Modified exponential model for carbon prediction in the end blowing stage of basic oxygen furnace converter

Several common models for carbon prediction were discussed based on an off-gas analysis of the basic oxygen furnace (BOF) process, and the basic principles, advantages and disadvantages of three exponential decay models with different correction algorithm were analyzed respectively. An improved exponential model of “critical carbon content fitting + update curves simultaneously” algorithm was established by combining the advantages of previous algorithms. Firstly, the historical decarburization curve in the end blowing stage and the critical carbon content in the bath were obtained by exponentially fitting the decarburization data of historical heats. Secondly, the reference decarburization curve was obtained by replacing the corresponding parameter of the historical decarburization curve with the maximum specific decarburization rate in the middle blowing stage of the real-time heat. Subsequently, the specific decarburization rates of the historical decarburization curve and the reference decarburization curve were converted to dimensionless values within the range of 0 to 1 by normalizing. Then, a multi-point correction method was used to correct the calculation results of the carbon content in the bath and repeatedly modify the key parameters of the calculated decarburization curve, according to the deviation of the removed carbon amount between the normalized reference curve and the actual process. The key parameters of the calculated and the reference decarburization curves were updated simultaneously after each calculation step, and the calculation error of the carbon content prediction decreased gradually through iterative calculation. Finally, the carbon content in the bath were precisely predicted in the end blowing stage of the BOF process. Plant trials were carried out in a BOF converter to demonstrate the performance of the proposed models. The results show that the new model exhibits better adaptability and higher accuracy than the other ones. The hit ratio of the new model to predict the end-point carbon content reaches 90% within a tolerance of ±0.02%.