Adaptive Mold Level Control in a Continuous Steel Slab Casting Process

2015 
Surface quality of the steel slabs in a Continuous Casting (C.C.) process is mainly determined by the stability of the Mold Level Control (MLC). Excessive mold level fluctuations lead to an additional machine scarfing process which is applied to the slabs to avoid surface defects. A conventional Proportional-Integral-Derivative (PID) controller is not robust enough to deal with the disturbances caused by the hydraulic pressure hunting in a withdrawing system, submerged entry nozzle clogging, or unsteady bulging. Different types of mold level fluctuations can be observed online through the realization of Fast Fourier Transform (FFT) algorithms. An adaptive control logic was developed based on the calculations of the Accumulated Mold Level (Acc.ML) deviations and the results of mold level FFT, which enables choosing a phase-lead compensator, a phase-lag compensator, a sliding mode controller or a standing wave filter automatically in a MLC loop to restrain fluctuations. At China Steel (CSC), the value of Acc.ML deviations in a minute determines if a slab is going to be machined or not. Each of these control methods corresponds to a specific type of mold level fluctuation. Depending on the statistics, the phase-lead compensator is mostly triggered during the silicon steel casting, and the sliding mode controller is normally run in a peritectic steel casting. The phase-lag compensator lowers both the mold level fluctuating frequency and the Acc.ML deviations in most kinds of steel casting. A low-pass filter is adopted to eliminate the influence of the surface standing waves on the flow control actuator. These adaptive MLC methods were resident online in CSC #5 and #4 slab C.C. since 2011 and 2012 respectively, and also tested in #2 slab C.C. at Dragon Steel (DSC) in 2013. The stability is improved apparently after the accomplishment of the adaptive MLC in a continuous steel slab casting process.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []