Search Documents

By P Holtham, K C. Runge, E Tabosa

Traditionally, comminution and flotation have been surveyed, analysed and optimised separately. This often results in less than optimal outcomes because of the complex interactions between the two processes. Flotation recovery is a strong function of the valuable mineral particle size distribution with lower recovery of ultra-fines due to poor flotation kinetics, optimum recovery for the intermediate sized particles and lower recovery for the coarser particles due to poorer liberation. Thus, to optimise flotation recovery and grade it is usually best to grind to a fine size. This requirement needs to be balanced by the cost of grinding finer and the revenue that results from an increase in grinding throughput, which often results in a coarsening of the flotation feed.This paper demonstrates a technique which can be used to predict the effect of a change in feed grind size on flotation performance. It can be used to determine the optimum grind size for a particular operation which optimises total circuit profitability. Analysis is relatively easy to perform, as a minimum, requiring only size by assay data of the feed and product streams of the process. Case studies will be presented that demonstrate the usefulness of this technique in feed grind applications. Potential use of the technique for regrind evaluation is discussed. The limitations associated with this technique in certain circuit scenarios will also be outlined.CITATION:Runge, K C, Tabosa, E and Holtham, P, 2014. Integrated optimisation of grinding and flotation circuits, in Proceedings 12th AusIMM Mill Operators’ Conference 2014 , pp 77–84 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Sep 1, 2014

By H S. Mitri, A Sainoki

Destress blasting techniques have been widely employed in underground mines for the purpose of preconditioning highly stressed rock mass. Appropriate applications of the technique successfully contribute to increasing mine safety (eg mitigating the risks for rock bursts). This paper presents a simulation technique to estimate the extent of damage zones induced by destress blasting. A 3D numerical model with a single blasthole is constructed with FLAC3D code, whereby static analysis is carried out in which horizontal and vertical stresses are applied to the model in order to simulate stress conditions in a deep underground mine. Afterwards, dynamic analysis is performed whilst applying time-varying blast pressure along the blasthole. The blast pressure profile takes into consideration an instantaneous increase and gradual decrease in blast pressure as well as the propagation of detonation along the blasthole. In this paper, constitutive models allowing for shear and tensile strengths of rock mass that are dependent upon strain rate are newly implemented into FLAC3D code with C++ programming language, whereby blast-induced damage is assessed by the extent of tensile and shear failure zones. The results obtained from the dynamic analysis show significant decrease in the extent of yielding zones induced by destress blasting with increasing the magnitude of in situ stress. They indicate that considering in situ stress state is indispensable in order to propose proper blast design for each level in underground mines. Based on these results, the relationship between the extent of the damage zone and depth at which destress blasting is performed is established. Furthermore, stress dissipation induced by destress blasting is examined with static analyses, in which the modulus of elasticity is decreased according to a rock fragmentation factor, a, only for the area where tensile and shear failures takes place during the dynamic analyses. A model parametrical study with respect to a is undertaken. The relationship between the rock fragmentation factor and the stress dissipation in the yielding zones is developed from the parametrical study. The developed relationship can be useful when destress blasting is designed for massive rock masses.CITATION:Sainoki, A and Mitri, H S, 2015. Simulation of damage zones induced by destress blasting, in Proceedings 11th International Symposium on Rock Fragmentation by Blasting, pp 245–256 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Aug 24, 2015

By D J. Thobaven

Project financing challenges caused by declining commodity prices have heightened the need to improve project metrics such as minimising capital payback and maximising net present value (NPV). Khnaiguiyah Mining Company (KMC) recognised the need to improve early cash flows at their Khnaiguiyah zinc-copper project after initial studies showed the project could not provide the desired financial outcomes necessary to compete for investment dollars in the current market.A significant improvement to the project cash flow in early years was realised through simultaneous optimisation of cut-off grade, mining sequence and strategic stockpiling. This improvement was achieved by accelerating mining in the pits, stockpiling lower value material and feeding higher value material directly to the plant. The schedule required higher mining and stockpile capacities than the sequential mine plan although minimal change occurred in the definition of ore and waste as measured by tonnes and grade. The practicality of the strategic schedule was demonstrated by implementation in the production plan with excellent conformance.CITATION:Thobaven, D J, 2014. An application of simultaneous optimisation – improving the payback period of a zinc-copper open cut project, in Proceedings Orebody Modelling and Strategic Mine Planning Symposium 2014 , pp 137–144 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Nov 24, 2014

