Patent Application
20250041876
The present disclosure relates to froth flotation of ores.
Froth flotation is used to separate components of mineral ore. In a froth flotation process, a slurry of ground ore is agitated in a large tank, target minerals are selectively rendered selectively hydrophobic and/or hydrophilic, and hydrophobic components are attached to bubbles which levitate to the surface and are removed. The efficacy of froth flotation is dependent on, among other factors, the types and amounts of chemicals used to treat the ore, the configuration and duration of processing in the tank, and other parameters.
Some aspects of this disclosure describe methods. For example, according to some mineral feed flotation methods, first measurement results are received from a characterization system in an ore processing plant. The first measurement results characterize a sample of mineral flotation feed. A first set of mineral characteristics of the sample of mineral flotation feed are determined based on the first measurement results. Respective similarities are determined between the first set of mineral characteristics and a plurality of predetermined sets of mineral characteristics that are each associated with one or more corresponding sets of flotation treatment parameters. A first set of flotation treatment parameters is selected based on the respective similarities. The first set of flotation treatment parameters is transmitted to a flotation processing system for separation of mineral components according to the first set of flotation treatment parameters.
Implementations of this and other methods can have any one or more of at least the following characteristics.
In any or all implementations, the plurality of predetermined sets of mineral characteristics include a plurality of clusters of characteristics.
In any or all implementations, the method includes receiving a plurality of measurement results characterizing a corresponding plurality of first mineral samples, determining a plurality of first sets of mineral characteristics corresponding to the plurality of measurement results, and clustering the plurality of first sets of mineral characteristics into the plurality of clusters of characteristics.
In any of the foregoing or other implementations, the method includes performing one or more measurement processes on the plurality of first mineral samples, to obtain the plurality of measurement results.
In any or all implementations, determining the plurality of first sets of mineral characteristics includes performing principal component analysis or linear discriminant analysis on the plurality of measurement results.
In any or all implementations, clustering the plurality of first sets of mineral characteristics includes passing the plurality of first sets of mineral characteristics through a decision tree.
In any or all implementations, a rule in the decision tree includes a decision based on a concentration of a first mineral component in the sample.
In any or all implementations, a first cluster of the plurality of clusters includes a first set of mineral characteristics corresponding to a first ore type and a second set of mineral characteristics corresponding to a second ore type that is different from the first ore type.
In any of the foregoing or other implementations, a first cluster of the plurality of clusters includes a first set of mineral characteristics corresponding to a first ore type, and a second cluster of the plurality of clusters includes a second set of mineral characteristics corresponding to the first ore type.
In any or all implementations, determining the respective similarities includes determining that the first set of mineral characteristics falls within a first cluster of characteristics of the plurality of clusters of characteristics. The first set of flotation treatment parameters is a set of flotation treatment parameters corresponding to the first cluster of characteristics.
In any of the foregoing or other implementations, determining the respective similarities includes determining that the first set of mineral characteristics falls outside the plurality of clusters of characteristics. Determining the first set of flotation treatment parameters includes determining a combination of at least two sets of flotation treatment parameters corresponding to at least two clusters of characteristics of the plurality of clusters of characteristics.
In any of the foregoing or other implementations, determining the first set of mineral characteristics based on the first measurement results includes at least one of determining principal components of the first measurement results or performing linear discriminant analysis on the first measurement results.
In any of the foregoing or other implementations, determining the first set of mineral characteristics includes mapping the first measurement results to a parameter space of the plurality of predetermined sets of mineral characteristics.
In any or all implementations, the first measurement results characterize the sample of mineral flotation feed when the sample of mineral flotation feed is in a slurry form.
In any or all implementations, the first sample of mineral flotation feed includes a ball mill cyclone overflow.
In any or all implementations, the first sample of mineral flotation feed includes at least one of a semi-autogenous mill feed, a semi-autogenous mill output, or a ball mill cyclone inlet feed.
In some implementations, the first sample of mineral flotation feed has not been subject to separation in a flotation cell.
In any or all implementations, the first set of flotation treatment parameters includes at least one of: a reagent composition, a collector type, a collector dosage, a collector addition point, a frother type, a frother dosage, a frother addition point, a modifier type, a modifier dose, a modifier addition point, a slurry pH, a residence time, a solids density, or a flotation process flow.
In some implementations, the sample of mineral flotation feed includes a combination of multiple ore types.
In any or all implementations, the first measurement results include at least one of an X-ray diffraction pattern, an X-ray fluorescence spectrum, a near-infrared spectroscopy spectrum, or magnetic resonance intensity at a given frequency.
In any or all implementations, the method includes performing one or more measurement processes on the sample of mineral flotation feed, to obtain the first measurement results.
In any or all implementations, the method includes separating the mineral components according to the first set of flotation treatment parameters.
In any or all implementations, the method includes determining a second plurality of predetermined sets of mineral characteristics by re-clustering the plurality of predetermined sets of mineral characteristics with the first set of mineral characteristics.
In any or all implementations, the method includes obtaining an additional set of mineral characteristics of an additional sample of mineral flotation feed, determining respective additional similarities between the additional set of mineral characteristics and the second plurality of predetermined sets of mineral characteristics, selecting an additional set of flotation treatment parameters based on the respective additional similarities, and transmitting the additional set of flotation treatment parameters to the flotation processing system for separation of mineral components according to the additional set of flotation treatment parameters.
