Thesis (MScIng)--Stellenbosch University, 2007. / Some digitised pages may appear illegible due to the condition of the original hard copy. / ENGLISH ABSTRACT: The uranium leaching behaviour of ore from three different mines in the Vaal River region, namely
Kopanang, Great Noligwa and Moab Khotsong, was investigated. The aim of the work was to
characterise the uranium leaching behaviour of the different ores. It involved a full mineralogical
evaluation of the ore as well as determining optimum operating conditions for maximum extraction of
uranium, relating the extractions to mineralogical features of the ores.
The major bulk minerals in the three Vaal River ores have been identified and consist primarily of
quartz (70-80 %), with lesser amount~ of muscovite (8-11%). Moab Khotsong and Noligwa ore
samples were found to be very similar, and Kopanang is different from the two ores. Kopanang ore
has less pyrite, quartz, and chlorite than the other two ores, but contains more pyrophyllite. Chlorite is
an acid consumer and a Fe2+/Fe3+ producer in uranium leaching. The difference in mineralogy of the
three ores definitely reflected in the reagent profiles during acid leaching of uranium, specifically acid
consumption.
With regards to uranium occurrence, bulk uranium analysis showed that 80-90 % of the uranium in
the ores is contained as uraninite, 8-19 % as brannerite, and the balance as traces of coffinite and
uranium phosphates. Uranium grain sizes were found to be very small, with 50 % of the particles
passing 19.4, 21.3 and 23.2 IJm for Kopanang, Noligwa and Moab Khotsong respectively. The degree
of liberation of the uranium-bearing minerals was low, between 11 and 45 %, and expectedly
increased as particle size decreased. However, between 87 and 93 % of the uraninite particles and 71
to 86 % of the brannerite particles have more than 10 % of their surfaces exposed, and even higher
proportions have more than 5 % of their surfaces exposed.
The primary experiments were designed within practically attainable boundaries, although special
experiments were performed outside the boundaries during subsequent mechanistic studies. H2S04
addition varied between 10 and 25 kg/t, temperature varied between 40 and 60°C, and Mn02 addition
varied between 2 and 4 kg/t (100 percent Mn02). Mn02 was added as pyrolusite, 1.5 hours after
addition of acid. A relative pulp density of 1.55 was used. Leaching times of 24 and 48 hours were
investigated. The ore was milled to 80 % -75 IJm, even though coarser grinding was also investigated.
It was found that 60-90 % dissolution could be achieved as leaching conditions were varied within
the operating window. Acid addition had the greatest influence on final uranium extractions. Practically
useful results, within the operating window attainable in a plant, can be summarised as follows:
• Uranium dissolutions are 80-85 % for Great Noligwa and Moab Khotsong ores, and 85-90 % for Kopanang ore
• Required residence time = 24 hrs
• To achieve the above dissolutions, Kopanang ore requires at least 11 kg/t acid, while acid
requirements for Great Noligwa and Moab Khotsong ores are 14 and 16 kg/t respectively
• MnOz addition can be kept at a minimum, because sufficient Fe can be leached from the ore
(meaning only enough MnOz for conversion of ferrous to ferric needs to be added)
• The chlorite dissolution reaction consumes acid, generates ferric and ferrous ions, but also
generates dissolved silica. Therefore, the kinetics of this reaction must be understood in order
to control it to the lowest extent possible, without impairing the availability of dissolved iron.
An understanding of the factors responsible for the leaching behaviour of uranium-bearing ores is
critical in achieving optimal uranium recoveries. This is particularly important in light of the fact that
dissolutions higher than 90 % are very difficult to achieve under the normal operating conditions
employed on the South African acid leaching plants.
While solubility limitations were initially suspected, it was found that the reason for the existence of the
upper limit for dissolution was most likely due to the presence of aerially locked uranium minerals
(uranium minerals that is not exposed to the leaching environment). However, residue analysis
showed that most of the uraninite dissolved and that the major fraction of unleached uranium existed
as brannerite which is in fact exposed to the leaching environment. Therefore, the slow leaching
kinetics / intrinsic inertness of brannerite was considered as the limiting factor for not achieving
optimum recoveries.
To achieve optimal extraction, considering the mineralogical characteristics of the ores (in terms of a
plant's flow/operational perspective), a diagnostic leaching approach was followed. A mineralogyleachability
explanation is presented to rationalise the difficulty in exceeding 90 % dissolution from low
grade uranium ores on the basis of a novel diagnostic leaching method. More specifically, to
determine the interrelationship between mineralogy, mineral liberation and the leaching behaviour of
uranium, a methodology was developed for unlocking uranium by a combination of chemical (drastic
leaching of minerals associated with the residual uranium) and physical (fine grinding to increase area
exposure and liberation) methods.
