Search Documents

By G J. Harbort, K Lam, C Sola

The relationship between the geology of porphyry copper deposits and their processing performance is an area of interest. Being able to predict comminution properties of the ore from physical characteristics and geological descriptions would allow an increase in the precision of capital cost estimates, operating cost estimates and throughput rates. This geometallurgical approach would reduce the requirements for expensive comminution tests.As the comminution circuit is the highest capital cost item in the plant and a possible throughput bottleneck, understanding the variability in hardness within the orebody is critical. A geometallurgical model allows calculation of ore hardness values in mining blocks from ore characterisation data. This model subsequently uses a combination of simulation and power-based approaches to relate ore hardness and flotation feed size to grinding circuit throughput.This paper reviews the geometallurgical models developed for seven porphyry copper deposits and discusses the strengths and shortfalls of the methodology and similarities and differences between the different deposits.CITATION:Harbort, G J, Lam, K and Sola, C, 2013. The use of geometallurgy to estimate comminution parameters within porphyry copper deposits, in Proceedings The Second AusIMM International Geometallurgy Conference (GeoMet) 2013 , pp 217-230 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Sep 29, 2013

By K L. Olzard

At the WMC Resources Phosphate Hill operation, monitoring the quality of the ore delivered by the mine to the beneficiation plant is of prime importance to ensure the fertilizer is produced cost-effectively through the various stages of production and meets customer quality requirements. P2O5 and Fe2O3 are among the critical-to-quality elements. Fe2O3 is a key performance driver as it is an impurity that adversely impacts some of the chemical processes. The ore is supplied to the plant on a conveyor belt that is equipped with an automatic sampler. Increments are taken off the flowing ore stream at regular time intervals. A variographic experiment was conducted to investigate sampling strategies capable of estimating ore properties over a 12-hour shift to a nominated level of precision. The combination of individual increments into a shift composite was included among the strategies that were assessed. The experiment involved collecting data at regular time intervals and preparing variograms to characterise the feed-grade variability on the conveyor belt. Duplicates were also collected to assist in quantifying the precision of the subsampling stages of the sampling protocol and assay accuracy was monitored with matrix-matched standards. Various sampling schemes were evaluated by combining geostatistical techniques and the sampling theory of Pierre Gy. The following parameters were varied: the number of increments, the increment-size reduction leading to the composite, the masses of the subsequent sampling stages and the number of analytical determinations. This paper describes the planning of the variogram experiment, the various phases of data collection and the methodology used to process the data. It discusses the development of some of the sampling schemes that were considered.

Jan 1, 2005

By Thompson IR, Solomons SS

Underground mining with the cut and fill method has been declining during the past decade or so, being replaced mostly by sublevel open stoping methods. During 1983 the Australian Mineral Industries Research Association Ltd. (AMIRA) commissioned a review of the method, with the aim of compiling the state-of-the-art, indicating future trends and recommending topics for research that would be needed to support this future mining. Mines using the cut and fill method in Australia and three other countries were visited to compile current practices and expected developments.Cut and fill is mostly used where geological and ground conditions are unsuitable for open or caving methods. These range from narrow, irregular orebodies to wide, massive orebodies where pillars are retained in ore. There is significant potential for improvement and further development of the method. It may then be used in a broader range of conditions and for orebodies that may not otherwise be viable to mine.

Jan 1, 1985

Study on Pure Mineral Flotation Tests of Low-Grade Phosphate Ore in Zhijin, Guizhou Province, China

Sep 13, 2010

By A J. Moore

Blasting operations in open cut mines results in airblast, ground vibration and rock movement with the potential for wild flyrock. In some mines there is a trend towards blasting closer to established infrastructure as extraction boundaries alter with increased resource prices and changed economic circumstances. Mine infrastructure in close proximity to blasting has the potential to be damaged by ground vibration and wild flyrock especially. To provide guidance for the determination of appropriate buffer distances and blast specifications, this paper describes methodologies that may be used to determine the potential for damage to infrastructure by predicting:dynamic blast;near field ground vibration levels;dynamic blast strains induced in infrastructure by consideration of ground vibration,ground strain and response amplification;potential for damage by comparing induced strains with allowable working strainsfrom structure design codes and serviceability requirements; andflyrock throw distances based on models developed from flyrock investigations.This report demonstrates the use of strain prediction methodologies developed during Australian Coal Association Research Programme (ACARP) Project C14057 for infrastructure such as conveyor belts, transmission poles and towers, buildings and bridges.

Nov 8, 2011

By S Watson, L Martinez

Mine projects are complex opportunities that demand a constant assessment of uncertainty, ie risk and upside potential. This is because the value of a mine project is typically influenced by many underlying economic and physical uncertainties, such as metal/mineral prices, metal/mineral grades, costs, schedules, quantities, and environmental issues, amongst others, which are not known with much certainty at the beginning of the project. But, is uncertainty an enemy or an ally of the mine planner/analyst? To answer this question, this paper first assesses and elaborates upon the general concept of uncertainty. It then looks at the different sources of uncertainty in a mining project, such as the orebody, costs and product price uncertainties, and their effect on the final project value. Next, a general?review?of current mine evaluation techniques highlighting their limitations when dealing with uncertainty is made. Finally, an extended evaluation and risk management framework that addresses the existing uncertainty in the (greenfield/brownfield) mine evaluation process is proposed. One important characteristic of this framework is that it uses and manages uncertainty in order to generate additional value to the project; this is done through operational and managerial flexibility, which allow mine planners to minimise/maximise project risk/upside-potential giving a more realistic estimate of the projectÆs market value - this is shown in practical and simple examples occurring at different stages of a mine project, such as exploration and operation. Final conclusions are provided at the end.CITATION:Martinez, L and Watson, S, 2012. Uncertainty - An important source of (additional) value for mine projects, in Proceedings Project Evaluation 2012 , pp 131-140 (The Australasian Institute of Mining and Metallurgy: Melbourne).

May 24, 2012

By R G. Skirrow, M Cooper, J Claoué-Long, K Czarnota, H Dulfer, J Thorne, D C. Champion

"An EXTENDED ABSTRACT is available for download. A full-length paper was not prepared for this presentation. Tholeiitic intrusion-hosted nickel (PGE-Cu) sulfide deposits are under-represented in Australia’s inventory of known resources of this deposit type, relative to regions globally with comparable geology. To support discovery of major new deposits, a continental scale analysis of Australia’s potential for tholeiitic intrusion-hosted nickel (PGE-Cu) sulfide deposits is being undertaken. The analysis applies a mineral systems approach to map the four essential components of the ore-forming systems:1. sources of ore constituents2. crustal and mantle lithospheric architecture3. energy sources or drivers of the ore-forming systems4. gradients in ore depositional physico-chemical parameters.The results reveal several regions of Australia with potential for discovery of large tholeiitic intrusion-hosted nickel (PGE-Cu) sulfide deposits, mostly around the margins of Archean and Proterozoic lithospheric blocks. These include the south-east and northern margins of the Yilgarn Craton, the south-west and south-east margins of the Kimberley Block, the southern margin of the Arunta Province in central Australia, the western flank of the Gawler Craton in South Australia, and possibly some areas within the Palaeozoic Lachlan Orogen of eastern Australia.CITATION:Skirrow, R G, Dulfer, H, Czarnota, K, Champion, D C, Thorne, J, Cooper, M and Claoué-Long, J, 2015. Australia’s potential for nickel sulfide ore systems – a new continental-scale time-space framework for exploration, in Proceedings PACRIM 2015 Congress, pp 679–684 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

Mar 18, 2015

By Kendall W. T C

This paper provides a brief overview of the history of the Chaffers Dewatering Project which was successfully commissioned in August 1993. Until the end of 1992 the `Superpit' and the deep mines of the Golden Mile had all been dewatered by a staged pumping system from the 28 level in the Lake View Shaft, which was removed at that time when the Lake View Mine was absorbed into the Superpit. Feasibility analysis had indicated that adaptation of oil industry technology offered the greatest potential for a low cost solution to the problem of dewatering the open pit and deep mines. A KCGM engineer and one of the authors visited the Jackson oilfields to look at the deep well pumping technology in use there in early-1991. In the final analysis a new dewatering station was established in the 23 level crushing station of the Chaffers Shaft using conventional plunger pumps, but the rising main was screwed 178 mm OD (7 inch) casing the suspended from the surface to a depth of 686 metres, and was installed using oilfield personnel and methods. Only the second installation of its kind in Australia, the project achieved the dewatering objective at about half of the capital cost of the more conventional approach and at lower operating costs as well.