Some implementations of the present disclosure describe a mineral feed flotation method that includes receiving first measurement results from a characterization system in an ore processing plant. The first measurement results characterize a sample of mineral flotation feed slurry. The method includes determining a first set of characteristics, the first set of characteristics including one or more mineral characteristics of the sample of mineral flotation feed slurry, the one or more mineral characteristics based on the first measurement results, and one or more flotation system characteristics, the one or more flotation system characteristics based on a state of a flotation processing system. The method includes, based on the one or more mineral characteristics and the one or more flotation system characteristics, associating the first set of characteristics with a first predetermined grouping of characteristics from among a plurality of predetermined groupings of characteristics, each predetermined grouping of characteristics associated with one or more corresponding sets of flotation treatment parameters. The method includes, based on associating the first set of characteristics with the first predetermined grouping of characteristics, selecting a first set of flotation treatment parameters that is associated with the first predetermined grouping of characteristics; and providing the first set of flotation treatment parameters to the flotation processing system for separation of mineral components according to the first set of flotation treatment parameters.
Implementations of this and other methods can have some or all of at least the following characteristics.
In any or all implementations, associating the first set of characteristics with the first predetermined grouping of characteristics includes passing the first set of characteristics through a decision tree.
In some implementations, a rule in the decision tree includes a decision based on at least one of a concentration of a mineral component in the sample, or a particle size distribution of the sample of the mineral flotation feed slurry.
In some implementations, a rule in the decision tree includes a decision based on at least one of the one or more flotation system characteristics.
In any or all implementations, the one or more flotation system characteristics include at least one of: a water balance in the flotation processing system, a number of available flotation processing stages available in the flotation processing system, a throughput of mineral flotation feed through the flotation processing system, a circulating load value of the flotation processing system, an availability of one or more reagents, or a sump capacity of the flotation processing system.
In any or all implementations, the sample of mineral flotation feed slurry includes ball mill cyclone overflow.
In some implementations, the sample of mineral flotation feed slurry has not been subject to separation in a flotation cell.
In any or all implementations, the first measurement results include an X-ray diffraction pattern.
In any or all implementations, the method includes performing one or more measurement processes on the sample of mineral flotation feed, to obtain the first measurement results.
In any or all implementations, the method includes separating the mineral components according to the first set of flotation treatment parameters.
In any or all implementations, the method includes receiving real-time data indicating a current state of the flotation processing system; and determining the one or more flotation system characteristics based on the real-time data.
In any or all implementations, the plurality of predetermined groupings of characteristics include a plurality of clusters of characteristics.
In some implementations, the method includes receiving a plurality of measurement results characterizing a corresponding plurality of first mineral samples; determining a plurality of first sets of mineral characteristics corresponding to the plurality of measurement results; and clustering the plurality of first sets of mineral characteristics into the plurality of clusters of characteristics.
In some implementations, determining the plurality of first sets of mineral characteristics includes performing principal component analysis or linear discriminant analysis on the plurality of measurement results.
In some implementations, a first cluster of the plurality of clusters includes a first set of mineral characteristics corresponding to a first ore type and a second set of mineral characteristics corresponding to a second ore type that is different from the first ore type.
In some implementations, a first cluster of the plurality of clusters includes a first set of mineral characteristics corresponding to a first ore type, and a second cluster of the plurality of clusters includes a second set of mineral characteristics corresponding to the first ore type.
In any or all implementations, associating the first set of characteristics with the first predetermined grouping of characteristics includes determining a similarity between the first set of characteristics and characteristics of the first predetermined grouping of characteristics.
In any or all implementations, determining the first set of characteristics based on the first measurement results includes at least one of determining principal components of the first measurement results or performing linear discriminant analysis on the first measurement results.
In any or all implementations, determining the first set of characteristics includes mapping the first measurement results to a parameter space of the plurality of predetermined groupings of mineral characteristics.
In any or all implementations, the first set of flotation treatment parameters includes at least one of: a reagent composition, a collector type, a collector dosage, a collector addition point, a frother type, a frother dosage, a frother addition point, a modifier type, a modifier dose, a modifier addition point, a slurry pH, a residence time, a solids density, or a flotation process flow.
In any or all implementations, the method includes determining a second plurality of predetermined groupings of characteristics by re-clustering the plurality of predetermined groupings of mineral characteristics with the first set of characteristics.
In any of the foregoing or other implementations, this and other mineral feed flotation methods can be implemented as, for example, instructions on a non-transitory, computer-readable storage medium, and/or as instruction on a storage medium in a system including one or more processors.
Implementations according to this disclosure can help to realize one or more advantages. In any or all implementations, mineral extraction recovery and/or grade can be improved. In any or all implementations, process parameters can be customized based on plant state for more efficient and/or cost-effective processing. In any or all implementations, process parameters can be determined in a more timely fashion. In any of the foregoing or other implementations, more relevant process parameters can be determined, e.g., for different ore samples classified as same or different ore types.
This summary may not necessarily list all characteristics and, therefore, subcombinations of these characteristics or elements may also constitute one or more implementations contemplated by the disclosure, the details of which are set forth in the accompanying drawings and the description below. Other aspects, features and advantages will be apparent from the description and drawings, and from the claims.
This disclosure relates to mineral flotation process customization. During froth flotation, components of mineral ore are selectively separated based on hydrophobicity/hydrophilicity (e.g., differential wettability) in order to extract one or more target minerals. As described herein, mineral flotation feed is characterized at an ore processing plant, and the characterization results are used to determine customized process parameters for froth flotation treatment based on comparison to predefined groupings of characteristics, for example, mineral characteristics and/or flotation system characteristics. This customization can provide an increase in mineral extraction efficiency, along with other benefits. While multiple implementations of a mineral feed flotation process/method are described throughout the disclosure, it will be understood that any feature described with respect to one aspect or one implementation of a mineral feed flotation process is interchangeable with another aspect or implementation of the process, unless otherwise stated, or such interchangeable feature would otherwise make the process inoperable. Accordingly, any description of a mineral feed flotation process, even though described in relation to a specific implementation or drawing, is applicable to and interchangeable with other implementations or drawings of the methods and processes herein described.
