WO2008061289A1 - Method and apparatus for monitoring a froth phase - Google Patents

Abstract

A method for monitoring a froth phase of a liquid system including: detecting acoustic emissions in the ultrasonic frequency range using a sensor located within the froth phase; and analysing the acoustic emissions.

Description

METHOD AND APPARATUS FOR MONITORING A FROTH PHASE

FIELD OF THE INVENTION

The present invention relates primarily to a method for monitoring a froth phase. The invention also relates to an apparatus for use in such a method. More particularly, the invention has application in the field of froth flotation and, as such, provides a method for monitoring a froth phase of a froth flotation cell. It will, however, be understood that such an application is merely exemplary of a field of use that the method of the invention is well suited for. While particular reference will be made hereafter to such an application, it should be realised that the invention may have much broader application. Such broader applications are considered within the scope and ambit of the present invention.

BACKGROUND TO THE INVENTION

Flotation is the method most widely used to selectively separate valuable minerals from other material. Flotation is a wet process in which a slurry of fine valuable minerals and gangue material is fed into an agitated tank. A key element of the process involves the addition of air bubbles to the feed and into the tank. The bubbles selectively collect particles of the minerals of interest and raise them to the surface of the liquid within the tank. To facilitate this task chemical reagents are added during processing. The floating bubbles with particles of valuable mineral attached create a froth layer that may be recovered from the tank to form a concentrate. The particles that do not rise to the froth layer are rejected to a tailings stream.

The tanks used such processes are known as flotation cells which can differ in their mechanism but share an important attribute. In all flotation cells two distinguishable phases are formed. The majority of the volume of the cell is occupied by the pulp phase, a turbulent phase where particles are suspended in the liquid within the cell and are colliding continuously with air bubbles. The second phase, the froth phase, floats on top of the pulp phase and is composed of mineralized bubbles attached to each other.

The objective of the flotation process is to separate and recover valuable minerals from non-valuable minerals in an effective manner. Loss of valuable minerals to the tailing stream will lead to loss of profits. Also, recovery of non- valuable, or gangue material will reduce the quality or grade of the product recovered. Furthermore, since the minerals fed to the flotation cell are not totally liberated, as some of them are found in composite particles, absolute recovery of valuable mineral without contamination is practically impossible. The performance of the system is therefore generally a trade off between recovery and grade. The froth phase plays an important role in the grade- recovery trade-off, because not only the valuable particles attached to the rising bubbles arrive to this phase, but also gangue particles that find their way up to the surface by different mechanisms.

The froth phase is inherent to flotation systems and helps to displace the concentrate of valuable mineral out of the cell. However, as mentioned earlier, the froth phase contains non-valuable or waste material that needs to be rejected before collection of the valuable mineral in the froth overflow. A deep froth phase in a flotation cell may assist in upgrading the end product. In other words, by increasing the size (depth) of the froth phase more gangue material can drain downwards and return to the pulp phase. The recovery and grade of the froth phase are governed by the froth behaviour that is linked directly to its structure and stability. The froth is composed of congregated mineralised bubbles having an internal network of liquid channels. Non-attached particles, valuable and non-valuable, are free to transit these channels and, due to gravity, generally return to the pulp phase. Conversely, gangue particles that are trapped within this structure may not be free to travel through the liquid channels. To liberate gangue particles from such entrapment, the froth needs to be unstable thereby facilitating their drainage, a task that can be achieved by the coalescence of bubbles within the froth. However, too much coalescence also encourages the detachment of valuable particles from the bubble walls which may result in their loss. Generally, it is not sufficient to have high pulp recovery if the froth recovery is poor. All the effort put into the optimisation of the pulp phase would be in vain. As a consequence, researchers and industry have realised that more attention needs to be given to the froth phase. More recently, researchers have been studying the froth phase with added interest, trying to model its behaviour by mathematical functions. It is believed that this may render the performance of the froth more predictable. Tools and techniques are also being developed to characterise and measured the froth phase efficiency. Mostly, the methods being developed are indirect. The over-riding issue is that the froth phase is very delicate and remains in this physical state for a limited period of time which makes direct sampling difficult.

Study of the froth phase has been slowed by a lack of appropriate equipment. There is a need for the development of customised tools for this purpose. Some studies have been based on visual identifiers recognisable by experienced technicians when looking at the surface of the froth. It was in this way that froth vision technology was developed, which has been in the field for more than one decade. Reaching maturity, this technology offers commercial alternatives that are finding their way into the industry. Even though it is a step forward, it seems that these methods still do not offer sufficient answers for the study and monitoring of the froth phase. The constraint is that froth vision equipment does not provide any information on what is happening beneath the surface of the froth.

In the current environment of increasingly high metal prices companies can continue to operate with processing inefficiencies that result in high operational costs. However, this may not always be the case and substantial long term advantages may be achieved through increasing processing efficiencies. One way to improve process optimisation and cost reduction is through crucial variable monitoring and its use in process control. For example, a tool that provides information on the froth phase and its performance, and ideally sends the information directly to the operators in the control room, may substantially improve profitability of a plant. Such a tool should preferably be non-labour intensive, low maintenance and able to generate on-line information so it can be incorporated in the automatic process control of the plant.

Researchers may also benefit from new tools that help them to explore what is happening within the froth phase. That is, tools which allow them to validate, improve or reject models and to propose new hypothesis. Current methods can be complemented and supplemented by new techniques and equipment that supply a broader picture.

The present invention has been developed on the premise that acoustic emissions may be used to monitor the froth phase of a system, for example as seen in froth flotation cells. As such, the acoustic emissions detected may be analysed to improve the performance of the system.

Initial investigations explored the possibility of recording the sounds produced by froth in a five litre batch flotation cell using a hydrophone. Figure 1 illustrates the set-up of the system used. In particular, the system 10 included a hydrophone 11 immersed in the froth phase 12, an amplifier and filter 13, an acquisition board 14 and a computer 15.

Sounds were acquired using a Bruel & Kjaer (B&K) type 8103 hydrophone followed by amplification and filtration using a Nexus B&K 2690 device. The processed acoustic signals were recorded on a computer connected to a DAQCard-6062E acquisition card (DAQ) from National Instruments. The sounds were stored in digital format (time, voltage pairs) for further analysis using a program designed in LabView V7.

The signal, previously stored, was passed through a band pass filter of variable frequencies. Afterwards, the program extracted the FFT values and the two signals, pre and post-processed, were plotted together to identify any differences. The frequency parameters of the band-pass filter were repeatedly altered in an attempt to obtain only the signal produced by the bursting of bubbles. In every situation the data was plotted and analysed in time and frequency domains in an effort to observe differences depending on variable conditions, such as air flow rate and impeller speed. However, due to the noise and vibration produced by the batch cell, it was not possible to isolate the sound produced by the bursting of bubbles. Sounds were recorded outside and inside the cell and the signals plotted in time and frequency domains. Figure 2 shows acoustic signals in time and frequency domains acquired with a hydrophone immersed in the froth and outside the cell (background noise). The background noise produced by the impeller cannot be distinguished from the signal captured in the froth zone.

Sounds produced by bursting bubbles are expected to be in the frequency range 3 to 10 KHz, which is audible to the human ear. However, this frequency range is shared by the environmental noises found in industrial plants, produced for instance, by industrial motors, vibration machinery and so on. As the power or amplitude of such environmental noises is larger than the amplitude of the sounds produced by bursting bubbles it is not possible to extract or detect the weaker signal by examining the audible frequency spectrum. The same consideration applies when using conventional microphones or hydrophones that operate on the broadband audible frequency spectrum.

Because of the poor results obtained, as discussed above, the analysis was extended to higher frequencies (higher than 10 KHz). At these higher frequencies small differences were observed in the experiments with air being introduced to the system and without (flotation cell running with and without bubbles). Referring to Figure 3, the amplitude of the signal obtained with air introduced to the system appears to be higher than the signal corresponding to the experiment without air. However, the distinction is not clear. SUMMARY OF THE INVENTION

According to one aspect of the invention there is provided a method for monitoring a froth phase of a liquid system including: detecting acoustic emissions in the ultrasonic frequency range using a sensor located within the froth phase; and analysing the acoustic emissions.

It has been found that the detection of acoustic emissions emitted from the froth phase in the ultrasonic frequency range advantageously eliminates issues related to background noise and provides a good indication of events occurring within the froth of the froth phase, including bursting/coalescence of bubbles and so on.

The acoustic emissions are detected in the ultrasonic frequency range, for example at a frequency of above 20 kHz. Preferably, the emissions are detected at a frequency in the range of from 20 to 75 kHz. In certain embodiments there may be advantages in detecting acoustic emissions in the range of from 30 to 50 kHz. It has been found to be particularly advantageous to detect acoustic emissions at an ultrasonic frequency of 40 kHz.

The acoustic emissions may be detected in a single position within the froth phase, or at a plurality of positions within the froth phase. Generally, the acoustic emissions will be detected at a plurality of positions within the froth phase. If so, the acoustic emissions are preferably detected in an array of positions in at least one horizontal plane and/or vertical plane of the froth phase, the positions being at a distance of up to 5 cm from one another.

It will generally be necessary to amplify the acoustic emission signals that are emitted from the froth phase. As such, it is preferred that the detected acoustic emissions be amplified prior to analysis. This will be dealt with in more detail below with reference to the accompanying figures and examples. The technique or methods for the analysis of the detected acoustic emissions is not particularly limited. This may be somewhat dependent on the system being monitored. In one embodiment the method of analysis includes filtering of acoustic emission signals detected at the resonant frequency of the sensor and calculation of the Root Mean Squared (RMS) values in cycles of 1 second. This provides an analogue signal corresponding to the RMS value in one second cycles of the filtered acoustic emission signal. The acoustic emission sampling in this instance is preferably at a sampling frequency of at least 10 S/s. In an alternative embodiment, the acoustic emissions are acquired at a high sampling frequency, preferably of at least 150 kS/s, and are then digitally processed to filter the signals and calculate RMS values. In an industrial environment, the analysis may include filtering the acoustic emission signals at the resonant frequency of the sensor, calculation of RMS values in cycles of 1 second and subsequent processing using a microcontroller, for example to convert the analogue signal to digital form for final processing of the RMS values. Again, more detail on the analysis of the acoustic emission signals detected will be provided below with reference to the drawings and examples.

