The digital twin is the virtual and digital copy of a factory allowing monitoring, post-mortem analyses, simulations, stress tests, machine learning and much more.
As a Lean practitioner having started his Lean experience in the 1980s, I faced the difficulty to get engineers, techs and sometimes foremen to the shopfloor to assess and understand the situation, and support the operators.
With the growing ability of the digital twin to get closer to the physical reality, my prospective question is: can the digital twin become the new gemba?
Not familiar with digital twin? Checkout: https://en.wikipedia.org/wiki/Digital_twin
Is the digital twin the new gemba?
My first answer would be no, for 2 reasons :
- The digital twin does not comply to the definition of gemba
- The digital twin is like Magritte’s pipe: a representation but not the real thing
“Gemba” (Japanese word) was long translated into the “real place” or “where it happens”, in Lean parlance where value is (supposed to be) created. This emphasis on the reality is a warning against the spreadsheet analyses or the tendency to trust blindly data.
Until recently, data capture was mostly manual and tedious, thus limited, and error prone. Trusting data meant taking chances by trusting a limited set of data to draw conclusions and/or being misled during analysis because of all the limitations and errors.
Compared to data analysis, going to the gemba and perceive the situation by oneself always gives a way better understanding and appreciation of what happened or is going on. The human body with its natural sensors can grasp more of the reality than the best and most accurate description.
Intuition can work better in this environment as more senses are stimulated. Links are established between distinct phenomenons or events, something that an analyst may not perceive when working only on a dataset.
This doesn’t mean that data are useless, simply they may not be used exclusively in order to widen the scope of the investigations and analyses.
The digital twin
Now what is a digital twin? A really big collection of data that allows to simulate the real-world twin, with growing fidelity and getting closer to reality. But it is still data regardless of their amount, and not the real place.
On the one hand immersing oneself into the virtual reality may allow exploring it in ways that are impossible in real world. The digital simulation can take an analyst virtually into a working machine and show the dynamics of its various components. Digital monitoring can even capture what the human sensors are unable to, thus ”augmenting” humans.
But on the other hand the sensors collecting data, however performant, capture only a limited reality. For the time being and as far as I know there is no smell in the digital twin, no sound like the squeaking or screeching, and heat can be measured but is not “played back” as heat in simulation, or is it?
There is no life in the digital twin, no unexpected insects, birds or rodents causing unforeseen problems. Vibrations can be measured and simulated, but is it as subtle as real vibrations and trepidations? Air flow will only exist in the digital twin if sensors or the digital model was designed to measure or simulate it. And the leaking roof problem may not exist in the digital model. Such examples are still countless.
In short, virtual reality may be closer to reality now than ever before, but still isn’t reality. And if it is not reality, it is not gemba.
Magritte’s painting of a pipe is a pretty realistic picture of this smoking accessory, with the caption “this is not a pipe”. Many of those recognizing a pipe would think of course this is a pipe! but the painter wanted to remind viewers that the pipe is only the picture of a pipe, not a real pipe. So is the digital twin. A pretty good representation of reality but not the reality.
The digital twin could become the gemba of the future
Now, in order for the digital twin to be considered the new gemba, it suffice to change the definition and get consensus about it in the (lean) community.
If a majority of people and lean practitioners consider the digital twin worth to be considered as gemba, or the actual definition of gemba being altered to encompass the digital twin, so be it.
I would not support this idea for the sake of clarity though. Think about the word “kanban” that for decades described a pull system in manufacturing and supply. Then the IT community adopted Lean principles and came up with its own version of kanban. It is built upon the same principles, describes a pull system, but is different enough to generate some misunderstandings.