Particle size distribution (e.g., fineness distribution), as determined by, for example, crushing/grinding duration, can be adjusted and may affect subsequent flotation results. For example, larger or heavier particles may be less likely to attach to air bubbles in a flotation cell. The slurry solids density (e.g., density of particulate in the slurry compared to solvent) is another important parameter: a lower density of particles in the slurry can lead to lower froth stability, which can be compensated for by appropriate reagents. Chemical crushing/grinding aids may be added to improve grinding efficiency, the appropriate choice of crushing/grinding aid dependent on parameters (e.g., hardness) of the ore.
The slurry then undergoes conditioning 104, during which one or more reagents are added to the slurry to prepare the slurry for flotation separation. While various types and combinations of reagents can be applied, in general the reagents are classified into three types. “Collector” reagents adsorb selectively to particular minerals within ore to render the minerals hydrophobic, such that the hydrophobic minerals subsequently adhere to bubbles as described in further detail below. Collectors are generally classified into sulfide collectors and non-sulfide collectors, with common examples including xanthate salts (e.g., potassium amyl xanthate, sodium isobutyl xanthate, or sodium isopropyl xanthate), sulfur-based ligands such as dithiophosphates or dithiocarbamates, fatty amines, fatty acids, and carboxylates. Collector type(s) added, collector dose(s), and collector addition point(s) can each modify the efficacy of subsequent flotation treatment.
“Frother” reagents act to promote froth/foam formation during treatment in flotations cells and to stabilize the froths/foams. Examples of frothers include pine oil, some alcohols (e.g., methyl isobutyl carbinol), polyglycols, and ethers thereof. As noted for collectors, frother type(s) added, frother dose(s), and frother addition point(s) can each modify the efficacy of subsequent flotation treatment.
One or more “modifier” reagents may also be added, which may have various effects. One notable modifier class consists of pH modifiers that adjust the pH in flotation cells and/or in slurries themselves, thereby modifying the effects of collectors, e.g., adjusting the selectively of certain collectors with respect to certain minerals. Examples of pH modifiers include calcium hydroxide (Ca(OH)2), soda ash (Na2CO3), caustic soda (NaOH), sulfuric acid (H2SO4), and hydrochloric acid (HCl). Other modifiers can be used, e.g., anionic modifiers, such as phosphates, silicates, and carbonates, and organic modifiers such as thickeners (e.g., dextrin, starch, glue, and carboxymethyl cellulose). Similarly to collectors and frothers, modifier type(s) added, modifier dose(s), and modifier addition point(s) can each modify the efficacy of subsequent flotation treatment.
“Addition point,” in reference to a reagent, refers to a timing or location of addition of the reagent, e.g., prior to entry into a first flotation cell (e.g., during ore milling/grinding), into a flotation cell itself, after processing in at least one flotation cell and before processing in another flotation cell, at another specified point during flotation processing, or a combination thereof.
After conditioning, the slurry, including any added reagents, is introduced into a flotation cell for flotation separation 106. As shown in
In practice, overall flotation processing may include multiple stages of separation. For example, either or both of the concentrate 216 and the tailings 210 may be further processed, e.g., by optional further grinding and/or reagent addition followed by introduction into further flotation cells. After “roughing” in a first flotation cell, the produced concentrate 216 is subject to “cleaning” in one or more further flotation cells, each additional stage producing a higher resulting grade of concentrate, e.g., a concentrate with a higher concentration of a target mineral. The tailings 210 are likewise “scavenged” in one or more further flotation cells to recover any of the target mineral that was not separated into the concentrate 216 during roughing. Each additional stage of processing can increase the recovery grade of target minerals.
In some implementations, in addition to analysis of the flotation feed as described throughout this disclosure, the concentrate 216 and/or tail 210 are themselves measured (e.g., by XRD) to determine concentrate and/or tail mineral characteristics, and the concentrate and/or tail mineral characteristics are used to determine flotation process parameters in accordance with this disclosure. For example, the concentrate and/or tail mineral characteristics can be combined with mineral characteristics determined from the flotation feed to obtain overall mineral characteristics for use in selection of a grouping of characteristics for process parameter determination.
Among the process parameters associated with flotation separation 106 are residence time within each cell, mixing parameters (e.g., speed or configuration of mechanical mixing), air parameters (e.g., amount or size of bubbles), cell temperature, cell pH, and the general process flow of the flotation separation 106, such as the overall number of stages, the number of cleaning/scavenging stages, concentration redirection to different parts of the flotation processing plant, cell activation/de-activation, and/or other parameters.
The goal of flotation processing is to improve plant performance, e.g., to extract, from a given portion of ore/feed, as much of a target mineral as possible, while minimizing contamination of the target mineral with other, undesired components within the ore, at an acceptable cost. The process efficiency is determined by the parameters of the flotation processing, such as the slurry processing parameters of slurry formation 102, the reagent parameters of conditioning 104, and the flotation separation parameters of flotation separation 106, as shown in
However, in general, the mineral makeup of flotation feed is not known when flotation processing begins, e.g., when ore from a mine is initially crushed, ground, treated, and/or separated. Ore mined from a given mining operation is not compositionally homogenous but, rather, will have varying makeup depending on, for example, the pit from which the ore was mined, the specific area within the pit, and the depth within the pit. This variance can be characterized to some extent, but rough data will not capture local variations. Moreover, ore mined from different areas of a mine is often mixed together when being crushed and/or when entering an ore processing plant, in combinations that can be difficult to predict or determine. Therefore, while ore processing plant managers may attempt to optimize process parameters based on predicted ore composition or based on a general knowledge of where the ore was mined, the results are often poor. For example, in typical practice, a reagent chemistry may be determined for the next six months of flotation processing based on knowledge of pits to be preferentially mined during those six months. This reagent chemistry may provide acceptable results, but it will not be optimized for the day-to-day or hour-to-hour variations in ore to be processed.
