Digital Twins for Improved Manufacturing
This project has received funding from the European Union‘s Horizon 2020 Research and Innovation Program under grant agreement No. 958357 and it is an initiative of the Factories-of-the-Future (FoF) Public Private Partnership.
The aim of the project InterQ is optimizing the quality of industrial manufacturing with the improved use of data. To reliably optimize manufacturing processes, virtual and physical sensor data needs to be gathered. This data is subsequently used to measure, predict, and control the quality of manufactured products, the manufacturing process and to evaluate the quality of the gathered data itself. The developments of the project will be applied to high-added value industrial applications such as windmill gearboxes, aero engines, and electrical cars. Within the project, ModuleWorks is working on a digital twin which can predict and show the part quality based on process simulations.
Data silos are a central issue of today’s manufacturing industry. A data silo occurs if information about a system is stored in isolation so that data cannot be shared with other related systems. In this way, data is not adequately distributed but instead divided across different systems. An increase of system information would theoretically enable data-based insights to improve manufacturing. Unfortunately, current manufacturing systems store obtained data mostly in an isolated way, creating a data silo.
Data silos are not only a technologically undesirable state but also an economical threat. As technology is advancing and manufacturing is improving, economic success will be largely dependent on high quality production and this, in turn, is closely related to the capability of maximizing the benefits of the digital revolution with the ensuing increase of data to assure this quality.
To address both the technological and the economical aspect of the data silo problem, several data-related obstacles must be overcome. Since large amounts of data with little information are transferred, valuable information on the product quality is lacking. Furthermore, data might be unreliable. From a practical perspective, product quality optimization in production is reactive as alarms about product quality are only raised after a part is produced. This also means that quality faults are only noticed after production and not related to their cause during the process events. Additionally, exchange between supply chain actors does not take place. Lastly, a lack of product traceability prevents following the life cycle of a product during manufacturing, usage and maintenance phases.
To address the previously mentioned problems, the project InterQ aims to overcome data-related obstacles to optimal manufacturing with the use of Digital Technologies and Artificial Intelligence and to apply them to high-added value industrial applications such as windmill gearboxes, aero engines, and electrical cars. This involves digital twins, virtual sensors, decision support systems, data fusion and a distributed ledger. These technologies are subsequently used improve the quality of manufactured products by linking the process, product and data quality across the supply chain.
To tackle the challenges of the project, the Process, Product and Data Quality Hallmark (shortened to PPD hallmark) has been put forward. Similarly, the developments of the project are spread across five modules concerning process, product, data, zero defect manufacturing and a trusted framework.
The PPD hallmark ensures the quality of each operation. To achieve improved product quality, physical and virtual sensors will be recording data about manufacturing conditions and information about machine states will be gathered. Product quality will be approached by in-process sensors and traditional metrology measures to measure the quality of the final part. Data quality will be evaluated by syntactic, semantic and pragmatic aspects which will enable certifying the reliability of the process and product data.
The modules provide a schematic view of the developments concerning the process, product and data quality with the improvement, resulting in zero defect manufacturing and data exchange in a secure environment via the trusted framework.
The process module monitors the manufacturing with new physical and virtual sensors which can measure close to the cutting point and estimate critical process characteristics providing more meaningful data. Possible deviations are detected and avoided by a fingerprint of the machine or the process during normal production. Machine operator knowledge is incorporated to increase automatic data processing.
The product module replaces manual quality inspections with new sensors and processing techniques to provide reliable data. Two digital twins will be implemented: one metrology-based digital twin to predict global production quality and one prediction digital twin giving an estimation on the product quality after each production step by considering process variables and measurements.
The data module is a result of the previous modules on the process and the product. In current manufacturing, data can be corrupted or inconsistent for example due to malfunctioning sensors, unproper signal processing or communication problems with data gathering elements. But since data is the basis for the optimization of algorithms and processes, reliable data is of the utmost importance. For InterQ, data is evaluated on two validation layers. The first layer checks the ontological and temporal consistency of data, while the second uses historical and statistical analyses to detect unusual values.
The ZeroDefect module is again a result of the previous modules which provide reliable data of processes and products. This enables assessing the product quality virtually without relying solely on measurements. Additionally, artificial intelligence production optimizations are used to improve the process by controlling geometrical deviations, surface finish and surface integrity of products.
All modules are embedded in the TrustedFramework which allows safe data exchange using cryptographic protocols and trusted data sharing mechanisms. In this way, a common hurdle of current manufacturing processes is overcome. Traditionally, that data on one production plant and one specific process is collected, neglecting the sequence and diversity of manufacturing processes. This problem is reinforced by a lack of trust between different actors among the supply chain as they do not trust each other enough to disclose their data. The InterQ TrustedFramework faces this problem from a new paradigm that is based on the traceability, integrity and trust of the gathered data to finally provide an environment for overcoming the data silo.
ModuleWorks’ main contribution to the project is building a digital twin which will be able to predict and show the part quality from the process simulations. Beside this, ModuleWorks is partially involved in the work packages on the process, product, and data. The overall result of the project are new solutions to predict the final quality of the product, a holistic approach manufacturing and the surrounding data by using digital twins.