SOFTWARE IMPLEMENTATION OF AN IMPEDANCE PI CONTROLLER FOR AN ELECTRIC ARC FURNACE CONTROL SYSTEM USING THE GO PROGRAMMING LANGUAGE

Abstract

Electrode control in electric arc furnace (EAF) operation represents one of the most technically demanding problems in steelmaking. The inherently unstable behaviour of the electric arc – its sensitivity to scrap composition, inter-electrode gap geometry, and thermal conditions – results in current and voltage fluctuations, increased specific energy consumption, and accelerated electrode wear. Despite a considerable body of research on mathematical modelling and controller design for EAFs, most published work relies on specialised software environments such as MATLAB/Simulink and does not address the direct deployment of control algorithms in industrial automation systems. The question of implementing control logic using general-purpose programming languages suitable for production environments remains insufficiently explored.

The objective of this work is to develop an impedance-based PI control algorithm for the EAF electrode positioning system and to implement it in the Go programming language, followed by analysis of the closed-loop system's dynamic behaviour.

The mathematical model is based on the Cassie-Mayr arc equations, which describe arc conductance dynamics under both transient and steady-state conditions. The controller is implemented in discrete time with a fixed integration step to ensure numerical stability. An anti-windup mechanism – clamping the accumulated error sum within defined bounds – is applied to prevent integrator saturation under control signal constraints. The software is structured as a modular Go application comprising dedicated components for control signal computation, simulation loop execution, and CSV-based result logging. The choice of Go is justified by its practical advantages for industrial automation: compilation to native executable code, absence of runtime dependencies, support for modular architecture, and high computational performance. Simulation results are visualised using Python/Matplotlib, ensuring reproducibility and clarity of the obtained data.

Numerical simulation results show that the developed system achieves a stable transient response with a settling time of 0.15-0.2 s and overshoot within acceptable limits. A residual steady-state error of 2-3% is attributed to simplifications in the adopted arc model and can be reduced by introducing adaptive gain correction. The smooth control signal profile, free from abrupt changes, confirms the algorithm's suitability for real-world application with respect to mechanical loading on the electrode drive. The findings demonstrate the viability of Go as an implementation language for control algorithms within digital EAF automation systems, as well as for the development of digital twins of electrotechnological processes.