everis, the multinational consulting firm that is part of the NTT DATA Group, has positioned itself as a strategic technological partner of the Dutch subsidiary of SEKISUI CHEMICAL group, a market leader with over 70 years’ experience in glass interlayer technology. The company manufactures glass components for the automobile and architectural industries. On this occasion, it decided to start up—along with the consultancy—a predictive quality management platform, which will be part of a digital transformation process in the environment of the new digital manufacturing technology: Industry 4.0.
Both companies joined forces to create a functional platform that centralizes data, maximizing calculations, connectivity, analysis and intelligence, man-machine interactions and digital to physical conversions. The fruit of this collaboration was a differential model that represents a competitive advantage in the industrial sector to maximize revenues and profits.
SEKISUI S-LEC B.V. is a company that employs unique fine particle, adhesion and precise synthesis technologies to develop and supply high-performance intermediate materials for a wide range of fields, including electronics, mobility, construction materials and infrastructures. In this Industry 4.0 context, the manufacturer sought an effective method to predict product quality according to—for example—the status of the production line or the environment.
A three-phase project
Starting from a viability study and a minimum viable product (MVP) for a pilot production line, the goal of this project was to build a comprehensive and scalable prediction and management solution that was user friendly and cross-cutting. To this end, it employed the most advanced AI technologies on a data analysis platform that considered operating constraints.
For project development, everis therefore established three well-differentiated phases. In the first (exploration and viability), sensor data were analyzed to verify that they could be used to develop the future model. This is a common practice in industrial control engineering due to the wide range of sensors and procurement standards for sensor data. To this end, it analyzed the data obtained from the sensor (data quality, validity and consistency assessment) and conducted a viability study of process modelling (number of available variables, required precision…).
In the second phase to assess the viability of the model, its viability was evaluated and validated after development in offline mode (not connected directly to the sensor) in different everis test environments. Finally, during the deployment phase, the platform was implemented to view the predictions on a real-time digital dashboard for its daily use.
Precision, prediction, differential model
The outcome was over 200 sensor data examined to identify the most relevant attributes. The developed model has a prediction precision of 94%, representing a 50% improvement over the baseline. The process capability index (CPK), a measure of the ability to produce output within the specification limits, was increased a 34%. In addition, the use of automatic control techniques based on the model's predictions can boost the enhancement to 50%.
This focus translates into a differential model and represents a competitive advantage in the sector to maximize revenues and profits in today’s highly competitive and low-margin markets. Compliance with high quality standards can reduce both inhouse costs (problems associated with the product before delivery, such as scarcity, waste and delays) and external costs (arising after delivery due to withdrawals and warranty costs).
According to Martin van Neer, Senior process engineer from SEKISUI S-LEC B.V., “We are very happy with how the collaboration with the everis team went and we’re convinced that this will surely extend our close partnership in the future. On our side, the project was mainly to show a PoC of the technology and investigate the potential benefits. Considering the outcome of this project, the technology we’ve implemented has proven to be beneficial for us in our continuous search to increase the quality of our products.”
For his part, Miguel Angel Fuentes de la Fuente, Head of Industry 4.0 from everis, states that “from our viewpoint, initiatives related to Industry 4.0 require comprehensive solutions adapted to the specific needs and challenges of the industrial environment”. The executive believes that “these solutions must be based on three core areas: advanced complex analytics and algorithms for industrial processes, Big Data architectures and operational methodology. This is undoubtedly one of the mainstays of this important project’s success.”
everis and SEKISUI S-LEC B.V. will keep working together to evolve the solution with functionalities with other sophisticated features to improve prediction precision and adapt it to a wide variety of manufacturing process scenarios.