Ph. D. Project
Robust control of an electric furnace to optimize energy consumption
2023/10/01 - 2026/09/30

The RIVA Group company manages an electric steelworks in Neuves Maisons called Société des Aciers d'armature pour le concrete (SAM). Backed by
its environmental commitment and faced with rising energy prices, electricity management has become a major challenge for RIVA Group. In the steel
industry, this increase has a direct impact on production given the high need for electric furnaces in energy (several MWh). Good regulation of the
electrode system is crucial in order to reduce this cost, the electric furnace needing a lot of power in order to be able to raise the temperature of the steel
to its melting point (technology known as "electric arc "). With this in mind, it is essential to modify and adapt existing systems in order to meet new
economic and energy challenges, and this requires efficient regulation of the electrodes.

The purpose of this thesis subject is therefore to propose optimization strategies for the management of the energy consumption of an electric furnace in a
steelworks by mobilizing the theories and tools of automatic control.

Initially, it is necessary to finely model the installation and to identify the actuators and sensors currently accessible in order to take stock of the available
control means and those which would possibly be interesting to have. In a second step, it will be necessary, based on the modeling carried out, to
reconstruct the unmeasured variables whose knowledge is relevant to implement control laws with a view to optimizing energy consumption via the
synthesis of observers of 'state. In a third step, control laws based on these observers will be developed in order to regulate the temperature of the electric
furnace while minimizing energy consumption. Finally, in a fourth step, it will be necessary to validate and test the proposed control laws.

Research project

In this doctoral research work, we therefore consider the problem of controlling a complex system under the constraint of optimizing the energy
consumed, this complex system being an electric furnace of the RIVA Group located at the Société des Aciers reinforcement for concrete at Neuves

In view of the motivations set out above, the proposed research project is divided into four parts, each of which includes a study of the literature and the
state of the art.

1) Modeling of the installation.
• Based on the laws of physics and the literature, the goal is to design a nonlinear model with Matlab/Simulink that is both simple enough to be
implemented efficiently and detailed enough to faithfully reproduce the dynamic behavior of installation.
• The design of this model will be based on the inventory of existing actuators and sensors and will make it possible to take stock of the available control
means. Proposals can thus be made on the actuators and sensors which it would be interesting to have available.
• The proposed nonlinear model will make it possible to develop linear models at relevant operating points for process control.

2) Estimation of unmeasured variables.
• Following the modeling mentioned above, it will be necessary to determine whether the knowledge of unmeasured variables is relevant to act on the
control of the energy consumed.
• Using techniques based on Kalman-Bucy filtering, linear and nonlinear observers will be synthesized to estimate these unmeasured variables.

3) Synthesis of control laws to reduce the energy consumed.
• The proposed control laws may be based on one of the following optimization criteria: H2, H-infinity or L1. These control laws will be based on the
observers mentioned above.
• The choice of control laws that will be relevant for this research project is an open question. Several options can be studied depending on the progress
of the work: gain sequencing or not, PID type structure or no structure, linear or non-linear, etc.
• The robustness of the proposed control laws will be studied to take into account the inevitable uncertainties on the values of the parameters of the

4) Validation of the proposed control laws.
• This part concerns the implementation of the control laws proposed on the nonlinear model of the installation with Matlab/Simulink.
• The robustness and performance of these control laws will be evaluated and the energy gains obtained in simulation will be quantified.
Control Identification Diagnosis