Technology Blogs by SAP
Learn how to extend and personalize SAP applications. Follow the SAP technology blog for insights into SAP BTP, ABAP, SAP Analytics Cloud, SAP HANA, and more.
Showing results for 
Search instead for 
Did you mean: 

A business outcome-driven approach to digital transformation

The notion of a digital twin has recently taken center stage to describe a live digital representation of a connected physical asset. While the idea of creating a computerized companion, which represents the actual state of an object dates back more than a decade [read the original paper], the rapid proliferation of the industrial internet of things (IoT) enables a scalable and cost-efficient implementation even for equipment with production cost as low as $100 while capturing Petabytes of live data from complex assets. However, connectivity and storage of IoT data are only the technical requirements to capture the full economic benefit of digital twins.

This blog lays out a broader vision for digital twins and how they relate to business outcomes while a second part describes an implementation approach.

A live digital representation of a connected physical object

Building on the historic definition, a digital twin is a live digital representation (or software model) of a connected physical object. Physical objects such as industrial goods, machines, trains, trucks, wind rotors, appliances, but also factories, farms, or buildings. Objects are first conceived in a design and engineering stage, a stage in which typically models of the future product or asset are created. Analysis is performed by means of assumptions regarding how the physical asset will perform once placed in operation. Virtual prototyping may be used to validate the design. Still, a design model or a virtual prototype is not a digital twin. As for their human twin counterparts, the real asset and the digital twin are born from the same mold, as reflective manifestations in the physical and digital world, respectively.

The following five characteristics define what a digital twin is:

  • Identity – a digital twin represents a single, unique physical asset; though we would prefer a 1-to-1 cardinality between asset and twin, business outcome considerations may imply a 1-to-N cardinality, each describing a different dimension; the digital twin can be instantiated at the creation (e.g. engineering, production, configuration, installation) and lasts until the asset is retired (or beyond if historic data is required)

  • Representation – capturing the essential physical manifestation of the real asset in a digital format, typically using formats such as CAD, MES, or engineering models with corresponding metadata and classifications; traceability between digital twin and physical object is ensured through serialization

  • State and events – reflecting real asset state (e.g. condition, location, speed, environment) in (near) real time providing information on properties that describe various aspects that the digital twin is built for; the digital twin triggers alerts and events on behalf of the object; historic state is typically recorded

  • Context – describing operating context such as physical installation, ownership, reference to financial or asset management information, roles and business partner involved, service level agreements, service work performed on the asset

  • Interaction – all properties above must be securely and digitally queried in a software system by other systems (via APIs) or users

A number of optional characteristics can further enrich or implement a digital twin:

  • Composition – the composition (or bill of material) of the object is often represented in the digital twin as a hierarchy, graph, grid, or multi-dimensional model; different user context may require re-composition (e.g. operations view vs. maintenance view)

  • Simulation and behavior – physics-based models (e.g. Newtonian structural analysis, computational fluid dynamics), statistical models (e.g. classification, anomaly detection, health scoring), or machine learning algorithms (e.g. for video analysis) provide insight and foresight into the behavior of the object beyond what is measurable (e.g. using virtual sensors)

  • Control – remotely controlling the physical object via actuators exposed by the digital twin; this may also involve triggering physical transactions

  • Distribution – the digital twin data model does not need to reside in one software system, but may be distributed between the edge (in proximity to the physical object) and central instances (e.g. in the cloud)

  • Visualization and analytics – a user experience (web, mobile, conversational) to monitor state or condition (including simulated or derived information), to query historic behavior, to receive alerts and notifications; this may involve a 2D or 3D representation (including augmented reality overlaying physical and digital world), replay of past behavior, or forecast of future behavior

Digital twins are built with a purpose in mind. A few examples used in industrial or commercial environments include:

  • Digital twin for connected product: Manufacturers create digital twins for their products to follow through the entire lifecycle from design, production, delivery, usage and service at customer to end of life

  • Digital twin for connected asset (or vehicle in logistics): Owners / operators use digital twins for their assets or their fleet of vehicles to track, maintain, optimize, and manage assets

  • Digital twin for production plant (or store in retail): Manufacturers (or retailers) map their factory (or store) to manage and improve production and operations

  • Digital twin for infrastructure (e.g. utilities grids, smart city): Like digital twin for assets, however, with focus on grid-like infrastructure to represent flow of vehicles (in cities), water, waste water, or gas

Digital twin for product

As a first example, consider a manufacturer of commercial or industrial goods. Building on an established after-sales business with support, repair, and spare part services (which account for 30-40% of total revenue), a connected product promises to drive additional value along the lifecycle of a product.

Most manufacturers start from improving the services efficiency. Digital twins for connected products enable more reactive issue resolution with remote diagnostics. The right resources (service technicians, tools, and spare parts) are more efficiently dispatched and can find the right solution using historic and actual data recorded in the digital twin. A joint view on the digital twin from customer staff, field services, and remote support from the manufacturers simplifies collaboration.

An API-based access to digital twin allows to create mobile applications and conversational interfaces for field services and remote support staff while ensuring proper access rights to critical data. These can be augmented with virtual reality, live video streaming, virtual repair and assembly views, or prescriptive instructions.

As manufacturers strive to differentiate in the market with additional services, connected products allow to develop data-driven business models and offer value-added services like benchmarking, machine optimization, or prescriptive maintenance.

A digital twin is also the prerequisite to develop new product-as-a-service business models which are based on actual product usage or consumption.

Based on usage information and condition monitoring, the sales or service department of the manufacturer can take a proactive approach at the end of life of the equipment and recommend the right replacement product.