By Y Cui, G Cao, H Zong, J Zheng, J Wang, S Yue, J Zhao, B Li

The iron (Fe) concentration in nickel slag is high, making its recovery and utilisation a hot and challenging topic. In the nickel smelting process, using CaO instead of SiO2 to adjust slag’s composition, and then directly reducing Fe from the output of molten slag is a more feasible method in practice for Fe recovery in nickel slag. The adjustment of slag’s composition in the nickel smelting process will have an impact on the distribution of impurity elements such as Pb, Zn, and As. In this paper, the slag adjustment of nickel smelting process was carried out with the oxygen-enriched topblowing system. The distribution of impurity elements in the smelting products with each composition slag was conducted by using thermodynamic software. Experimental verification was performed with rich oxygen top-blowing, and the distribution characteristics of impurity elements in the smelting products were characterised to explore the differences in mineral phases of impurity elements with each smelting conditions, elucidating the distribution patterns of Pb, Zn, and As in each phase. The results indicate that after slag adjustment, impurity elements are more easily to volatilisation and removal into the gas, reducing their total concentration in slag and matte. In the slag, the existence form of Pb, Zn is changed from silicate to ferrate, and As is changed from arsenate to low-valent arsenic oxide. It is easier to decompose the impurity-containing mineral phase and remove the impurity elements in the subsequent iron extraction process.

Jun 19, 2024

By G Reynolds, L Lu

Queensland Pacific Metals (QPM) is developing its flagship Townsville Energy Chemicals Hub (TECH) facility to produce nickel and cobalt sulfates from imported high-grade nickel laterite ore for nickel-cobalt-aluminium and nickel-cobalt-manganese Li-ion batteries. The TECH facility is expected to use a patented leaching, recovery and recycling process called the DNi Process™. It uses the nitric acid leaching process for extracting nickel, cobalt and other valuable metals from laterite ore. Apart from the nickel and cobalt sulfates, the TECH Project will also produce high purity alumina, hematite (~640 ktpa) and other by-products. The hematite by-product precipitated from the DNi Process™ (hematite precipitate hereafter), is high in Fe and particularly low in impurities. QPM is working to identify potential applications for these high-purity by-products to generate additional revenue and to achieve its zero-waste ambition. As such, QPM has engaged CSIRO to evaluate various agglomeration options to produce low-cost, hematitic products suitable as ironmaking feed materials. This paper will first reveal the unique characteristics of the hematite precipitate arising from the TECH Project and then examine the sintering and pelletising processes for preparing the hematite precipitate as an ironmaking feed.

Sep 18, 2023

By A Späth

Active explosion barriers are used to contain and suppress a methane gas and/or coal dust explosion propagation, including localised mine fires. Early research indicated that these active explosion barriers could be used on a mechanised cutting machine, making it the only technology that could suppress an explosion during its propagation phase at the cutting face. Thereby not only adding protection during mining operations but also reducing the development of toxic gases, fatal temperature increase and fatal overpressure build-up. During their evolution over the last two decades, further advantages of Active explosion barriers were proven in mine operations in countries such as South Africa and China. Characteristics such as effectivity, efficiency, mobility and space restrictions allow these systems to be used in various locations within a mine operation, including highwall operations. This improved technology contribution to safer coal mining was recognised by mine operators, workers and regulators. The Queensland’s Recognised Standard 21 (Resources Safety and Health Queensland (RSHQ), 2021) stipulated active explosion barriers as a technology to ensure the safety within their underground coalmines. The mobility and flexible implementation of the active explosion barriers allow a feasible and practical solution for all critical locations for explosion and fire management in an underground coalmine. The conformance to the Recognised Standard can be achieved all underground coal mining operations such as Bord and Pillar, and Longwall mining. This paper describes the conformance, implementation and use of Active explosion barriers for Longwall mining and development as well as Bord and Pillar mining in Queensland’s underground coalmines.

