Understanding Hybrid Twins: A New Digital Frontier in Engineering Innovation
Elevate Your PLM Strategy with Hybrid Twins: Create Faster, Cheaper, Disruptive, and Scalable Innovations
In the world of engineering and technology, the concept of a "twin" is pivotal for predicting outcomes, optimizing designs, and ensuring efficient operations. While digital twins have been a staple in these endeavors, a more sophisticated and effective approach has emerged: the Hybrid Twin. But what exactly is a Hybrid Twin, and how does it differ from its digital counterpart? Let's delve into these questions and explore the exciting applications and advantages of this cutting-edge technology.
What is Meant by Hybrid Twin?
A Hybrid Twin is an advanced engineering model that combines the strengths of both numerical simulations and real-world experimental data to create a more accurate and reliable representation of complex systems. By combining both physics-based models and data-driven algorithms, engineers can simulate and predict real-world phenomena. Unlike digital twins, which primarily rely on data and often require vast amounts of it, Hybrid Twins incorporate fundamental physical laws and mechanics alongside with machine learning technologies into their models. This dual approach enhances the accuracy and reliability of simulations, making it possible to predict complex behaviors with fewer data points.
Here’s how it works:
- Numerical Simulations: Mathematical models simulate how a system behaves based on mathematical formulas. They are quick and cost-effective but not always 100% accurate because they rely on assumptions and simplifications about the real world.
- Experimental Data: This is real-world data collected from physical experiments. It’s highly accurate but can be time-consuming and expensive to gather.
What is the Difference Between Hybrid Twin and Digital Twin?
The primary difference between a Hybrid Twin and a Digital Twin lies in their underlying methodologies. In a hybrid twin, the key element that differentiates it from a traditional digital twin is the integration of machine learning models that are trained using real-world experimental data to correct the inaccuracies of numerical simulations.
Digital Twins focus on creating a virtual model based solely on data collected from sensors and other sources. While effective, this approach can be limited by the availability and quality of data.
In contrast, a hybrid twin goes a step further by:
- Recognizing and correcting errors in the simulation model: The hybrid twin acknowledges that the numerical simulations have an inherent error due to simplifications and assumptions. It uses real-world experimental data to identify and correct these errors.
- Using machine learning to fill gaps: The hybrid twin applies machine learning algorithms to learn from the experimental data and adjust the simulation model, effectively "filling in the gaps" between predicted and actual outcomes.
Therefore, the machine learning correction model that specifically addresses the ignorance or error in the numerical simulations is the part that is not typically present in a traditional digital twin. This makes the hybrid twin a more accurate and reliable tool for complex systems where pure simulations may fall short, allowing for more precise simulations, even with minimal data. This fusion of this so-called 'smart data' and physics not only bridges the gap between predictions and reality but also reduces the computational resources required, making it more efficient and sustainable.
Which Types and Digital Examples of Hybrid Twins Exist?
Hybrid Twins find applications across various industries, from automotive R&D to manufacturing. Here are two recent use cases from Keysight CAE customers:
- Hybrid Twins in vehicle development | Hybrid Twins https://www.esi-group.com/blog/understanding-hybrid-twins-a-new-digital-frontier-in-engineering-innovation?__hstc=181257784.b4233e71c271fe80e9630e45fddfe144.1757668918962.1763638112674.1765351637370.33&__hssc=181257784.5.1765351637370&__hsfp=2653731370are very powerful in crash simulation scenarios. By employing cutting-edge methods like sPGD and ReCUR for model order reduction, OEMs can optimize e.g. vehicle-side member reinforcement, significantly cutting down simulation time and costs. This democratization of the design process enables also non-specialist engineers to engage in high-level simulations.
- Hybrid Twins in vehicle manufacturing |Utilized in manufacturing engineering, Hybrid Twins can e.g. optimize laser brazing and weld quality control. By combining experimental and numerical data, this approach ensures the strength and durability of components, crucial for the manufacturing process.
Beyond these examples, Hybrid Twins play a critical role in elevating companies' Product Lifecycle Management (PLM) strategies. They facilitate faster, cheaper, disruptive, and scalable innovations. By enabling real-time simulations and predictive maintenance, Hybrid Twins streamline the product development cycle, reduce time-to-market, and lower costs. This agility allows companies to rapidly adapt to market changes and introduce groundbreaking products, maintaining a competitive edge. Moreover, the scalability of Hybrid Twins supports the transition from component-level analysis to system-of-systems thinking, crucial for industries like smart cities and industrial automation.
How do you Make a Hybrid Twin Model?
Hybrid Twins are powered by several key technologies. Hybrid AI combines physics-based models with data-driven approaches, enhancing the accuracy and explainability of simulations. Model order reduction techniques, such as those used in the Renault partnership, simplify complex models, making real-time simulations possible. Smart data from machine learning, a concept that focuses on using the right amount of data rather than overwhelming quantities, ensures that the Hybrid Twin remains efficient and effective.
In a Hybrid Twin, you start with numerical simulations to model the system. However, because these simulations aren’t perfect, you gather some experimental data to improve the model. This real-world data is used to train a machine learning model that identifies and corrects the gaps between the simulation predictions and actual outcomes. The hybrid twin then uses this corrected model to provide a more accurate and reliable prediction of the system’s behavior.
This method is particularly useful in complex fields, where traditional models might struggle to capture every detail accurately. By using both simulation data and real-world data, a hybrid twin offers the best of both worlds, leading to better predictions and decisions in engineering and manufacturing processes.
What are the Main Advantages of Hybrid Twins?
The advantages of Hybrid Twins are manifold:
- First, they offer a higher level of explainability, as the underlying mechanisms of the models are better understood due to the integration of known physical laws. This transparency is crucial for error detection and quality assurance.
- Second, the reduced reliance on extensive datasets translates to lower energy consumption, making it an environmentally friendly option.
- Finally, Hybrid Twins enable faster, more accurate simulations, which can lead to significant cost savings and innovation in product design and manufacturing.
Paving the Way to Unparalleled Accuracy, Efficiency, and Sustainability
In essence, Hybrid Twins represent the next evolution in engineering and simulation, offering unparalleled accuracy, efficiency, and sustainability. By integrating physics-based models and data-driven algorithms, they enhance PLM strategies, enabling companies to innovate faster, more cost-effectively, and at a larger scale. This capability is crucial for achieving disruptive advancements and managing complex systems in industries such as automotive, manufacturing, and beyond.