A digital twin is a digital representation (digital model) of actual physical objects, systems, or processes synchronized at a specified frequency and fidelity. 

Combining algorithms with real-time, historical, and synthetic data, a digital twin represents the past, present, and future states of physical objects, systems, or processes. 

Depending on the object of the digital twin, whether it is a physical thing, system, process, or all together, a company can leverage it for different purposes. 

• Simulate the behavior of the object under various external circumstances, 
• Do virtual experimentation, verify performance hypothesis, and optimize the object performance with different configuration 
• Monitor the object’s actual state  
• Predict the object’s future performance  
• Proactively maintain the object minimizing the cost 

A digital twin has several key elements:

Digital Twin Platform Components

• Entity metadata, a model describing a twinned object, depending on the scope, can have physical components details, behavioral characteristics, cross components interaction details.   
• Synthetic data representing a state at a time of a twinned object. Such data is often time-series and might be real or synthetic/generated.  
• Analytical models are algorithms describing the state transition of twinned objects and processes in time. These algorithms answer questions of what if, what is now, and what is next.  
• Other software components are visualization engines, monitoring dashboards, integration layers, etc. 

It might take from 2 months for relatively simple cases and up to 6 months for large and complicated systems/processes/physical objects. 

Depending on the use case complexity and scope of the digital twin, which includes 

• The complexity of a physical object, system, process, or combination.
• Availability of data and its quality.
• The number of integrations.  

• Physical twin: a digital replica of a physical asset, such as a machine or a building. 

• Process twin: a digital replica of a process, such as a manufacturing process or supply chain.
• Operations digital twin: an operations digital twin is a digital model of a company’s end-to-end processes and operations required to produce value or deliver a service. 

• Functional twin: a digital replica of a system or component, such as a power generator or an aircraft engine. 

• Conceptual twin: a digital replica of a concept or design, such as a city plan or a new product.

Depending on the fidelity, it can be a simple object/process of a very detailed system model with many components.

• Physical twin: a digital twin of a wind turbine can be used to monitor its performance and predict when maintenance is needed.

• Process twin: a digital twin of a supply chain can be used to optimize delivery routes and inventory levels.
• Operations digital twin: a digital twin of the end-to-end customer journey of a bank can be used to optimize the flow span across several product lines. 

• Functional twin: a digital twin of an aircraft engine can be used to optimize its performance and predict when maintenance is needed.

• Conceptual twin: a digital twin of a city can be used to simulate and optimize traffic flow and energy consumption.

The cost structure of the digital twin is comprised of two elements: 

Cost of implementation:

Often, it is higher than the cost for the platform, and it will range a lot depending on the implementation service provider rates. Some digital twin platform companies provide service by themselves, like GE, Siemens, etc., vs. other leverage service provider partner ecosystems, Microsoft, NVidia, etc. 

Cost for a platform:

Different platforms embrace different pricing models. 

For instance, Microsoft’s digital twin platforms bill for resource consumption similar to cloud services – you pay for what you use.

Players such as AnyLogic and VSOptima use monthly/annual subscription fees.

An operations digital twin is a digital model of a company’s end-to-end processes and operations required to produce value or deliver a service.

Usually, the digital twin implementation project is comprised of three steps:

Step 1. Define the objectives and scope for a digital twin

• What is the overall vision for digital twins?
• Which digital twins give our enterprise the greatest leverage and opportunities for reuse?
• What is the total value at stake?
• What are the highest-value, most feasible use cases we should deliver first, and what is the process to attribute value to digital twins?

Step 2. Analysis and Design a digital twin

• What is the physical assets/system/process boundaries?
• What data layers and attributes do we need to collect?
• How will we source and model the data?
• What models will be built on top of the data? What is end-state architecture?
• How will the project team work together with business users to deliver the digital twin and use cases it supports?

Step 3. Build the model and integrate a digital twin

• Build and calibrate the model
• Integrate the model with data sources to make a digital twin
• Configure predictive capabilities (horizon, fidelity, and frequency) 
• Onboarding the digital twin operating team

DT is used to increase the efficiency of asset management and service delivery by quick identification of system issues and to predict future performance under various scenarios. 

• In the case of a heavy-asset industry (e.g., manufacturing, utilities, natural resources, or facilities management), the state of the asset, its proactive maintenance, and utilization are the core of business efficiency. Often businesses in such industries benefit from asset and equipment digital twins since they enable monitoring of the asset state, performance prediction, and prescribe actions given the asset performance function. 
• In service-focused industries (e.g., healthcare providers, retail and supply chain, finance, insurance, etc.), the service operations performance becomes a core management focus for the business. In this case, such companies can benefit from operations digital twin since it will reflect the current state of end-to-end service delivery, identify issues, suggest mitigation actions, and predict future performance under various scenarios.     

The key functions of the digital twins tangent to the twinned object (physical thing, system, or process) are the following:

• Describe what happened with the twinned object.
• Diagnose why such a thing happens (ed) with the twinned object. 
• Predict with high accuracy what is going to happen with the twinned object. 
• Prescribe what action should be taken to address identified issues of the twinned object. 

A digital twin platform is a set of integrated services, applications, dashboards, and other digital twin subsystems that are designed to be used to implement and run a digital twin.

Gartner outlines the blueprint of such a platform, which often is a combination of various solutions from multiple vendors.

Key building blocks are data capabilities to work with unstructured, relational, and time-series data, Integration capabilities, IoT platform, Analytics tools, Engineering tools, Simulation software, process discovery software, Security capabilities, application delivery, etc.  

According to the state of digital twin 2022 report, executives of companies implementing digital twins see the following benefits: 

• Improved customer satisfaction, since digital twin gives companies the advantage of understanding better customer needs and behavioral patterns and how products/services satisfy them. 
• Improved product quality since a company has the actual state of an asset or a service; therefore, can proactively mitigate issues or potential advert events.
• Reduced time to market, empowered with data and simulation, a company can accelerate the innovation cycle through experimentation in the digital world that is cheaper, faster, and more accurate.  
• Inform and drive sustainability efforts since a company has an end-to-end view of the operations and optimizes it for various factors, including sustainability. 
• Enhance supply chain agility and resilience since a company can find the optimal supply chain configuration under different scenarios, including ones that don’t exist in historical data. 
• Enable new business models, converting products into services business or providing additional service business on top of the product.  
• Drive operational efficiency and improve productivity – reduction of downtime in asset management or reduction of operational waste in services and processes obviously brings about better operational efficiency and productivity.

Advantages of digital twin:

• Provide real-time visibility on assets/service performance.
• Enable and accelerate continuous improvement cycles.
• Business model innovation – products as a service
• Operational efficiency and productivity improvement
• Better product/service quality

Disadvantages of digital twin:

• The success of technology is often dependent on internet connectivity, especially for physical assets.
• Security and privacy are at stake.
• The adoption journey might take some time. 
• Adoption inequality among organizations.