Modelling

Models are designed using applied engineering and technological skills and skills in basic sciences such as physics and chemistry. A real-life boiler’s model can be created virtually or a real-life plant’s model can be digitally created. Models can be made of aircraft engines, and power plants.

Such digital replicas are tested by simulation to see whether they meet the performance expectations of the original design. There could be a deviation, which could be analyzed and could be resolved. There is no need to suffer a breakdown.

Models could be made of products. Similarly models could be made of equipment and the entire plant. The whole enterprise could also be modelled — with its business processes, say manufacturing, supply chain, sales etc. Here there are people involved. It makes the model uncertain. Apart from physics, chemistry and engineering, you do require knowledge of human psychology to factor them in the model.

When models are made, they are called digital twins. A digital twin is essentially a computer programme that uses real-world data to create simulations. These simulations predict how a product/process will perform. Thus a twin is a connected virtual replica of an in-service physical asset.

The emergence of sensors on physical products in the last decade and the ability to process the data have made modelling easier.

Digital Twins

Projects have become capital-intensive. That is the reason why digital twins are crucial. There are three types of digital twins — product twins, production twins, and performance twins.

There could be combination and integration of these three. It results into a digital thread. It is a thread since it is woven bringing data from all stages of product life cycle (PLC).

These twins evolve and update continuously. There is a closed-loop of feedback in a virtual environment. It results into product, production and performance optimization at minimal cost.

In developing twins, the first step is conceptualization of the entity in terms of its organisational context and existing knowledge-base. There are interviews and focus-group sessions. As a second step, the twin is validated by tying up the twin to reality. It is tested to see whether it behaves as it does in reality. Validation is at the individual level and relationship level. As the last step, the twin is subjected to simulations — both quantitative and qualitative and do decision-making checks.

The model developed could be wobbly to begin with, but it evolves over a period of time. It can be tweaked and matured.

On a digital twin, one can run hundreds of what-if and if-what analyses for decision-making.

Benefits

Product digital twins validate product performance — how a product performs under different conditions If adjustments are to be made, they are made in the virtual world and in real-world the product performs exactly as expected.

There is no need for multiple prototypes. The development time is reduced. Twins improve the quality of the final product. There are faster iterations to feedback.

Production digital twins validate manufacturing before it is brought on the shop floor. It simulates the process. The production methodology is developed which remains efficient under various conditions.

Costly downtime to equipment could be curtailed. We can predict preventive maintenance.

Manufacturing operations become faster, efficient and reliable.

Performance digital twins use the vast amount of data from products and plants to get actionable insights for decision-making. The performance twins are leveraged to create business opportunities, aggregation and analysis.

Cube Model

There are 252 different classifications, but all of then do not occur in reality. Out of 30 plus digital twins case studies, it has been inferred there are three common digital twins.

AI and ML

These enable the creation of digital twins which are dimensionally accurate 3-D models. which can be upgraded quickly. It can be a 3D or 2D model.

Coinage

The term has been coined by Michael Grieves, University of Michigan in 2002. However, without calling it a digital twin, NASA has used the twin models for around 50 years prior to that. Gartner selected Digital Twins as one Significant Technology Trend for 2017 (in 2016).

Applications

Digital twins are used in manufacturing, automobile manufacturing, healthcare, energy systems, hotels and urban planning (say a City Twin).

Microsoft’s Azure Digital Twins is an IoT platform that enables creation of digital representation of real-world things, places, business processes and people.