Quantifying texture for acid rock drainage characterisation and prediction

 

Show simple item record

dc.contributor.advisor Becker, Megan
dc.contributor.advisor Broadhurst, Jennifer
dc.contributor.advisor Harrison, Susan
dc.contributor.advisor Bradshaw, Deidre
dc.contributor.author Guseva, Olga
dc.date.accessioned 2021-01-25T08:24:12Z
dc.date.available 2021-01-25T08:24:12Z
dc.date.issued 2020
dc.identifier.citation Guseva, O. 2020. Quantifying texture for acid rock drainage characterisation and prediction. . ,Faculty of Engineering and the Built Environment ,Department of Chemical Engineering. http://hdl.handle.net/11427/32658 en_ZA
dc.identifier.uri http://hdl.handle.net/11427/32658
dc.description.abstract Minerals, metals and mining are the cornerstone of technological development and play an essential role in achieving the United Nations Sustainable Development Goals. Mining, however, is not a wastefree process, with mine wastes being a source of a host of environmental problems. One of these is acid rock drainage (ARD), which forms through a complex series of acid formation (mainly via sulfide oxidation), neutralisation (primarily by carbonates) and gangue mineral dissolution reactions in waste rock and tailings storage piles. The resulting drainage waters are often acidic, highly saline and may contain elevated levels of deleterious elements. Effective ARD mitigation requires accurate ARD characterisation and prediction strategies. To date, standard guidelines recommend a suite of geochemical static (characterisation) and kinetic (prediction) tests. Characterisation tests such as acid-base accounting (ABA) and net acid generation (NAG) tests provide a quick and relatively inexpensive estimate of the "worst case" scenario for acid formation and neutralisation, while kinetic tests (commonly humidity or column leach tests) aim to predict the longterm weathering potential of waste material. The UCT biokinetic test (not currently industry standard practice) was developed to address the effect of microorganisms on ARD formation and allow for the collection of relative kinetic data on neutralisation and acidification within a shortened time period. None of these tests, however, account for the additional layer of complexity introduced by mineral texture, which describes the interrelationship of mineral grains to one another, their shapes and sizes, with some frequently studied textural parameters including mineral liberation, association, grain size distribution and particle size. Mineralogical and textural analyses are infrequently practiced in the context of ARD assessment due to the difficulty in obtaining statistically sound quantitative textural data, high costs of measurement, and standard ARD assessment protocols recommending (rather than necessitating) these assessments. An ARD assessment approach that includes static, kinetic, mineralogical and textural assessments has nonetheless been suggested by several researchers. This project assessed the dominating textural parameters on the scales of kinetic (humidity cell) test (HCT) feed material (meso-scale) and characterisation (static and UCT batch biokinetic) test (SCT) feed material (micro-scale) using four waste rock samples (A, B, C and D) from a greenstone belt gold deposit as a case study. More specifically, the study aimed to assess the role of mineralogy and texture in the ARD assessment "toolbox" and to investigate the role of coarse material sampling for ARD assessment. Data sets collected included the PSD of the micro- and meso-scale material, sample chemistry data obtained from XRF spectrometry and LECO total sulfur, bulk mineralogy data from QXRD and QEMSCAN, as well as textural and mineralogical data from QEMSCAN for sized and unsized micro- and meso-scale material. ARD-specific data sets included results of geochemical characterisation tests such as ANC and single-addition NAG tests, the UCT batch biokinetic test with and without pH control for samples C and D, as well as prediction test data from water-fed and modified humidity cell tests. The geochemical static tests performed on samples A, B, C and D classified them as PAF, PAF, uncertain and NAF, respectively. Non-pH-controlled UCT batch biokinetic tests remained circumneutral for samples B, C and D over the duration of 90 days, while for sample A the pH became acidic over time. The pH-controlled tests demonstrated a steady depletion of neutralisation potential over the first 30 days. Humidity cell test results demonstrated no acidic leachate formation for waterfed tests over 40 weeks, while modified tests showed a decreasing pH over time as the neutralisation capacity was reduced. The mineralogy was important for the interpretation of test results on both the micro-and meso-scales and was assessed in terms of both discrete minerals and reactivity groupings (Fe-Sulfide, other sulfide, dissolving (carbonate), fast weathering, intermediate weathering, slow weathering, inert and other). For sample A the mineralogy was dominated by