The above described method provides a means for monitoring the froth phase of a liquid system. Consequently, according to another aspect of the invention there is provided a liquid system including: a vessel containing a froth phase; a sensor for detecting acoustic emissions in the ultrasonic frequency range, the sensor being located within the froth phase; and means for analysing the detected acoustic emissions.

One particular application of the invention is in froth flotation. As such, in one embodiment the vessel is a froth flotation vessel including a froth phase and a liquid phase. It will, however, be understood that the invention has broader application and is not necessarily limited to this field of application.

Although a more detailed account of the set-up of the sensor will be provided below, in a preferred embodiment the sensor includes an ultrasonic transducer and an amplifier. The ultrasonic transducer may, for example, be an ultrasonic piezo-electric transducer with a resonant frequency of 40 kHz.

As noted above, the analysis method will be discussed in more detail below. Generally, the means for analysing the detected acoustic emissions includes a data acquisition module and a computer.

As was the case with the method described above, the system may include a single sensor, or any number of sensors as desired. Generally, the system includes a plurality of sensors located at a plurality of positions within the froth phase of the liquid system. If so, once again, the sensors are generally located in an array of positions in at least one horizontal plane and/or vertical plane of the froth phase, the positions being at a distance of up to 5 cm from one another.

In some circumstances the operation of the system in question gives rise to vibrations that may be mechanically transmitted to the sensor. It may be that those vibrations have a negative impact on the ability to effectively detect the acoustic emissions being emitted from the froth phase. In such cases, it is preferred that the sensor be mechanically isolated from any vibrations resulting from operation of the system. Accordingly, the system may include one or more vibration attenuating devices that comprise a suitable insulating material for attenuating vibrations generated by mechanical operation of the system.

It will be appreciated that the method for monitoring a froth phase in a liquid system and the system described above may be used as a means for providing vital information about the system and its efficiencies or deficiencies. As such, according to yet another aspect of the invention there is provided a method for controlling operation of a liquid system having a froth phase, the method including: detecting acoustic emissions in the ultrasonic frequency range using a sensor located within the froth phase; analysing the acoustic emissions; and adjusting at least one operating parameter of the liquid system based on the analysis of the acoustic emissions.

It will be appreciated that the preferred and optional features described above are equally applicable to this aspect of the invention. As such, further mention of those features will not be dealt with in more detail here.

In a particular embodiment of the invention, the liquid system is a froth flotation system and the adjustment of at least one operating parameter of the system includes adjusting at least one of feed rate, f rather addition rate, collector addition rate, impeller speed and air flow rate.

DETAILED DESCRIPTION OF THE INVENTION

The present invention will now be described in more detail with reference to the accompanying drawing. It should be appreciated, however, that the following detailed description is provided for reference only and should not be construed as limiting on the invention in any way. In the drawings:

Figure 1 illustrates an experimental set-up using a hydrophone for detecting acoustic emissions;

Figure 2 illustrates a plot of signals acquired using the set-up illustrated in Figure 1 at frequencies in the range of from 3 to 10 kHz; (a) acoustic signal outside the cell, (b) frequency spectrum outside the cell, (c) acoustic signal inside the froth and (d) frequency spectrum inside the froth; Figure 3 illustrates a plot of signals acquired using the set-up illustrated in Figure 1 at higher frequencies (i.e. above 10 kHz); (a) frequency spectrum of experiment with air and (b) frequency spectrum of experiment without air;

Figure 4 illustrates an embodiment of a system according to the invention; Figure 5A-5C illustrate a number of options for the analysis of the acoustic emissions detected by the sensor; Figure 6 illustrates a plot of signals acquired using the set-up according to the invention; (a) froth sensor signal inside the froth zone and (b) froth sensor signal outside the cell/froth (background noise signal);

Figure 7 illustrates a plot of RMS acoustic emission signals across a froth phase in a froth flotation cell;

Figure 8 illustrates a plot of RMS acoustic emission signals acquired in a lab Jameson Cell;

Figure 9 illustrates a diagram of a pilot plant rig;

Figure 10 is a photograph of a sensor in position in a flotation cell; Figure 1 1 illustrates a plot of sensor signal over time at different collector dosages;

Figure 12 illustrates a plot of average RMS values over time at different collector dosages;

Figure 13 illustrates a plot of standard deviation of RMS values over time at different collector dosages;

Figure 14 illustrates a plot of sensor response compared with metallurgical performance;

Figure 15 illustrates a plot of sensor signal over time following a frother dosage change from 40 ppm to 10 ppm; Figure 16 illustrates a plot of sensor signal before and after frother dosage change;

Figure 17 illustrates a plot of average RMS and standard RMS over time;

Figure 18 illustrates a vibration attenuation device; Figure 19 illustrates a plot of sensor signals over time acquired outside a flotation cell and inside the froth phase of two flotation cells of different locations;

Figure 20 illustrates a plot of froth Acoustic (FA) and velocity (FV) measurements over time in two cells in different flotation banks with a summary of the three outputs (the standard deviation has been magnified five times). PR1 : primary rougher Cell 1 ; SC5: slow cleaner cell 5.