Jan 1, 1995

By D Davis

The prediction of ground-borne noise levels from tunnel construction activities requires knowledge of the input vibration forces from the vibration source(s) into the cutting face, knowledge of the vibration transmissibility through the surrounding medium, and knowledge of the receiver building’s vibro-acoustic response. The conversion of this vibration energy into sound pressure within the receiver building depends upon the vibration coupling of the soil and the building’s foundations, the structure’s response including floor-to-floor attenuation and amplifi cation due to resonance, the radiation effi ciencies of the internal structural elements, and the room’s acoustical properties. Beginning with a known input vibration spectrum at the cutting face, it is possible to predict the resultant ground-borne noise levels within a receiver building using either analytical and/or empirical techniques, or by numerical modelling. This paper presents an overview of the various methods available to estimate ground-borne noise levels from tunnel construction activities and discusses the advantages and limitations of each.

Mar 8, 2011

In a ætypicalÆ gold deposit each tonne of ore will contain tens of millions of very fine gold particles, whereas; at Bendigo the majority of gold in each tonne of ore is present in just five to 15 very coarse gold particles, a distribution more akin to diamond deposits. At Bendigo the minimum sample size required for good assay repeatability is estimated at 4.5 t for each assay. Drill core samples are several orders of magnitude too small to generate representative assays.The drilling data are characterised by a nugget effect approaching 90 per cent, a very skewed distribution and, on average, under-report in situ grade by 50 per cent to 75 per cent. Since 2002 Bendigo Mining NL has completed 47 000 m of diamond drilling, 1200 m of on reef development, collected and processed 100 bulk samples (150 t each), 800 mini bulk samples (50 kg each) and processed in excess of 25 000 t of ore through a pilot plant. The results of this work have provided the information required to develop and verify a robust grade estimation procedure. No standard geostatistical manipulation of the drilling assay data can generate an accurate grade estimate as the samples and associated assays are not representative of the ore being sampled. The recognition of this fact was critical as a radical transformation of the raw drill assay data was required to develop a workable methodology for grade estimation in such a high nugget environment. The aim of the transformation was to reduce the nugget effect and high variance and so approach a lognormal assay distribution. Ideally the average drill grade would approximate actual in situ grade and the assay distribution would reflect actual grade distribution within the reefs. Following the data transformation simple block modeling within the defined resources allows grade estimates to be made that can be used in the estimation of Indicated Resources and hence Probable Reserves. The grade estimates have been checked by extensive bulk sampling and processing of 25 000 t of ore (three per cent of total resource) through a pilot plant with planned full scale mining to provide the ultimate verification.

Jan 1, 2004

The construction of mine headframes of reinforced concrete is not a new departure as many structures have been built of this, material particularly in Canada and South Africa. This type of construction has been satisfactorily tested in service and is now accepted by the mining fraternity.The two headframes at New Broken Hill Consolidated Limited are the first structures of this type to be built in Australia, and provide an interesting application of reinforced concrete design and construction.Reasons for Adoption of ConcreteDuring 1947 the writer visited D.S.A. and Canada and inspected three headframes, of reinforced concrete construction.Leading details are as follows:(i) Hollinger Gold Mines. (Timmins, Ontario) : Height of sheaves 110 ft. Shaft has provision for 4 skips each of 12,000 lb. capacity. Structure incorporates surface crushing plant.(ii) International Nickel Corporation-Murray Shaft (Sudbury): Height to sheaves 153 ft. 6 in. Shaft provides for 2-12 ton skips and 2 cages. Surface crushing plant built integrally with frame.(iii) Asbestos-Johns Manviae (Quebec): Height of sheaves 116 ft. Shaft provides for 2 skips 200 cu. ft.

Jan 1, 1956

By S Wilkinson, S Black, L van den Berg

"The mine ventilation system plays an important role in mitigating human exposure to diesel particulate emissions (DPM), exhaust gases and heat in underground mines. The mine’s secondary ventilation systems are critical for diluting contaminants when working at the face of a heading. To improve these systems it is important to understand and visualise the localised air flow profiles. We have considered an approach that includes both tracer gas measurement–modelling to analyse this complex flow behaviour in a heading. The use of tracer gas enables reliable measurements of flow behaviour and differentiation between the contributions of various contaminant sources including DPM. A pilot study was conducted to demonstrate how tracer gas can be used to directly measure the effectiveness of ventilation dilution in a heading during shotcrete activity. This activity is considered to be one of the more challenging areas for diesel exhaust exposure management. Exposure sources include the diesel engine of the spraying equipment (Spraymech) and concrete agitator truck (Agi) that supplies the shotcrete. During operation both the Spraymech and Agi operators are located outside the equipment’s air-conditioned cabins and both are exposed to diesel exhaust. In the pilot study tracer gas was injected directly into the Spraymech exhaust stream. This allowed thorough mixing, ensuring the tracer gas was able to track exhaust gases and particulates released to the heading. The release was repeated at three different headings with varying degrees of secondary ventilation controls. The study showed that small changes to the secondary ventilation system can significantly impact the contaminant concentration. It also demonstrated that tracer gas can be used to isolate different contaminant sources. The intent is to use this technique in a more comprehensive study to develop best practice guidelines for secondary ventilation installations. This paper presents the technique used and the preliminary results from the pilot study. CITATION: Wilkinson, S, van den Berg, L and Black, S, 2015. Application of tracer gas flow profiling to improve secondary ventilation practice – a pilot study, in Proceedings The Australian Mine Ventilation Conference, pp 345–352 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

Aug 31, 2015

Dynamic Simulation of an Alumina Refinery

Sep 13, 2010

By R F. Sandenbergh

Due to environmental considerations, mineral processing plants are increasingly recycling water to decrease the impacts of wastewater on environment and the depletion of fresh water resources. Industrial experience has shown that increasing recycling water in flotation typically negatively impacts the recovery and selectivity, especially in the selective flotation of minerals from complex sulfide ores. The major reason for the reduced selectivity is typically the inadvertent activation of sphalerite or pyrite by the heavy metal ions (HMI) in solution. These HMI are often produced during milling and flotation and absorbed by other minerals. In the normal flotation pH value range (8 - 10), these HMI are often in the form of precipitates or complex ions that may re-appear in solution or become active should conditions favour this. In this paper, microflotation, batch flotation and XPS surface measurement were used to investigate the use of HMI absorbents additions to the pulp to improve flotation selectivity of a complex sulfide ore. It has been demonstrated that HMI removal does improve flotation selectivity by facilitating more effective depression of the sphalerite. Microflotation and XPS surface measurements have shown that HMI absorbents compete with sphalerite to absorb HMI, and thus reduce the surface concentration of HMI on those sulfides. The practical use of HMI absorbents in complex sulfide ore flotation is also discussed.

Jan 1, 2005

Knowledge of the hydrogeological setting of a mine forms an integral part of a mining operation. Groundwater impacts, initially, on mine feasibility. Then, on commencement of mining, it interacts on the operational components of dewatering, water supply, tailings leachates, mine wall stability, stormwater run-off, water disposal, impact on regional environment and mine closure planning. The recording of hydrogeological information is frequently overlooked during resource and reserve definition as well as during the course of mining when prime opportunity exists for the collection and collation of this data. In recent times, the exclusion of hydrogeological data has been exacerbated by use of high air pressure drill rigs that exclude water inflow. The recording of water data has been further inhibited by the use of computer coded drill logs that do not include the site geologistÆs or drillerÆs comments on drilling problems and observations associated with groundwater intersections. Geologists now infrequently directly supervise drill rigs, but visit the rig only intermittently to evaluate drilling progress. They do not see first hand drill performance characteristics at water make zones. Drill cuttings, as drill chips and cores, are now mostly logged away from the drill site so that wet sample locations, visual water flow evidence and verbal comments from the driller are not recorded. As a result, mine geologists now do not direct adequate attention to the delineation of lithologies and structures that store and transmit water. This paper provides examples of water sighting and recording techniques applicable to a mining operation. Examples are also provided where hydrogeological conditions have been disregarded with serious impact on the progress of mining. The recording of hydrogeological data should be maintained and formatted by the site geologists into a useful database for review as and when required by specialists in hydrogeology, geotechnical and mining engineering and the environment. The information can be coded by the site geologist for inclusion into the mine database. A groundwater dataset, compiled during the progress of field investigations, will save the considerable cost of re-drilling and hole re-entry testing which would otherwise be required to provide base level hydrogeological information required for progress of mining. Apart from working mine use, a hydrogeological dataset is a necessary component for planning mining closure and mine legacy conditions.