the inert (quartz), slow weathering (magnetite, plagioclasealbite) and intermediate weathering (Fe-amphibole) categories, with lesser contributions from the FeSulfide (pyrrhotite), dissolving (calcite) and fast weathering (epidote) groups. The main groups contributing to the sample B mineralogy were the slow weathering (plagioclase-albite, magnetite, Kfeldspar), inert (quartz) and intermediate weathering (Fe-mica, chlorite) groups, followed by Fe-Sulfide (pyrite), dissolving (calcite) and fast weathering (epidote) mineral groups. Sample C mineralogy comprised predominantly inert (quartz, titanite), Fe-Sulfide (pyrrhotite), dissolving (calcite) and intermediate weathering (Fe-mica, chlorite) minerals, with lesser contributions from slow weathering (K-feldspar) and fast weathering (epidote) minerals. Sample D comprised intermediate weathering (Feamphibole, chlorite, Fe-mica) and slow weathering minerals, with lesser contributions from slow weathering (magnetite), dissolving (calcite) and Fe-Sulfide (pyrrhotite) minerals. Textural parameters (liberation and association, grain size distribution and liberation spectrum) were evaluated for the FeSulfide and dissolving minerals. On the micro-scale, a large portion of the Fe-Sulfide and dissolving minerals in the samples was found in the liberated category (50%) of the texturally significant size fractions (>1mm), which comprised predominantly locked Fe-Sulfide and dissolving minerals. Evidence of a bimodal distribution was, however, found for sample C via the liberation spectrum and grain size distribution (early liberation size of 8mm), which accounted for the larger degree of liberation observed in the >1mm size fractions, and a larger degree of liberation for the sample overall. The association of Fe-Sulfide and dissolving minerals for all samples was found to be primarily to inert, intermediate weathering and slow weathering minerals, with a larger degree of association of Fe-Sulfide to dissolving minerals observed in sample C. On the micro-scale the mineralogy helped inform the placement of the samples on the geochemical classification plot based on the Fe-Sulfide, dissolving and intermediate weathering mineral contents. For the non-pH-controlled UCT batch biokinetic test, the presence and abundance of calcite was thought to dictate the PAF/NAF nature of the test, as even relatively low amounts of calcite rendered the pH circumneutral for the duration of the test (sample B). For pH-controlled tests, however, the calcite was depleted over time, which led to a favourable acidic environment for the acidophilic bacteria used in the batch biokinetic test. For both the geochemical characterisation and the pH-controlled UCT batch biokinetic tests there was evidence to suggest the contribution of intermediate weathering (Feamphibole, Fe-mica, chlorite) and slow weathering (magnetite) minerals to the neutralisation potential in the sample. On the meso-scale the effects of mineralogy were most prominent for the modified humidity cell tests, which showed some pH fluctuations and a steady depletion of the primary neutralisation potential. The pH fluctuation after the depletion of the dissolving minerals was attributed to the dissolution of intermediate weathering minerals over the 40 weeks of the tests. These effects were not observed during the 40 weeks of the water-fed experiments. Given sufficient time for the latter test, however, it would be expected that upon the onset of acidification, similar effects of the mineralogy on the leachate quality would be observed as in the modified tests. Knowledge of the Fe-sulfide and dissolving mineral texture yielded several insights. on the micro-scale, the liberation and grain size distribution data provided an indication that a sample-customised grinding size should be established to ensure adequate "worst case" scenario determination via characterisation tests, as material with fine Fe-Sulfide or dissolving mineral grains may not be fully liberated at the recommended 75µm top size. On the meso-scale, the texture yielded insight into the circumneutral behaviour of the water-fed HCT, as most of the acid-forming minerals were contained in size fractions where the liberation was either limited or negligible, with predominant association to slow weathering, intermediate weathering and inert minerals. These findings highlighted the importance of considering mineralogy, texture and the PSD of the material for HCT result interpretation. When considering texture as a parameter for ARD assessment, the potential for sampling and mineralogical errors arose due to the coarse material size (specifically on the meso-scale) and the limitations on the number of particles that could be assessed. Quantitative mineralogy and texture data allowed for the quantitative assessment of the sampling and mineralogical errors, which were investigated through Pierre Gy's fundamental sampling error (FSE) equation, the binomial distribution approximation and the plotting of confidence intervals over