This disclosure describes flotation processes that incorporate measurements of flotation feed to determine flotation process parameters based on comparisons of determined mineral characteristics and/or other types of characteristics, such as flotation system characteristics, to predetermined groupings of characteristics, improving process performance
One or more portions of ore 302a, 302b are crushed by a crusher 303. The ore 302a, 302b is crushed and mixed together to produce crusher output with contributions from both portions of ore 302a, 302b. As described above, the crusher output may have particle sizes on the millimeter scale, for example. Optionally, one or more chemicals such as grinding aids, reagents, and solvents (e.g., water) may be added to the crusher output (305) to produce SAG (semi-autogenous grinding) mill input (a slurry) that is ground in a SAG mill 307 to produce SAG mill output. Alternatively, the SAG mill input may be dry. Optionally, one or more chemicals such grinding aids, reagents, and solvents may be added to the SAG mill output (309) to produce ball mill input that is ground in a ball mill 311 to produce ball mill output. Optionally, one or more chemicals such as grinding aids, reagents, and solvents may be added to the ball mill output (313) to produce cell input. Although the addition of chemicals at any one of the indicated process points 305, 309, 313 is optional, the cell input itself is a slurry that includes added chemicals. Chemicals may instead or additionally be added during crushing or grinding itself.
Both the SAG mill 307 and the ball mill 311 grind particles using balls (e.g., metal and/or ceramic balls) in a rotating cylinder. The friction and impact of the balls on the particles, as the balls and particles are together lifted and dropped repeatedly by rotation of the cylinder, grind the particles into progressively smaller sizes. Typically, ball mills have higher ball charges (proportion of volume filled by the grinding balls) than SAG mills, and accordingly ball mills generally produce more finely-ground particles than do SAG mills. Note that the combination of two mills shown in
The ball mill 311 engages in a partially cyclical process with a ball mill cyclone 321 (e.g., a hydrocyclone), which receives output from the ball mill 311. At the cyclone, as shown in
During this process 300, the flotation feed can be sampled at any one or more of the illustrated sampling points 315. Specifically, in this example, any one or more of the crusher output, the SAG mill input, the SAG mill output, the ball mill input, the ball mill output, the ball mill cyclone overflow, the ball mill cyclone underflow, or the cell input may be sampled, and the sampling may occur before and/or after a slurry is formed. Sampling is not limited to these portions of the process; other examples of processes may include additional, or alternative, processing steps, and the outputs/inputs of those other processing steps may be sampled.
In some implementations, it is desirable to sample the slurry when the slurry is more similar (e.g., compositionally and/or morphologically) to what will be fed to flotation cells, so that measurements of the sample are more reflective of the material to be processed and, therefore, can be used to determine more effective process parameters for flotation separation. For at least this reason, in some implementations the cell input and/or the ball mill cyclone overflow (which may be equivalent) are sampled.
However, in some cases, it can be desirable to sample the flotation feed earlier in the process, so that process parameters can be derived more quickly and, therefore, can be used to process flotation feed that is more similar to the sampled flotation feed. For this reason, in some implementations, the ball mill input, the ball mill cyclone underflow, the SAG mill output, the SAG mill input, and/or the crusher output are sampled.
In some implementations, the portions of ore 302a, 302b have been mined at different mine locations or from different mines and therefore have different compositions. It is unlikely that either of these compositions will be known exactly. Even if approximate data exists to characterize ore composition at a given mine or at a given location in a mine, this data is unlikely to account for real-world compositional variations. In addition, even if the composition of both ore 302a and ore 302b is known, a typical ore processing plant handles many tons of ore per day, often from disparate sources, and ore from different sources may be combined in an unsystematic way, such that the particular ore combination being processed at any given moment is unlikely to be known. In the absence of direct compositional knowledge of flotation feed composition, flotation process parameters are based on estimates, e.g., estimates reflective of an average ore composition in a mine.
These estimates may be based on one or more “ore types,” an ore type being an ore composition that is considered to be representative of or characteristic of a species of ore, e.g., a species of ore known to be found in a given mine (e.g., in a given area of a pit of a mine). However, ore types are archetypical rather than directly descriptive of the ore present in feed/slurry at any given time (which may be a mixture of ore types or not described well by a combination of known ore types, such as not well described by a weighted average of known ore types). Classification into ore types may be irrelevant or of little use for determining effective process parameters, for example, because ore types as currently and typically defined may not correspond to clusters of mineral characteristics, and/or, for example, because the parameters by which ore types are currently and typically defined are not the most relevant parameters for differentiating mineral characteristics in a mineral characteristic space. Determination of process parameters based on ore types is at best an inflexible and sub-optimal approach. The use of mineral characteristic similarities can, by comparison, allow for more flexibility and higher flotation recovery, grade, and/or performance.
To allow for improved flotation process parameter determination in the example process 300, the one or more samples 317 of the flotation feed are subject to one or more measurement processes 308, as shown in
According to some implementations of the present disclosure, it has been recognized that XRD is a particularly useful measurement process to apply to flotation feed slurries, such as ball mill cyclone overflow. XRD data can be gathered relatively quickly, and characteristics of the XRD data, such as peak-related information (e.g., peak location, magnitude, and/or width parameter), have proven to be useful for determining mineral characteristics of the flotation feed slurry, such as compositional information.