Diagnostic leaching results indicated that to improve uranium dissolution beyond 90 %, uneconomical
conditions (residence time between 48 and 72 hours, constant pH = 1 or Eh = 700 mV) need to be
considered and it will still not necessarily be possible to increase beyond 95%. The maximum obtained
is 98% using nitric acid digestion at evaluated temperatures (900C).
Mineralogical analysis indication that it is possible to leaching brannerite but, the leaching kinetics
thereof is very slow. Based on the diagnostic leaching tests the following is recommended for the tree ores tested: Sulphuric acid leaching must be used for treating Kopanang ore. If the brannerite
concentration of Noligwa and Moab Khotsong ore is < 20 % sulphuric acid leaching is recommended
but if the brannerite concentration> 20 % other leaching methods will be recommended (Le. pressure
leaching or using a different leaching reagent).
A non-linear decision tree model was developed for modelling of the experimental data and is
presented in this thesis. In this context it would be more realistic to determine a range for expected
recovery rather than trying to determine an exact value. Therefore, a classification tree model was
used. Using this method, just evaluating the tree shows that: For uranium dissolution higher than 70%
a residence time longer that 17 hours is required including an acid addition higher that 11.35 kg/t for
Noligwa and Moab Khotsong ore while lower acid concentration can be tolerated for Kopanang ore. It
is proven that that model can with 86% accuracy separate that into the various classes. Through cross
validation it is also proven to be a representative model with an average success rate of 84 % for
classification of data with a standard deviation of 2.8%. This model can be used to predict the
expected dissolution range based only on the operating parameters within the Vaal River context.
Since the model is based on laboratory experimental data it can not necessarily be used for plant
operation/optimisation purposes. It is recommended to develop an equivalent model using pilot plant
data to develop a more accurate model which can also be used for optimisation purposes.
Although it is not part of the original scope of the project (based on the information gained though out
the project), a proposal of an empirical model simulator is also presented. Since development of the
simulator is still in the developing stages, this thesis will only include the algorithm as well as a basic
model predictor which can be used for future research. The ideal is to develop a simulator has the
ability to predict uranium leaching behaviour and reagent consumption based on mineralogy.
Gold extractions obtained by forward leaching (direct cyanide leaching of gold) were compared with
those obtained by reverse leaching (sulphuric acid leaching of uranium followed by cyanide leaching
of gold) for three different Vaal River ores, to quantify the benefits of reverse leaching option. Reverse
leaching of gold recovery by between 3 and 4 percentage points, improving total gold recovery to
98 %. A gold benefit of between 0.4 and 0.6 g/t was measured. The exact financial gain is dependant
on the gold price and other economic factors, but an estimated benefit for treating an average of
240000 t ore/month is in the order of R 14000 OOO/month. The reverse leaching operation for the
recovery of both gold and uranium is therefore a financially justified process route, because recovery
of uranium will ensure that the costs are lower than the revenue increment. / AFRIKAANSE OPSOMMING: Die logingsgedrag van uraan erts vanaf drie verskillende myne in the Vaalrivier omgewing, naamlik
Kopanang, Noligwa en Moab Khotsong is ondersoek. Die doel van die studie is die karakteriseering
van die uraan logingsgedrag en behels 'n indiepte mineralogiese evalueering van die verskillende
ertse, die bepaling van die optimale bedryfskondisies vir die ekstraksie van uraan as ook die bepaling
van 'n verband tussen ekstraksie en mineralogiese eienskappe van die erts.
Die grootmaat minerale van die drie Vaalrivier ertse is geidentifiseer en bestaan hoofsaaklik uit kwarts
(70 - 80 %), met 'n laer konsentrasie muskoviet (8 - 11 %). Daar is gevind dat die Moab Khotsong en
Noligwa erts monsters tot 'n groot mate ooreenstem, terwyl Kopanang erts daarvan verskil. Kopanang
erts bevat minder piriet, kwarts en chloriet maar meer pirofilliet in vergelyking met die ander twee
ertse. Chloriet is 'n suur verbruiker en 'n Fe2+/Fe3
+ verskaffer in uraan logings prosesse. Dit blyk uit die
resulte dat die verskil in die mineralogie van die drie ertse definitief reflekteer op die reagense profiele
tydens suur loging van uraan, veral suur verbruik.
Die grootmaatanalise, in terme van die aanwesigheid van uraan, toon dat 80 - 90 % van die uraan in
die erts voorkom as uraniniet, 8 - 19 % as branneriet en die balans kom voor as koffiniet en uraanfosfate.