Oct 12, 2022

Subsidence prediction methods can be divided into two basic types: empirical and analytical. Empirical methods are based largely on a combination of experience, intuition and observations from a large number of observed subsidence data. They are simple in their application, but they are not based on any principle of mechanics. Analytical methods are based on equivalent material modelling principles, where the ground surrounding the extraction is mathematically represented by idealised materials that deform according to the principles of continuum mechanics. Wide applicability and improved understanding are the principal advantages claimed for this approach. In NSW, significant progress has been achieved in the area of subsidence prediction during the last ten years. Empirical prediction curves are available in NSW for predicting the ground movement in the Southern and Newcastle Coalfields. A' comparison of subsidence values predicted from these curves with the actual monitored values indicates that the curves are adequate for use in conjunction with the existing damage assessment methods. Future research for improving the existing empirical prediction curves can only be justified if it is cost effective. There are several numerical models currently in use in NSW for predicting subsidence. The models do not appear to be superior in their ability to predict subsidence when compared with the empirical curves. the performance of these models will no doubt improve along with advances in defining strata properties and failure mechanisms.

Jan 1, 1991

By M Brady, O Salisbury, C Claassen, B Beamish, H Baxter

Monitoring for signs of elevated spontaneous combustion in the active goaf behind the longwall face is crucial for early intervention of any potential heating incidents. Typically, trigger action response plans (TARPs) are established at the start of longwall mining and often are only reviewed when triggers are activated frequently and become ‘nuisance alarms’, or in the worst case scenario after a heating event. Narrabri Mine has adopted a proactive approach of reviewing the longwall active goaf spontaneous combustion monitoring TARP at the commencement of each longwall block and prior to reaching a known ‘mylonite zone’ of crushed coal, which is associated with increased coal oxidation. The review is based on information recorded from the monitoring data of the previous longwall blocks as well as interpretation of relevant spontaneous combustion propensity results for the coal. This has resulted in the mine implementing a zoned TARP approach for each longwall block covering both the longwall retreat through normal undisturbed coal and the zone where the goaf is influenced by the increased coal oxidation associated with the mylonite coal. The consequence of this approach is that it has removed the issue of having nuisance triggers being activated during mining through a known increased oxidation zone, while maintaining a high awareness of indicators of spontaneous combustion.CITATION:Baxter, H, Claassen, C, Salisbury, O, Brady, M and Beamish, B, 2015. Implementation of a zoned trigger action response plan associated with changing mine conditions at Narrabri Mine, in Proceedings The Australian Mine Ventilation Conference, pp 113–120 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Aug 31, 2015

By M N. Mian

The Cayeli volcanogenic massive sulphide deposit is located in Black Sea region of Turkey. Cayeli Bakir Isletmeleri (CBI) was formed as a joint venture between Inmet (Canada) and Eti Holding (Turkey). CBI is processing complex copper/zinc sulphide ores. The Cayeli concentrator was commissioned in 1994, with a design capacity of 600 000 tonnes/year. The grinding circuit was a classical rod mill/ball mill combination. The rod mill was converted to a primary ball mill in May 1997. The reasons for conversion were to use mill power more efficiently and eliminate feed interruptions during rod charging. The grinding circuit now has a primary ball mill in open circuit, and a secondary ball mill in closed circuit with hydrocyclones. Classification was originally done in two stages with hydrocyclones. Both mill discharges were combined and pumped to a primary cyclopack having two, 500 mm diameter cyclones. Primary cyclone overflow was pumped to a secondary cyclopack for further classification. The secondary cyclopack had ten, 200 mm cyclones. Both unit underflows were fed to a secondary ball mill by gravity. The secondary cyclone overflow, typically 62-64 per cent passing 36 ¦m, at around 42 per cent solids (w/w) was fed to flotation circuit. In order to increase circuit capacity to 1 000 000 tonnes/year and to improve control over the grinding circuit, it was decided to convert the classification circuit from two stage to one stage. For this purpose, six Warman Cavex cyclones having 400 mm diameter were installed. Now, the combined primary and secondary ball mill discharge was classified in single stage with the Warman cyclones in closed circuit with the secondary ball mill. The cyclone overflow goes directly to the copper rougher circuit as flotation feed. The grinding circuit is producing a flotation feed of 70 per cent passing 36 ¦m. The change over to single stage classification at Cayeli was successful. A mill throughput of 1 000 000 tonne/year is being achieved with better control over classification, and improved grind size. Results and comparison between the two classification circuits are presented in the paper.