Jan 1, 2003

By Casley G. E, Pritchard J

A single slurry fed fluosolids roaster was chosen to replace six multiple hearth roasters in the Copper Smelter at Mount Isa as part of the overall program to increase production from 100 000 to 155 000 tonnes blister copper per annum. The paper describes factors influencing the choice of the unit, problems encountered during the commissioning and details of the present roaster operation.

Jan 1, 1974

By G Le Juge, K Waldock

The current suite of bulk mining explosives has been available to the industry for over 20 years. The flexibility and performance of ammonium nitrate based emulsion blend explosives were accepted by the mining industry in the early 1980s. These water resistant explosives rapidly displaced the water gels and slurries developed during the previous two decades. However, since the 1980s there has been no equivalent step change to the basic formulation or application of bulk explosives, although there has been a continuous process of refinement with respect to stability, handling characteristics and most significantly, cost. In the 1980s explosives supply contracts were awarded based on technical evaluations, usually accompanied by field trials. Technical personnel from both mining companies and explosives suppliers dominated the negotiations and there was a strong focus on explosive energy and blasting requirements. Today, explosives from alternative suppliers are difficult to differentiate on the basis of energy so the focus has shifted to aspects of service and price û bulk mining explosives are now largely treated as a commodity and the negotiations are dominated by commercial personnel. Commercial pressures from large mining companies have seen a substantial reduction in the real price of explosives. However, this has come at a cost. In order to reduce manufacturing costs, emulsion formulations have changed. In general they are not only cheaper, but also less energetic. Powder factors have had to increase to achieve the same blasting results, especially in hard ground. It is debatable whether the mining companies have achieved all of the economies they were seeking. This paper discusses these issues and argues that the mining industry would benefit from taking a more technically driven approach to meeting their blasting requirements. Recent trends towards more open contracts between the mining companies and their suppliers can be extended to explosives. To achieve this, mining companies will need to acquire sufficient technical knowledge to take a more pro-active approach to specify the characteristics of the explosives required to maximise the economic performance of their operations.

Jan 1, 2004

By K F. Tang, J-W Li

"An EXTENDED ABSTRACT is available for download. A full-length paper was not prepared for this presentation. The Xiong’ershan district in the southern margin of the North China Craton (NCC) contains more than 20 gold deposits hosted in the late Neoarchean to early Palaeoproterozoic amphibolite facies metamorphic rocks or late Palaeoproterozoic unmetamorphosed volcanic rocks. Gold mineralisation consists mainly of auriferous quartz veins and disseminated ores in hydrothermally altered zones, with additional gold mineralised breccia pipes. Gold orebodies are strictly controlled by secondary- or higher-order brittle/ductile-brittle faults. Isotopic dating reveals two separate episodes of gold mineralisation: the Shanggong and Dianfang deposits formed in the range of 236 ± 1.4 to 202 ± 18 Ma, whereas the Gongyu, Qianhe and Qiyugou deposits formed between 135.0 ± 0.5 Ma and 124.0 ± 1.0 Ma. Stable isotope and noble gas geochemistry studies indicate that the Triassic gold deposits were precipitated from deeply-sourced, low-salinity fluids at 250–350°C, but whether the fluids were mainly derived from metamorphic devolatilisation or magmatic degassing remains unresolved. In contrast, the Early Cretaceous deposits most likely precipitated from a dominantly magmatic fluid, although metamorphic and meteoric components cannot be ruled out. Results from this study, when combined with regional geological and geochemical data, suggest that the Triassic gold mineralisation was generated by the deformation associated with the collision between the North China and Yangtze cratons, forming the Qinling-Dabie orogen. In contrast, the Early Cretaceous gold veining was a result of lithospheric destruction of the NCC, a process initiated by plate subduction from both the south and north since late Palaeozoic but ultimately triggered by the westward subduction of the palaeo-Pacific plate in the late Mesozoic. This study helps to reconcile the long-term debate on the timing, genesis and geodynamic setting of gold mineralisation in the Xiong’ershan and other districts along the southern margin of the NCC.CITATION:Li, J-W and Tang, K F, 2015. Gold deposits in the Xiong’ershan District, Southern North China Craton – products of Triassic deformation and Cretaceous craton destruction, in Proceedings PACRIM 2015 Congress, pp 341–344 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

