Research Projects

Smart Sensoring and Predictive Maintenance in Steel Manufacturing (SUPREME)


Start: 01-02-2017
End: 30-06-2021


Tata Steel, Semiotic Labs, IJssel Technologie, M2i

This project is funded by Stichting Technologiewetenschappen STW


Smart Industry program


PhD1 (vacancy), PDEng (vacancy), Richard Loendersloot, Tiedo Tinga

PhD2 (vacancy), Nirvana Meratnia, Paul Havinga


Maintenance is vital in ensuring the availability, reliability and cost effectiveness of technical systems like Tata Steel’s production facilities. However, the degradation of systems is a dynamic process, governed by changes in both the system and its environment. To save on maintenance costs (replacements not too early) and increase systems availability (replacements not too late), the challenge is to achieve just-in-time maintenance. We therefore address the following research question: “How can advanced sensing technology and modelling of degradation and failure processes be used to develop a predictive maintenance concept for production systems?” The approach followed to answer this consists of four important steps.

Firstly the appropriate sensors must be selected to collect the relevant data (what, how and where to measure). Contrary to many other approaches we do not just deploy a large amount of sensors (big data approach), but aim to understand the critical failures of the system, assess the most relevant parameters and select the required sensor type, location and quality.

The second step will be to collect the required data in an efficient and flexible manner. While wireless communication in industrial environments is extremely challenging due to the high variability of the radio channel in terms of error rates, channel downtimes and error bursts caused by various equipment, one of the key challenges in developing a wireless sensor network is to introduce reliable and robust networking protocols that have real-time capability. This will be achieved by focusing on distributed scheduling for radio communication that enables autonomous operation by providing self-organization to adapt to changing requirements, available resources, and environmental conditions (e.g. interference levels). In addition, optimizing RF power and energy efficiency is of paramount interest to ensure reliable communication.

The third step is the development of models, based on the physics of failure, to predict the critical failures in the system due to e.g. wear, fatigue, corrosion or creep. The collected data on usage, loads and environmental conditions will be used as input for the models. As opposed to the common approach of data analytics, which heavily rely on training with historic data, a physical model-based approach has the advantage that not only previously encountered failures can be predicted.

The final step is to combine the data collection with the model development and demonstrate the integral concept on a real production facility. Tata Steel has offered the HIsarna steel production pilot plant as case study for this purpose. Once the prediction of component degradation has been realized, the information can easily be used for maintenance process optimization (i.e. planning / clustering) and solving logistic and resource planning challenges.