Measurement And Application Methods For Friction And Lubrication Behavior In Slab Continuous Casting Mold

Mold/slab friction behavior and lubrication behavior have a great impact on the slab surface quality. The real-time detection of the complex friction behaviour, the inspection of main factors and control methods for mold/slab friction effect, and reasonable regulation and control of lubrication behavior between mold and initial solidified shell, are of great significance to the reduction of slab surface defect and the improvement of slab quality.In the present work, based on a slab continuous caster equipped with hydraulic oscillators, combining with field experiment conditions and testing environment, the basic measuring principle and general idea of friction are described. With the respect to the casting process for conventional slab, wide and thick slab continuous casting, the main process parameters influencing on the friction such as the casting speed, casting temperature, slab section, and mold oscillation modes are tested and analyzed. The response of friction to the variation and change of mold level, submerged entry nozzle, slab width and steel grades are discussed. Besides, based on the detected results of friction during start-up, casting speed increasing and decreasing process, the fluctuation characteristics of friction during casting speed changing are also studied. Simultaneously, the detected results of friction under sinusoidal, non-sinusoidal oscillation mode and control model among casting speed, frequency and stroke are also investigated and analyzed.Basing on the measured mold friction data, the periodical variations and characteristics of mold friction are studied. The results of transient friction in time domain are analyzed under different parameters and different oscillation modes. In addition, the results before breakout are also illustrated. Using Fast Fourier Transformation analysis, mold friction in the time domain is transformed into the signal in frequency domain, the response characteristics of friction during sinusoidal oscillation and non-sinusoidal oscillation, changing casting speed, before breakout and changing ladle nozzle are discussed. Moreover, the fundamental frequency and the amplitudes of friction signal in the frequency domain are investigated with the help of frequency spectral analysis. The sensitivities of friction in frequency domain to the variation of the operational parameters and abnormal conditions are investigated.By collecting multiple groups of slag compositions and properties, the neural network training samples of slag composition-performance are established. With the help of neural network toolbox in MATLAB software, the prediction model of BP neural network is built, and the slag viscosity and melting temperature are predicted. Through training and testing network model, the number of hidden layer nodes and learning rate are reasonable modified and optimized. The predicted results are identical with the actual values, the errors between them are small, which indicate the feasibility of the neural network to model and predict the complicated nonlinear relationship between slag composition and performance.