Rareș Curatu is Industrial Automation & Machinery Industry manager at MathWorks.
What is the primary focus of software-defined automation (SDA)?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: Software-defined automation (SDA) increases industrial automation systems' flexibility, adaptability and intelligence, as described by concepts such as flexible manufacturing, autonomous automation and Industry 4.0. By decoupling automation logic from task-specific, fixed-configuration hardware, SDA allows production processes and workflows to be updated in software, with minimal manual intervention to reconfigure the manufacturing line. This enables rapid response to changing requirements, reduces downtime and improves traceability and production quality.
What are the primary benefits of software-defined automation?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: SDA creates the opportunity for machine builders to innovate, differentiate and deliver long-term value. SDA systems stand out ahead of “conventional” automaton systems.
SDA expands functionality with advanced features such as predictive maintenance, data analytics, anomaly detection, virtual sensors and high-precision control. For example, Coca-Cola developed a virtual pressure sensor with machine learning to improve beverage dispenser diagnostics.
It creates new business models: The software-centric approach supports value-added services like remote diagnostics, continuous software updates and data-driven maintenance contracts. For example, Aerzen Digital Systems used anomaly detection with server-based AI solutions to deploy models rapidly in a DevOps pipeline.
Better machines improve overall equipment effectiveness (OEE). Intelligent automation increases uptime, throughput and product quality by reducing manual interventions and enabling real-time optimization. For example, DHGE and DMG MORI developed AI applications for predicting machine tool failure risks.
SDA brings greater system functionality and more opportunities for machine builders to build better products and differentiate through software, features, services and quality.
How does software-defined automation figure in the convergence of IT and OT?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: SDA solution architectures can involve edge and cloud components, bringing IT and OT systems closer together. On the factory floor, high-performance industrial controllers run advanced algorithms and connect machines to plant and enterprise networks, blurring traditional IT/OT boundaries.
SDA solution architectures, high data volumes and high volumes and complexity of software demand an evolution of how systems and software are developed. SDA software-development practices are more similar to IT development practices than to traditional OT. Some of the changes include advanced analytics and AI, continuous improvement, digital twins and cybersecurity.
More measurement data creates an opportunity for data analytics and AI features. Latency-sensitive or real-time features are implemented on the edge, on industrial controllers. Algorithms that are not timing-critical can execute away from the factory floor, often using data features extracted on the edge.
If certain values, such as the temperature inside a motor, are hard to measure, AI with physics-based models can be used to develop virtual sensors that use a system model and other measurement data to infer a virtual sensor measurement.
If measurement data—for example, fault data or edge scenario data—is limited, physics-based models can be re-used for generating synthetic data. This data can be used to train AI models. For example, Mondi implemented statistics-based health monitoring and predictive maintenance for manufacturing processes with machine learning.
SDA systems don’t have to age the same way as traditional automation systems. Their flexibility and connectivity create the opportunity to develop perpetually upgradable machines.
Engineers perform model-in-the-loop, software-in-the-loop and hardware-in-the-loop tests to prototype, test and validate their designs before ever deploying code in production. Together with continuous integration/continuous deployment (CI/CD) programming practices, engineers can perform software updates that are well-tested and validated, addressing customer feedback and evolving requirements, to systems that are already in the field. For example, Krones built a reinforcement learning–based process control system that can be updated with AI models that are retrained remotely.
The data-rich SDA systems create an opportunity to deploy and operationalize digital twins. SDA often involves high, bi-directional data flow that can offer operators insights into processes. Reusing the models used for system development, machine builders can build live digital twins for real-time monitoring and optimization. For example, Atlas Copco minimizes cost of ownership using simulation and digital twins.
While not a new requirement for automation systems, SDA's software-rich and connected nature raises the bar for cybersecurity measures. Besides securing controllers and following network security best practices, cybersecurity must be built into the control software. For instance, robust control systems must detect and mitigate abnormal behavior and avoid chain reactions. This is an overview of how to deal with cybersecurity challenges in embedded system designs.