the Fe-Sulfide liberation data. The results showed that although tools such as Gy's "safety line" provide a useful quick means of sampling error assessment, this approach may yield excessively large sample mass requirements for coarse material. Calculating the sampling error from the textural and mineralogical data provided a useful tool to estimate sample representativeness. Additionally, the estimation of sampling errors may help in the planning of an appropriate sampling approach, which may ultimately provide a means to relate data sets to one another across scales based on how representative samples are of one another, and therefore of the parent lot. The current study showed how mineralogy and texture are not simply "tools" in the ARD assessment "toolbox", but rather a key means for interpreting characterisation and prediction test data. Additionally, the quantitative assessment of mineralogy and texture provided the opportunity to assess the materialspecific sampling error, which, in turn, may allow for the correlation of data sets across various scales and for the planning of appropriate sampling strategies. Recommendations for future work include: the quantitative assessment of the ARDI for meso-scale material; the assessment of detailed characterisation and prediction test leachate chemistry; trace element assessment and deportment throughout UCT batch biokinetic and humidity cell testing; mineralogical and textural assessment on characterisation and prediction test residues during and after tests; an in-depth analysis of the minimal/optimal sample block/sub-sample mass required for minimal error; the assessment of samples using X-ray microcomputed tomography to assess and decrease the effects of stereological bias prevalent in 2D measurements; and the application of a similar texture and mineralogy assessment to additional waste types (such as coal wastes, or waste material containing non-Fe-bearing sulfides).
dc.subject acid rock drainage
dc.subject mine waste
dc.subject ARD characterisation
dc.subject acid based accounting
dc.title Quantifying texture for acid rock drainage characterisation and prediction
dc.type Master Thesis
dc.date.updated 2021-01-25T08:23:37Z
dc.language.rfc3066 eng
dc.publisher.faculty Faculty of Engineering and the Built Environment
dc.publisher.department Department of Chemical Engineering
dc.type.qualificationlevel Masters
dc.type.qualificationlevel MSc
dc.identifier.apacitation Guseva, O. (2020). <i>Quantifying texture for acid rock drainage characterisation and prediction</i>. (). ,Faculty of Engineering and the Built Environment ,Department of Chemical Engineering. Retrieved from http://hdl.handle.net/11427/32658 en_ZA
dc.identifier.chicagocitation Guseva, Olga. <i>"Quantifying texture for acid rock drainage characterisation and prediction."</i> ., ,Faculty of Engineering and the Built Environment ,Department of Chemical Engineering, 2020. http://hdl.handle.net/11427/32658 en_ZA
dc.identifier.vancouvercitation Guseva O. Quantifying texture for acid rock drainage characterisation and prediction. []. ,Faculty of Engineering and the Built Environment ,Department of Chemical Engineering, 2020 [cited yyyy month dd]. Available from: http://hdl.handle.net/11427/32658 en_ZA
dc.identifier.ris TY - Master Thesis AU - Guseva, Olga AB - Minerals, metals and mining are the cornerstone of technological development and play an essential role in achieving the United Nations Sustainable Development Goals. Mining, however, is not a wastefree process, with mine wastes being a source of a host of environmental problems. One of these is acid rock drainage (ARD), which forms through a complex series of acid formation (mainly via sulfide oxidation), neutralisation (primarily by carbonates) and gangue mineral dissolution reactions in waste rock and tailings storage piles. The resulting drainage waters are often acidic, highly saline and may contain elevated levels of deleterious elements. Effective ARD mitigation requires accurate ARD characterisation and prediction strategies. To date, standard guidelines recommend a suite of geochemical static (characterisation) and kinetic (prediction) tests. Characterisation tests such as acid-base accounting (ABA) and net acid generation (NAG) tests provide a quick and relatively inexpensive estimate of the "worst case" scenario for acid formation and neutralisation, while kinetic tests (commonly humidity or column leach tests) aim to predict the longterm weathering potential of waste material. The UCT biokinetic test (not currently industry standard practice) was developed to address the effect of microorganisms on ARD formation and allow for the collection of relative kinetic data on neutralisation and acidification within a shortened time period. None of these tests, however, account for the additional layer of complexity introduced by mineral texture, which describes the interrelationship of mineral grains to one another, their shapes and sizes, with some frequently studied textural parameters