Jan 1, 2003

By J Jackson, D C. Curry, A C. Keith

Australian Bureau of Statistics (ABS) data shows that since 2006 the balance of capital costs associated with mine development has become weighted greater than 75 per cent to enabling infrastructure and less than 25 per cent on process plant, equipment and machinery. The well-developed mine planning techniques including pit optimisation and scheduling and process plant design and optimisation have typically proceeded in a linear fashion from geology, mine planning, through processing, with some iterations back to the mine. Only then is attention paid to planning and costing of enabling infrastructure in any detail.Optimisation has typically involved developing cut-off grades, mine plans and schedules and ore processing that produces maximum economic ore. Infrastructure has then been sized æto suitÆ the mining and processing core business drivers.What the ABS data, together with global trends, indicates is that infrastructure can no longer be an afterthought if miners wish to achieve robust project value estimates, realistic ore to market costing and resource statements based on achievable economic cut-off grades.In addition, with such a large proportion of the costs originating from non-mining or processing activities, a mining operation can only be truly optimal if these aspects are given due consideration in the mine design stages, based on quality geometallurgical inputs so that the size and cost of enabling infrastructure is optimised to maximise project value.This paper outlines an iterative holistic approach to mining project development, which integrates the geometallurgical approach and enables infrastructure planning into project design and evaluation. It presents case studies showing reduced overall costs, increased net present value (NPV) and reduced project delivery timelines.CITATION:Curry, D C, Keith, A C and Jackson, J, 2013. Is a 2.5 per cent success rate good enough? Traditional mine development methods have the tail wagging the dog, in Proceedings The Second AusIMM International Geometallurgy Conference (GeoMet) 2013 , pp 113-128 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Sep 29, 2013

By G Hancock, A Kemp, I Taylor

The mine site is situated south east of Port Hedland in the Pilbara region of Western Australia. A detailed engineering design was undertaken for six irregularly shaped waste rock dumps at the site. The design was undertaken to ensure long-term stability of the dumps and optimise material movements for closure construction. The Watershed Erosion Prediction Program (WEPP) and SIBERIA landscape evolution modelling was undertaken for the landforms using regional and site-specific material characteristics to provide an indication of the maximum allowable slopes and corresponding slope lengths for long-term stability.The majority of the waste materials present in the waste rock dumps are considered likely to be subgrade banded iron formation (BIF) and chert, which is generally competent material. The presence of a large number of dolerite dyke swarms indicates appreciable volumes of dolerite were produced as waste from the pits. Geological records also indicate that the dolerite is logged as being heavily kaolinised in the geological logging, hence kaolinite is very likely to be present within the waste dumps on-site. This material has an elevated risk of being highly erodible/dispersive. Clays and shales are also likely to be present in the dumps and are considered to be less competent material (relative to the BIF/chert).The final design configuration of the largest existing ‘unplanned’ waste rock stockpile is compared to a hypothetical continuous ten degree outer embankment alternative. This paper will present a brief description of the alternatives for closure of three waste dumps in close proximity to each other and how the detailed design based on site-specific materials and conditions compare to an alternative approach blankly assuming that most materials will be erosionally stable at less than ten degrees.CITATION:Kemp, A, Taylor, I?and Hancock,?G,?2014. Landform design for Pilbara mine site – why plan and design waste rock dumps for closure based on site-specific conditions?, in Proceedings Life-of-Mine 2014 , pp 47–54 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Jul 16, 2014

By J J. L du Plessis, A Spaeth

The objective of this paper is to demonstrate the efficiency of machine-mounted active explosion suppression systems (AESS). Their proven effectiveness against methane explosions is highlighted, as well as their additional potential when used as an early warning system should frictional sparks occur during the coal-cutting operation at the coalface.Methane and coal dust explosions remain one of the primary and unpredictable risks in underground coal mining. Although technologies such as proactive ventilation, methane monitoring and wet-head cutting systems are widely used as preventative measures, there are no fail-safe controls able to reduce the risk to zero. None of these controls is designed to be able to counter the negligence associated with human error or to deal with rapidly changing face conditions which could lead to flammable gas ignitions. Machine-mounted AESS have been installed on various continuous miners as an additional control to further reduce the risk of ignitions and to act as a safety shield when all the other precautions taken have failed. This was the primary motive for the development of AESS. As an additional feature, the ExploSpot® AESS is being enhanced to identify frictional ignition sources at the coalface that reach a predefined monitoring level. This functionality may serve as a further effective preventative measure.As part of ensuring coalmine operational safety, the detection of frictional ignitions can play a huge role in creating heightened awareness of potentially dangerous situations. By being able to process the frictional spark burning pattern, the AESS could support the decision-making process regarding whether to continue the coal-cutting operation, whether there is a need for maintenance or whether the operation should be stopped completely due to the increased risk of a flammable gas ignition and an ensuing explosion.CITATION:Spaeth, A and du Plessis, J J L, 2015. Enhanced risk reduction using early warning technologies embedded in active explosion suppression systems, in Proceedings The Australian Mine Ventilation Conference, pp 315–322 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Aug 31, 2015