SDA brings IT and OT closer together, creating opportunities but also challenges and changes to how systems are engineered and software is developed. With the right tools and workflows, IT-like software development practices can be integrated into machine builders’ workflows and play a crucial role in developing the next generation of manufacturing equipment and machinery.
Which components will see the biggest impact from software-defined automation?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: Significant impact of SDA is being seen in controllers, including programmable logic controllers (PLCs), programmable automation controllers (PACs), industrial PCs (IPCs) and drives, components where advanced algorithms and software-driven features are increasingly integrated.
There is a growing trend toward the use of virtual PLCs, which are moving some edge functionality to the cloud. Among other benefits, using virtual PLCs allows factory managers to manage PLCs closer to how they manage IT assets, rather than OT assets. However, it is unlikely that all PLC functions will soon be moved to the cloud or away from the factory floor.
For the foreseeable future, high-speed, real-time and deterministic control algorithms will still be close to the factory floor on continuously higher-performance PLCs or even embedded on microcontrollers or field-programmable gate arrays (FPGAs) in the machines, gradually designing out traditional PLCs.
Get your subscription to Control Design’s daily newsletter.
In what ways does software-defined automation allow machine builders more flexibility in hardware selection and management?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: Simulation is a key enabler of flexibility in SDA. Engineers build physics-based machine models to size components and optimize the bill of materials before hardware commitment. The machine models are also used to test and prototype control algorithms in simulation. This reduces the need for physical prototypes and ensures designs meet performance and cost targets.
Simulation also benefits other roles, such as sales teams. Sales and pre-sales teams can use existing models to quickly evaluate configurations and demonstrate performance to customers, supporting tailored solutions and faster decisions.
With model-based design, algorithms are designed, tested and validated in simulation, using the machines’ physics-based models. The next step is deploying algorithms on controllers. Automatic code generation enables engineers to generate IEC 61131-3 or C/C++ code directly from simulation models and algorithms. This keeps intellectual property independent of specific PLC integrated development environments (IDEs) and provides flexibility in control hardware selection. This streamlines deployment and ensures code equivalency and consistency between simulation and real-world operation.
Simulation and code generation support machine builders from engineering to sales and deployment, empowering them to optimize designs, better serve customers and decouple their solutions from specific control hardware platforms. For example, Atlas Copco minimizes the cost of ownership using simulation and digital twins.
How can machine builders prepare for and leverage software-defined automation?
Rareș Curatu, manager, Industrial Automation & Machinery Industry, MathWorks: As software complexity grows, machine builders adopt model-based design. Model-based design puts models at the center of the engineering and development process. Engineers simulate the machinery, most often using physics-based models. With the help of the machinery models and simulation, engineers develop algorithms, such as control algorithms. In simulation, they can iterate the design; test and validate algorithms; and perform virtual commissioning.
Automatic code generation, such as IEC 61131-3 languages or C/C++, helps deploy validated algorithms directly to industrial controllers, decoupling software IP from specific PLCs and IDEs and supporting hardware flexibility.
Model-based design can shorten development cycles and reduce development time by 50% or more. For instance, Shibaura Mechatronics, when developing a new control algorithm, has reduced the time it took engineers to adjust parameters from half a day to 10 seconds.
With model-based design, teams can try out new ideas, explore the design-space and validate concepts faster; shorten development cycles, reduce the need for physical prototypes and reduce time-to-market; maintain a digital thread for traceability from requirements to implementation and test; enable agile, iterative development for higher-quality, feature-rich software; decouple the solution from specific control platforms or IDEs; reuse models as digital twins, for AI, analytics, anomaly detection and optimization; design resilient algorithms, for example, by simulating cyber attacks, as part of cybersecurity measures.
Model-based design, including code generation for industrial controllers, provides a robust foundation for managing complexity, accelerating innovation and delivering reliable, high-performance software-defined automation solutions.
With the right tools and development workflows, machine builders can capitalize on these opportunities without completely re-skilling their existing design and engineering departments.
