Study On Arsenic Flow In SKS Lead Smelting Process

Arsenic is a wide spread element in lead smelting process, which often flow out of the smelting process and harm natural environment and human body by forms of gas, liquid and solid wastes. As our nation’s lead smelting industry has long been facing difficulties in discovering the distribution and whereabouts of arsenic contaminants, the whole industry have not formed a technical system aiming at controlling and managing arsenic contaminants. This text was based on an investigation on arsenic flow features in a typical SKS lead smelting process, and built up computer models and system to simulate arsenic flow in SKS process. Thus, the arsenic flow in the typical SKS smelting process was continuously calculated and predicted, in order to discover the distribution and output feature of arsenic, and provide scientific basis of treating arsenic. The contents and results were as follows:(1) Arsenic content features were manually audited, which was based on on-site investigation, sampling and analysis of arsenic in input&output materials. The results demonstrated the wide distribution of arsenic in the investaged system. Among them, soot dust from the the blast furnace, crude zinc oxide from the fuming furnace, anode mud from the electrolysis, and matte from the flame furnace were main outputs of arsenic, which took up about60%to89%of the system’s arsenic.(2) Neural network models of arsenic in SKS smelting process were established, which simulated the amount of arsenic flow in soot dust, rude zinc oxide, anode mud, and matte. The models were tested by MSE, MAE, A, and R tests, with average results of0.0055,0.0578,1.565%, and0.4755, respectively. The results illustrated fine modeling qualities.(3) Linear, exponential and logarithmic regression models were built and put through model tests. The MSE, MAE, A, and R tests values of linear regression models were respectively0.0068,0.0865,1.3252%, and0.5770in average, which indicated better modeling quality than exponential and logarithmic regression models. Hence, linear regression models were selected to be the building method of regression models.(4) The modeling and prediction of arsenic flow were conducted by combining neural network models and regression models. The arsenic flow in soot dust, rude zinc oxide, anode mud, and matte were simulated by neural network models, while the arsenic flow in other materials were modeled by linear regressin models. The results showed that90%of the arsenic has entered the segment of blast furnace and fuming furnace, and48.1%of the arsenic has entered the segment of lead bullion refining, and28%of the arsenic has entered the segment of flame furnace. The model results were compared with manual auditing results, which were in good accordance. The results indicated that the models have fine modeling and predicting quality, which were suitable to deploy in automatic auditing arsenic flow in SKS lead smelting process.(5) An audit and diagnose system of arsenic flow in lead smelting process was constructed, and put into use of arsenic flow auditing and prediction. The system audited and predicted91.96%of the arsenic flow compared to the input amount. Through applying the system in automatic arsenic auditing and prediction of the lead smelting process, the system solved the difficulties of continuously auditing that manual auditing held.