Thomas Lenne¹, Guy Druart¹, Marcello Bentivegni^1, ⋆
⋆ : marcello.bentivegni@vallourec.com
¹ Vallourec Research Center France
Mots clés : thermal simulation, furnace piloting, level 2 models, piloting algorithm
Résumé :

In the premium steelmaking industry, the heat treatment of steel products is a key process to obtain the good final physical properties and in a correct range of values. Over the years, Vallourec has known a paradigm change in his production organization linked to the carbon steel market evolution. Reduced-order sizes, simultaneous heating of different products and frequent stoppages made the production more discontinuous than in the past and increase the critical quality-related phases. The necessary changes for furnace management requires now improved flexibility and dynamism to ensure the heating quality of the products but bring opportunities for energy savings. One of Vallourec’s Industry 4.0 targets is a complete management and control of the production. Furnace level 2 modelling is one of the keystones for process data generation: in a heating process we know the charging and discharging state by sensors, what is unknown is the physical evolution in between. The Vallourec L2 model fulfils this lack.

The mastering of the processes requires the best comprehension of the temperature field’s evolution and the correlated mechanism. All thermal processes are submitted to the same partial-differential equation, the well-known heat equation. Between the heating, the natural cooling or the quenching of a product, the only mathematical difference comes from the applied boundary conditions which are related to the considered thermal flows on the limits (radiation, convection, conduction). This equation can be analytically solved only in some particular cases which are not relevant for industrial purposes. Some dedicated commercial software can solve the heat equation in a very precise way, but they are not adapted to the specific needs of real-time computation and aren’t consistent with the measurement precision in the plants. The necessity of a model dedicated to the thermal physic, to the specific geometries and the furnace configurations is based on these plant-oriented requirements. The choice of the numerical methods has been based on criteria such as calculation speed, maintainability, the ability to improve and precision. The models are then tested on temperature uniformity survey trials as a validation gauge.

Most Level 2 models for furnaces are comparing the real-time simulation results with ideal “static” heating curves to adapt the piloting of the furnace. This method showed great results and robustness over the years but showed a lack of flexibility when a deviation from the ideal production situation occurs. The new generation of Vallourec Level 2 models for billets reheating furnaces, uses another principle and removes the ideal heating curve as an ideal case that rarely happens. Based on the real-time simulation results and the on-line production data, the computed forecast of the future discharging state of the products allows the software to provide a dynamic prediction of the heating quality with irregular production parameters. For sure these results are obtained with an higher computational cost that were not affordable in the past but are acceptable today with the most recent evolution of information technology, as for example parallelization of the simulations.

doi : https://doi.org/10.25855/SFT2022-032

PDF : download