Mar 18, 2015

By H Li

"An EXTENDED ABSTRACT is available for download. A full-length paper was not prepared for this presentation. The central Nanling region in South China experienced multistage granite intrusive events and corresponding metallogenic processes during the Early Yanshanian period (190–140 Ma). In this study, a wide variety of published data on age determination, petrological characteristics, major and trace elements compositions, whole rock Sr–Nd contents and zircon Hf isotopic composition of the representative granite composites were systematically collected to estimate their cooling-uplift rates, granite evolution processes and mineral deposit formation. These granites have been classified into two groups based on their age and geochemical variations: Group 1 (180–165 Ma) and Group 2 (165–150 Ma). Major and trace element geochemistry suggests that Group 2 experienced more intense magmatic evolution and differentiation than Group 1. In addition, Group 1 granitoids are classified as A1-type alkaline rocks with faster cooling and uplift rates, whereas Group 2 granites are considered as A2-type aluminous rocks with slower cooling and ascent rates. Furthermore, the Group 1 granitoids have younger Nd and Hf two-stage model ages but higher eNd(t) and eHf(t) values, indicating influence from crust–mantle interaction. In contrast, the Group 2 granites show different values, suggesting stronger upper crustal contamination or assimilation. The slowly cooled and ascended Group 2 granites experienced much heavier crustal contamination (or assimilation) and crystal fractionation, contributing to the large W–Sn mineralisation in the central Nanling region. The Early Mesoproterozoic (1.6–1.4 Ga) crustal basement of the Cathaysia Block in South China is the ore source bed for the world-famous Nanling W–Sn polymetallic mineralisation.CITATION:Li, H, 2015. Where is the source of the world-famous Nanling W–Sn polymetallic mineralisation?, in Proceedings PACRIM 2015 Congress, pp 619–622 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

Mar 18, 2015

By D Kentwell

IGO Ltd’s (IGO) Nova Operation (Nova) is 160 km east-north-east of the town of Norseman in Western Australia (WA). Discovered by Sirius Resources NL (Sirius) in 2012 and acquired by IGO in 2015, the Nova-Bollinger Deposit (Nova-Bollinger) occurs within the Fraser Zone of the AlbanyFraser Orogen. Nova-Bollinger is a complex nickel-copper-cobalt (Ni-Cu-Co) magmatic intrusive sulfide deposit, which is characterised by high-grade folded and undulating breccia lodes that host the bulk of the payable metals. Significant sulfide mineralisation is also found in several magmatic intrusive styles including net-textured, disseminated and blebby styles. A smaller but material volume of sulfides has also been remobilised into the deposit’s footwall rocks as ‘stringer’ bands in metasediments, or as discrete massive sulfide lenses. When interpolating Mineral Resource estimate (MRE) block grades in the presence of locally varying grade connectivity trends, such as folds or undulations, the sample search neighbourhood, which is used to identify the composites subset that is used for each block estimate, should be oriented to match the local trends. In theory, the most robust method of estimation where locally varying angles (LVAs) are applied to search neighbourhoods, are techniques where the composites and the blocks to be estimated are transformed to an unfolded coordinate system (UCS). However, UCS methods can be tedious to implement and require subjective decisions to be made that cannot always be fully defended. An alternative LVA method commonly used in the industry is dynamic anisotropy (DA), which is much simpler to implement and is often assumed (but not that often demonstrated) to result in similar results to UCS-style estimation. In 2018, Nova’s Mine Geology team (MGT) estimated the grades and density of the Nova-Bollinger MRE using the UCS approach. For the 2020 MRE update, Nova’s MGT tested and compared the USC grade interpolation method with the DA approach. As applied to the Nova-Bollinger’s 2020 MRE, the comparison, advantages and disadvantages of the two estimation methods are the subject of this paper

Mar 22, 2022