Daar is gevind dat die uraan korrelgrootte baie klein is, met 50 % van die partikels kleiner as
19.4, 21.3 en 23.2 IJm vir Kopanang, Noligwa en Moab Khotsong respektiewelik. Die graad van
bevryding van die uraan bevattende minerale was laag (tussen 11 en 45 %) en neem na verwagting
toe soos partikel grootte afneem. Nietemin, tussen 87 en 93 % van die uraniniet partikels en 71 _ 86
% van die branneriet partikels het meer as 10 % van die mineraal oppervlakte blootgestel aan die
logings omgewing met selfs hoer proporsies met meer as 5 % blootgestelde oppervlak area.
Die primere eksperimente was ontwerp binne die praktiese grense haalbaar op 'n aanleg. Spesiale
eksperimente was wei ook uitgevoer buite die grense, vir opeenvolgende meganistiese studies. H
2
S0
4
byvoeging was gevarieer tussen 10 en 25 kg/t, temperatuur gevarieer tussen 40 en 60°C en Mn02
byvoeging is gevarieer tussen 2 en 4 kg/t (100 persent Mn02)' Mn02 is toegevoeg as pirolusiet 1.5 uur
na die toevoeging van die suur. 'n Relatiewe pulp digtheid van 1.55 was gebruik. 'n Loging residensie
tyd van 24 en 48 uur was ondersoek. Die erts was gemaal tot 80 % -75 IJm alhoewel growwer maling
wei ook ondersoek is.
Daar is gevind dat 60 - 90 % oplossing wei bereikbaar is binne die bedryfsgense. Dit blyk dat die suur
byvoeging die grootste invloed het op die finale uraan ekstraksie. Praktiese nuttige informasie, binne
die bedryfsgense van 'n aanleg, kan as volg opgesom word:
• Uraan oplossings tussen 80 - 85 % vir Great Noligwa en Moab Khotsong erts en tussen 85 -
90 % vir Kopanang erts.
• Vereiste residensietyd = 24 uur
• Om begenoemde oplossings te bereik, benodig Kopanang erts ten minste 11 kg/t suur, terwyl
die suur vereistes vir Noligwa en Moab Khotsong erts 14 en 16 kg/t is onderskeidelik
• Mn02 toevoeging kan tot 'n minimum beperk word aangesien voldoende Fe geloog word
vanaf die erts (dus sleg genoeg Mn02 vir die omskakeling van Fe2+ na Fe3+ word benodig)
• Die chloriet oplossings reaksie verbruik suur en genereer Fe3
+ en Fe2+ as ook silika in
oplossing. Dit is dus belangrik om die kinetika van die reaksie te verstaan om sodoende die
reaksie tot 'n groot mate te onderdruk sonder om die beskikbare yster in oplossing te
beinvloed
'n Deeglike begrip van die faktore verantwoordelik vir die logingskarakteristieke van uraan bevattende ertse is krities om 'n optimale uraan opbrengs te bereik. Dit is veral belangrik aangesien oplossings
hoer as 90 % moeilik is om te bereik onder die normale beheer kondisies wat gebruik word op Suid
Afrikaanse suurlogingsaanlegte.
Terwyl oplosbaarheidsbeperkings oorspronklik verdink was, is daar gevind dat die bestaan van 'n hoer
limiet vir uraan oplosbaarheid waarskynlik 'n gevolg is van die teenwoordigheid van area geslote
uraan minerale (uraan minerale wat nie aan die logings area blootgestel is nie). Nietemin, residu
analiese toon dat die meeste van die uraniniet opgelos het en dat die grootste fraksie van ongeloogde
uraan voorkom as brannerite, wat wei bloot gestel is aan die logings omgewing. Gebaseer op die
bevindings word die stadige loging kinetika / intrisieke traagheid van brannerite gereken as die
beperkende faktor vir optimum uraan herwinnig.
Om optimale ekstraksie te bereik, gelet op die mineralogiese karakteristieke van die erts (in terme van
'n aanleg se vloei / beheer perspektief), was 'n diagnostieke logingsbenadering gevolg. 'n
Mineralogiese-uitloging verduideliking word bespreek om die beperking van oplossings hoer as 90 %
te bespreek vir lae graad uraan ertse wat gebaseer is op 'n stap vir stap diagnostiese metode. Meer
spesifiek, om die interaksie tussen die mineralogie, mineraal bevryding en logingsgedrag van uraan te
bepaal is 'n metodologie ontwikkel om uraan minerale te bevry deur 'n kombinasie van chemise
(drastiese loging van minerale geassosieer met die residu uraan) en fisiese (fyn maling om die area
van blootstelling en bevryding van uraan minerale te vergroot) metodes.
