Digital Twins Are Here
Published on : Saturday 04-09-2021
It is important to recognise the distinction between the different types of digital twins and understand the role of a digital thread.
Digital twins – digital representations of real-world entities such as sensors, devices, machines, systems, and even people – have evolved from a niche concept to an integral part of the industrial landscape.
In fact, due to the rapid adoption of digital twins, ABI Research forecasts that the industrial digital twin market will grow from US dollar 3.5 billion in 2021 to US dollar 33.9 billion in 2030, at a 29 percent Compound Annual Growth Rate (CAGR).
However, despite the growing adoption and prevalence of digital twins, there are still many misconceptions about the term. The issue stems from the fact that digital twins are not a technology, but rather a composition of solutions aimed at bridging the physical and digital worlds, from design through simulation, manufacturing, assembly, and after-sales service and support.
Defining digital twins
The term digital twin gets thrown around casually, but its important to recognise the distinction between the different types of digital twins and understand the role of a digital thread.
Digital Twin: Digital twins are digital representations of real-world entities (sensors, devices, machines, processes, complex systems, and even people/persons/living beings or entire facilities) deployed to drive business outcomes.
They provide connectivity, metadata management, data management, increasingly advanced analytics, and often integration with business applications and process systems.
Digital Thread: A digital thread is the glue that holds a digital twin together and allows for communication and interaction between the digital twin components. It is a record of a product or systems lifetime, from its creation to its removal. It is an enabling technology (for digital twins) that follows designs from their earliest ideation (digital definition, such as CAD) through real-world operation.
The four types of digital twins
Basic Digital Twin: A true digital twin requires a near real-time representation of a physical asset. It stores data and insights from monitoring to allow engineers and other employees or partners to extract insights. This is more than a simple system for alerting because of the data storage capabilities coupled with near real-time asset visibility and, importantly, context to derive insights from parts or processes that cannot be easily observed.
Intelligent Twin: Intelligent twins take the next step beyond basic storage and visibility to perform and provide actionable analytics. This level of twin includes pattern recognition and can enable higher-value use cases such as twin-based predictive and prescriptive maintenance. The use of ML on collected data and more advanced analytics on single or fleets of endpoints are common traits of this level of maturity.
Simulation Twin: Simulation twins are characterised by physical or physics-based simulations that take collections of sensor data and run simulations on monitored components to determine the real-world impact on factors like durability and wear, process and performance, or overall design.
Executable Twin: The general premise and key attribute of this most advanced stage of digital twin is that twin models include environmental context that spans not only the asset twin, but also the environment in which it operates. Ultimately, this means instituting closed-loop quality and likely 3D and spatial mapping, which can be achieved with fixed or mobile sensor data (e.g., video). The ability to confidently perform or implement commands on a physical model because of digital simulation is another notable characteristic.
Putting digital twins to work
Digital twins provide value to a wide range of stakeholders throughout the technology industry, including:
1. Brands and OEMs: Digital twins enable improved customer service, new revenue streams, and entirely new business models (e.g., Outcome-as-a-Service).
2. Plant Managers: Digital twins drive better business process visibility across the facility and between systems by uniting critical data feeds, from design through manufacturing.
3. Frontline Workers: Digital twins provide and catalogue real-time information for more efficient operations and better demand response.
4. Operational Executives, such as Lean Leaders: Digital twins allow rapid, data-driven insights for the health and performance of assets and processes, with continuously updated sensor values and states.
5. Engineers: Digital twins allow more rapid deployment of new applications in addition to simulations and scenario planning for business process optimisation.
Key applications
The true game-changing value of digital twins is in the unification of digital and physical assets for better operational continuity and ultimately better outcomes. The following applications and use cases are some of the most prominent.
Design: While many companies create computer-aided design (CAD) models before building prototypes, these models do not really fit the definition of a digital twin because their physical twin has yet to be built. So, in this case, they fill the role of a digital prototype. Companies use these digital prototypes for engineering and testing purposes, such as verifying and inspecting the overall 3D design and making sure all parts fit together.
Simulate and Test: Once the engineering team has a basic design, team members can run as many physics-based simulations on the digital prototype as needed. This allows engineers to understand the limits of products in a field environment and make adjustments without wasting physical resources. Once the physical prototype is built, the actual digital twin can verify performance of the physical prototype in near real-time to validate the simulations. These simulations can continue when the physical twin is in operation by using predictive analytics for failing parts, providing insight for engineers on how to improve the design.
Instruct and Build: Tablets and smart glasses with augmented reality (AR) applications can alert and overlay a digital twin on the assembly line to show instructions in context for hands-free access to information that can assist complex or customised assembly and engineering changes. This increases productivity and improves quality by minimising errors and promotes on-the-job training for a better allocation of resources.
Monitor and Maintain: Traditionally, monitoring and maintaining industrial equipment meant routine inspections and tests based on historical and physically-observed operating performance. Now, with the IIoT, sensors monitor performance and health in real-time. The ability to assess equipment, measure status, and perform troubleshooting in a virtualised and sensor network-enabled environment alleviates the need to staff in-person inspections.