By R Wedmaier

Henty Gold Mine is an underground operation producing 100 000 ounces of gold per annum. The original mining method used was flatbacking with paste fill. However, a non-entry method û uphole benching û was introduced in 1998 to recover the deteriorating crown pillars. Initial challenges were encountered, including poor drilling accuracy, difficulty in charging holes and poor rise performance. These challenges were overcome by a Continuous Improvement Team, consisting of mining engineers, supervisors and miners. Following this, the Team implemented further changes aimed at reducing ore loss and dilution. A second area addressed by the Team was development blasting, where innovative detonator technology known as the ExelTM DevelDetTM was introduced and rapidly accepted by the operators. Continued use of the ExelTM DevelDetTM resulted in a number of benefits, such as easier bogging, less perimeter damage, faster charging and hook-up and decreased wastage. Finally, improvements were made in the area of explosive handling, where competency-based training of operators was conducted and manual handling of ANFO has been significantly reduced by the introduction of the MiniMoverTM into the underground magazine.

Jan 1, 2001

By Morrison AL

The volume changes occurring during the reduction by hydrogen of single, high-grade hematite pellets under isothermal and non- isothermal conditions were investigated using a time-lapse photographic technique. The major cause of swelling during iso- thermal reduction at temperatures in the range 600-1000¦C, was the disruptive stresses set up during the transformation of hematite to mag- netite. The severity of the disruption was temperature dependent; the degree of swelling increased with increases in the reduction temp erature. Under non-isothermal reduction conditions, in which the pellet temperatures were increased at constant heating rates over the temperature range 500-1000¦C, minor swelling (6-8%) occurred in the initial stages of reduction followed by pellet contractions of up to 15% in the final reduction stages. The low initial swelling and high final pellet shrinkages were due to a com- bination of the temperature dependence of the swelling mechanism, and the sintering of fine pores in the reduced iron in the latter stages of reduction at higher temperatures.

Jan 1, 1980

By D Pienaar, B McFadzean, C T. O’Connor

Collector and frother molecules in sulfide mineral flotation are thought to have separate roles in the flotation process, where the collector selectively hydrophobizes the mineral surface and the frother adsorbs onto the bubble surface, inhibiting bubble coalescence. Molecular interactions have been observed between these molecules which allow them to co-adsorb at each of the respective interfaces, improving pulp phase recovery. This study investigated the molecular interactions between sodium diethyl dithiophosphate (SDEDTP) and hexanol, and the mechanism by which these interactions improve the recovery of galena. In this study each of the respective interfaces were investigated for the adsorption of SDEDTP. The adsorption of SDEDTP on galena was measured by UV-Vis spectrophotometry and isothermal titration calorimetry. At the air-water interface foam stability and pulp bubble coalescence was used to quantitatively discern if there are interactions between SDEDTP and hexanol. Microflotation and induction time experiments were performed to determine the effect of SDEDTP, hexanol and their mixtures on mineral recovery and bubble-particle attachment. The results of all these tests were collated to determine if SDEDTP-hexanol mixtures synergistically improve sulfide mineral recovery. It was found that SDEDTP reduced bubble coalescence in the pulp demonstrating activity at the airwater interface. Together SDEDTP and hexanol synergistically enhanced foam stability. This indicated that there are attractive molecular interactions between these reagents. Even though DTP did not adsorb onto galena, DTP-frother mixtures synergistically improved the recovery and attachment probability of this mineral. Adsorption studies with aeration showed that DTP in the presence of a frother transfers from the air-water interface to the galena mineral-water interface during bubble-particle collision. It was proposed that the transfer of hexanol through the thin film enhanced the kinetics of film thinning between the bubble and the particle by destabilising the thin film thus improving pulp phase mineral recovery.

