openZDM: The open platform for zero-defect manufacturing

Manufacturers face the constant challenge of trying to produce high-quality products both efficiently and sustainably. Each of these aspects – sustainability and quality – is significantly affected by defects that arise during the manufacturing process. Many manufacturers use traditional quality control methods – physical examination, statistical process control, reactive maintenance, and equipment calibration – which in combination can be quite effective in reducing defects. However, they all have one characteristic in common: they address defects after they have happened.

In recent years, the advent of the Industry 4.0 concept has led to the supplementation of these quality control methods with digital technologies like AI, machine learning, and advanced data analytics. Beyond these advances in software, great progress has also been made in hardware such as intelligent sensors and embedded IoT systems. These advances have allowed a proactive quality control paradigm to emerge that emphasises preventing defects before they occur. This zero-defect, right-first-time approach is integral to modern manufacturing strategies for improving efficiency and reducing waste.

Adoption of the zero-defect manufacturing (ZDM) philosophy and innovative digital solutions will be vital in helping to realise the widespread shift towards greater sustainability in manufacturing. Nevertheless, such solutions, like high fidelity digital twins, data-driven analysis, and cutting-edge non-destructive inspection techniques, require specific domain knowledge, and their implementation can be prohibitive in both time and cost for manufacturers.

It is in this context that the openZDM project was established. Funded by the European Union under the Horizon Europe programme framework, openZDM will develop and demonstrate a platform designed to enable ZDM through data-driven quality control. The platform integrates advanced ICT solutions and innovative, non-destructive testing, thus setting the foundations for an integrated system that can be used by a large variety of industries.

The project will increase the flexibility of organisations and their capacity to innovate and continuously control their product quality while increasing production efficiency and competitiveness. This addresses the grand challenge of sustainable manufacturing: to deliver high-quality products while minimising waste and energy consumption.

“After the COVID-19 pandemic, many manufacturing companies were required to recalibrate and have faced the challenges of meeting modern customer demands for things such as personalised products while preserving high-quality production rates in their facilities,” says Dr Nikos Nikolakis, project coordinator of openZDM. “We want to show through our project that proactive quality control systems can help them in this respect.”

“We also want to show that by properly harnessing the enormous volume of data that is produced in modern manufacturing, which is often stored in isolated silos in different systems that are not properly connected or correlated, we can generate meaningful insights that can be used to target aspects of quality control.”

The project will provide an end-to-end proactive quality control system for defect identification that uses a variety of non-destructive inspection systems, deployed in-line, that collect data and measure different aspects of the production process depending on the defects that have been identified in the project pilots. The openZDM platform will ingest data coming from sources across the shop floor, which will then be further analysed by several software applications.

AI-driven software applications will identify defects and patterns to prevent them from occurring. A tool allowing the construction of digital twins will allow for feed-forward simulations which enable alternative scenarios to be simulated, analysed and evaluated, after which recommendations can be given for direct process control, avoiding future error recurrence.

The openZDM project will test and finalise its technological tools in five industrial pilots from different sectors in real-life operational conditions. VDL Weweler in the Netherlands manufactures steel parts, Volkswagen Autoeuropa in Portugal manufactures and assembles passenger vehicles, Aptiv in Italy assembles battery modules for electric vehicles, Vidrala in Spain produces glass bottles, and Sonae Arauco in Spain manufactures decorative wooden panels.

“These pilots have been selected because together they present a wide spectrum of manufacturing setups, with differing levels of digitalisation and complexity which will allow us to extensively test our system and instil it with extra robustness and performance,” explains Dr Nikolakis.

“Furthermore, the defects that arise in these pilots are all of a different nature, which will require different inspection systems and sensors; therefore different methods of analysis.”

The project has already completed its requirements collection phase, in which initial data from the pilot sites is collected to provide a baseline for future comparison and validation of the platform. The reference architecture of the overall system has also been designed. OpenZDM is now in the process of designing, refining and implementing its set of solutions, including non-destructive inspection systems, a unified data model for the entire system, a digital twin tool that allows users to create digital twins from scratch or load existing ones, and several data-driven quality assessment modules that use AI methods to find patterns and generate insights about the production status.

“One of the really exciting aspects of openZDM is what we call the “what-if” analysis tool, which uses digital twin technology to allow us to examine alternative future scenarios and production settings with different objectives in mind,” Dr Nikolakis states. “The user can feed in their specific objectives – minimising material waste, minimising cycle time, or reducing its environmental footprint, for instance – and then the tool can evaluate different possibilities and provide recommendations for how the manufacturing system, based on its digital twin can be adjusted to meet these objectives.”

“Of course, given this is based on AI methods, there is a level of uncertainty which means that final decisions are currently still left to human operators, but we envisage future scenarios whereby this tool will be able to proactively and directly control manufacturing assets to improve their performance.”

The European Commission has in recent years sought to drive what it calls “circular economy” strategies, which aim to reduce waste by improving coordination between stakeholders all the way across the supply and value chain. “In order to achieve these kinds of goals, data exchange is needed between these different entities,” explains Dr Nikolakis. “With all of the data being collected in our project, we have seen an opportunity to contribute to such circular economy strategies by implementing an IDS – industrial data space – connector, which allows companies to upload and share data with one another. Additionally, the Asset Administration Shell (AAS) technology allows for interoperability and seameless data exchange across manufacturing entities.”

“For instance, a manufacturing company using a robot could share its data with the robot manufacturer, allowing them to see how the robot behaves in different scenarios and conditions, ultimately allowing them to improve the robot as a product. This data sharing provides a win-win scenario for both the robot manufacturer and the user.”

The full pipeline of openZDM has now been fully realised in a proof of concept, so the remaining years of the project will be spent increasing the maturity of the separate components, ensuring that by the end it is well-positioned for commercialisation, enabling industry adoption.

“More and more companies are investing in digital solutions, but at the moment these efforts are isolated,” says Dr Nikolakis. “Through the openZDM system, we are showing that by fusing these together, analysing them and generating meaningful insights, we can have an impact not only for companies in the manufacturing industries but also in other sectors.”