By Pat Dixon, PE, PMP
President of DPAS, (DPAS-INC.com)
While as a business owner I need to obey Jim's maxim to spin the invoice printer, I also engage in pro bono activities. One of them is to serve as chair of the Technical Advisory Committee for Digital Transformation with the International Society of Automation (ISA).
This week ISA introduced new forums to foster discussion amongst professionals in automation. In the Digital Transformation forum, the current topics are Artificial Intelligence, Data Science, Digital Strategy, Digital Twin, Smart Manufacturing, and Virtual, Augmented, and Mixed Reality. More topics may be added in the future, but these topics are clearly hot in industry today.
In many of my prior articles in this newsletter I have addressed defining terms. We seem to have a tower of Babel in industry where terms like Digital Twin mean entirely different things depending on who is saying it and what the application is. For example, ISA has a very helpful artificial intelligence bot on its website called MIMO. Like any chatbot, it responds to your questions by searching for answers within the training dataset it was given. If you ask MIMO to define Digital Twin, you get:
- A process model must adapt from realtime data, and the learning is unsupervised (no human review, approval, or intervention)
- There is automatic flow of data from the digital twin back to the process that can influence or control the physical process
- It must encompass the entire process
There are no applications in the manufacturing industry that meet those requirements:
- Realtime data can be noisy, inaccurate, or invalid. There are times when the process is down, and the data is useless. If a human does not pre-process the data to remove noise and outliers, the model learns garbage. I know of no applications in industry that entirely replace subject matter or data analytics engineers with unsupervised software pre-processing.
- It is very risky to automatically connect an adapting model to a physical process. Every deployment I know of requires a human in the loop to evaluate and approve predictive process models before they can impact the process.
- Most applications of realtime prediction models are applied to unit processes, not an entire process. In a paper mill there can be such applications for each paper machine, a coater kitchen, a lime kiln, each evaporator set, each continuous digester, and each boiler. In a chemical facility the process model may apply to one reactor. If a PID loop is not big enough to be considered a digital twin, how many inputs and outputs are required to make it big enough? I see no clear definition of the size of a digital twin.
Therefore, digital twins as defined by ISA do not exist
I am not alone in this opinion. I asked Dr Michael Grieves, who is credited for introducing the concept of digital twin in 2002, to render his opinion. He agreed that the answer MIMO gives does not match what he originally defined as a digital twin.
Perhaps clarity on defining terms is not as important as spinning an invoice printer, but they are related. If I sell a project only to find the customer had very different expectations at the outcome, it will hurt my reputation and my business. I try to be careful to manage expectations so that my customers clearly know what is being provided. If I am providing a digital twin, what does the customer think they will get?
I believe ISA can be the one place where automation professionals can provide clarity on terms. I encourage you to join the forums at ISA.org and make your pro bono contribution to automation.
