration points for data tools, including data and AI platforms.
Connecting the Twin
Next, the 3D model is exposed and attached to data sources in two principal ways:
Connecting directly to enterprise systems of record and operational systems through API’s and connectors; these near-real time connections can be bi-directional, with the ability to “write-back” to systems of record.
Live, real time integration to data and analytics platforms: data warehouses, data lakes, AI and machine learning These connections allow the 3D digital twin to display highly intelligent AI-generated insights, predictions, and recommended actions.
An important distinction here: data connections tend to cause replication of data, which leads to complexity and additional efforts to sort out which is the correct version. The 3D digital twin should reference and reflect these systems of record. Rather than replicating the data it is displaying, it merely reflects operational data in a visual context.
Immersion into the Industrial Metaverse with AR and VR
Once the fully operationalized 3D model is built, it can be used and consumed in many ways, depending on the job role of the user. Of course, operations teams, inspection teams, and other engineering or plant personnel can view the model from their laptop or take it in the field with them on mobile devices. Exciting new ways to use these models involve connecting Augmented Reality and Virtual Reality devices to allow users to experience their data while walking around through the virtual plant floor. Training, rehearsal of complex process, and evaluation of future changes being contemplated are all situations that are driving more and more AR/VR type usage of the 3D Digital Twin.
3D Digital Twins can visualize data from source such as:
Manufacturing Execution Systems (MES)
Geographical Information Systems (GIS)
Engineering Document Management Systems (EDMS), for P&IDs and isometric drawings
Data historians
Computerized Maintenance Management Systems (CMMS)
IIoT sensors
Live video or image feeds
Interactive manuals and training videos
How to get started
As manufacturers and other operations leaders build a more and more connected digital world model for their operations, a 3D visual digital twin can be a foundation on which better remote management, more effective training, more comprehensive planning can be built. What is the right approach to getting started? Three important things to keep in mind are:
Don’t boil the ocean: take an incremental approach, focusing on one area, processing center, or production line that is selected for a purpose.
Connect each piece into a coherent view of the whole: while incremental projects make the cost and complexity more manageable, remembering the “don’t throw anything away” maxim, each new project should be connected back to a complete and holistic visual model of the whole facility, plant, or operation.
Connect, don’t replicate: the visual digital twin is not the same as your data systems of record, nor is it an analytics or data platform. While live connections will help operationalize the model, good digital twin architecture ensures data architecture does not get needlessly complex or redundant.
