An evaluation of liberation models for the prediction of diagnostic leaching results

Annandale, Gideon Johannes (1995)

Thesis (M. Ing.) -- University of Stellenbosch, 1995.

Thesis

ENGLISH ABSTRACT: The interrelationship between mineral liberation and leaching behaviour of a gold ore is ill defined and lacks understanding mainly due to the complexity of the individually processes, i.e. leaching and mineral liberation. An understanding of this relationship could result in lower costs, since an increase in the efficiency of gold dissolution and a decrease in costs related to the crushing and grinding operations, could be expected. Diagnostic leaching was used as the analytical method to investigate the mineralogy (gold deportment) of various gold-bearing ores designating from different goldfields in South Africa. A thorough literature survey was also conducted to determine the availability of theoretical models which could describe the relationship between liberation and leaching behaviour. From the experimental results obtained, it was concluded that the leachability of an ore is not only dependent on the degree of liberation of that ore when comminuted, but that it is also very much dependent on the degree of exposure of gold grains throughout the whole of the ore. It was found that all the theoretical liberation and exposure models underestimated the amount of leachable gold in each particle size fraction, mainly due to the neglecting of exposed gold grains situated in minor cracks and fractures. It became evident in this research, that in order to model the leachability of gold fundamentally, the data obtained from diagnostic leaching tests alone are insufficient. Data concerning the amounts of gold bearing minerals, present in the ores as well as SEM data of the gold grains, are needed together with diagnostic leaching results. However, an empirical model was developed that predicts the degree of liberation (leachability) of an ore milled to 70% -75μm fairly accurately as a first attempt to model this highly complex process. The degree of liberation (leachability) was found to be a function of the particle size and the gold deportment of the unmilled ore. It was decided to extend the King liberation model and the exposure model developed by Hsih and co-workers empirically to provide for the fraction of gold not directly related to liberation. Both the models showed good agreements with the experimental results. Neural network analysis of the diagnostic leaching results were done with partial success, but unfortunately more data is still needed to assist in the development of an accurate leaching or liberation model.

AFRIKAANSE OPSOMMING: Weens die komplekse aard van die mineraalvrystelling- en logings prosesse, word die verwantskap tussen mineraal vrystelling en die logingsgedrag van 'n goud erts nie goed verstaan nie. Aangesien hoë kostes verbonde is aan die vergruising- en die maalproses van die ertse, is dit belangrik om hierdie verwantskap goed te verstaan. Dit kan ook tot gevolg hê dat die effektiwiteit waarmee die goud opgelos word, verhoog word, wat op sy beurt tot laer kostes sal lei. Die analitiese metode, diagnostiese loging, is gebruik om die mineralogie (goud verspreiding relatief tot die verskillende minerale, in die erts) van ertse uit verskeie Suid-Afrikaanse goudvelde te bepaal. 'n Deeglike literatuurstudie het die verskillende teoretiese mineraal vrystellings modelle getoon, wat vrystelling met logingsgedrag in verband probeer bring. Die eksperimentele resultate het getoon dat die loogbaarheid van 'n erts nie slegs afhanklik is van die graad van vrystelling nie, maar ook van die mate waartoe die goudkorrels aan die logingsvloeistof blootgestel is. Daar is tot die gevolgtrekking gekom dat al die teoretiese mineraal vrystellingsmodelle, die hoeveelheid goud in elke partikelgrootte fraksie totaal en al onderskat. Dit kan hoofsaaklik toegeskryf word aan die hoeveelheid goud wat barste en krake in die erts vul, wat nie so seer vrygestel is nie, maar wel loogbaar is. Dit blyk duidelik uit die navorsing, dat die diagnostiese logingsdata alleen, onvoldoende is om die loogbaarheid van goud fundamenteel te modelleer. Data in verband met die hoeveelheid van elke mineraal wat in die erts teenwoordig is asook SEM data van die goudkorrels in die verskillende minerale, word ook sodanig benodig. Desnieteenstaande, is daar as 'n eerste probeerslag, 'n emperiese model voorgestel wat die loogbaarheid van 'n gemaalde erts (gemaal tot 70% -75 μm) redelik goed voorspel. Die graad waartoe die goud loogbaar is (mineraal-vrystelling), is 'n funksie van die partikelgrootte en die goudverdeling in die ongemaalde erts. Twee teoretiese modelle, nl die King vrystellingsmodel en die Hsih en ander-model is ook empiries aangepas om sekere tekortkominge te oorbrug. Die modelle het goed met die eksperimentele waardes ooreengestem. Analises van die diagnostiese logings resultate is met behulp van neurale netwerke gedoen. Alhoewel dit gedeeltelik suksesvol was, word heelwat meer data benodig om die loogbaarheid van goud akkuraat te voorspel.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/54645
This item appears in the following collections: