(Non) Identical Twins
People generally like models. Toys are models of the real thing, or of what we would like to be real. The world is filled with miniature towns and cities depicting the ‘real world’ to varying degrees of accuracy. Many models also aimed to capture and predict the movement of things, from early clockwork models of the solar system to Bill Phillips’ hydraulic computer modelling the flow of money.
The advent of the computer off course made it possible to start developing models in bits and bytes rather than brass and glass but only with the current level of cheap and powerful computing, combined with cheap and accurate sensors has it become possible to make a model accurate enough for it to be a ‘twin’.
All so called twins, however, are not created equal.
The intent of a digital twin goes beyond a model, even beyond BIM. A twin is a dynamic model that accepts feedback from its ‘real-world’ physical twin. Through sensors for pressure, heat or flow it can adjust its behavior to mimic the real world, incorporating the influence from wear or blocked routes. In return the digital version can then be used to predict future behavior.
But like real life twins; even if they look the same, move the same and sound the same, it doesn’t mean they are the same.
The ‘easiest’ digital twins are those of contained systems, an engine or internal flows of a plant (using ‘easy’ a bit loosely here). Most systems though are not fully contained, they are subject to external influences which can be modeled but make prediction much harder but, using multiple scenarios still potentially useful, especially if scenarios can converge over time with additional data.
The ‘digital twin’ is the latest step in our desire to understand, model and predict the world around us. A digital version of the machine or process we work with that can help us optimise it and predict the outcome of a ‘what if’ scenario.
Digital twins are now very real, but still limited. They are however not as ubiquitous as the hype around them because no matter how detailed a model you may have, if it cannot adjust its behavior based on real-world inputs its not a twin.
An improved version of the hydraulic economics machine, dubbed the Glooper, didn’t just model the economy, it influenced it. I wonder if we’ll get our models that far.
You might also like:
I am overstressed – take action now!