including mineral liberation, association, grain size distribution and particle size. Mineralogical and textural analyses are infrequently practiced in the context of ARD assessment due to the difficulty in obtaining statistically sound quantitative textural data, high costs of measurement, and standard ARD assessment protocols recommending (rather than necessitating) these assessments. An ARD assessment approach that includes static, kinetic, mineralogical and textural assessments has nonetheless been suggested by several researchers. This project assessed the dominating textural parameters on the scales of kinetic (humidity cell) test (HCT) feed material (meso-scale) and characterisation (static and UCT batch biokinetic) test (SCT) feed material (micro-scale) using four waste rock samples (A, B, C and D) from a greenstone belt gold deposit as a case study. More specifically, the study aimed to assess the role of mineralogy and texture in the ARD assessment "toolbox" and to investigate the role of coarse material sampling for ARD assessment. Data sets collected included the PSD of the micro- and meso-scale material, sample chemistry data obtained from XRF spectrometry and LECO total sulfur, bulk mineralogy data from QXRD and QEMSCAN, as well as textural and mineralogical data from QEMSCAN for sized and unsized micro- and meso-scale material. ARD-specific data sets included results of geochemical characterisation tests such as ANC and single-addition NAG tests, the UCT batch biokinetic test with and without pH control for samples C and D, as well as prediction test data from water-fed and modified humidity cell tests. The geochemical static tests performed on samples A, B, C and D classified them as PAF, PAF, uncertain and NAF, respectively. Non-pH-controlled UCT batch biokinetic tests remained circumneutral for samples B, C and D over the duration of 90 days, while for sample A the pH became acidic over time. The pH-controlled tests demonstrated a steady depletion of neutralisation potential over the first 30 days. Humidity cell test results demonstrated no acidic leachate formation for waterfed tests over 40 weeks, while modified tests showed a decreasing pH over time as the neutralisation capacity was reduced. The mineralogy was important for the interpretation of test results on both the micro-and meso-scales and was assessed in terms of both discrete minerals and reactivity groupings (Fe-Sulfide, other sulfide, dissolving (carbonate), fast weathering, intermediate weathering, slow weathering, inert and other). For sample A the mineralogy was dominated by the inert (quartz), slow weathering (magnetite, plagioclasealbite) and intermediate weathering (Fe-amphibole) categories, with lesser contributions from the FeSulfide (pyrrhotite), dissolving (calcite) and fast weathering (epidote) groups. The main groups contributing to the sample B mineralogy were the slow weathering (plagioclase-albite, magnetite, Kfeldspar), inert (quartz) and intermediate weathering (Fe-mica, chlorite) groups, followed by Fe-Sulfide (pyrite), dissolving (calcite) and fast weathering (epidote) mineral groups. Sample C mineralogy comprised predominantly inert (quartz, titanite), Fe-Sulfide (pyrrhotite), dissolving (calcite) and intermediate weathering (Fe-mica, chlorite) minerals, with lesser contributions from slow weathering (K-feldspar) and fast weathering (epidote) minerals. Sample D comprised intermediate weathering (Feamphibole, chlorite, Fe-mica) and slow weathering minerals, with lesser contributions from slow weathering (magnetite), dissolving (calcite) and Fe-Sulfide (pyrrhotite) minerals. Textural parameters (liberation and association, grain size distribution and liberation spectrum) were evaluated for the FeSulfide and dissolving minerals. On the micro-scale, a large portion of the Fe-Sulfide and dissolving minerals in the samples was found in the liberated category (50%) of the texturally significant size fractions (>1mm), which comprised predominantly locked Fe-Sulfide and dissolving minerals. Evidence of a bimodal distribution was, however, found for sample C via the liberation spectrum and grain size distribution (early liberation size of 8mm), which accounted for the larger degree of liberation observed in the >1mm size fractions, and a larger degree of liberation for the sample overall. The association of Fe-Sulfide and dissolving minerals for all samples was found to be primarily to inert, intermediate weathering and slow weathering minerals, with a larger degree of association of Fe-Sulfide to dissolving minerals observed in sample C. On the micro-scale the mineralogy helped inform the placement of the samples on the geochemical classification plot based on the Fe-Sulfide, dissolving and intermediate weathering mineral contents. For the non-pH-controlled UCT batch biokinetic test, the presence and abundance of calcite was thought to dictate the PAF/NAF nature of the test, as even relatively low amounts of calcite rendered the pH circumneutral for the duration