Making factories more efficient and products more customer-oriented

Advancing through Digital Twins

Connecting real products with digital models.

This article appeared in the May 2017 Issue of Smart Engineering.

Digital twins are an intensely discussed topic in science and research. The connection between digital models and real-time data promises a variety of applications that can improve the efficiency of factories and the customer-orientation of products. Developments throughout the course of the 4th Industrial Revolution are bringing digital twins closer to the focus of businesses that want to profit from the actual benefits of theoretical concepts. Now processes and systems need to be prepared for the use of digital twins. This article presents the realizable benefits of digital twins and identifies implementation challenges and solutions.

The Concept of Digital Twins

The use of digital models in industrial companies has been a standard practice for a long time. These models are utilized in product development (CAD), digital assurance (DMU, simulation), production (CAM), visualization and marketing purposes (high quality renderings). The same applies to the planning of complex production facilities, visualization (design review) collision checks (collision detecting), and to determine performance and ideal operating conditions (material flow simulation). If both worlds are connected, applications such as the offline programming of robots and CNC systems are well-known. This list could be adapted, expanded and modified to fit specific industries and individual requirements. Digital models are ubiquitous in many companies. All of the applications described here have something in common – they are predominantly used before the specific use of the product or means of production, and at best, during intermediate phases, such as in changes to system configuration.

The concept of the digital twin goes beyond this understanding of the model and begins in the use phase of a product or a production plant. It consists of three essential components: the real product, the digital representation and the (new) connection between both of these worlds. By linking the actual status with the digital model, a virtual model of reality can be created that is not valid for an initial configuration, but can collect the actual status of a real product. Therefore, studies and tests can be conducted in the digital model and, for example, better configurations for the current situation of the real product can be determined through simulations. If a real product and its digital model are now linked together in both directions, these configurations and settings can be implemented in the real product, thereby allowing for targeted modifications and optimizations in the product characteristics. This optimization cycle is represented in Figure 1.

Figure 1: The Concept of the Digital Twin  

The optimization cycle in the digital twin can be described in the following steps:

  • The starting point is the use of the product.
  • Data about the product and usage is collected while the product is being used.
  • The collected data is stored either centralized or decentralized, i.e. in the cloud.
  • If the data is not stored in a central location, it must be transferred to the digital image.
  • In the digital world, data is evaluated and analyzed. In the simplest cases, usage data is linked directly to variables in digital models; in more complex cases, data must first be analyzed and adapted for use in the digital model.
  • In addition to the stored variables from real situations, product parameters are also varied. Only products with a degree of freedom enable optimizations. 
  • Future product behavior is predicted by changed parameters and settings by simulating product features with adapted parameters.
  • By evaluating the simulations, the degree of improvement can be determined and recommendations for action can be derived for the real product. Steps 5 through 7 are repeated until the desired optimizations are achieved.
  • The varied parameters are saved, just as the previously collected data.
  • If necessary, the parameters must be transferred to the real world.
  • After receiving information from the simulation, the parameters have to be configured in the real product.
  • The product is used with optimized parameters and settings.

The entire optimization cycle can be automated in many places and thus expedited. Today, data collection and data transfer can occur practically in real-time. Depending upon the application, the simulation and feedback of information can also be automated.

The use of digital twins is always aimed at improving the performance of real products – whether in the consumer market or in the B2B area, for example, production plants and entire production systems.


Digital Twins in the Consumer Market

For suppliers in the consumer market, the use of digital twins results in a series of new business models. As a result, product performance can be continuously improved, even after products have been purchased. Today, additional revenues are generated when software modifications or upgrades enable new functions for the customer after the purchase. For example, the e-mobility provider Tesla relies on software configurations in order to differentiate between the performance classes within a product line. If a digital twin is also used, optimizations and new functions can be offered based on individual use. This individual increase in product benefit can result after objective aspects (i.e. product performance), or after subjective aspects (individualization). Selling the digital twin itself is also conceivable.  The customer can make optimizations to the product independently and rely on the digital twin’s ability to make simulations and predictions.

