If you’ve been involved in a Digital Transformation, you might have heard the term “Digital Twin” being thrown about. Not many folks understand what it really means, so I’m going to break it down for you here.
Digital twinning is a fancy way of saying that you’re creating a model of a process that a company can use for simulation.
What good are digital twins? There are several uses cases:
- They allow you to test decisions before you make them in the real world.
- They allow you to generate Reaction Plans in advance of potential disasters.
- They allow you to foresee and plan for events that might otherwise be considered black swans.
- They allow you to very quickly train up a suggestion engine (using Machine Intelligence [MI]).
Let’s say you own a mine, and you’d like to be able to create a suggestion engine that will help mining executives know the results of some of their decisions in advance. In mining, you have a “mill”, which breaks large rocks into smaller rocks. This is often the most expensive part of the mine and a single mill will cost upwards of 1 billion dollars, so you want to keep it running and maxed out as much as you can. Now, let’s say that mill goes down. The mill is a bottleneck, so at this point it might make sense to stand down a shovel or a few trucks in order to perform maintenance on them. A good mining executive will likely make this decision very quickly. But what if you had a machine intelligence (MI) plugged into the mill, that could alert dispatchers to start idling trucks as soon as a problem is detected? What if you could alert dispatchers before a problem is detected??
With a digital twin, this is possible. An MI is able to generate a model of the mill, and the trucks, and then run simulations on that model to detect when there might be a problem and advise pre-emptive action.
A good first step in building a digital twin is to properly instrument your site. At a previous organization, we attempted to build a digital twin, but ultimately failed because we didn’t have enough data to construct a coherent model. If you’d like to build a digital twin at some future point, start to think about data now.
That’s it for today. We’re going to cover how to build a digital twin for your organization in a future article, which I’ll link here when it’s up. Until then, have a great day!
