Digital Twins
Digital twins, no not Mario & Luigi or the Property Brothers, although both could be considered digital twins if they are viewed on any digital media, but ‘real’ digital twins, meaning digital constructs that are used in analytics and predictive analysis. Since 2002, when Michael Grieves introduced the concept[1] at an engineering conference. NASA began using the idea in 2010 when it created simulations of space capsules and vehicles for testing, and has since been cited as one of the top 10 technology trends of the last few years.
So what is a digital twin? Simply a digital twin is a mathematical model of an object. Static mathematical models (aka Monte Carlo simulation) represent an object at a given point in time and have been around for centuries, but dynamic mathematical models are time dependent and therefore change as parameters change over time, and while external observational changes could be made to dynamic models, only recently has technology been available to make those changes without human intervention. The data for such models can be as simple as the mathematical description of the physical characteristics of a golf ball or as complex as the entire workings of a manufacturing plant, but the most important characteristic of a digital twin is that it is dynamic and can be used to simulate or mimic the actual object or objects on which it is based, without affecting the object itself.
Digital twins have innumerable uses in a wide variety of applications, ranging from stress testing to workflow, and can be used both before a product is developed, during that development for product scenario simulation, and after a product is developed for testing and modifications to extend the product’s abilities, with data from sensors that monitor the product in actual use feeding back information to the twin. This also includes real-world historical data that can be added to the twin, all of which can be used to predict how the product will work and in what ways it can be improved, without the myriad prototypes and static models that are usually needed during product development and improvement.
There are many technologies on which the digital twin concept relies. AI, analytics, and machine learning, and advances in these areas can certainly go toward increasing the value of digital twins across a broad number of applications and industries, but while IoT and sensor technology is usually given short shrift relative to flashier technology, the ability to feed real-time information from working products to digital twins is a major step forward that will add considerably to their value, and that data will allow faster product development and more responsive products as the digital twins gather more information.
Of course, we can’t have wires running from sensors on equipment running across the country to product development or testing labs, so how will all of that real-time information get to digital twins? 5G would be the ideal mechanism for moving the data as close to real-time as possible and the highest transport speeds in 5G are those using mmWave spectrum. While the necessity for mmWave speed and low latency are less for the typical mobile user, the bandwidth of mmWave does allow for lots of data to be sent without congestion or bottlenecks, which bolsters the case for mmWave private networks in industrial or business settings, particularly where digital twin applications are more commonplace.
By mimicking physical assets, process operations, and frameworks, digital twins paired with 5G IoT transport can allow employees to view equipment in remote locations in real-time to solve production or maintenance issues without travel, such as in oil refineries, and can help to improve automotive design for vehicles from trucks to Formula racecars, and can solve some of the more recent global supply chain issues by creating and defining more efficient logistic networks. Building maintenance and space optimization can be simplified using a digital twin and in retail, sensor information can be fed to a digital twin to predict customer behavior and the financial impact of a wide variety of scenarios, so while technologies like AR/VR get much of the headlines, digital twin software is the more practical side of the digital world and with growth estimates between 25% and 35% over the next few years, the opportunities for the expanding use of digital twins seems obvious, especially as IoT sensor and data transport technologies improve.
[1] The actual digital twin concept came from “Mirror Worlds” by David Gelerner in 1991.
So what is a digital twin? Simply a digital twin is a mathematical model of an object. Static mathematical models (aka Monte Carlo simulation) represent an object at a given point in time and have been around for centuries, but dynamic mathematical models are time dependent and therefore change as parameters change over time, and while external observational changes could be made to dynamic models, only recently has technology been available to make those changes without human intervention. The data for such models can be as simple as the mathematical description of the physical characteristics of a golf ball or as complex as the entire workings of a manufacturing plant, but the most important characteristic of a digital twin is that it is dynamic and can be used to simulate or mimic the actual object or objects on which it is based, without affecting the object itself.
Digital twins have innumerable uses in a wide variety of applications, ranging from stress testing to workflow, and can be used both before a product is developed, during that development for product scenario simulation, and after a product is developed for testing and modifications to extend the product’s abilities, with data from sensors that monitor the product in actual use feeding back information to the twin. This also includes real-world historical data that can be added to the twin, all of which can be used to predict how the product will work and in what ways it can be improved, without the myriad prototypes and static models that are usually needed during product development and improvement.
There are many technologies on which the digital twin concept relies. AI, analytics, and machine learning, and advances in these areas can certainly go toward increasing the value of digital twins across a broad number of applications and industries, but while IoT and sensor technology is usually given short shrift relative to flashier technology, the ability to feed real-time information from working products to digital twins is a major step forward that will add considerably to their value, and that data will allow faster product development and more responsive products as the digital twins gather more information.
Of course, we can’t have wires running from sensors on equipment running across the country to product development or testing labs, so how will all of that real-time information get to digital twins? 5G would be the ideal mechanism for moving the data as close to real-time as possible and the highest transport speeds in 5G are those using mmWave spectrum. While the necessity for mmWave speed and low latency are less for the typical mobile user, the bandwidth of mmWave does allow for lots of data to be sent without congestion or bottlenecks, which bolsters the case for mmWave private networks in industrial or business settings, particularly where digital twin applications are more commonplace.
By mimicking physical assets, process operations, and frameworks, digital twins paired with 5G IoT transport can allow employees to view equipment in remote locations in real-time to solve production or maintenance issues without travel, such as in oil refineries, and can help to improve automotive design for vehicles from trucks to Formula racecars, and can solve some of the more recent global supply chain issues by creating and defining more efficient logistic networks. Building maintenance and space optimization can be simplified using a digital twin and in retail, sensor information can be fed to a digital twin to predict customer behavior and the financial impact of a wide variety of scenarios, so while technologies like AR/VR get much of the headlines, digital twin software is the more practical side of the digital world and with growth estimates between 25% and 35% over the next few years, the opportunities for the expanding use of digital twins seems obvious, especially as IoT sensor and data transport technologies improve.
[1] The actual digital twin concept came from “Mirror Worlds” by David Gelerner in 1991.
A simplified View of the Digital Twin Landscape - Source: AccuCities.com
RSS Feed