2.5.2. Multivariate statistics in the sintering process

being the best results found when coke size is 0.25–3 mm [28]. The fraction 1–3.15 mm is more economical in terms of consumption [28–30]. The reduction in coke breeze consumption is being carried out mainly by replacing coke breeze by new renewable solid fuels as biochar,

Fe total (%) (min. 51, max. 61, typical > 56 [20]): iron ore market is mainly composed by hematite (Fe2O3), goethite (α-FeOOH) and magnetite (Fe3O4), with low impurities content (alkalis, sulfur and phosphorus) and iron average content of 60–65% in 2016 (around 40% in 1940) [2]. However, the depletion of rich iron ore mines will lead to exploitation of complex chemistry ores and low-grade iron ore mines, considering in some cases the bioprocessing [5].

Basicity index (typical 1.70 [20]): reduction degradation index (RDI) decreases when basicity increases [20]. Obviously, basicity index has importance in the hardness and reducibility of the sinter as expresses the relation CaO/SiO2. See [20] for a deeper explanation on sinter basicity

FeO content (%) (min. 4.0, max. 11 [20]): this parameter informs about the thermal state of the sintering process [20] and about coke rate. RDI is reduced when FeO content is increased [20, 34]. The problem is that increasing FeO content affects coke rate, deteriorates the sinter

Al2O3 content (%) (min. 0.6, max. 1.8, typical 1.35 [20]): sinter quality is related with Al2O3 content as when Al2O3 content is increased, RDI is higher. This point is under discussion as Kumar et al. observed that with 10–10.5% CaO an increase of 0.1% in Al2O3 increases RDI 2 points [34], while Hiesch observed no relation between RDI and Al2O3 [37]. Alumina is also related with reducibility, but the same as for RDI, there are different opinions, Yu et al. observed that maximum reducibility is reached when Al2O3 content is 2.5% in the sinter ore [38], while Umadevi et al. observed that reducibility increased when alumina content is raised

MgO content (%) (min. 0.7, max. 2.2, typical 1.65 [20]): MgO is related with the obtaining of an optimum blast furnace slag (flowability and desulphurization). For that reason, MgO is added as dolomite or dunite, directly or through the sinter. It was observed an increase in the utilization of dolomite (containing CaO and MgO) or other MgO-bearing materials with low SiO2 (as dunite) due to the use of iron ores with high SiO2 content. There are mixed opinions regarding how this constituent affects reducibility, RDI, Tumbler and productivity. Umadevi et al. observed that reducibility was decreased with increasing MgO addition (in low and high silica iron ore fines) [40]. Kalenga and Garbers-Craig observed an increase in Tumbler index and coke rate with increasing MgO content [41]. Bhagat observed that varying MgO content within certain limits (1.5–2.5%) did not affect productivity and Tumbler, while coke breeze consumption was reduced [42]. On their behalf, Umadevi et al. observed that increasing MgO content from 1.4 to 2.6% increased heat consumption, and productivity was reduced [43]. And Yadav et al. observed that raising MgO content from 1.75 to 3.25% caused a deterioration in productivity and an increase in Tumbler index [44]. With regard to how MgO content affects RDI, there are several opinions, and in this way, [41, 43, 45] observed that RDI improved when

reducibility [35, 36], and can also affect blast furnace productivity [35].

biomass, or charcoal [31–33].

68 Iron Ores and Iron Oxide Materials

index and sinter structure.

from 2 to 5.5% [39].

Data analysis is nowadays widely used with the purpose of correlating variables and, in this way, evaluating the effect of them in a certain process, whose have a higher influence, and how changing them will affect the sintering process. We have studied what variables are and what weight have them on three sinter quality indices as Tumbler Index, Reduction Degradation Index, and Reducibility Index. The objective of our work was to estimate them by using multivariate statistics, and thus avoiding long and expensive laboratory tests. Once each index was defined by a mathematical equation whose parameters are easily measured, we used fuzzy inference to develop a fuzzy control system that could warn in case of deviations from the conditions that ensured the best quality indices.

For our work, we used real data from a sintering machine. The variables with a higher weight for each quality index are: RDI (% < 3.15 mm): fines (%), fluxes (%), Fe2O3 (%), FeO (%), CaO (%), S (%), 10–20 mm (%), average particle size (mm), productivity (t/h), basicity index, and abrasion (% < 0.6 mm); Tumbler Index (% > 6.35 mm): Fe2O3 (%), CaO (%), MgO (%), MnO (%), P2O5 (%), 0–10 mm (%), 10–20 mm (%), basicity index, and maximum temperature in box prior to sinter discharge (�C); and, RI: Recoveries from other stages of the ironmaking and steelmaking process (%), FeO (%), P2O5 (%), MgO (%), S (%), Al2O3 (%) and >50 mm (%).

We elaborated a mathematical equation that correlated the RDI, the TI and the RI with those variables mentioned above, and in this way, it is possible to predict the indices by using these easily measured variables.