Die diagnostiese resultate wys daarop dat om uraan oplossing te verhoog bo 90 %, onekonomiese
kondisies (residensie tyd tussen 48 en 72 uur, konstante pH =1 of Eh = 700 mV) oorweeg moet word
en dit nie noodwendig oplossings hoer as 95 % bewerkstellig nie. Die maksimum ekstrasie wat bereik
was is 98 % tydens salpetersuur vertering by hoe temperature (90°C).
Mineralogiese analieses wys daarop dat dit wei moontlik is om branneriet te loog maar dat die loging
kinetika baie stadig is. Gebaseer op die diagnostieke logings resultate word die volgende aanbeveel:
Swawelsuur loging vir die behandeling van Kopanang erts. Indien die konsentrasie van branneriet in
Noligwa and Moab Khotsong erts laer is as 20 % sal swawelsuur loging aanbeveel word maar indien
dit verhoog (>20 %) moet ander logings tegnieke oorweeg word (bv. drukloging of die gebruik van
ander logings reagense)
'n Nie linieere besluit boom model was ontwikkel vir die modelering van die eksperimentele data en
word weergegee in die tesis. Aangesien daar meer waarde Ie in die bepaling van 'n verwagte uraan
oplossings in 'n bepaalde gebied, eerder as om die presiese waarde te probeer bepaal is dit as 'n
klassifikasie model hanteer. Deur hierdie metode te volg kan daar, gebaseer op die ontwikkelde boom
die volgende afleidings gemaak word: Om oplossing bo 70% te bereik is 'n residensie tyd van langer
as 17 uur benodig as ook 'n suur toevoeging van meer as 11.35 kg/t vir Noligwa en moab Khotsong
erts terwyl minder suur benodig word vir Kopanang erts. Daar is getoon dat die model met 86 %
akkuraatheid die data kan skei in die verskillende klasse. Kruis valideering van die model toon dat dit
verteenwoordigend is en gemiddeld 84 % sukses behaal in die klassifiseering van data met 'n
standard afwyking van 2.8 %. Hierdie model kan gebruik word om die verwagte oplossings gebied te
voorspel gebaseer op die beheer veranderlikes binne the Vaalriver konteks. Aangesien die model
gebaseer is op laboratorium gegenereerde data kan dit nie noodwendig gebruik word vir aanleg
beheer doeleindes of optimiseering nie. Daar word dus aanbeveel om dieselfde model te ontwikkel vir
proef aanleg data om 'n meer akkurate model te ontwikkel wat gebruik kan word vir optimiseerings doeleindes.
Alhoewel dit nie deel was van die oorspronklike omvang van die projek nie (gebaseer op die
informasie versamel deur die verloop van die projek) word 'n voorstel van 'n empiriese model
simulator bespreek in die tesis. Aangesien die ontwikkeling van die simulator nog in die ontwikkelings
fase is sal die tesis slegs die algoritme as ook 'n basiese model voorspeller simulator insluit wat vir
toekomstige navorsing gebruik kan word. Die ideaal is om 'n simulator te ontwikkel wat die vermoee
het om die uraan logingsgedrag te voorspel as ook die reagense verbruik gebaseer alleenlik op
mineralogie.
Goud ekstraksies verkrygbaar deur voorwaartse loging (direkte sian ide loging van goud) is vergelyk
met ekstraksies verkry deur terugwaartse loging (swawelsuur loging van uraan gevolg deur sianide
loging van goud) vir die drie verskillende Vaal rivier ertse. Die hoof doel is die kwantifiseering van die
voordele ten opsigte van terugwaartse loging. Terugwaartse loging van gold verhoog die goud
opbrengs tussen 3 en 4 persentasie punte wat lei tot 'n totale gold herwinning van 98 %. 'n Goud wins
van tussen 0.4 en 0.6 g/t is bepaal. Die presiese finansieele wins is sterk afhanklik van die goud prys
en ander ekonomieses faktore, maar 'n geskatte wins vir behandeling van gemiddeld 240000 t
erts/maand is in die orde van R 14000 OOO/maand. Daarom is die terugwaartse logingsmetode vir die
herwinnig van beide goud en uraan 'n finansieel geregverdigde proses roete.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:sun/oai:scholar.sun.ac.za:10019.1/50694 |
| Date | 12 1900 |
| Creators | Lottering, Maria Johanna |
| Contributors | Lorenzen, Leon, Stellenbosch University. Faculty of Engineering. Dept. of Process Engineering. |
| Publisher | Stellenbosch : Stellenbosch University |
| Source Sets | South African National ETD Portal |
| Language | en_ZA |
| Detected Language | English |
| Type | Thesis |
| Format | 88 pages : illustrations |
| Rights | Stellenbosch University |
Page generated in 0.004 seconds