Development, practical verification and implementation of stochastic models of electric arc phenomenon
About the project
Project manager: Maciej Klimas, MSc
Project supervisor: Dariusz Grabowski, PhD, DSc
Source of financing: Ministry of Science and Education as part of the “Diamond Grant” programme, VIII edition
Grant amount: 220 000 PLN
Project timeline: November 2019 – May 2023
Project number: 0139/DIA/2019/48
The project concerns an issue of modelling of the electric arc phenomenon, occurring in electric arc furnaces, in particular. Electric arc furnaces are widely used in the steelmaking industry. Unfortunately, the nonlinear and stochastic character of such a load can result in many problems in the power system, such as, among others, voltage sags, swells, and harmonics. Due to the high share of electric arc furnaces in worldwide steel production and recycling, engineers need to conduct simulations of the power system taking into account the phenomena related to the arc itself. It is necessary for those calculations to be as reliable as possible in order to apply effective mitigation means to the aforementioned problems. Modeling the electric arc furnace is also essential during the design phase of a new facility in order to estimate the influence of such loads on the power quality. This project is dedicated to the issue of modelling of the electric arc phenomenon, primarily including its stochastic nature.
The scope of the project includes electrical measurements conducted in both industrial facilities and in a laboratory setting. Laboratory measurements are intended to be conducted on a laboratory stand designed and constructed for examination of the electric arc phenomenon. Such construction will allow for reflecting the processes occurring in electric arc furnaces, but in a safe scale. Suitable equipment will also provide methods for examination of the electrical parameters of the phenomena that take place inside of the furnace. The data gathered during the measurements will serve as the base for arc modelling with several different theoretical approaches. The first approach will include stochastic differential equations. Next, a black-box model will be constructed. It will use artificial neural networks that have been trained to reflect the shape of the input and output data waveforms. The third model will be built using the Hammerstein-Wiener model fitted with a fractional differential equation in its dynamic part. The last proposed approach will apply chaotic systems to reflect changes in furnace characteristic. All the developed models will then be compared to assess their agreement with reality and usefulness in transient simulations of the power system.
The final results of the project research will consist of electric arc models implemented in Matlab software, along with comparative analysis and measures reflecting their similarity to the real electric arc phenomenon occurring in the electric arc furnaces. In addition, at least one of the best models will be implemented in the EMTP-ATP software for transient simulations of electrical circuits. EMTP-ATP is a software widely used by electrical engineers around the world and due to that fact the models developed in the project would be beneficial for the simulations considering circuits with nonlinear and stochastic loads such as electric arc furnaces.
International collaboration
As part of work dedicated to development of stochastic models of the electric arc furnace, we have established a collaboration with the team led by Professor Starkloff from Institute of Stochastics, Faculty of Mathematics and Computer Science at Technische Universität Bergakademie Freiberg (https://tu-freiberg.de/fakult1/sto). Joint work includes development and analysis of stochastic models in which the deterministic part is extended with random component allowing for representation of stochastic changes in the furnace characteristic.
Laboratory
One of the project objectives concerns the construction of a laboratory station, designed for the study of the electric arc phenomenon.
Ultimately, the laboratory station is intended to be a source of measurement data which would include recordings of not only voltage and current waveforms but also a corresponding image of an arc column obtained with a fast industrial camera. The modular construction of the laboratory station allows testing of the electric arc occurring on both single-phase and there-phase furnaces.
Publications
-
M. Klimas, D. Grabowski, “Identification of nonstationary parameters of electric arc furnace model using Monte Carlo approach,” 2020 Progress in Applied Electrical Engineering (PAEE), 2020, pp. 1-6, doi: 10.1109/PAEE50669.2020.9158732.
-
M. Klimas, D. Grabowski, “Application of shallow neural networks in electric arc furnace modelling,” 2021 IEEE International Conference on Environment and Electrical Engineering and 2021 IEEE Industrial and Commercial Power Systems Europe (EEEIC / I&CPS Europe), 2021, pp. 1-6, doi: 10.1109/EEEIC/ICPSEurope51590.2021.9584512.
-
M. Klimas, D. Grabowski, “Application of Long Short-Term Memory Neural Networks for Electric Arc Furnace Modelling,” Intelligent Data Engineering and Automated Learning – IDEAL 2021, 2021, Lecture Notes in Computer Science, vol 13113. Springer, Cham., pp 166-175, doi: 10.1007/978-3-030-91608-4_17
-
M. Dietz, D. Grabowski, M. Klimas, H. -J. Starkloff, “Estimation and analysis of the electric arc furnace model coefficients,” in IEEE Transactions on Power Delivery, doi: 10.1109/TPWRD.2022.3163815.
-
M. Klimas, D. Grabowski, “Application of shallow neural networks in electric arc furnace modeling,” in IEEE Transactions on Industry Applications, doi: 10.1109/TIA.2022.3180004.