Measurement results are then analyzed by a computing system 310 (in some implementations, in conjunction with other data, such as flotation system characteristics) to determine flotation process parameters 319. The flotation process parameters 319 are sent to a flotation processing system 312 that performs froth flotation processing based on the flotation process parameters 319. The flotation process parameters 319 may instead or additionally be transmitted to systems that are used earlier in the process 300, such as the crusher 303, the mills 307, 311, and/or chemical dispensing tools that form the slurry or dispense other reagents. As described throughout this disclosure, the use of process parameters that are more optimized for the particular composition of the flotation feed is likely to lead to improved process efficiency. Because of delays associated with sample extraction, sample measurement, parameter determination, parameter transmission, and/or process reconfiguration, the flotation feed that is processed according to the determined process parameters may not be identical to the flotation feed that is sampled, e.g., may be flotation feed that arrives at the ore processing plant tens of minutes after the sampled flotation feed. However, the processed flotation feed is likely to be similar to the sampled flotation feed, such that the determined process parameters are likely to be effective.
The flotation process parameters can be determined based on (i) mineral characteristics, which are based on the measurement of the flotation feed as described above, and/or (ii) flotation system characteristics, which describe a state (e.g., a current state and/or a predicted future state) of the flotation processing system that performs flotation. The synthesized analysis of these two types of characteristics can provide highly efficient flotation that is responsive to both mineral and system variation. In some implementations, as described below, predetermined groupings of characteristics are groupings of both mineral characteristics and flotation system characteristics; the resulting joint association/clustering process can provide better process parameters than would be obtained by groupings/clusters of mineral characteristics alone.
In some implementations according to this disclosure, to determine process parameters, a set of characteristics (which can include either or both of mineral characteristics and flotation system characteristics) are associated with one of multiple predetermined groupings of characteristics. For example, a predetermined grouping can be a cluster (e.g., a cluster in a multi-dimensional characteristic space) and/or a bin (e.g., a bin corresponding to a leaf of a decision tree). Each grouping corresponds to a set of process parameters. Accordingly, when the set of characteristics has been associated with a particular grouping, the process parameters corresponding to the grouping can be selected and used for flotation processing.
The computing system 310 that performs process parameter determination may be, in various implementations, a local computing system, e.g., a computing system located at the ore processing plant; a remote computing system such as one or more remote servers (e.g., cloud computing servers) configured to communicate with equipment at the ore processing plant over the Internet or another network connection; or both local and remote. For example, some data processing and/or analysis operations may be performed locally, and the results of these operations may be transmitted to a remote computing system for further processing. Flotation process parameters determined at the remote computing system may be transmitted back to the local computing system for operational implementation.
Optimal parameter determination can be based on pre-processed prior sample data that has been analyzed and clustered into predetermined groupings at least partly prior to processing of the one or more portions of ore 302a, 302b.
In some implementations, each prior ore sample 400a, 400b, 400c is also tested in various flotation processes, with various flotation process parameters, to determine optimal flotation process parameters for each prior ore sample 400a, 400b, 400c. Testing may involve laboratory separation tests conducted on each prior ore sample 400a, 400b, 400c for a number of different experimental conditions, and/or processing in a full flotation system, e.g., in an ore processing plant.
Measurement results for each prior ore sample 400a, 400b, 400c are processed to obtain mineral characteristics derived from the measurement results. For example, based on measurement data (e.g., in the case of XRD, an X-ray diffraction pattern, and/or, in the case of near-infrared spectroscopy, a near-infrared spectroscopy spectrum), mineral characteristics, such as an elemental composition of each prior ore sample 400a, 400b, 400c, are determined.
In some implementations, each prior ore sample 400a, 400b, 400c is associated with one or more flotation system characteristics. For example, each prior ore sample 400a, 400b, 400c can be processed under a certain set of system conditions, e.g., water balance, number of stages, feed throughput, and/or another flotation system characteristic as described herein.
The correspondence between each prior ore sample 400a, 400b, 400c and one or more characteristics (e.g., mineral characteristics and/or flotation system characteristics) represents a mapping between each prior ore sample 400a, 400b, 400c and a characteristic space 402. Each dimension of the characteristic space 402 (e.g., as c1 and c2 in
In some implementations, mineral characteristic determination includes specialized, physics-based analyses of the measurement results. For example, XRD data may be analyzed by a Rietveld method or a comparable method to obtain peak parameters, e.g., peak locations and full-width half-maxima. These peak parameters can be used directly as mineral characteristics (e.g., mineral characteristics to be clustered or otherwise sorted or analyzed), and/or the peak parameters can be compared to libraries of known peak parameters to determine mineral characteristics such as mineral types and concentrations.
In some implementations, mineral characteristic determination includes statistical, machine-learning-based, or other analyses of raw measurement data. A representative example of such methods is principal component analysis (PCA), which can be applied to measurement results (either as raw, as-obtained measurement data or after further processing, e.g., Rietveld processing) to obtain principal component values. For example, in the characteristic space 402, c1 and c2 may represent principal components extracted from measurement data of each prior ore sample 400a, 400b, 400c. PCA maps the measurement results to corresponding points 404a, 404b, 404c in the characteristic space 402 so as to emphasize variance between data points along the axes c1 and c2. Although the example characteristic space 402 has two axes, in various implementations three, four, or more than four-dimensional decomposition may be used.
As noted above, flotation system characteristics describe a state of the flotation processing system. For example, the flotation system characteristics can represent constraints on flotation processing, and/or can represent system changes that affect the optimal processing of a given portion of flotation feed. Because of the inclusion of flotation system characteristics in some implementations according to this disclosure, flotation feed having a particular set of mineral characteristics may be processed differently at different times, based on different sets of flotation system characteristics at the different times. For example, a reagent chemistry that is effective for treating a particular mix of minerals when the flotation system is operating at low throughput, may be less effective for treating the same particular mix of minerals when the flotation system is operating at high throughput. Accordingly, different sets of process parameters (including different reagent chemistries) can be used in the two situations.
