Statistical Variability of the Measurable Objects That Appear during Continuous Casting of Steel

Samples of the liquid metal temperatures during continuous casting of steel are analyzed. The samples are formed from the liquid metal temperatures measured by a thermocouple in a steel-teeming ladle and the data obtained by five consecutive measurements of the liquid metal temperature in a tundish. The measurements in the steel-teeming ladle are made after the stage of steel degassing before a metal is fed for continuous casting. The temperature measurements in the tundish are carried out throughout the entire casting period, which includes all heats assigned to a series. A hypothesis about a normal distribution of a set of the metal temperatures in both the steel-teeming ladle and the tundish is advanced. This hypotheses is verified using the following three tests of goodness of fit: the Pearson χ 2 test, the Kolmogorov λ test, and the Shapiro–Wilk W test. The hypothesis of a normal temperature distribution in the steel-teeming ladle is shown to be consistent with the experimental data for all three tests of goodness of fit and, hence, to be accepted. The hypothesis of a normal temperature distribution during measurements in the tundish should be rejected as not consistent with the experimental data, and all three tests of goodness of fit show the need to reject this hypothesis. The main factors that lead to a deviation of the temperature distribution from a normal law are as follows: an increased cooling rate of the metal in the tundish and the introduction of slag-forming and heat-insulating mixtures onto the metal surface. When the number of temperature measurements in the tundish increases, the absolute statistics of the goodness-of-fit tests are above (Pearson and Kolmogorov goodness-of-fit tests) and below (Shapiro–Wilk test) a critical boundary, and the temperature distribution in the tundish tends toward a normal distribution.