Figure 21 illustrates a plot of average sensor signals obtained every minute for different cells in the secondary rougher bank and a summary of the overall average and standard deviation values;

Figure 22 illustrates a plot of sensor measurements over time along the Zn scavenger bank and a summary of average and standard deviation signals;

Figure 23 illustrates a plot of sensor and plant signals over time in a zinc scavenger cell;

Figure 24 illustrates a plot of the collector dosage signal and the inverted sensor signal over time;

Figure 25 illustrates a plot of sensor signals recorded for different MIBC concentrations over time; Figure 26 illustrates a plot of sensor and metallurgical response to changes in frother (MIBC) concentration; and

Figure 27 illustrates a plot of the inverted sensor signal against water recovery % for different frother types and concentrations.

Figures 1 through 3 have been dealt with above and will therefore not be considered in detail here. However, as already noted, it is clear from these figures that the signals received from within the froth phase 12 are not well distinguished from the background noise. The results are slightly improved using a higher frequency, but are still not of particular use in monitoring the system 10.

Turning to Figure 4, a system 40 is provided that includes a number of basic modules. The first module 41 is a sensor that contains an ultrasonic transducer 42 and an amplifier 43. The sensor is the only component that is immersed in the froth phase 44. Generally, two stages of amplification will be provided by the amplifier 43 of the first module 41. The components are provided in the from of a small printed circuit board (e.g. 25 mm x 50 mm) which is enclosed and sealed for waterproof protection by a stainless steel cylinder (not shown) with an external diameter of 30 mm and length of 60mm. The cylinder has one orifice for the ultrasonic transducer 42 and an outlet for a cable that transports power and the amplified signal to the next module.

The second module 45 may take a number of forms. These are generally illustrated in Figure 4, and more specifically illustrated in Figures 5A through 5C. For example, the second module may include an intermediate stage of processing and data acquisition that converts the signal into a digital format. The amplified acoustic emission (AE) signal is processed by filtering 48 the signal at the resonant frequency of the transducer 42 (e.g. 40 KHz) and calculating the RMS (Root Mean Square) value 49 in cycles of one second. The signal processing of the second module 45 can be implemented either by hardware, before the signal is digitised by a Data Acquisition (DAQ) module 46 and sent on to a computer 47 (see Figure 5B) or can be implemented by software once the signal is in digital format in the computer 47 (see Figure 5A).

Software implementation adds flexibility in the construction of the processing blocks, because it allows easy and quick adjustment of the tuning parameters which are very important in the initial stages of experimentation and sensor development. Furthermore, a separate module for DAQ and signal processing via software allows the trial of different probes and transducing methods.

For that reason, the initial system has been built around a National Instruments card (model DAQCard-6062E) for data acquisition and the program LabView for signal processing implementation. The card was inserted in the PCMIA compartment of a laptop, which was connected through a cable to a connection board. The board was enclosed in an aluminium heavy duty box, advantageously rendering it dust proof and splash resistant. The box had a cable connecting it to the PC and two circular plug connectors, type M 12, allowing connection and disconnection of the sensor while the cables were constantly protected from water or contacting other voltage sources.

For the system illustrated in Figure 5A, the acquisition card 5A2 is advantageously fast enough to capture signals in the frequency of interest. Therefore a card with a maximum sampling frequency of 500 kS/s (kilo samples per second) was used. The sampling frequency selected to acquire the signals from the AE sensor 5A1 was set at 150 kHz, allowing for good resolution and sufficient sampling points of the ultrasonic signal at 40 KHz without demanding a high amount of computational power.

The signal was sampled and the Root Mean Squared (RMS) value of the signal over one second duration was extracted, plotted and stored. Just one RMS value of processed data per second was stored. In this way the analysis is made on-line capturing more information in less storage space and using less computation time. Therefore, a file of processed AE consists of a set of pairs (time, values) in time increments of one second. The stored file can then be opened and analysed in software 5A3, such as Excel.

Using the system illustrated in Figure 5B, acoustic emission signals detected are filtered 5B2 at the resonant frequency of the sensor 5B1 and the Root Mean Squared (RMS) values are calculated 5B3 in cycles of 1 second. This provides an analogue signal corresponding to the RMS value in one second cycles of the filtered acoustic emission signal. The acoustic emission sampling in this instance may be at a much lower sampling frequency than that for the set-up illustrated in Figure 5A. For example, this may be as low as 10 S/s. An acquisition card 5B4 and final processing software 5B5 are included in the system.

