During the past two decades many efforts have been made to increase the control of blast furnace conditions to ensure a homogeneous product. The dissections on blast furnaces by various iron and steel companies in Japan in the early 70s provided valuable information on the high temperature properties of charging material. Standard tests (ISO) to determine ore, sinter and pellet qualities only provide information of up to 1100°C . By using the REAS apparatus - a high temperature reduction vessel that simulates the blast furnace process from stockline to melting - the high temperature properties of burden materials have been investigated. The REAS process not only provides an insight into the reactions occurring during the softening and melting process but a range of indices with which to judge the blast furnace performance. Since 1993 new developments started and a test method for Iscor blast furnaces was specifically developed. Although certain indices have been established, uncertainties around the melting mechanisms still existed. These uncertainties include: • Why does the maximum pressure over the sample bed vary extensively between different samples? • Why does a temperature decrease occur only in certain samples and what determines the extent of the temperature decrease? • Which low melting phase forms that causes the initial rise in pressure drop over the sample bed? Four tests were performed on a mixture of Sishen and Thabazimbi ore to determine the phase changes in the test sample. During the reduction of the iron ore, five distinct phases are present. Above 1200°C two liquid phases, an alkali rich phase and a liquid phase with a fayalite composition is present. The rest of the iron reports at different stages in various forms of metallic iron and wustite. Small amounts of a high melting oxide phase, hercynite, also occurs. Softening of the sample is said to occur when the ΔP over the sample bed increases by more than 200 mm H2O. For the specific tests evaluated, this occurred at 1200°C. At this temperature, the liquid with a fayalite composition as well as the alkali rich liquid are present. The formation of the low melting fayalite phase with a high viscosity appears to cause the sudden rise in ΔP. A temperature arrest occurs at the same time supporting the suggestion that liquid formation is responsible for the pressure increase. The results indicate that the mechanisms responsible for the observed pressure drop (decreased gas permeability) and dripping may well be different from those given in the literature. The literature mechanisms emphasise the importance of the amount of FeO available to act as flux for the silica which is present as gangue; hence a greater degree of (indirect) reduction below the melting point of fayalite gives poorer fluxing of silica since less FeO is available. However, the charge materials considered in this study appear to be of substantially higher grade than those used in the previous work. For this reason, there does not appear to be any shortage of FeO to act as flux. This abundance of FeO, and the observation that the peak in pressure drop is not associated with any great change in the amount of liquid, together imply that the literature mechanism regarding changes in the amount and composition of the liquid (i.e. becoming more Si02-rich and viscous as the FeO is reduced) cannot explain the pressure fluctuations observed here. Rather, the increase in pressure appears to be a joint effect of liquid being present (giving the first increase in pressure) and compaction of the sample. Loss of voidage in the sample by this substantial amount of compaction appears the likely cause of the pressure increase. The subsequent decrease in the pressure drop is probably associated with lower viscosity as the sample temperature increases. The importance of compaction means that the amount of indirect reduction does playa role in the development of the pressure drop, but not for the reasons cited in the literature. Pure iron is more malleable than the oxides, and reduction gives a porous iron structure which is more easily compacted. F or both these reasons, the metallic product of indirect reduction favours compaction (and hence the pressure increase). The sharp increase in reduction rate close to the peak pressure presumably results from better contact between the remaining iron oxide (in the fayalite-based liquid, and wustite) with the coke reductant, so favouring direct reduction; this increased reduction (endothermic because of the Boudouard reaction) results in one of the noticeable temperature arrests on the sample temperature curve. The correspondence between the temperature arrests and the changes within the sample do imply that these arrests can be used to gain some information on the reduction mechanisms. However, the reliability of the temperature arrests as indicators of the state of the sample and the reaction conditions within the sample must be tested by further work. / Dissertation (MSc (Metallurgy))--University of Pretoria, 1998. / Materials Science and Metallurgical Engineering / unrestricted