News of global supply chain problems has been around for months – that Christmas gifts may not reach customers on time, dozens of container ships queuing in ports, and retailers trying to find alternative shipping methods. As the MIT Technology Review writes, one way to deal with this problem may be the proliferation of so-called digital twins.
By itself, this technology has been used for several years. The digital twin is a kind of virtual model of a real object, based on artificial intelligence and helping to optimize work. In order for the digital twin to properly reflect reality, it must receive a huge amount of data from its real prototype – this is done using various sensors, cameras and other data sources.
With the help of such systems, it is possible to simulate situations in production (or in logistics) and possible disruptions in work. AI can also provide guidance on how to deal with these violations.
However, until now, digital twins of limited objects have usually been created – from a single device to a group of enterprises.
Now we are talking about digital twins of supply chain systems as a whole – modeling an extremely complex system with numerous suppliers, complex transport networks, etc.
According to Hans Thalbauer, managing director of Google’s supply chain and logistics division, the main problem for logistics companies is the inability to predict events that may affect supply chains. “Anyone in the supply chain business will tell you that they lack the breadth of vision they need to make decisions,” says Talbauer.
It is this latitude that digital twins can provide. If such an AI-based digital twin receives enough data from its physical counterpart, it can replay potential failures over and over again and find the best option through trial and error.
The data that the digital twin must receive for normal forecasting is very different. Information about the company itself, its suppliers, their stocks and planned delivery dates, data on consumer behavior based on financial forecasts and market research. And also information about the world as a whole, for example, about the likelihood of natural disasters and the geopolitical situation in different regions.
For example, it can test the assumption that Taiwan will experience a drought and water shortages will lead to a reduction in semiconductor production. The digital twin will predict the likelihood of such an event, track its impact on different parts of the supply chain, and suggest options on how to at least mitigate the blow. Even with the use of AI, such analysis is still a very difficult task.
So, the automaker Ford has more than 50 factories around the world, annually they use 35 billion parts to produce 6 million cars. These parts come from 1.4 thousand suppliers from 4.4 thousand factories, and in some cases supply chains come into play. A stress test of a system requires checking all of its elements and exploring different options – what to do in the event of a failure at a particular plant or at a particular supplier.
Amazon is already using digital twins for supply chains. Such technologies are being developed by Google, FedEx and DHL.
There are also companies specializing in digital twins for logistics systems, such as Pathmind or Deliverr.
Pathmind’s digital twins work with existing supply chain management tools and data from a particular company. They use this data to model various problems in the system and options for solving problems. According to Pathmind founder Chris Nicholson, the resulting synthetic data – that is, not directly obtained, but generated by AI based on this direct data – helps predict a variety of scenarios, such as a pandemic, and learn to minimize their damage to supply chains.
“This is the answer to the question: why is AI smart? He just lives longer than we do in all these different worlds, many of which never existed, ”he says.
As the head of the data analysis laboratory of the Massachusetts Institute of Technology David Simchi-Levy notes, one of the important goals of logistics companies with the onset of the pandemic and the crisis was to increase the resiliency of the supply chain, that is, the ability of the system to maintain stability and operability in the event of failure of some of its elements. Companies are willing to invest in this, but they want to maintain a balance by strengthening the supply chain enough without spending too much money – and from this point of view, digital twins can also help.
“We are seeing more and more companies stress testing their supply chains using digital twins,” says Professor Simchi-Levy.
At the same time, experts admit that technology alone will not help here.
“Technology will not solve these problems. They won’t help ships carry more containers. But digital twins will help us detect problems before they happen, ”says Professor Simchi-Levy.
Difficulty is also the lag of some elements of the supply chain in terms of digitizing data. “Many of the world’s ports do business on paper, you are very lucky if they use PDF or email. And these are the big operators, not the New Hampshire candle manufacturer. Without digitalization, AI will not be able to work, ”says Mr Nicholson.