The practical benefits for providers of digital twins in the consumer market include the following:  

  • Competitive advantages through continuous performance improvement of existing products
  • Product customization to increase customer benefits
  • Additional sales of the digital twin

Digital Twins in B2B

The use of digital twins in production plants also aims to constantly improve plant performance. For example, the process data of a milling machine is recorded throughout its entire use phase and transferred to the digital image. The data is used to simulate alternative production scenarios and production parameters. The parameters that have been simulated and optimized are transferred back to the plant and production can be continued with them. Optimized plant configurations can be determined to extend the service life of tools or for maximum productivity during the available production time until the next defined maintenance interval. In the future, these simulations can basically occur in real-time based on real data and enable continuous optimization. For example, a decrease in specific performance indicators could automatically trigger a simulation with changed parameters and lead to an optimization.

When considering an entire production network, performance limits and bottlenecks can be identified early on and countermeasures can be initiated by linking system data with the planned production program. Therefore, any optimizations that may be implemented with an APS system can incorporate additional criteria such as tool life, wear or energy consumption. Interesting business models for plant manufacturers also result from selling a machine’s production capacity as a service instead of the machine as hardware. The manufacturer then benefits from the increased performance and can transfer findings from a customized application to the applications of other customers.

An overview of the practical benefits of digital twins with a focus on production plants or production systems:

  • Continuously improved production efficiency
  • Automatic/ semi-automatic simulation and optimization of production parameters
  • Early recognition of performance limits or bottlenecks
  • Selling a product’s service, therefore profiting from continuously improved performance.

Requirements and Challenges

The requirements for the development of digital twins are highly dependent upon individual applications and can be evaluated from three different aspects (see Figure 2):

1. Performance of the Digital Model

The required performance of the digital model must be adapted to the intended purpose of the digital twin. All of the product features that have a significant influence on control variables and performance numbers must be mapped in the model. In addition, the digital image must be able to process the captured data of the real product. Variables, interfaces, types of storage and storage location have to be adapted to individual cases and, if necessary, changed or created in the existing models.

2. The Ability to Capture Data

The next requirement is the capability for collecting data and information in a real product. Relevant information and data are highly dependent on the application. In order to enable the acquisition of data, the necessary sensor technology must be provided in product development or retrofitted in a production plant.

3. The Ability to Process

Finally, the real product must also have the necessary control mechanisms to be able to implement the ascertained optimization. The optimized parameters must be received and processed by the product/plant and the configuration must be able to be adjusted.


Figure 2: Requirements for realizing the potential of digital twins

Realizing Potentials

The introduction of digital twins begins with defining the target state. The required performance of the digital image, the data/information acquisition capability and the necessary ability to process parameters and variables are determined (No. 4 in Figure 2). Next, the current position in the three dimensions is determined (No. 1). If the starting point and target state are known, the necessary development and expansion activities are transferred to a roadmap. Initially, it is conceivable to increase the performance of the digital image (from 1 to 2), e.g. by detailing digital models and anchoring variables in the model. Next, options for data acquisition can be created in reality (from 2 to 3). For example, force measuring sensors are provided or retrofitted in tool holders for this purpose. The last step is to create the options for processing the parameters (from 3 to 4). The first applications of the digital twin for consumer products and production goods are already real. As early as 2015, PTC introduced a mountain bike whose springs and shock absorber parameters can be electronically controlled. Real data from current or past rides are recorded and the spring behavior is optimized through simulations. Of course, this application also provides the driver the ability to configure via a smartphone app. In the area of B2B applications, General Electric demonstrated a wind farm that constantly self-optimizes based on current weather and wind data. Although the exact implementation details remain unknown, this shows that the topic of the digital twin is already being accepted in many ways.

The digitalization of the product presents users with a number of challenges: Large amounts of data must be recorded and processed within the shortest possible time, and the structure of the digital twin and the real product must be kept continuously up-to-date, and both models must be synchronized in the software configuration.


The concept of the digital twin offers great potential for industrial use. The prevalence of digital twins is increasing, and digitalization trends and the 4th Industrial Revolution are promoting feasibility. The continuous miniaturization of sensors and processors, for example, makes it possible to increase the availability of real-time status data; the networking of all production and living environments makes it possible to link digital twins with real products. Increasing computing power and the ubiquity of powerful processors (for example in controlling units, cars, smartphones, etc.) make it possible to perform the calculations required in the simulation at any time with high performance. All of these developments promote the feasibility and ultimately, the spread of digital twins. The opportunities that result from increased performance, adapted product behavior and finally, also from data generation are drivers for the spread of digital twins. It is high time to address this issue and make your own products and means of production fit for the future!


Your Contacts for Digital Twins

To the top