Moreover, in some implementations, it can be beneficial to associate sets of characteristics that include both mineral characteristics and flotation system characteristics with predetermined groupings of characteristics. That is, rather than incorporation of flotation system characteristics separately from mineral characteristics—for example, as a supplementary analysis step—the flotation system characteristics and the mineral characteristics are treated substantially or entirely equivalently, both being used to sort data corresponding to a sample of feed into a predetermined grouping.
Examples of flotation system characteristics can include, without limitations: states of tanks/processing stages (e.g., a number and/or capacity of available tanks/stages), a throughput of feed through the flotation processing system, a circulating load value (e.g., a ratio of an amount of solids going through a mill divided by an amount of solids going through a processing circuit), availability/usage of one or more resources consumed for flotation processing (e.g., water balance and/or reagent quantities), and/or sump capacity of the flotation processing system, e.g., how full one or more sumps of the flotation processing system are. In some implementations, the flotation system characteristics may be relatively difficult to modify at-will, or may be undesirable to modify, as opposed to flotation process parameters, which may be more readily configurable.
In some implementations, the flotation system characteristics include current flotation system characteristics indicating a current state of the flotation processing system. For example, in some implementations, the flotation system characteristics are determined in real-time or near-real-time. The computing system 310 can receive real-time sensor data indicative of the flotation system characteristics and use at least the flotation system characteristics to select a predetermined grouping of characteristics. For example, the real-time sensor data can be obtained from one or more sensors, such as temperature systems, flow-rate sensors, and/or other types of sensors of the flotation processing system.
In some implementations, the flotation system characteristics include future flotation system characteristics indicating a predicted future state of the flotation processing system. For example,
Continuing with the description of determining the predetermined groupings, following characteristic determination, the characteristics are clustered. As shown in
Each cluster 410 is assigned one or more corresponding sets of flotation process parameters according to which ore assigned to that cluster 410 should be processed. For example, as shown in
Assignment of flotation process parameters to clusters 410 can be performed in various ways. In some implementations, optimal flotation process parameters have been previously determined for a point in each cluster 410, and those optimal flotation process parameters are used directly for the entire cluster 410. In some implementations, different optimal flotation process parameters have been previously determined for multiple points in each cluster 410, and the different optimal flotation process parameters are combined (e.g., averaged or otherwise combined in a weighted combination) to determine process parameters for the entire cluster 410. In some implementations, optimal flotation process parameters may not be known in advance for points in some or all of the clusters 410, and flotation process parameters assigned to the clusters 410 can be determined in one or more other ways, e.g., based on literature values or known best practices.
In some implementations, linear discriminant analysis (LDA) is used instead of or in addition to PCA. LDA may be used in a supervised analysis to improve dimensionality reduction (e.g., mapping of measurement results to characteristics) and/or to improve clustering (e.g., mapping of characteristic to established clusters). Supervised training using LDA can be based on previously-obtained data, such as recovery rates.
In some implementations, cluster identification and assignment is performed by a decision tree method. As shown in
The decision tree 430 can be generated using a decision tree classification model, in some implementations based on known mineral recovery rates and/or other results of flotation processing, such as output purity. For example, a regression model (e.g., incorporating a random forest approach) may be used to identify aspects of measurement results that are most critical for recovery outcomes. Bins (buckets) are prepared (e.g., low, medium, and high recovery bins), and decision tree node parameters (splittings) are determined using a visual and/or algorithmic approach, such as based on purity tests (e.g., Gini impurity tests) where the objective is to split the original dataset of all ore samples into subsets having high purities. Characteristics corresponding to the prior ore samples 400a, 400b, 400c are used to train/generate the clustering decision tree 430, and the trained clustering decision tree 430 is subsequently applied to new flotation feed samples to determine flotation processing parameters.
Other characteristics may instead or additionally be used as bases on which to sort ore samples in the decision tree. For example, in some implementations, PCA, LDA, and/or another statistical method is used to identify mineral characteristics of the ore samples (e.g., principal components), and the statistically-determined mineral characteristics are used to sort the ore samples in the decision tree 430.
As a result of the series of decision nodes 432, the prior ore samples are sorted into clusters 434, e.g., cluster 0, 1, 2, 3, 4, 5, or 6. Each cluster 434 is assigned one or more corresponding sets of flotation process parameters (e.g., as described above in reference to
It is noteworthy that the clusters need not (but, in some implementations, may) correspond to known, previously-defined ore types. Two ore samples that are thought to be of the same ore type (e.g., based on mined location and/or analyses different from analyses used to sort ore samples into the clusters) may be grouped into different clusters corresponding to different flotation process parameters. Different flotation process parameters are therefore determined for the two ore samples, allowing them to be processed more efficiently than if one set of flotation process parameters applied to both samples.
Correspondingly, in some cases, ore processed at two different times corresponding to two different ore types may be grouped into the same cluster, e.g., based on recognition of previously-deemphasized similarities in composition or another characteristics, and/or based on flotation system characteristics. The ore processed at the two different times can correspondingly be processed using the same or similar flotation process parameters that are configured to be optimized for the ore, improving process efficiency. An analysis that took into account only ore type might process the ore more differently at the two different times, leading to worse process efficiency for some or all of the ore.
