In Zer0fect we provide innovative solutions tailored to each individual case. Our deep understanding of the technologies and new needs is our secret to provide sustainable solutions. Understanding in depth the definitions and methodologies and having the ability to step back and see the big picture allow us to create innovative solutions. “There is a solutions for almost every problem, we simply do not know it yet”, if you trust us we will find and provide a solution for your case!

ZWS academy is an interactive environment with multiple functionalities. More specifically, is a space where we share knowledge through our blog and at the same time, we can interact with you through the interactive questions and answers. In the interactive questions and answers section you can post a scientific question in areas around Zerofect domains, and we can have a discussion around it. The interactive questions and answers section is dedicated mainly for students, but also professionals are welcome to post a question. Also, in ZWS academy we share our research results.

ZDM is a holistic approach for ensuring both process and product quality by reducing defects through corrective, preventive, and predictive techniques, using mainly data-driven technologies and guaranteeing that no defective products leave the production site and reach the customer, aiming at higher manufacturing sustainability (source_1 & source_2). There are three main pairs of ZDM strategies, therefore, when designing the implementation of ZDM, it is clear what to do: if a defect is detected then it must be repaired if possible. In cases where repair is not possible, a preventative action should be taken to avoid future defects. The ultimate tool of ZDM is defect prediction: if a defect is predicted that will occur in the near future, then manufacturers can act before it has been generated and thus prevent it. Finally, it is recommended that manufacturers combine some of the three pairs to achieve higher efficiency and increase sustainability. To conclude, the ZDM approach as it is defined can be considered a user-friendly method because it relies on logical and understandable statements for planning the implementation (source_1).

ZDM has evolved from current preventive and corrective actions based on data collected from past production and defects, to the capacity to predict when defects will occur and to learn from them by analyzing both the past and the present and attempt to avoid them before even they created. To perform the comparison authors have used the results from a recent (2020) review article that analyses all the traditional QI methods (Psarommatis, Prouvost, et al. 2020). Traditional QI methods analyze only the past to improve the future, which in many cases is still a human-based activity. Therefore, there is a loss of potentially important information from the present. Not analyzing the present creates an inertia between the occurrence of an event and the identification of an improvement linked to it. In this sense, ZDM provides an additional predictive layer to traditional approaches and looks for faster, more reliable, and adaptive methods of ensuring defect-free products by looking at data from both the process and the product. In fact, LM, SS, Lean Six Sigma (L6S), the TOC, and TQM do not learn from defects but rather simply remove them, and neither do they utilize the full capacity of Industry 4.0 technologies. Traditional quality approaches are suitable for mass production since they focus on the improvement of processes from standardization. While standardization from every perspective is essential for QI (and thus for ZDM), ZDM looks for benefits with diversity/variability. The key point is to ensure that no defective product reach the customer while maintaining the ability to produce and inspect products using sustainable methods, even for highly customized products. Contrary to traditional approaches, ZDM reaches for the causality link between defects and its cause to achieve multilevel-control closed loops using past and real-time data; moreover, it must be integrated into the production process right from the beginning as opposed to trying to address the issues at a later stage (Psarommatis, May, et al. 2020). ZDM is the evolution of other methods. It includes some of the core concepts of traditional QI methods but the implementation is changing. The standard Six Sigma methodology embraces ZDM as one of its core concepts, defining it as allowing a maximum of 3.4 defects per million products, since achieving zero defects in a real context is practically impossible. To achieve this, the evolution of Industry 4.0-enabling, data-driven innovation leads to an easier and more efficient implementation of the ZDM concept, due to the availability of the required amount of data for techniques such as machine learning to work properly. ZDM is also interested in continuous improvement (to reach zero defects) and can also make use of continuous (quality) improvement (CI) tool such as DMAIC, FMEA, and SPC. To summarize ZDM in an holistic approach the utilizes the best practices of all the QI methods while it exploits the notion of defect prediction something that is not the case of the traditional QI. Source

ZDM can be implemented in two ways product oriented and process oriented. Each approach characterizes the starting point of the analysis. In the product-oriented approach to ZDM, the starting point is product quality, and the measured quality is checked against specifications. If the product quality meets the standards, the corresponding machine’s status is deemed healthy; if it does not, then the corresponding machine needs either maintenance and/or (re)calibration or tuning based on the defect data. By contrast, in the process-oriented approach, the health status of the machine is monitored; if it is good, then the produced product will be of good quality. In both cases, the implementation of ZDM is a closed-loop process that has its feedback loop either on the product or on the process quality, leading to maintenance and/or machine (re)calibration and/or tuning in case of low quality. The selection between product and process oriented ZDM is not an easy and straight forward process and therefore investigation is required for selecting the proper ZDM approach for a specific use case (source).