of the test (sample B). For pH-controlled tests, however, the calcite was depleted over time, which led to a favourable acidic environment for the acidophilic bacteria used in the batch biokinetic test. For both the geochemical characterisation and the pH-controlled UCT batch biokinetic tests there was evidence to suggest the contribution of intermediate weathering (Feamphibole, Fe-mica, chlorite) and slow weathering (magnetite) minerals to the neutralisation potential in the sample. On the meso-scale the effects of mineralogy were most prominent for the modified humidity cell tests, which showed some pH fluctuations and a steady depletion of the primary neutralisation potential. The pH fluctuation after the depletion of the dissolving minerals was attributed to the dissolution of intermediate weathering minerals over the 40 weeks of the tests. These effects were not observed during the 40 weeks of the water-fed experiments. Given sufficient time for the latter test, however, it would be expected that upon the onset of acidification, similar effects of the mineralogy on the leachate quality would be observed as in the modified tests. Knowledge of the Fe-sulfide and dissolving mineral texture yielded several insights. on the micro-scale, the liberation and grain size distribution data provided an indication that a sample-customised grinding size should be established to ensure adequate "worst case" scenario determination via characterisation tests, as material with fine Fe-Sulfide or dissolving mineral grains may not be fully liberated at the recommended 75µm top size. On the meso-scale, the texture yielded insight into the circumneutral behaviour of the water-fed HCT, as most of the acid-forming minerals were contained in size fractions where the liberation was either limited or negligible, with predominant association to slow weathering, intermediate weathering and inert minerals. These findings highlighted the importance of considering mineralogy, texture and the PSD of the material for HCT result interpretation. When considering texture as a parameter for ARD assessment, the potential for sampling and mineralogical errors arose due to the coarse material size (specifically on the meso-scale) and the limitations on the number of particles that could be assessed. Quantitative mineralogy and texture data allowed for the quantitative assessment of the sampling and mineralogical errors, which were investigated through Pierre Gy's fundamental sampling error (FSE) equation, the binomial distribution approximation and the plotting of confidence intervals over the Fe-Sulfide liberation data. The results showed that although tools such as Gy's "safety line" provide a useful quick means of sampling error assessment, this approach may yield excessively large sample mass requirements for coarse material. Calculating the sampling error from the textural and mineralogical data provided a useful tool to estimate sample representativeness. Additionally, the estimation of sampling errors may help in the planning of an appropriate sampling approach, which may ultimately provide a means to relate data sets to one another across scales based on how representative samples are of one another, and therefore of the parent lot. The current study showed how mineralogy and texture are not simply "tools" in the ARD assessment "toolbox", but rather a key means for interpreting characterisation and prediction test data. Additionally, the quantitative assessment of mineralogy and texture provided the opportunity to assess the materialspecific sampling error, which, in turn, may allow for the correlation of data sets across various scales and for the planning of appropriate sampling strategies. Recommendations for future work include: the quantitative assessment of the ARDI for meso-scale material; the assessment of detailed characterisation and prediction test leachate chemistry; trace element assessment and deportment throughout UCT batch biokinetic and humidity cell testing; mineralogical and textural assessment on characterisation and prediction test residues during and after tests; an in-depth analysis of the minimal/optimal sample block/sub-sample mass required for minimal error; the assessment of samples using X-ray microcomputed tomography to assess and decrease the effects of stereological bias prevalent in 2D measurements; and the application of a similar texture and mineralogy assessment to additional waste types (such as coal wastes, or waste material containing non-Fe-bearing sulfides). DA - 2020 DB - OpenUCT DP - University of Cape Town KW - acid rock drainage KW - mine waste KW - ARD characterisation KW - acid based accounting LK - https://open.uct.ac.za PY - 2020 T1 - Quantifying texture for acid rock drainage characterisation and prediction TI - Quantifying texture for acid rock drainage characterisation and prediction UR - http://hdl.handle.net/11427/32658 ER - en_ZA


Files in this item

This item appears in the following Collection(s)

Show simple item record