As described in reference to
As shown in
In some implementations, when characteristics of a flotation feed sample fall within a particular cluster, the previously-defined flotation process parameters assigned to that cluster are directly identified and used to process further flotation feed. For example, in response to identification of the first cluster 604, process parameters can be altered so that subsequent flotation feed is ground to a particle size of 30 μm and treated with N-alkoxycarbonyl, S-n alkyl thionocarbamate as a collector reagent. The subsequent flotation feed may be, for example, feed that enters an ore processing plant soon after (e.g., 10 minutes after, 30 minutes after, or one hour after) the flotation feed from which one or more samples were extracted to obtain the first set of characteristics 602a.
In some implementations, the previously-defined flotation process parameters assigned to the corresponding cluster are modified to obtain further-customized process parameters. For example, relative similarities of the characteristics of the sample(s) to characteristics of prior ore samples assigned to the cluster may be determined, and the previously-defined flotation process parameters for the cluster can be modified to more closely match predetermined optimal parameters for the most similar prior ore sample, rather than merely exactly matching the previously-defined flotation process parameters defined broadly for the entirely cluster.
In some cases, characteristics corresponding to flotation feed do not fall neatly into any particular predefined cluster. For example, as shown in
If new ore samples are determined to not fall within an existing cluster, clustering may be re-performed using characteristics of the new ore samples as part of the clustering process (e.g., as input data to the PCA analysis, LDA analysis, and/or decision-tree generation process, and corresponding cluster generation), so that future ore having the same characteristics as, or similar characteristics to, the new ore samples will be grouped into a cluster with the new ore samples and processed appropriately. The effectiveness of characteristic sorting can therefore be improved over time. In some implementations, experiments may be conducted to identify optimal flotation process parameters for processing the new ore samples.
In some implementations (e.g., implementations in which clusters correspond to predetermined groupings of mineral characteristics), a cluster is assigned multiple sets of flotation process parameters. These multiple sets may correspond to different operating states (e.g., flotation system characteristics) of an ore processing plant that can necessitate or suggest different treatment types. For example, a cluster may be associated with two sets of flotation process parameters that include at least partly different reagents, process flows, etc. If the ore processing plant has low supplies of a given reagent (an example of a flotation system characteristics), a set of flotation process parameters that does not include that reagent can be selected. If a high throughput is desired (another example of a flotation system characteristic), a set of flotation process parameters that will process the flotation feed more quickly (e.g., by an altered process flow) can be selected. By association of multiple sets of flotation process parameters with clusters of characteristics, process flexibility can be improved.
In some cases, the result of the aforementioned analysis will be that a “default” set of flotation process parameters are selected. For example, if characteristics do not fall into a cluster (e.g., as described for characteristics 602b), the system may revert to a default set of flotation process parameters that are reasonably effective at treating different types of mineral samples. Default flotation process parameters may be set equipment-side, so that, for example, an analyzing computing system (such as the computing system 310) may transmit not the default process parameter set itself but, rather, an instruction for flotation processing equipment to proceed with the equipment's locally-stored default operations. In some implementations, the default flotation process parameters are stored at and transmitted by the analyzing computing system. Default flotation process parameters may instead or additionally be associated with one or more clusters of mineral characteristics, so that, if a newly-analyzed sample is determined to fall within one of those clusters, an instruction is sent for flotation processing equipment to use default process parameters.
In some cases, characteristics of an extracted ore sample correspond to two or more predefined clusters. For example, as shown in
Flotation process parameters are provided to a flotation processing system for treatment of flotation feed characterized by the measurement-derived mineral characteristics. In some implementations, the process parameters are provided to a user terminal (e.g., a personal computer, a mobile device, a kiosk, or an application-specific user interface system) for review by a flotation processing system operator. The operator can configure the flotation processing system in accordance with the flotation process parameters. In some implementations, the operator makes decisions in a “validation” phase in which the process parameters appear on the user terminal as a recommendation. The operator is permitted to either agree with the recommendation or overrule the recommendation and substitute the operator's own selected process parameters. In some implementations, a machine learning model is trained using these operator decisions as training data. The training data can include one or more of flotation system characteristics (e.g., a state of the ore processing plant), information about the flotation feed (e.g., measurement results and/or mineral characteristics), characteristics similarity information (e.g., one or more clusters in which the characteristics are grouped), one or more sets of flotation process parameters recommended by the similarity-based analysis, and/or other data. The training data is labeled with the operator validation decisions. A machine learning model trained in this way can be used to recommend future operator decisions and, eventually, fully supplant the operator.
In some implementations, at least some of the flotation process parameters are transmitted to an automatic processing system that automatically acts in accordance with the flotation process parameters. For example, as shown in
Other flotation processing systems may instead or additionally receive flotation process parameters to guide operations during flotation processing. Grinding and crushing mills, flotation cells and associated systems/mechanisms, and plumbing systems (e.g., valves) are examples of industrial systems that may automatically adjust their operational behavior based on received flotation process parameters and/or that may be adjusted by a tool operator based on the flotation process parameters.
In some implementations, the determined reagent chemistry 700 is based on an identification of one or more problematic minerals in the flotation feed. Non-value species can interfere with flotation, resulting in lower-grade product. The deleterious effects of the non-value species can be controlled by the addition of special reagents to target the non-value species and compensate for their inclusion. One or more clusters can be associated with one or more problematic minerals and, correspondingly, associated with flotation process parameters to compensate for the one or more problematic minerals.
