The concept of the Digital Twin continues to receive a mixed reception when it is raised in conversation. There are a multitude of factors that have given rise to this ‘love-hate’ relationship, ranging from conceptual stretching to mis-selling, to conflation and a certain amount of dissonance between vendors, users and consultants. Underlying all these factors is a common theme, a fundamental lack of understanding!
To address this lack of understanding and to refresh the understanding of those who believe they are well versed in all things digital and twinning, we have decided to compile a repository of our work on Digital Twins from the last 8 or so years. The series covers the fundamentals (basics) and the journey from the definitions of what is and what is not a Digital Twin to technical approaches for cost modelling of a Digital Twin project and methods for appraising and quantifying the benefits.
This first blog set the scene for the series including extracts from a talk given to an international audience by Prof Hicks three years ago. Our second blog covers what we call the ‘definitions dilemma’. So named because it is only when using these definitions correctly that organisations must come to accept that they might not have a Digital Twin after all! But, the difference between a Digital Twin, Digital Shadow and Virtual Twin are not the only terms to be misused. As a consequence, we present detailed, unambiguous definitions of all things digital that relate in full, or part, to Digital Twins.
| Term | Definition |
| Digital Twin | Digital Twin consists of three components: a physical product, a virtual representation of that product, and the bi-directional data connections that feed data from the physical product to the virtual representation, and information and processes from the virtual representation to the physical product. |
| Digital Shadow | A digital twin is not a shadow, because shadows – even though a wide range of data is available – do not provide interoperability between the real and physical world. The digital shadow is limited to collecting and presenting data, while a real digital twin can interact with a real-world object. |
| Digital Thread | A data and/or information flow between systems and/or people in an organisation that is systematic, consistent and auditable delivering the right information at the right time to the right location through the right mechanism. |
| Surrogate Model | Analytical model(s) that approximate the input/output response of complex systems when the actual relationship is computationally expensive to evaluate. These models approximate the behaviour of the underlying complex simulations to an acceptable precision while also being computationally cheaper to evaluate. |
| Multi-physics model | Models the physics of more than one physical property and the interactions between properties. Examples include thermo-mechanical models and coupled CFD lift and heat transfer models. |
| Simulation (Environment) | The modelling of a system (using one or more models) with a set of input parameters to produce a set of output parameters. The environment is considered as part of the system. |
| Grey box | A grey-box simulation is a hybrid modelling technique that combines elements of both white-box (physical) and black-box (data-driven) models. In a grey-box model, parts of the system’s structure are known based on first principles (e.g., physics-based equations), while other parts are inferred from data, allowing for flexibility and adaptation where full knowledge of the system’s behaviour is unavailable. |
| Black box | Input-output models based purely on data with no representation of the underlying physical characteristics of a system. |
| Condition Monitoring | Assessing the condition (periodically and/or in real-time) of an asset through the analysis of asset sensor data and/or inspection techniques (typically non-destructive). |
| Anomaly Detection | The identification of distinct or abnormal occurrences or features within a signal (data stream) that do not conform to known patterns. |
| Diagnostics | The ability to identify the causal conditions for a feature identified in a dataset. |
| Prognostics | The ability to forecast future events/actions based on some current information. |
| Model-in-loop | The use of a model within a control-loop of a system where the information is used to affect human or AI-driven decision-making. Often used interchangeably with model-based testing. However, MBT is more generalisable. |
| Model Based Control | Model-based control facilitates the development of complex systems using a model-based design approach. Among the notable examples of application in control engineering is model predictive control (MPC), where it is usually adopted to address the complex behaviour of nonlinear dynamic systems. |
| Software-in-the-loop | Executing software on a virtual platform that mimics the target hardware, enabling thorough testing and validation without requiring the actual physical hardware. |
| Emulation | Ability to reproduce/imitate an asset typically in VR/AR. The fidelity of the emulation can vary. |