With regard to the set-up illustrated in Figure 5C, in an industrial environment, the analysis may include filtering 5C2 the acoustic emission signals at the resonant frequency of the sensor 5C1, calculating RMS values 5C3 in cycles of 1 second and subsequent processing using a microcontroller 5C4, for example to convert the analogue signal to digital form for final processing of the RMS values 5C5 before output in industrial format 5C6.

During initial testing it was found that the sensor may be sensitive to impacts or vibration transmitted during operation of the system. It was found that a sponge or other form of insulation may be used to isolate the vibration. In a particular embodiment, a vibration attenuation device 80 for attenuating the vibration felt by the sensor is illustrated in Figure 18. The device 80 is cylindrical in shape and is made up of connecting caps 81 and 82 which are connected to a central piece of silicon rubber foam 83 via metal holding structures 84 and 85. The device 80 may be connected at any appropriate location between a vibration source and the sensor.

EXAMPLES

Experiments on Lead/Zinc Ore

A sensor was subjected to preliminary tests in an industrial plant processing lead/zinc ore. Initial measurements were performed on the flotation lead rougher bank. The sensor was immersed in the froth zone of the first rougher flotation cell at different depths in the froth. It was also put above the froth surface suspended in air in order to acquire background noise and establish differences between the measured signals. The signals were acquired at a sampling frequency of 150 kS/s over a period of one second, then filtered and finally plotted with a time span of one second. These results are illustrated in Figure 6.

The RMS value of the background signal shown is 14 mV and for the signal obtained inside the froth is 38 mV. Without wanting to be bound by theory, it is proposed that the RMS value of acoustic emission signals is proportional to the local froth activity and instability.

The measurements show differences in the waveforms when the sensor was located outside the cell, producing a non changing signal, and when immersed in the froth, picking up activity in the form of short burst envelopes, as can be seen in Figure 6. It can be said that the spikes are due to activity occurring only in the froth.

The spikes seen in Figure 6, when the sensor is immersed, are due to activity in the froth produced by bubble coalescence and bursting, and may also be attributed to friction between the bubbles, particles and the sensor (in particular a sensing face of the transducer). As such, it cannot be said that the signal observed is generated only by bubble coalescence or bursting disregarding other events that may be happening in the froth. Comparing the signals observed in Figure 6 with the signals observed using the hydrophone and illustrated in Figures 2 and 3, it can be seen that the sensor of the invention is far superior in its capability to isolate undesired background noise compared with the hydrophone.

The processing of the RMS value over one second cycles was chosen because it is long enough to summarize activity occurring at high frequency, yet short enough to capture real time trends. This way, only important processed information is stored facilitating the storage and analysis of otherwise large amounts of data, as can be observed in Figure 7. In this figure 20 RMS points are plotted, each point represents 150 thousand samples processed and summarised in a single RMS value point. Another technique that could be used consists of counting peak values that exceed a selected threshold per unit of time. However, this technique is subjective to the selection of the threshold and more difficult to implement. On the other hand, RMS calculation can be easily implemented in software, using LabView for instance, and hardware with the use of integrated circuits developed for this specific application.

The RMS value was chosen as a quantitative parameter because it encapsulates crucial information such as the amount of power per unit of time dissipated in the froth in the form of acoustic emissions. The calculation of RMS values is a common practice in instrumentation when working with signals that show a random behaviour and the information is modulated in a variable or "ac" manner.

During the initial tests, it was discovered that acoustic emissions are sensed only in close proximity to the sensor. Thanks to its size, the sensor allows for localized measurements with vertical spatial resolution of approximately 5 cm in the froth. Taking advantage of this property a simple froth profile was performed on the first rougher flotation cell of the Pb flotation bank. Every measurement consisted of one second sampling at 150 kS/s and was repeated for twenty times for every point. Then, the sensor was located at different vertical distances across the froth in the same cell. At least another twenty measurements per location were obtained. The raw data was stored and processed to extract its RMS value. The results are shown in Figure 7 and Table 1.

Table 1 shows that the RMS value averaged over 20 repeats per geographical point differ across the froth for this particular cell. As indicated before, the RMS value of AE signals in the froth can be interpreted as a quantitative measurement of local froth stability. Table 1 Average RMS values across the froth in a flotation cell

Depth below Out of the 5 cm 10 cm 15 cm 25 cm Froth cell

Average RMS 14 mV 33 mV 29 mV 10 mV 21 mV value

It will be appreciated that these results carry minimal weight because too few samples were taken to give a real representation over a significant period of time. That is, during this preliminary testing only 20 samples were taken equivalent to 20 seconds of sampling. Conversely, for the final experiments samples were taken over a period of more than two minutes. Furthermore, during the above tests there was not metallurgical data available for correlations to be performed. However, these results do illustrate that the froth sensor gives a "dead" signal when it is located out of the froth and produces an active signal when it is located within the froth. In this sense the froth sensor is capable of isolating the environmental or background noise from activity taking place only in the froth.

