By Pat Dixon, PE, PMP
President of DPAS, (DPAS-INC.com)
In 1984 during my 3rd year in college I got a co-op job at a far-away paper mill. That meant I needed to buy my first car. That car died in the middle of nowhere on my way to that paper mill. I had the usual display of analog gauges on the dashboard, but I had no way of knowing that the fuel pump was going to fail.
Today, cars are loaded with data. Sensors have proliferated throughout vehicles and our dashboards present a full suite of information which just might tell you that you better fix a fuel pump before you are stranded.
This data does not have to be isolated to your vehicle. In a connected IoT world, data from your car could be monitored by the manufacturer to determine that components, like fuel pumps, are a problem. Using this data with a digital simulation of the car can help diagnose the root problem and allow testing of a solution. Such a solution could provide higher quality to drive sales and mitigate costly recalls.
Unlike cars, our paper mills have been instrumented for a long time. Sensors and the display of information are not new to us. In prior articles I have addressed the difference between IoT for consumer applications that previously lacked instrumentation (like cars) and the IIoT (industrial) applications for facilities that have been instrumented for a long time but were not previously connected through the public infrastructure of the internet. Yet, there is a perception that we in the industrial sector are technology laggards.
We are on the cutting edge of technology. I recently viewed a webinar demonstrating the ability to design a manufacturing process and simulate it with dynamic behavior, allowing development and testing of automation solutions before the facility actually exists. This application was described as a "Digital Twin".
Digital Twin is a very hot buzzword in the Industry 4.0 lexicon. However, even the people giving the presentation I just mentioned had a hard time defining what a Digital Twin is, even though they were calling their solution a Digital Twin.
Simulating processes is what many of us have been doing for years. Whether it is a simple tieback of IO or a sophisticated multi-variable and dynamic process model, simulation of real processes has been a fundamental component of automation. It is how we can safely test solutions and ensure we have covered scenarios before putting a solution into a real process. The growth of artificial intelligence techniques, such as neural networks, have provided a way to develop simulation models based on the ubiquitous data in our facilities.
Is a digital simulation system a Digital Twin? Well, it is certainly digital and it is a twin in the sense that it behaves like the real thing. However, simulation systems that we have used for years, including the one described in the webinar, were not connected through the internet to live process data.
According to WikiPedia, the webinar I saw that was describing a Digital Twin was not a Digital Twin:
"First and foremost, the technology enables connectivity between the physical component and its digital counterpart. The basis of Digital Twins is based on this connection, without it, Digital Twin technology would not exist."
If the connection of the digital representation of physical component like a paper machine to the actual paper machine is required in order to be considered a Digital Twin, does this connection have to use the public internet to meet that definition? If you have a control system already wired to your instrumentation, do you get a Digital Twin if you send that data to another node on your network where your paper machine simulation lives?
Imagine you have such a system. What does your paper machine simulation do with that data? You already have a real paper machine out there and you can see the data it gives you. Simulators are most commonly used to figure out what will happen if the data differs from what you are actually getting without impacting the process. Feeding the same process data into a simulator would only be used to compare the accuracy of the model.
This would be one of the uses for a Digital Twin. We all know that a process model has a shelf life and requires new process data to keep it current. If the Digital Twin is continuously connected to the real paper machine, it could use adaptive techniques to retrain a neural network model of the machine. Of course, the danger with any adaptive technique is to determine what data helps or hurts a model (noise, downtime, etc). For that reason, continuously adaptive a model may not be a good idea. Data scientists and process experts are usually involved in data preprocessing to carefully select the data used to train a model. Like all technology, this field is evolving to move toward a more automated means of selecting this data. Theoretically, a continuously adapting model could be automatically maintained, and be disconnected to play scenarios with test data when changes are being considered.
The common component of the Digital Twin examples I have mentioned so far has been simulation. However, does a Digital Twin require simulation?
Getting back to the car, the windshield serves as our view to the real thing. Cars today have GPS capability that transmit location through the internet. Couple with traffic data, a virtual display can be superimposed on the windshield showing a virtual road to follow. You are seeing both the real and digital versions of the road simultaneously. This is being described in a white paper by TATA Consultancy "Digital Twin in the Automotive Industry: Driving Physical-Digital Convergence" as an example of a Digital Twin in the auto industry, yet where is the simulation? By this definition, industrial facilities have had Digital Twins ever since the first live process graphics were deployed on computer screens in our control rooms. We are way ahead of the curve!
Therefore, we have confusion in this term. When someone is describing their Digital Twin solution to you, it may or may not have:
- Simulation
- Connectivity to live realtime process data
- Connectivity through the internet
As we drive down the Industry 4.0 highway, we need to be sure that the fuel pump is reliable. The fuel pump of this marketplace is our understanding of the technology and the terminology we use. Otherwise, we could be left stranded in the middle of nowhere.