In some implementations, a feedback system is incorporated to change future flotation process parameter determination based on present results. For example, if it determined that extraction efficiency is low using flotation process parameters associated with a given cluster (e.g., based on a low flotation yield), then the flotation process parameters associated with the cluster can be adjusted, and/or the cluster can be split into multiple clusters or combined with another existing cluster. New clusters can be created when it is determined that existing clusters do not correlate well with certain sets of characteristics. Models (such as clustering models and decision tree models) can be provided with an active learning functionality to adjust over time and provide more accurate clustering and more beneficial process parameter correlation. Data obtained from analyzing and processing new flotation feed samples (e.g., characteristics of the new flotation feed samples, and/or results of performing flotation separation on the new flotation feed samples) can serve as input data for revising a set of identified clusters, decision nodes in a decision tree, and other models used for future flotation process parameter determination.
The result of this flotation process parameter determination based on comparisons of characteristics to predetermined groupings of characteristics can be a highly responsive and customizable flotation separation process. Live flotation feed measurements are used to determine suitable collector types, dosages, and addition points; frother types, dosages, and addition points; modifier types and addition points; slurry pH at various points in the plant; and/or other parameters, these parameters then directly being used for treatment of the flotation feed. In some cases, the overall process flow can itself be determined, e.g., disabling a bank of flotation cells for clusters that are associated with faster floating. Adjustments can be made in real time or near real time, e.g., at intervals of less than two hours, less than one hour, less than thirty minutes, or more often.
Each ore sample A, B, C, and D belonged to the same ore type X, such that the ore samples would typically be processed using the same processing parameters. However, XRD measurements of the ore samples were obtained and processed using PCA (e.g., PCA performed on raw θ−2θ X-ray diffraction spectra) to associate the ore samples with predefined clusters. It was found that ore samples A and B were grouped into Cluster 1 while ore samples C and D were grouped into Cluster 2. Samples A and C were processed using Reagent A, while samples B and D were processed using Reagent B. It was found that, for Cluster 1, Reagent A increased copper recovery by 3% compared to using Reagent B, while, for Cluster 2, Reagent B increased copper recovery by 6% compared to using Reagent A. Based on this, an association can be established between Cluster 1 and Reagent A, and between Cluster 2 and Reagent B, and future samples sorted into either cluster can be processed accordingly. Conventional flotation process parameter selection-based, for example, on predefined ore types corresponding to mined locations, without a mineral characteristic similarity-based analysis including predefined clusters-would not optimize flotation process parameters based on measurement-derived clusters, and therefore would not realize these copper recovery gains for both Cluster A and Cluster B.
In some implementations, taking measurements of flotation feed itself, and basing flotation process parameters on those measurements, provides improvements compared to basing flotation process parameters on other streams such as concentrates (e.g., rougher concentrates or final concentrates). For example, a concentrate obtained from flotation treatment can be analyzed and used to infer the composition of flotation feed that was processed to generate the concentrate, rather than measuring the flotation feed itself. However, first, the concentrate measuring method is a highly delayed method: the concentrate measurement can only provide information about flotation feed that has already undergone at least one round of separation in a flotation cell, e.g., flotation feed that began to be processed multiple hours or more before the concentrate measurements were taken. This increased delay means that derived flotation process parameters may be already out-of-date by the time they are derived, with to-be-processed flotation feed significantly different in composition from the flotation feed that was measured to device the flotation process parameters. When the outdated flotation process parameters are used, efficiency of flotation separation may be reduced. Second, analysis of a concentrate can only provide information on minerals floated in the concentrate, missing information about other minerals (e.g., certain silicates) that may be useful for making a decision on flotation process parameters such as reagents or PH levels.
In addition, XRD measurements on flotation feed itself can be more accurate than measurements made on concentrates, allowing for more accurate flotation process parameter determination.
As shown in
Based on the one or more mineral characteristics and the one or more flotation system characteristics, the first set of characteristics is associated with a first predetermined grouping of characteristics from among a plurality of predetermined groupings of characteristics (1106). Each predetermined grouping of characteristics is associated with one or more corresponding sets of flotation treatment parameters
Based on associating the first set of characteristics with the first predetermined grouping of characteristics, a first set of flotation treatment parameters that is associated with the first predetermined grouping of characteristics is selected (1108).
The first set of flotation treatment parameters are provided to the flotation processing system for separation of mineral components according to the first set of flotation treatment parameters (1110).
In some implementations, the method 1100 is performed by a computing system, such as computing system 310, and can include any of the processes described with respect to
Various implementations of the systems and techniques described here (such as computing system 310) can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
These computer programs (also known as programs, software, software applications or code) include machine instructions for a programmable processor, and can be implemented in a high-level procedural and/or object-oriented programming language, and/or in assembly/machine language. As used herein, the terms “machine-readable medium” “computer-readable medium” refers to any computer program product, apparatus and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well. For example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback). Input from the user can be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (“LAN”), a wide area network (“WAN”), and the Internet.
The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
For a system of one or more computers to be configured to perform particular operations or actions means that the system has installed on it software, firmware, hardware, or a combination of them that in operation cause the system to perform the operations or actions. For one or more computer programs to be configured to perform particular operations or actions means that the one or more programs include instructions that, when executed by data processing apparatus, cause the apparatus to perform the operations or actions.
Although a few implementations have been described in detail above, other modifications are possible. Logic flows depicted in the figures do not require the particular order shown, or sequential order, to achieve desirable results. In addition, other actions may be provided, or actions may be eliminated, from the described flows or methods, and other components, features, or elements may be added to, selected from one or more member of a group including any two or more of the listed elements, components, or features, or removed from, the described systems or methods. Accordingly, other implementations are within the scope of the following claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 22166967.4 | Apr 2022 | EP | regional |
This application claims priorities to U.S. Provisional Patent Application No. 63/288,908, filed on Dec. 13, 2021 and to European Patent Application No. 22166967.4, filed on Apr. 6, 2022, the whole content of each of these applications being incorporated herein by reference for all purposes.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2022/085723 | 12/13/2022 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 63288908 | Dec 2021 | US |