It is interesting to note the differences across the froth as illustrated in Table 1. It is especially interesting to note that one location in the middle of the froth (15 cm) produced a signal of lower magnitude than the background noise, suggesting zones of low activity across the froth. This kind of information would be difficult to obtain with non-contact instrumentation. Another point of interest is the oscillatory behaviour of the acoustic emissions in the froth observed in Figure 7. This oscillation can also be recognized in the further experiments below and appears to be related to the stability of the froth.

Experiments on Coal Flotation Using a Lab Jameson Cell

The AE sensor was tested in a laboratory continuous Jameson cell used for fine coal flotation. The sensor was put in the froth phase of the cell at a fixed distance below the surface (5 cm) to compare the signal produced by two flotation tests of dissimilar froth stability. Every test consisted of five concentrates per coal sample. The froth height (pulp level) and reagent additions were adjusted trough the test to keep a constant overflow. Concentrates were collected at 1 , 2V₂, 5, 7V₂ and 11 minutes. Every concentrate was analysed for ash content and yield product.

Figure 8 shows the continuous monitoring (RMS values processed every second) of froth acoustic emissions for two tests on a lab continuous Jameson Cell and is compared to the ash content of the first four concentrates. The solid lines correspond to the moving averages of 20 periods of RMS AE signals.

The two runs had different froth conditions, with the second test being more stable than the First test. As seen in the Figure, the sensor is able to differentiate between two different conditions of froth stability in an on-line manner.

Comments from the operator and final yield products results are exposed below:

Run 1 : Coal had very little "fines" in the sample. Coal floated readily but required some frother through to maintain adequate froth conditions for the "coarse" fraction. Expect high overall yield and combustibles recovery but moderate ash content in the concentrates perhaps from "composite" or "oxidized" coal;

Run 2: More "fines" present in this sample with more stable froth conditions as a result. Anticipate high overall yield and combustibles recovery at lower ash content in the concentrates than for the F/L ROM (first run) sample. Table 2 Lab Jameson Cell Flotation Test Summary

In conclusion, the coal sample of the second run gave a better performance than the sample of the First run. The second run presented higher yield, less ash content and less reagent addition than the first run. The high stability of the second run was caused by a high amount of "fines" in the feed. The froth sensor was able to measure the froth stability of the two runs showing significant differences in an on-line manner with rapid sampling and real time processing.

Main Experimental Details and Procedures

In order to establish the capabilities of the sensor further experimentation was required. The purpose of the following experiments was to evaluate the sensor response over a range of froth conditions and compare the sensor response to the metallurgical performance of the system. Optimally, the selected experiment needed to generate a wide range of froth behaviour and stability. Furthermore, the selected experiment should be controllable such that a single parameter that changes the froth zone can be varied while other parameters can be kept constant to ease the subsequent analytical process.

From the literature, it can be generally appreciated that the best way to achieve this desired result is to operate a controlled flotation cell changing only the frother addition rate or the collector addition rate. The variation of these parameters significantly impacts on the froth phase. Ideally, real ore should be used in this experiment. Moreover, to determine the potential use of the sensor on an industrial scale on site the experiment should be performed under similar conditions to those seen in industrial cells. A lab scale cell should have a performance similar to an industrial cell. This may be achieved using cells with Sb values of at least 60 s^'1 which generally provide a froth zone comparable to an industrial system.

Change in Collector Dosage - Experiments

All of the above characteristics were met by an experiment designed to model the impact of staged collector addition in flotation circuits and to determine the size-by-liberation floatability distribution of an industrial ore. The main aim of this experiment was to develop a methodology and model to predict changes in floatability with changes in collector addition.

The main experiment was completed at a lead/zinc ore mine site using a pilot plant rig 90 as illustrated in Figure 9. The rig consisted of a 40 litre continuous flotation cell 901 and three sumps of 230 litres each. Two sumps 902 and 903 were used for this experiment allowing continuos flotation of 460 litres of slurry.

For this experiment it was a requirement that the feed 904 for each test be free from reagents. Therefore, the source feed was taken from the secondary cyclone overflow prior to the lead conditioning tank where reagents are added. The conditioning tank fed the lead rougher bank. At the point before the conditioning tank, the level of reagents was very low which was due only to the process water being used. The operating conditions for these tests are summarized in Table 3.

Four tests per experiment were conducted in random order, one test at no collector addition and three tests with collector dosages at 5, 15 and 40 grams per tonne. The experiments were repeated in the same manner and simultaneous sensor signal was acquired for three experiments. For these experiments the slurry was used as it came from the source, thus the solids content and feed grade were maintained at the same level at the plant operating point.

Air 905 was added to the bottom of the cell by the impeller mechanism 906 and fixed at a rate of 110 L/min. The froth depth was kept constant for all the tests. The distance between the lip launder 907 and the pulp weir 908 was 7 cm. However, the real froth depth was 9 cm with 2 cm of froth constantly above the overflow lip.

Collector dosage 909 was added as required directly into the sumps 902 and 903 and the slurry was conditioned in two sumps of the rig with stirrers that kept the solids in suspension. After conditioning time and electrochemical measurements the slurry was fed into the cell by a peristaltic pump 910 operating at 8 L/min. The frother 911 was added on-line to the feed via a small peristaltic pump 912 that allowed slow addition rates to keep the desired frother concentration.