Aug 24, 2022

By Smith N. S, Rholl S. A, Otterness R. E

The Bureau of Mines is conducting research to develop blast designs for the control of fragmentation and rock displacement and the reduction of overall mining costs. This Bureau experiment, a program of 25 reduced-scale blasts conducted at the University of Missouri-Rolla's experimental quarry, determined the effects on fragmentation of changes in burden, spacing, explosive diameter, and stemming-explosive length. Combined with previous work at this experimental quarry examining air gap, delay, and decoupling effects, this experiment provided a data base of over 50 shots. The fragmented rock, 1000 to 3000 kg from each 3- to 4-hole blast in the 1 m bench of massive dolomite, was completely screened to assess fragmentation. The explosive used for all the tests was extra dynamite initiated by seismic caps fired from a sequential blasting machine with delays of 5.2- to 13-ms/m of burden. Empirical equations were developed which predicted the effect of shot design parameters on the fragmentation. These formulas predicted the average fragment size of each of the 25-shots to within 13 percent. For spacing to burden ratios of 1, 1.4 and 2, there were optimum burden to explosive diameter ratios (B/de) of 30, 26 and 20, respectively. Decreasing the B/d. ratio beyond these levels resulted in little or no reduction in the average size. For B/de ratios above these levels the average size increased but, at least a 15 percent reduction in the average size results when the shot design parameters are optimized.

Jan 1, 1990

By R LEATHAM

Long hole drilling underground has come a long way since 'down the hole' surface drills were first taken underground. The underground miners have progressively developed the use of long hole stope drilling from the 'quarry blast' to a precise and selective removal of ore while leaving the remaining structure as undamaged as possible. To reach this position has involved the development of equipment and explosives to give the desired results. This paper reviews some aspects of long hole stope production drilling.

Jan 1, 1992

By C M. Stewart, P Bowen

McArthur River Mine, in the Gulf Country of the Northern Territory, exploits one of worldÆs largest deposits of zinc, lead and silver. It was officially opened in September 1995, 40 years after its discovery in 1955. This paper describes the geological, engineering and metallurgical assessments which made up the feasibility study. It then relates how these study components compare with the first two years of operation. The challenges that had to be overcome before the McArthur River Mine became economically viable included the metallurgical problems associated with extremely fine grained ore and the development of an effective mining process.

Jan 1, 1998

Minerals processing operations use a significant quantity of fossil carbon to provide energy and as reductant. Greenhouse gas emissions from the use of this fossil carbon contribute to the global increase in anthropogenic CO2 emissions to the atmosphere. In the spirit of fostering thinking about a step change in this situation, this paper explores opportunities for the use of sustainable carbon from biomass sources in a range of mineral processing operations. The exploration shows that many common processes are contenders for fossil carbon replacement and that feasible scientific research by minerals processing technologists would add to deeper understanding of the potential for a sustainable bio-carbon based industry. It is also shown that there are many dilemmas on the social, environmental and economic fronts to be solved before a biomass based industry could become a reality.

Jan 1, 2004

By Y Tang, G Si

The analysis of rock fragment sizes plays a pivotal role in various stages of geotechnical and mining engineering. Traditional case-based methods often falter in addressing the variability in rock fragment images due to fluctuating geological conditions and imaging environments. To counter this challenge, AI-based models necessitate training on expansive and diverse data sets, encompassing a wide range of conditions and fragment types. However, the creation of such data sets is often hampered by financial and environmental constraints. This paper introduces a novel approach for rock fragment recognition, which leverages the power of AI for data augmentation, automated annotation and prediction. The methodology integrates three key components: Style-Based Generator Architecture for Generative Adversarial Networks (StyleGAN) for data augmentation, the Segment Anything Model (SAM) for automated annotation, and the multiscale version of you only look once (YOLO-MS) real-time instance segmentation model. The proposed method begins by training a new StyleGAN model using the original data set. It then delves into exploring the latent vectors of different rock fragment images to generate specific ‘style representations’. These latent vectors form the foundational elements of each image in a high-dimensional space, encapsulating critical features such as size, shape, texture, and colour variations. The next step involves conducting quality checks on the generated images to ensure their realism and suitability as training data. Subsequently, the augmented image set is annotated using modified SAM module, followed by training the YOLO-MS real-time instance segmentation model. The performance of this model, trained on the augmented data set, is compared with a model trained on the original data set, particularly in terms of generalisation abilities in new scenarios. The results exhibit a significant enhancement in accuracy, robustness, and generalisability of the model trained on the augmented data set, underscoring the potential of StyleGAN as a powerful tool for data augmentation in geotechnical engineering image analysis.

Sep 1, 2024