Today's business models are increasingly demanding digital twins, including their component objects and processes and live data on their activities, as companies look to grapple with the complexity of operational roadblocks that the Internet of Things (IoT) continuously creates.
Digital twins bring new ways to visualize, simulate, optimize, and remotely control systems and processes for rethinking and reconfiguring the seemingly humdrum routines of business activity to respond to fickle markets.
Digital Twins: From Space to the Commercial World
Digital twins had a providential debut with NASA’s Apollo 13 mission whose oxygen tanks exploded 200,000 miles away from Earth. As if preordained, the digital twin, far less sophisticated than those currently used, of the spacecraft was the linchpin to its rescue mission; the model mirroring it, the data on its flight path, and its remote-control systems helped to orchestrate its return to Earth.
Increasingly, NASA’s innovation is finding its way into the commercial world where IoT is the substrate of automated continuous operation systems that are composed of electronics, software, communications, and hardware. (Read What Are the Top Driving Forces for the Internet of Things (IoT)?)
Digital twins, 3D or 2D dashboards, visualize and simulate the integration of these diverse components before building and installing systems.
The models created by digital twins encourage cross-functional collaboration as they can be easily shared. They also help to compare designs and results to align them before catastrophic errors occur on the shop floor.
According to a recent survey by Gartner, 13% of the organizations implementing IoT are using digital twins, and 62% are in the process of either setting them up or plan to do so shortly.
“Well known technology companies such as IBM, Microsoft, and SAP notably have combined digital twin simulation capabilities with data analytics, while and IoT and engineering simulation software providers, such as DS (Dassault Systèmes), GE, PTC, and Siemens, have acquired companies or entered into partnerships to advance their CAD and or digital twin simulation services, on top on their IoT platforms,” says Elson Sutanto, Principal Analyst at Juniper Research.
A typical use case for digital twins is supply chains, labyrinths fraught with the risk of damage, theft, or spoilage as valuable products, such as electronics or pharmaceutical products, wind their way over multiple means of transportation, warehouses, and distribution centers and across several geo-climatic conditions.
It can be overwhelming to think and try to pinpoint the influence of variables on risks in supply chains, if they are observable at all, as varied as the quality of packaging, environmental factors, and law enforcement.
“Granular data on micro-variables like temperature and humidity, gathered from sensors, and the need to control their impact on outcomes has increased the demand for digital twins,” says Keith Kirkpatrick, Principal Analyst at Tractica.
Resource allocation in a labyrinth-like railway network, such as scheduling of the maintenance of locomotives, can go awry without the granular data fed by sensors and controls that digital twins bring. Situational awareness on the availability of capacity is gained by adding the numbers across the network including the utilization of repair shops, wear and tear of assets and the urgency of repairing them, and travel paths of the trains.
Only then does it become possible to simulate decision options and apportion resources. The initial calculations are redone in response to accidents, unexpected events and changes in weather to keep operations on even keel.
Alstom uses digital twins to simulate the options for scheduling the maintenance of assets for train services running from London to Glasgow and Edinburgh and monitors the network in real-time to make tactical adjustments.
The On-Demand Services of Digital Twins
Not every commercial organization can afford to install thousands of sensors and maintain them or have the capability to master analytics and the software that goes into digital twins. Organizations like Hitachi Vantara see this as an opportunity to shift away from hardware sales into services.
“We provide rail transportation as a service on the Intercity Express program in the UK which is responsible for meeting the operational consistency for the service level agreement to be profitable,” says Bjorn Andersson, Senior Director, IoT Solutions Marketing of Hitachi Vantara.
Hitachi, as the manufacturer of the railway transportation equipment, in this instance, "had the knowledge and design data to be able to do simulations ahead of deployment,” Andersson said.
Hitachi also installed sensors in all rail cars and related systems to monitor the condition of the equipment and to repair machines for consistent performance.
“We keep costs low by using sensors installed for other purposes such as surveillance cameras which provide data on how well a process or machine is working, how humans are interacting with the machines, inflow of material to a production line, etc.” Andersson added. (Read Why Superintelligent AIs Won't Destroy Humans Anytime Soon.)
Information Models for Digital Twins
To be sure, digital twins have mainly been used in pilots of specific applications and are not necessarily repeatable and scalable at this stage, we spoke to Dimitri Volkmann, a digital transformation consultant, who has experience working with digital twins.
“First of all, Digital twins need an information model and taxonomy, of the kind SQL achieved in the 1990s for software, to make them scalable and repeatable. It will start with a standard description drawn from a data description model; their primary data will characterize them with their attributes which sets the stage for extracting insights and intelligence about performance, simulation, and predictions." Volkmann said.
"Any application can then benefit from the data by accessing it with the same descriptors and query it for insights. Currently, every digital twin pilot has its own taxonomy and often proprietary data model, which makes it hard to integrate them.”
“A platform will best achieve the first task of creating digital representations of physical objects, their data descriptors, and taxonomy. After creating an inventory of digital proxies and their associated data, it becomes possible to have simulations and predictions and remotely control physical systems, using analytics and AI/machine learning."
"As the number of digital twins increases, it becomes possible to learn continuously for an expanded range of phenomenon.” Volkmann added.
What We've Learned About Digital Twins
Digital twins help to distill the intelligence from the maze of the physical world of complex systems. Not only do they help to gain situational awareness at all times but they also assist to control objects to achieve goals.
In a world of intersecting technologies and self-driven systems (Read 7 Autonomous Vehicle Myths Debunked), they make it possible to learn about the impacts of their individual elements and make corrections in real-time.
Above all, they bring transparency and flexibility to systems that were, until now, fixed and open the way for innovations in today's business models for operation management.