While most manufacturers have digitized design documents (CAD, PLM) and use CAE/CAM for engineering purposes, access to digital twin information from the services phase provides a different level of customer insight. The composition of an asset (bill of material) may be represented and visualized differently for various purposes: The product design team looks at an engineering view while the services team requires a view with detailed assembly and disassembly information.

Continuously ingesting and analyzing data not only allows to better design products based on actual usage, but also provides immediate feedback after production on potential warranty issues. At the same time, engineering will evolve the edge capabilities of the equipment using embedded systems and systems engineering techniques to lay the foundation for additional value-added services or product-as-a-services business models.

While the digital twin is often created during the design phase, instantiation and binding to a physical object happens during the production process. Serialization, hardware configuration, supplied components, and software versions are recorded as a reference for future traceability. Moreover, measured production and quality data supplement configuration information and provide the baseline for the service phase. With more sophisticated embedded systems, software-driven customization provides tailored products and after-sales services for individual customer needs.

The digital twin of the connected product provides a common link between manufacturers, customers, service partners and potentially financial services providers and insurers.

Digital twin for asset

The second example of a digital twin is a connected asset for an owner or operator of the asset. While the technical implementation may not vary between connected products and connected assets, the business perspective is different. An operator of an asset strives for overall equipment effectiveness and operational efficiency while improving asset performance and reducing maintenance cost. Managing risk and safety are equally important in many industries. Most industrial assets are built or assembled from individual products and require a holistic perspective.

A digital twin provides a live digital representation along the asset lifecycle. With a strong focus on efficiency, condition monitoring provides a real-time visibility into the state of an asset during operations and allows operations staff to react immediately to any anomaly, continuously improve, and gain insight and predict potential issues. For the maintenance crew, predictive maintenance delivers asset health scores and remaining lifetime information on a component level; however, most components are not inspected, maintained, and repaired individually, but as part of scheduled maintenance which includes resource allocation for people, tools, spare parts, and potentially workshop space; moreover, maintenance windows need to be aligned with operations phases.

Using historic data easily accessible through the digital twin, operators will be able to take a data-driven approach to asset disposal or reuse. During the planning phase, data allows to simulation asset performance and assess risks in detail, complementing and enhancing traditional reliability engineering methods. While emerging product-as-a-service models from manufacturers promise higher services share, operators adopting these models will need access to digital twins to define service definitions and service level agreements.

With an increasing amount of software in embedded systems, operators will need traceability and version management to ensure compliance. Digital twins provide a live representation of what is deployed in the plant.

The digital twin for connected assets serves also as the central integration point for various stakeholders working on or with the assets. These not only include operations and maintenance staff from the owner / operators, but also external service providers and contingent workers, insurers and other financial service providers. In the case of asset-as-a-service business models, the digital twin becomes the hand-over point between data used for providing services and data used for managing end-to-end operations at the operator.

With an API-enabled and managed digital twin, third party applications may leverage the information for different purposes. While data sovereignty and access rights need to be managed on a fine granular level to ensure compliance and data protection, they will be governed at a business level based on contractual relationships (and not on a per end user basis).

A network of connected products and assets

In today’s networked world with many different roles and relationships, bilateral setups like equipment manufacturer portals or supplier collaboration spaces will not scale when moving from orders and contracts to live streams and representations of connected products and assets. Several other business partners are involved to deliver contract services, rental services, outsourcing services and to enable pure SLA-based business models like product-as-a-service. In these business networks, partners play different roles and require a role-based access to digital twin and related asset information:

  • Manufacturers frequently update asset information with service bulletins, new spare part references and availability, firmware updates for embedded systems

  • Service providers act on behalf of customers or manufacturers and deliver services against defined SLAs

  • Insurers provide usage-based contracts adapted to individual risk profiles based on actual asset deployments

  • Asset owners or operators as primary data owners need to agree to data usage and proliferation (e.g. usage of specific or anonymous asset condition or environment data)

  • Suppliers are involved to review designs based on shared digital twin information and to trace component performance from inception to retirement

Moving from a product-centric to a service-centric economy will elevate the role of the digital twin from a data-driven model to the basis of financial accounting. Product-as-a-service providers require live information for proper revenue recognition and forecasting while operators consuming asset-as-a-service models look at SLA adherence with financial impact as well as quantifiable live asset risk and performance information.

Further examples of digital twins

The idea of digital twin as a live digital representation can be applied to more complex physical structures, often combining connected assets and products with specific business outcome in mind.

  • Production: Digital process control for operations teams is widely used in discrete manufacturing, mills, and process industries; maintenance teams have mostly relied on historic information, usage and failure data to schedule planned downtime; the combination of connected assets with quality and process information in real-time creates an end-to-end perspective for a production process

  • Retail: A digital twin of a retail store provides information about status, energy consumption, and asset health of equipment like refrigerators, HVAC, or vending machines; information used in the past either in the front office by store staff or central maintenance and store planning teams is made accessible via digital twin APIs to create persona-centric applications; the digital twin spans both immediate issue resolution as well as long-term equipment strategies

  • Utilities: Combining live information from grid operations (e.g. water flow) with connected asset information (e.g. pump stations) creates a digital representation to manage and maintain the infrastructure to deliver utility services; external service provider integration, compliance, and safety requirements are specific aspects of digital twins for utilities

  • Moving assets: Fleets of connected vehicles (e.g. forklifts, vans, trucks) or connected moving assets (like trains, ships, airplanes) add a spatial dimension to the digital twin; geo-fencing, track and trace, and other location-specific information (e.g. weather, road conditions) are commonly found when building solutions for logistics and transportation

  • Farming: A farm combines physical space with moving assets and potentially production facilities with the purpose of improving productivity and profitability

Ready to implement? Read part two on how to implement a digital twin for products or assets.

Do you share this vision for digital twins? Leave your comments below.
1 Comment