Attention members: billing update happening now. Learn more.

Digital Twins Calibrated with Operational Data Drive Efficiency

  • By Colm Gavin
  • Digital Transformation

By deploying a closed-loop digital twin throughout a system’s entire lifecycle, machine builders and end users can reduce costs and optimize efficiency.

As digital trends evolve throughout industrial automation, more individuals and organizations are realizing the benefits of modeling and simulation on their production systems. These practices can lower design and development costs and reduce time spent troubleshooting equipment in the field during commissioning. Perhaps less well known is an equally important benefit: simulation can also help continuously optimize operational efficiency, especially when production data is used to calibrate the model. This methodology is known as a closed-loop digital twin (CLDT).

A digital twin is a virtual representation of a physical asset. The CLDT expands on digital twins by using historical data to improve accuracy over time. Particularly in material handling and manufacturing, it is difficult to determine labor and machine utilization, and a CLDT can identify and provide insights and recommendations to improve the efficiency of these systems. CLDTs also provide benefits during the design and commissioning stages of a system’s lifecycle, but this article focuses on their usage during operation. While a system is in service, a CLDT helps users make informed decisions to adjust operations on the fly for improved efficiency.

Old models stagnate

In manufacturing and intralogistics environments, facility managers are faced with the difficult task of maintaining optimal key performance indicators (KPIs) despite daily and unplanned changes in employee numbers; large unexpected incoming or outgoing orders; and package-handling bottlenecks.

Facility models can help staff identify critical production points to achieve aggressive operational targets, but many of these models are rigid at best or inaccurate at worst. To maximize efficiency and productivity, staff need a model that can be tuned, but most facilities do not have the trained personnel or the time to manually perform these adjustments. Additionally, with so many control variables, it is difficult to know what to even simulate in a model. Models guided with artificial intelligence (AI) and machine learning (ML) can bring answers to these and other problems.

Insight-bearing digital twins

The digital twin methodology provides precise insight for optimizing parameters to meet and maintain KPIs. Expanding on this concept, a CLDT creates an accurate replica of the assets’ current states to forecast accuracy beyond that of a standard digital twin without feedback. CLDTs compare current state conditions with numerous adaptations to determine an optimal future state, aided by artificial intelligence and machine learning.

In this process of generating and evaluating potential future states, cloud computing power is immensely useful for simulation and data ingestion. In an ideal scenario, a machine builder or system integrator develops a CLDT to aid in the development process, and they then deliver it with their product, giving end users the ability to continue using it. As detailed below, CLDTs have benefits for machine builders and end users during the design, commissioning, and operation phases of the machine lifecycle.

Design phase

During the design phase, CLDTs empower machine builders and system integrators to persuasively demonstrate the effectiveness of their design to potential customers, equipped with statistics and visual interfaces to monitor virtual machine performance. These models and simulations can be used to convincingly predict the outputs end users are looking for, increasing confidence in designers’ proposals, and enabling them to win more bids.

By delivering a CLDT with their systems, machine builders can leverage the investment beyond the design phase, creating additional revenue streams through an ongoing services model. This gives them the ability to follow up and support products deployed in operation, and it extends and strengthens the relationship between machine builders and end users.

For end users, these digital twins increase profit margins by accurately forecasting capital and operating expenditures—along with throughput—providing a baseline from which to increase operational efficiency. They can do this through informed decision-making with the aid of a CLDT.

Commissioning phase

During development and commissioning, a digital twin combined with programmable logic control (PLC) and human-machine interface (HMI) simulation helps engineers spot bugs and inefficiencies in a machine before physical equipment and moving parts come into the picture. The addition of a digital twin to PLC and HMI simulation is a clear link between the automation programming logic and machine performance, complete with a virtual visualization of the machine. This setup makes it easy to find problems early on.

By identifying potential issues virtually and early in the commissioning phase, these tools limit unexpected and costly project changes. Additionally, greater virtual troubleshooting and testing abilities shorten required physical commissioning time, while reducing labor requirements and costs for integrators. 

Shorter and more effective commissioning benefits end users because systems start up and enter production on time, with reduced chances of costly change orders late in the commissioning phase. And unlike classic commissioning processes, hands-on digital twin plant simulation enables end users to begin operator training and equipment tuning programs well ahead of physical commissioning.

Operations phase

Throughout operations, a closed-loop system brings its greatest value with cloud collectors that ingest production data, providing continual fine tuning for optimal operational states. Simulation with a digital twin provides:

  • experimentation of multiple states, modeling hours of production and estimating results in a matter of seconds.
  • prediction of important KPIs—like throughput, utilization, and idle time.

Closing the loop and supplying a simulation with historical data greatly improves simulation accuracy.

