Natural convection in the molten core of a killed steel ingot

Abstract

As we have shown earlier [1-3], the molten steel in the core of an ingot is intensely agitated for a very long t ime while it is hardening (for a 7 mg killed steel ingot the convection.t ime is 80 rain). As the killed steel hardens, the layers of liquid meta l at the crystallization front descend; here, in the lower portion of the still molten core of the ingot there is a gradual accumulat ion of stationary mel t and the boundary between the convective meta l and the stationary metal moves upward with t ime. We can list some causes responsible for the observed mixing of the mel t while the ingot is hardening 1) Agitation of the mel t as a result of the mechanica l effect of a jet of meta l during pouring. However, calculation shows that it is not possible to explain the very long convection t ime by the effect of a meta l jet during pouring. 2) Agitation of the mel t as a result of natural heat convection. Here the meta l is molten because it is above the liquidus temperature and the colder layers at the crystallization front descend, forcing the lighter me ta l at the center of the ingot upward [% 5]. Clearly, when heat is removed, natural convection must cease. According to Ivantsov [6] and the exper imental data of Tageev and Gulyaev [7] it takes 10 rain for the excess heat to be removed from the center of an ingot weighing 7 nag so that it is quite impossible to explain the very long ingot convection t ime in terms of this factor. 3) Some researchers [8] assume that the basic principle responsible for convection in a hardening ingot is the shrinkage of the meta l during hardening. To verify this hypothesis a bismuth ingot weighing 13 kg was cast in a steel cylindrical mold. After the ingot was cast, the radioactive isotope thal l ium-204 was added to the mirror. As is clear from the autoradiogtams obtained from this ingot (Fig. 1~ the thal l ium isotope, being entrained by the convective flow, penetrates deeply into the ingot although nothing can be said about shrinkage, in this case, the metal at the side crystallization front ascends. One exception is bismuth, which expands rather than shrinks during hardening 4. Agitation of the mel t as a result of the appearance of concentrated impurity streams during the process which, naturally, can ex plain the observed descending meta l current. 5. Melt convection as a result of the formation of a nucleating solid phase at the crystallization front and the subsequent descent of the enriched me l t along the crystallization front. We shall calculate the change in the volume of the circulating mel t with t ime for a prismatic killed ingot, assuming that in addition to heating the steel above the liquidus line convection is also caused by the presence of solid-phase particles in the boundary layer. To do this the entire hardening period is divided into two stages. In the first stage of the process when overheating is still present, the mel t current descends along the crystallization front because of natural convection. Here flow and heat exchange are turbulent. After e l iminat ion of overheating, thexe is a descending flow of meta l along the crystallization front in the second stage of the process; this is due to the enr ichment of the boundary with solid-phase nuclei . Metal flow in this stage is laminar . The mel t enriched by solid-phase nuclei accumulates in t he lower portion of the ingot; when the concentration of the solid phase passes the meta l creep boundary, mixing in this region of the ingot ceases. On the basis of the data of Nekhenedzi [9] and Khvorinov [10] we can assume that tile steel is still capable of flowing in the viscous state when up to about 20% of the solid phase is present. The thermal balance equation for the second stage of the process can be written as "~ 0 Scrdz = --(5~.'4f'cO)'{dVcr. (1)