Table 3 Operating conditions for collector change tests

The procedure for each test can be explained by a series of steps as follows:

Step 1 : Clean thoroughly sumps 902 and 903 and cell 901 to eliminate residual collector from previous tests.

Step 2: Fill up both sumps 902 and 903 continuously stirring with slurry. Step 3: Add required collector dosage 909 into the sumps.

Step 4: Perform electrochemical measurements (Ph, Eh and dissolved O₂).

Step 5: Start floating and wait until steady state condition is achieved (20 minutes).

Step 6: Make metallurgical sampling (Feed, Concentrate and Tail flow rates). Step 7: Make gas dispersion measurements.

Step 8: Perform Bubble Load Measurement. Simultaneously put sensor in froth and acquire signal for the same period of time as the BLM (10 minutes).

The metallurgical sampling of the flows of interest was made with 10 litre buckets. Feed 913 and tails 914 were sampled for 10 seconds and concentrate for more than 30 seconds to obtained a suitable amount of mass. Step 7 was achieved by use of a bubble viewer device and an electronic Jg measurement device of the same kind developed by McGiII University researchers. Gas hold up was performed using a mechanical device. Since the air flow rate and frother addition was kept constant, not all the three gas dispersion measurements were performed on every test. Preliminary measurements confirmed that the gas dispersion characteristics were preserved through the tests.

The final step was done with a bubble load modified device. The collection tube used on this instrument was increased to 5 cm in diameter to reduce turbulence and coalescence of rising bubbles in the tube. Further, a nozzle of 1 cm in diameter was attached to the end of the tube where the collection occurs. The sensor was not put in the froth during the metallurgical sampling to avoid affecting the recovery. Because of limited time for each test the froth sensor was put in the froth simultaneously with the bubble load device even though this device could affect the froth performance. However, it was ensured that these conditions remained constant for all the tests completed. Figure 10 shows the location of the froth sensor 101 and where the bubble load sensor 102 was placed in the 40 litres cell.

Finally, the sensor was connected to the acquisition card in the laptop as described in the last section and signals were acquired between 8 to 10 minutes.

Figure 11 shows a plot of the signal produced by the sensor over time at different collector dosages.

It can be seen that at starvation (no collector added, 0 ppm), the signal produced by the sensor signal includes several spikes and high amplitude variations. At this dosage the material in the froth phase consists mainly of natural hydrophobic floatable mineral and low size gangue that reaches the froth by entrainment mechanisms. Also, froth overflow is reduced and visual observation confirmed that the bursting was cyclic and relatively intense, which is in agreement with the sensor signal.

Increasing the collector dosage (5, 15 and 40 parts per million - ppm) resulted in a lower and smoother signal with fewer spikes and spikes of lower amplitude. It is reiterated that the only parameter changed during the experiment was collector addition. Therefore, it can be said with confidence that the sensor is producing information related to changes in the froth phase activity resulting from change in collector addition.

As noted above, one way to summarize and quantify the information produced by the sensor is extracting the average value of the RMS signal over a period of time. This processed signal may then be used as the main output signal. It is suggested that the period of time should be long enough to smooth the short term variations and allow for comparison with subsequent data acquisition. For example, it is proposed here that the time period should be at least two minutes. Figure 12 shows the calculated average values of the RMS sensor signal over periods of 145 seconds versus collector addition.

It should be noted that closer inspection of Figure 11 indicates that extracting only the average value could hide information such as the instability or variability of the signal. Therefore, it is suggested that complementary processing techniques may also be applied. These techniques could process information over the same period of time and could include, but are not limited to, the extraction of the standard deviation value and frequency of the periodicity of the spikes as output complementary signals. Figure 13 illustrates such a potential complementary output sensor signal.

It is expected that changes in collector addition will affect the metallurgical performance of the process, as seen in Table 4.

Table 4. Metallurgical performance vs. changes in collector addition.

The overall recovery should increase as collector is added. However, up to what point the froth phase is responsible for the total recovery is difficult to ascertain. Figure 14 shows the relationship between overall recovery and the sensor signal produced exclusively by the froth phase, which allows for a clearer determination of the key role played by the froth phase.

From Figure 14 the relationship between the sensor signal outputs and the recovery achieved can be appreciated. Increasing collector dosage increases the overall recovery with the sensor signal following this pattern in an inverse manner. From this analysis the signals presented could be used as indicators of the froth performance and its contribution to the overall recovery of the system in question. Furthermore, this analysis provides on-line information that could potentially be used by an operator to control the flotation process based on a desired target for the sensor outputs.

Change in Frother Dosage - Experiments

Further experiments were performed to establish the sensor response to changes in the frother dosage. For this experiment collector dosage remained constant and the frother was added on-line to the feed. The frother flow-rate was controlled by a small peristaltic pump. The speed of the pump, being proportional to the frother flow-rate, was set initially at 40 ppm for a period that allowed a steady state to be reached. At that time the sensor was positioned to acquire information for a period of ten minutes. Subsequently, the speed of the pump was reduced instantly to simulate a step change in frother rate to 10ppm. Data was acquired again to capture the transient response of the froth phase produced by a step change of frother flow-rate. The results are illustrated in Figure 15.