The resulting CLDT enables fine tuning of operations by providing reports with optimal parameters, such as machine settings, manpower allocations, and shipping/receiving capacities, through a combination of simulations and AI/ML. Facility staff can set up automatic report generation at specific time frames—for example, before or during shifts, or in preparation for a daily staff meeting.

When human-based analysis is required to augment a decision-making process, easily understood model visualization provides insights into the way a facility operates. Visual simulation helps staff identify production bottlenecks and areas where excess resources are allocated. It enables staff to simulate multiple scenarios to answer situational questions—such as “What happens if there are fewer associates at a picking station?” or “What if too many robots are being sent to one area (e.g., picking) versus another (e.g., loading)?” CLDT software gives users the capability to adjust control variables and visualize their effects on operations (figure 1), quickly resolving these and other issues.

Figure 1. Users can adjust control parameters in software like Siemens Plant Simulation to visualize and determine effects on throughput and other KPIs.

Visual elements help users better understand the numbers and point out where changes need to be made. These software tools evaluate the best utilization of machines and labor—for instance, ensuring a warehouse has enough trucks available at the loading bay to handle outgoing shipments, but not too many, to avoid sending away unloaded trucks.

In situations involving a large number of parameters and theoretical combinations to experiment with, automated software helps eliminate redundant or impractical experiments by intelligently identifying those that are feasible. This can reduce thousands of combinations to tens or fewer, ultimately identifying the best set of parameters (figure 2).

Figure 2. Software helps eliminate examples that are not useful. Siemens Plant Simulation software—with the HEEDS design exploration and optimization engine—determines suitable variants and the best design.

Creating the calibrated CLDT

To calibrate the CLDT, the first step is connecting the digital twin with the automation equipment to feed data to the model. Edge devices are prime interfaces for data collection because they can preprocess machine data before sending it to the cloud for synchronization with the CLDT’s historical data-based optimization algorithm.

Users can build, operate, deploy, and maintain software solutions across multiple edge devices using managed edge apps. Numerous apps are available for analyzing machine data. Their ecosystem—including patch and version control—is managed through a central system deployed in the cloud or on premises.

Using industrial edge controllers reduces the number of devices connecting to machines on the plant floor. They also provide the means for an on-premises simulation solution, or advanced analysis in a cloud-based simulation.

With a simulation in the cloud on an open Internet of Things (IoT) cloud platform, users can map data from the plant floor to the digital twin model. This data is used to create insights for optimizing conditions and control variables on the production lines to maximize throughput and other KPIs. Some cloud platforms include a dedicated app for preparing and aggregating time series data into a simulation application (figure 3).

Figure 3. An open IoT cloud platform, such as Siemens Industrial Edge and Cloud native apps, empowers users to calibrate their digital twin with historical production data.
The model accuracy improves over time, as more data is collected and aligned with control variable inputs and predictions. In advanced AI- and ML-enabled configurations, the models can manipulate PLC parameters to improve operational efficiency, in addition to generating reports with suggested resource and asset reallocation. This type of calibrated CLDT is created as a turnkey system, specific to a machine or facility.


A manufacturing company recently added a high-volume production line at its facility, containing 75 machines, 25 pick-and-place robots, and conveyor belts to perform complex material handling operations. Coordinating the vast amount of equipment required careful planning to implement and optimize operations after commissioning. The project team developed a digital twin to aid design and commissioning efforts, and closed the loop once machines were installed, creating a calibrated CLDT (figure 4).

Figure 4. One manufacturing company optimized operations and predicted throughput with more than 99 percent accuracy using Siemens digital twin methodology with Plant Simulation and MindSphere applications.
Following calibration, the model predicted 105.26 jobs per hour for a certain shift. The actual shift yielded 105 jobs per hour, achieving model production accuracy of 99.75 percent.

CLDTs streamline operation

Digital twins are already widely accepted for identifying potential issues early in design and development, reducing error occurrence, and speeding up physical commissioning. Their potential value multiplies during operation. 

A calibrated CLDT reduces the time required for manually monitoring production data, and eliminates human guesswork involved in planning procedural changes and resource reallocation to increase efficiency. This translates to streamlined operation and higher profit margins, empowering manufacturers and intralogistics companies to become more competitive in demanding markets.

All figures courtesy of Siemens

Reader Feedback

We want to hear from you! Please send us your comments and questions about this topic to

Like This Article?

Subscribe Now!

About The Authors

Colm Gavin promotes digitalization topics with Siemens Digital Industries Software group for machine and line builders. Working for Siemens for 19 years, Gavin uses his experience in discrete manufacturing to assist companies in taking advantage of new innovations with Industry 4.0. He was previously responsible for the marketing of Siemens’ Totally Integrated Automation Portal software in the U.S. Gavin holds a BS in manufacturing engineering from Trinity College, Dublin, Ireland.