Your challenges when it comes to digital twins

The simulation of components, processes, and even entire plants can contribute to detecting possible faults even before start-up. After start-up, your digital twin can be used to constantly monitor the condition of the product or system and to predict pending failures.

Are you already planning a digital twin but facing problems with the implementation, architecture, development, interfaces, or the processing of huge quantities of data?

Breathe! We won’t leave you to cope all on your own.

Your benefits through using a digital twin

The use of a digital twin enables optimizations and troubleshooting across the entire product life cycle, thus saving considerably on costs. Examples of digital twins that are already in use today include the virtual model of Singapore for the optimization of city planning and the digital modeling of an airplane turbine at Boeing.

This allows the optimum design of component shapes with regard to flow behavior and strength characteristics using machine learning methods and algorithms during development. Faults and vulnerabilities are detected early or avoided altogether. The algorithms select simulation models with the greatest probability of success, which minimizes the work required for the creation of complex, expensive prototypes. This saves on costs and avoids wasting time, since the development process is streamlined.

A digital twin isn’t suitable only for implementation at the start of the product life cycle; instead, it can also be advantageous for an existing product or system already in use in production plants. Real data can be used to identify dependencies and optimize products or systems. On the basis of this, predictive maintenance can be introduced, thus identifying failures in advance so that maintenance work can be planned more effectively and downtime can be prevented.

How a digital twin works

Two application scenarios emerge when we look at the idea of a digital twin in relation to the age of the product in the product life cycle: A digital twin used in the planning/development phase and a digital twin for a product that is already in production.

A digital twin in the product planning/development phase

In order to acquire an advantage at the beginning of a product life cycle, it makes sense to develop a digital twin right from the very start. This enables the detection of faults at the start of the life cycle and means that you can test individual devices or entire systems without having to carry out expensive prototyping. Furthermore, it allows system interactions to be simulated in advance, enabling the testing of various product variants, for example. This results in major savings on time and costs in comparison to the customary approach. The digital twin created in this phase can be used later on in the product life cycle, too.

A digital twin for a product already in production

Once a product is already in production, there are two possibilities for implementing a digital twin. On the one hand, you can work with existing digital models that are then additionally linked with real sensor data, a database, and analysis software. On the other hand, you can use the real sensor data to model a digital twin with the help of machine learning.

A twin is used in the production phase to find vulnerabilities, detect general dependencies in the system, and to improve efficiency on the basis of the new information. In addition, the digital twin enables the introduction of predictive maintenance, which can determine the probability of failure on the basis of models and the real sensor data. This allows for the early identification of critical components and the performance of targeted maintenance tasks.

Furthermore, software updates or new products can be simulated in advance on a digital twin to illustrate the behavior of the machine or production line. Again, faults are identified in advance and can therefore be prevented, which increases efficiency and consequently reduces costs. 

Moreover, the sensor data can be used to constantly optimize the product, production line, or component. With the help of machine learning, correlations are identified early on from the local condition data, and improvements can be derived as a result.

In addition, personnel can be trained with the help of a digital twin and virtual reality technology. This means that new employees can be trained on virtual production plants and machines. This increases the learning pace and reduces error rates.

Because products are so diverse in nature and have differing levels of complexity, there are different types of digital twin, too. Thus, we can talk about part twins, product twins, and system twins:

Part twin

Digital twins of existing components have been in use for decades now. The design of components in the development phase using CAD programs, FEM simulation software, and/or flow simulations is now absolutely standard in industry. The use of machine learning methods can further optimize the process and restrict simulations to the truly essential elements. This shortens simulation times and enables new, unusual, and more efficient forms to be developed, thus lengthening the lifetime of a component, reducing the development time, and ultimately reducing costs.

And the use of the part twin doesn’t stop after the development phase.  In the next step of the product life cycle – start-up – further data on the part is collected using sensors. In this way, the performance of the part is monitored, and machine learning is used to derive improvements. This enables predictive maintenance and the constant improvement of the component.

Product twin

Unlike a part twin, a product twin comprises multiple components or multiple part twins. In addition to the parts themselves, these twins can include regulators and software. Like a part twin, such twins can be used to save on time and costs during development. Complex, high-level dependencies are simulated and detected and development parameters are set optimally.

During the product life phase, predictive maintenance, monitoring, software updates, and product simulation are comprehensively provided for. The constant supply of real data provides a good basis for meaningful predictive models, stress tests, and simulations.

System twin

A system twin exists at a higher level than a product twin and consists of multiple product twins. A system twin might depict a production line or even an entire plant. Above all, this kind of model can be applied to interaction with other plants and suppliers, logistics, and the capture of condition data for an entire plant.

Reference architecture

The figure below shows a top-level depiction of the reference architecture of a digital twin.

Reference architecture of a digital twin. Source: Own representation

As you can see, a digital twin is not simply a virtual model like a CAD model or a software simulation; instead, it consists of different software components: The digital model, data analysis software, databases, communication interfaces for sensors, and graphical interfaces. The challenge here is the processing of the high data volumes of sensors and simulators as well as the configuration and mapping of sensors.

Particularly in the field of plant engineering and machine construction, game4automation can provide support with the visualization and simulation processes as well as the handling of sensor data in order to help you to build a graphically sophisticated and efficient digital twin.

Our digital twin services

We can provide the advice you need to decide whether a digital twin will bring about lasting added value for your product and company.

On the basis of this, we’ll also help you to construct your digital twin. This might be a twin for a new product/system or a twin for an existing product/system. Together, we’ll develop the architecture and we’ll provide you with support with the software-side implementation: From database management to data analysis with machine learning and the visualization of data.

Here’s what you can expect from us:

  • The design of digital twins in the context of new business models
  • The selection of the required technologies on the basis of the existing system landscape
  • The implementation of your concepts – using agile methods and constant dialog

Digital twins are only one dimension of digitalization. Business models and data analyses are other dimensions.

Find out more and keep things clear in your head – we’ll help you to do so!

Your direct contact


Jonas Grundler

Director New Business Development
Table of contents
Jonas Grundler Director New Business Development