In general, near-zero-defect production is rare, and when achieved, it is for products that have been thoroughly studied, with must knowledge and time spent behind the scenes. To achieve this state using traditional QI requires years of observation and corrective actions from manufacturers, and most importantly that nothing is changed about the process or the product. For such products, ZDM might seems unnecessary, but what if the product changes? or what if the same process could happen more efficiently? Product life cycle has reduced significantly and customers require customized products; therefore, at some point the process or product will change and all the knowledge and improvements acquired over the years will not be applicable to the new situation. This is due to the fact that traditional QI methods do not learn from defects; they simply remove them from the production, ignoring the valuable information that can be acquired (Psarommatis, Prouvost, et al. 2020). The ZDM approach, on the other hand, uses advanced data-driven technologies and can thus learn from the defects or other types of failures and increase the resilience and flexibility of the production system. This means that in light of a change, the quality of the manufacturing process will not be affected and significant time will be saved from not applying the traditional QI methods to the new manufacturing circumstances. To summarize, manufacturers that have production stages with near zero defects can also benefit from the adoption of the ZDM approach. They will increase the resilience of their production, making it flexible to adapt to changes, thus reducing the time required to bring quality at the designed levels in the event of change. Furthermore, utilizing modern data driven technologies could increase their efficiency and sustainability significantly.

Real manufacturing environments are characterized by a high level of uncertainty, which makes it nearly impossible to reach actual zero-defects, meaning that no defects occur during the manufacturing process. One popular argument for ZDM not being feasible is the fact that variability constantly threatens the production system. Incoming raw material, for example, has an intrinsic complexity that brings variability and a probability of a defect appearing and propagating through the production line. The goal of ZDM is to reduce defects as much as possible, and it is compulsory that all products that leave the factory are defect-free. In other words, defects are kept in the factory, since it is virtually impossible to have incoming material, in-process material, and final products defect-free every time. This will have significant impact on the sustainability of the corresponding manufacturing system. This is due to the fact that if a defective part is sent to a customer, the customer will return the part and either ask for a new one or never order again, which will both impact the manufacturer. ZDM prevents this negative side effect by ensuring that no defective product leaves the factory. The usage of digital platforms can help to democratize ZDM through the anufacturing domain and assist companies in the adoption of the ZDM approach. Additionally, advanced data-driven technologies are emerging, revealing that implementations are reaching cost-effective values to be implemented in either SMEs or large enterprises (source).

To determine whether a specific lot or another amount is acceptable, 100% inspection is performed in manufacturing as part of “acceptance inspection.” It is more commonly deployed to satisfy legal or political requirements when the consequence of appearance of a defect is severe, either for safety reasons (e.g., the aviation industry) or contractual monetary penalties. Contrary to occasional misinterpretation, 100% does not refer to all quality characteristics of a part (or assembly) being inspected. Instead, it refers to a certain characteristic being 100% inspected on every manufactured part. Inspecting 100% of production is becoming key for smart factory to enable the early detection of defects and avoid its propagation by the proliferation of in-process inspection, also known as in-line inspection or quality control. The use of this level of inspection brings several disadvantages, such as a substantial cost increase, either by requiring additional personnel or measuring devices to perform the inspection, or by consuming time. Furthermore, 100% inspection is unreliable to implement when destructive inspection methods must be used, such as when legal equirements are imposed. However, the most significant disadvantage of 100% inspection is the fact that unless the inspection method is 100% effective, some defects will continue to appear since inspections errors can occur due to imperfect measurement systems. This is especially critical for human-based inspection (such as visual inspection), where the ability for defect detection can vary with tiredness, attention, laziness, or stress. This bias can be largely avoided with the use of faster, more accurate, and more reliable inspection systems, such as (but not exclusive to) machine vision and laser systems . Although several generic models for the cost of quality were already developed, it is still arguable if the optimum level for quality equals to zero-defects or not (source).

Traditional Quality Improvement (QI) methods utilize historical data in an attempt to improve the future without considering the present production status. This means that real-time data are not utilized, and therefore, critical information about a defect may be lost. This creates a significant amount of inertia between the time of the quality event detection and the time of assigning the mitigation action (Psarommatis, Prouvost, et al. 2020). This has as an effect of a delay in the response time in case of a quality issue as well as reduced accuracy of the QI method. ZDM resolves these issues because it utilizes both historical and real-time data, which are crucial for back-tracing the cause of a defect and learning from that event. ZDM can target the time of defect occurrence exactly and act immediately, whereas traditional QI cannot act immediately. Through utilizing both historical and real-time data, ZDM achieves quick response times as well as more accurate solutions with fewer resources, thus increasing the sustainability of the production system. Furthermore, because ZDM requires advanced data-driven approaches to implement, it takes advantage from another technology called a semantic framework. With the use of ontologies and other technologies, the semantic framework can enrich data with context and thus extract useful knowledge from existing data, which in other cases would remain unknown. This type of technology can boost the responsiveness and accuracy of ZDM even higher, something that traditional QI methods cannot take advantage of because they are not designed to be data-driven and utilize the Predict–Prevent pair to such an extent (source).