With digital twins, managers take the next step by aggregating sensor data for each asset and enriching it with historical, fleet, environmental, and situational data. By continuously analysing this data and increasing transparency between departments, plant operators can optimise operations. In the field, technicians can use AR technology to display an overlaid digital twin on a physical twin/asset to ensure they repair or replace the correct part the correct way.
Update and Improve: Machine learning software can draw insights as more data feeds into digital twins. Sophisticated digital twin models can take these insights and make suggestions on how to improve performance via software updates, new processes, or redesigned components. For connected devices, plant owners can send software updates over the air. At more advanced levels, digital twins act as digital masters. They command the physical twin based on analytic and predictive models.
The market today – and tomorrow
The pandemic led companies to focus on technologies that empower remote operations and reliability for customers, boost revenue, and maintain margins by keeping costs down. Digital twins play a role in all three—from training new operations staff before they step foot in a plant, to using simulations and real-time data for better business planning.
The main takeaway from research interviews is that digital twins have been pushed up the adoption curve, with the conversation shifting away from How do I build a digital twin of my entire enterprise? to How does a digital twin create value today? This change is less about deprioritising other investments and more about the need for faster time to value.
Other recent market developments include:
1. The emergence of standards and generally accepted definitions, allowing for data to be aggregated across dissimilar systems.
2. The growth and enhancements of edge intelligence, which leverages robust APIs and microservices to improve data aggregation and contextualisation for better decision making.
3. The integration of artificial intelligence (AI) and machine learning (ML), which allows for continuous predictions and dynamic optimisations, and the ability to push digital twin technology into lower-level components which can be aggregated into a comprehensive digital twin of an entire facility.
On the horizon
There are several key trends and advancements that will shape the future of Digital Twins.
Data and Model-Based Standards: While industry standards consortiums have formed, progress is needed to improve access portability, and ease of implementation.
Marketplaces: Industrial firms want more individual apps that yield robust and repeatable implementations.
New Data Sources: Integrating real-time data from virtual sensors, fixed and mobile video cameras, and frontline workers provides the necessary dimensions for full context twins.
New Business Models: Greater control of assets and their data allows new business paradigms, like power-by-the-hour in the transportation industry or Data-as-a-Service for reporting on the performance of equipment on behalf of clients.
A Continuation of AI/ML Inclusion: Adding a virtual digital model on top of a highly automated process creates new visibility into performance and enables companies to leverage AI to identify new correlations in addition to potential optimisation.
OEE Aggregation and Inclusion: Equipment-level digital twins will be aggregated into an operational process-centric view that identifies operational performance improvements rather than focusing on maintenance optimisation.
Executable Twins: Digital twins of spaces will converge with digital twins of products to create full-context executable twins. Example data sources include frontline workers, video/machine vision, and environmental sensor data.
Strategic recommendations for technology suppliers
Focus on End-to-End: Manufacturers want something that works. This means starting with a narrow set of applications that accelerate time-to-value. Solutions that do this will be holistic and end-to-end to support projects from pilot through scale, which may require partners.
Recognise the Evolving Role of Edge: Digital twins and IIoT specialists need to think about operating at the edge and reporting to the cloud. Edge, alongside 5G, enables low-latency capabilities like real-time sensor monitoring, Augmented Reality (AR), and robotics. The edge is also growing more encompassing with the convergence of Engineering Technology (ET), Information Technology (IT), and Operational Technology (OT).
Embrace Microservice Architectures: Microservice architectures allow for agnostic cloud deployments, as well as on-premises deployments when required. The use of Rest Application Programming Interfaces (APIs) can be used for out-of-the-box integration and accelerate time-to-value for customers. The key is to have enough data from the physical asset to feed into the digital twin so that the simulated output tracks tightly with the as-built asset.
Embed Simulation and AI: Simulation is key in both the design and manufacturing of next generation products. AI is important to automate data tasks impractical or tedious for users. Together, simulation and AI allow for greater degrees of not only automation, but also optimisation within and between departments.
Strategic recommendations for manufacturers
Take an Outcome-First Approach: Technology should not be deployed for technologys sake. Prioritise high-value use cases that provide a short and meaningful time-to-value. Once a plan is in place, IT needs to be able to execute fast and cost-effectively.
Put an IT Person on an OT Team: IT and OT professionals need to work together. For this reason, it is important to have an IT/OT liaison for the duration of a project. Doing so will ensure functional requirements are properly accounted for and KPIs achieved, with minimal rework.
Prioritise Continuous Learning: There are three dominant strategies for change management: 1) Greenfield, 2) Side-by-side, and 3) Gradual evolution. The most successful digital twin deployments understand the options and continuously re-evaluate progress.
Consider Scale, but Not Too Soon: Executive mandates are useful to get the organisation to corral around a certain way of thinking or business objective. This helps energise sentiment around digital transformation however, adoption works best with buy-in from the ground up.
Whitepaper courtesy: ABI Research. Download link: https://abi.link/3k5APLk.
Digital Twins is just one example of a technology transforming the manufacturing and industrial space. ABI Researchs Manufacturing and Industrial Service can help you keep up with all of them, providing you with the insight you need to understand the technology landscape, assess your market and your competitors, and make optimal investments. Contact: [email protected]