| Virtual Twin | Digital twins and virtual twins are both digital representations of real-world objects, but they differ in how they are used and what they can do. Virtual twins allow users to model, visualize, and simulate a system or environment. Virtual twins are often used to prototype and refine products and processes, and can be used to simulate complex scenarios without real-world consequences. Virtual twins are often used in manufacturing and product design, and can also be used in healthcare and urban planning. A Virtual Twin is a twin of the ‘physics’. It does not have any data interfaces either to or from the physical and is hence not a digital shadow or digital twin. |
| Device Twin | A device twin is a virtual model of a specific physical device, typically used in the Internet of Things (IoT) context. Device twins can be used to mimic data from sensors or other sources and can be used to simulate the behaviour of individual devices in an IoT system. |
| Virtual Product Development | Virtual product development (VPD) is the practice of developing and prototyping products in a completely digital 2D/3D environment. VPD has four main components: 1) virtual product design (3D shape, 2D graphics/copy); 2) virtual product simulation (drop test, crush test, etc.); 3) virtual product staging (retail space planning, consumer research and behaviour analysis); and, 4) digital manufacturing (process planning, assembly/filling virtualization, plant design). |
| Reduced order model | A computational model, often based on the physics and system dynamics but with only the minimum number of system orders to capture those dynamics in the range of interest. |
| State Estimation | Constructing a “true” value of the time-varying state of a system (i.e. position, speed, acceleration, current, voltage etc.) from the actual measured variables. This term captures a variety of techniques for handling uncertainty, sensor noise, systematic error etc. and allows us to handle the system ‘states’ with an associated uncertainty computed in a robust and rigorous manner. |
| Parameter Estimation | Determination or updating of unknowns within the system model or simulation. These may be completely unknown constants in a model, or initially known parameters which are time varying (or initially unknown parameters which are also time varying). |
| Artificial Intelligence | Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy. |
| Machine Learning | Machine learning (ML) is a branch of computer science that focuses on using data and algorithms to enable AI to imitate the way that humans learn, gradually improving its accuracy. |
| Feature Engineering | Is a preprocessing step in machine learning and statistical modelling which transforms raw data into a more effective set of inputs. Each input comprises several attributes, known as features. By providing models with relevant information, feature engineering significantly enhances their predictive accuracy and decision-making capability. |
| Virtual Model | Virtual model refers to a computer-based model. It is not necessarily a virtual twin. It could be a simplified representation/approximation. |
| Model Based Design | Model-Based Design (MBD) is an approach used in engineering, particularly in fields of control systems, embedded systems, and software development, to design and develop complex systems and products. It involves creating mathematical and visual models of the system’s behaviour and using these models throughout the development process to design, simulate, test, and implement the system. |
| Visual Programming Language | In computing, a visual programming language (visual programming system, VPL, or, VPS), also known as diagrammatic programming, graphical programming or block coding, is a programming language that lets users create programs by manipulating program elements graphically rather than by specifying them textually. |
| Text-based programming language | Text-based programming languages differ from visual-based (or block) programming in that it requires individual lines of code to be written. |
| Closed loop control | A closed-loop controller or feedback controller is a control loop which incorporates feedback, in contrast to an open-loop controller or non-feedback controller. A closed-loop controller uses feedback to control states or outputs of a dynamical system. |
| Open loop control | In control theory, an open-loop controller, also called a non-feedback controller, is a control loop part of a control system in which the control action (“input” to the system) is independent of the “process output”, which is the process variable that is being controlled. It does not use feedback to determine if its output has achieved the desired goal of the input command or process setpoint. |
| Model | A model can be either physical or virtual. The latter is sometimes called digital or computer-based. |
| Product Lifecycle Management | The process of managing the entire lifecycle of a product from its inception through the engineering (design and manufacture), in-service and eventual disposal. |
These definitions have been co-developed with our industrial partners and are each grounded in literature and international standards. All citations are available on request.
We hope that you find these useful and can start to build you own dictionary of terms that conforms with accepted definitions and standards. In the long-term, using a common language is mission critical for all stakeholders!
Please do get in contact with the team if you would like to know more and/or work with us. Watch out for our third blog that will cover the seminal papers on Digital Twins.