A Digital Twin is an executable real-time virtual model of physical reality. In another words, a Digital Twin is a digital representation, a digital mirror, of the real-world. They are uniquely useful to detect, prevent, predict, an optimize processes, systems, and objects which they do by using real-time data and
analytics.
Let’s look at one simple example. Think of your own car. You drive it around with your own style of driving which influences the amount of fuel you consume, the wear and tear of your tires and the status of your engine and other systems in the vehicle. You rely on the computer in the car to tell you when you need an oil change, tire rotation, and how many more miles you can ride before you run out of fuel.
How can the computer predict all this? You might have not been recording billions of pieces of data while you drive, but the car’s computer has. The computer has a well-defined Digital Twin, albeit simple so far, of the interaction of your driving and the car responses. The computer has been capturing all these data in real-time, and it uses that data to predict when you need to add fuel, change oil, and one day, it will predict when you are likely to have your next accident.
What technologies is the car using to create and manage the Digital Twin? We can say they fall into one
of these categories:
- Sensing, such as IoT (Internet of Things), and I would also add any ETL, Extract-Transformation-Load, of existing data from internal and external systems, to enable constant capturing of relevant data that is used to create the digital mirror of the reality
- Data Storing, in a massive scale, using Cloud to store all data of the digital twin to be accessible from anywhere from any system and maybe Blockchain to manage data transactions
- Computing to analyze and “think”, using all data to detect and describe trends and signals, to provide valuable insights, to make predictions and to provide prescriptive solutions
Executing using Cyber technology to connect back with the physical reality to change parameters or implement actions. It all sounds science fiction, but it is not. As Ron Garan said, we are limited by our imagination and our will to act. This entire Digital Twin concept comes from NASA in its attempt to improve simulation models in 2010, and today, Digital Twins are at the core of INDUSTRY 4.0, the fourth Industrial Revolution, introduced by team of German scientists in 2015 to promote the computerization of industrial Operations and Supply Chains.
There is no way out of this trend. Digital Twins are becoming a strategic imperative to build business
models that are dynamic, agile, resilient, and successful. Digital Twins can enable the four big
transformations that I propose for all business today:
- Move from Descriptive to Predictive
- It is not enough to describe reality through static charts, the data has to work for us to detect signals and make predictions
- Move from Sequential to Concurrent
- It is not enough to manage one process performed by one department at a time, there is a need to manage some processes concurrently, take Supply Chain as an example.
- Move from Deterministic to Stochastic
- It is not realistic to think that process parameters are static and do not change, we need to use the data to dynamically build the variability that we see in real life, take supplier lead times or freight lead time as an example.
- Move from Reactive to Proactive
- This is obvious. By now, all companies that did not proactively change their way of working and thinking, are paying for all disruptions from the past 24 months.
So, how do you think of, plan, and implement a Digital Twin strategy?
Again, the business case for Digital Twins will be strong based on the size of the benefits:
- Monitoring the physical system to preventing unplanned events and drive continual improvement. Examples:
- Monitoring oil changes in your car to prevent a major costly failure based on actual conditions of the oil rather than an oil change every 3000 miles.
- Monitoring cycles and status of components of a plane engine to prevent failure instead of relying on scheduled very costly monthly maintenance even if not necessary.
- Simulating the physical system under the influence of changing a series of parameters in the digital system before adopting the in the physical system. Examples:
- Simulating new parameters in a Digital Twin of a machine to understand what can go wrong before we implement them in real life
- Simulating supply chain performance by changing parameters in sourcing, logistics, manufacturing, distribution, etc before those changes are made
- Predicting about problems or future possible scenarios, analyzing processes in real-time. Examples:
- Predicting a machine part failure
- Predicting out of stocks based on the whole supply chain model
- Resolving problems originating from insufficient information by building analytic models that augment and complement reality. Examples:
- Resolving an optimization problem for a warehouse
- Resolving a freight routing optimization
- Learning. Examples:
- Learning what the optimal parameters are for a process from a recurring
Any of these examples illustrate the magnitude of benefits of Digital Twins. In a simple way, who would not pay to have a network single source of truth and a virtual representation of reality to try and test ideas?
Just citing GE: “using digital twin patterns, GE has realized these specific benefits: “93-99.49” increased reliability in less than 2 years, 40% reduced reactive maintenance in less than 1 year, 75% reduced time to achieve outcomes, $11M avoidance in lost production by detecting and preventing failures”.
Or in my personal experience, a double digit improvement of revenue and EBITDA in one year by making Supply Chain and Operational decisions assisted by a partially completed Digital Twin.
Regarding costs, we just need to remember that Physical cost increases on an annual basis are based on inflation (labor, materials, and overhead cost increases) and Digital costs keep decreasing exponentially with time, so the delta between physical reality and its digital twin keeps increasing over time.
It is unrealistic to think that a company can move from no real-time modelling to a full Digital Twin immediately. The process is long and incremental, but the benefits are obtained in every step of the way. As an example, this is the path I went through to implement a Digital twin for Supply Chain Management:
- Stabilize data/processes in ERP
- Stabilize data/processes in existing Transportation system
- Add functional system: Advanced planning system
- Add functional system: Warehouse Management system
- Add system of intelligence: Sales, Inventory, and Operations Planning Process SI&OP
- Add system of intelligence: Integrated Business Planning IBP
- Add functional system: Supplier/Procurement management process
- Add system of intelligence: Container tracking system
- Add system of intelligence: Available to Promise
Functional systems are packaged software that needed customization and configuration. Systems of Intelligence are internally developed systems using big data and analytics.
And it took 24 months from beginning to where we currently stand, now able to visualize our supply chain from end-to-end, able to predict future performance, and able to make decisions to mitigate risks. And I must add that it did help immensely through the Supply Chain disruptions in the past 2 years.
Every step had its own business case, capital expenditure request, implementation plan, and each one delivered better than expected. We are eating the elephant one bite at a time without burdening excessively our internal people and financial resources. One step at a time and never stopping.
We have the backbone already to bolt on new data vectors from sales, marketing, finance, HR, manufacturing, to increase the availability of integrated concurrent non-deterministic proactive decision-making models.
The only limit we have is our imagination and our willingness to act.
Pedro Caceres
pcaceres@comcast.net
(763) 412-8915
