Predictive control for air cooling in steel galvanising

  • Internship
  • IJmuiden

Website TATA Steel

2 Internship/MSc Thesis projects at Tata Steel

Dear Systems & Control students. We would like to inform you of the possibility of doing an internship at TATA Steel (3-6 months), which can be combined with an MSc thesis.

TATA Steel is one of Europe’s largest steel producers, employing over 80,000 people worldwide. One of the main products made by their Ijmuiden facility is galvanised steel strip of typically 0.3 to 3 mm thickness.

Currently, two projects are available in cooperation with their R&D department on Thermal Processes. In both cases, you will be supervised by Dr. R. (Riccardo) Ferrari from DCSC, and by a researcher from TATA Steel.

1) Air knife model

The galvanising lines in steel industry use air knives to control the zinc layer thickness. The air knife pressure is dependent on strip speed, zinc layer thickness and distance to the strip. Because there is always some deviation between model and measurement, adaptation is needed. At this moment a constant speed is needed for the moment of adaptation to find the steady state error. The reason is the dead time between actuator and measurement: the strip travels several meters before it reaches the measurement. This is a pure delay. The line is controlled by the in-house developed OSCAR system using Non-linear Model Predictive Control (NMPC). When OSCAR is in control speed is transient to match the temperature requirements and maximize throughput. The constant speed for some seconds is not necessary when we can take into account the time delay between air knives and measurement. The pure delay can be modeled by adding states to the state space model but considering the low sampling time, this will take a lot of states. Another possible solution is the use of a smith predictor in the control loop correcting for the pure delay.
In this assignment, the aim is to develop a solution for this problem using the best approach for dealing with this pure time delay.

2) Predictive control for air cooling in steel galvanising

Tata Steel’s R&D department has developed a dedicated control program for the hot-dip galvanising lines, based on first principle process models and a model predictive control strategy. The program calculated target settings for the various heating and cooling sections of the production line, as well as the line speed, to ensure the strip is at a suitable temperature in every phase of the galvanising process.
One of the main challenges in this development is that few strip temperature measurements are available and that both measurements and models are subject to errors. To account for this, adaptation routines are included in the program.
For the air knife control, the current version the adaptation routine requires a constant speed to find the steady state error. This is a consequence of the distance between air knives and the temperature sensor, which translates to a pure time delay. In practice, the line speed varies and therefore a correction for the time delay in the control loop would constitute an important improvement to the control program.
For this project you will be improving a model predictive control algorithm, designed to calculate target settings for various heating and cooling sections of the production line for the galvanising process of this strip.

For further inquiries feel free to contact us or Dr. R. Ferrari at:

To apply for this job email your details to