Figure 15 shows how the froth phase moves form a very stable state towards an instable state produced by a dramatic reduction in fother addition. Figure 16 shows the sensor signal before and after the change.

Using the proposed outputs mentioned above it is be possible to generate signals for continuous monitoring of a system that may, for example, be fed to the control room every 145 seconds. A reconstructed graph from this example is illustrated in Figure 17.

Such a signal would be valuable to an operator of the system as this gives an indication of the froth performance. It should be noted that in this case using the average signal would not be sufficient as an interpretative output, because the signal trend is towards a lower value of the initial condition. However, the standard deviation output tells another story. This indicates that the original acoustic emission signal before processing has significant spikes which can be associated with froth instability.

The above examples illustrate how well the sensor notices changes in the froth phase resulting from reagent addition changes. It will be understood, as described above, that there are a number of other variables that could affect the froth zone which can also be monitoring.

Experiments with the sensor in industrial sites

The sensor was tested at two sites, namely a lead-zinc concentrator in Australia and in a platinum concentrator in South Africa.

At the platinum concentrator, the sensor was tested in several cells in different parts of the circuit where cameras of a froth vision system were also in place. The site has clear differences in stability of the froth between the rougher and the slow cleaner banks. The froth of the rougher bank has bubbles that are more mineralised than the later sections of the circuit; in contrast, the froth of the slow cleaner bank shows bubbles with transparent windows, an indication of a lower amount of solids in the froth. The rougher bank operates with deep froth levels (around 50 cm) and the slow cleaner cell with shallower froth levels (around 35 cm). It has been observed that froth recoveries in the slow cleaner section are lower than in the rougher section, probably caused by the lower froth stability. In order to monitor the froth phase in detail, cameras were located on top of the froth close to the weir, providing froth velocity as the main output. The sensor was located close to the area of froth being monitored by the camera, and immersed 20 cm into the froth. Data was acquired over 10 minute intervals, with several repeats, for the first rougher cell and the Fifth cell of the slow cleaner bank.

The sensor was also tested along the secondary rougher bank in order to observe differences in the acoustic signals down the bank. For these measurements the sensor was positioned 20 cm below the froth surface; however, as each cell has a different froth level the location of the sensor in respect to the froth phase interface is different for all the experiments. Table 5 shows the froth levels and location of the sensor in the cells.

Table 5. Froth depth and sensor location across the secondary rougher bank at the platinum concentrator.

At the lead-zinc concentrator a variety of experiments were performed, including surveying a bank with the sensor, data acquisition in a single cell for a 12 hour period and a series of trials with different frother types and concentrations in a 3 m³ cell. In first instance, the sensor was placed in various cells down the zinc scavenger bank. The acquisition time varied between five and ten minutes per cell. The probe was immersed in some cells at two different froth depths to see how the sensor signal changed depending on the position in the froth phase. Table 6 summarises the sensor position and froth depths for this survey.

Table 6. Froth Depths and sensor location along the Zn Scavenger bank at the lead-zinc concentrator.

Since it was unknown how the probe would cope immersed in the froth phase for longer periods of measurement, the sensor was left for a 12 hour period (overnight) in the first cell of the Zn scavenger bank. The acquired data was compared to some other instrumentation outputs of the plant.

Finally, a series of experiments was carried out in a 3 m³ cell fed by the rougher tails, with a zinc grade of about 7%. During these experiments feed rate and air rate were kept constant for various frother types and concentrations at fixed froth depths.

Results - Platinum Concentrator

The cells selected for froth monitoring were the first rougher cell (receiving the plant feed) and the fifth cell of the slow cleaner bank (which processes slow floating material).

Figure 19 shows three measurements of one second cycle rms values of acoustic emissions acquired for two minutes. One of these measurements was made outside the flotation cell, which shows the background noise, clearly indicated by a non-changing signal. This characteristic is the main advantage of the sensor: its ability to isolate environmental noise caused by machinery in a flotation concentrator.

The other two measurements were made by immersing the probe in the froth phase of two different cells. It has been shown before (Tsatouhas et al., Minerals Engineering, 19(6-8): 774-783, 2006) that froth stability is lower in the latter stages of flotation where slow floating components are present, and this is confirmed by the results shown in Figure 19. The acoustic emission signal from the first rougher cell has a lower value overall and less "spikes" than the signal from the slow cleaner bank, which in contrast has higher values and more erratic behaviour. Without wishing to be bound by theory, it is believed that the acoustic signal is caused mainly by the coalescence of bubbles within the froth. Therefore, the first rougher cell has a lower coalescence rate producing a more stable froth than the slow cleaner cell. It can be said that stable froths produce lower acoustic emission signals with less spikes, interpreted as a constant coalesce rate in time. However, it seems that in cases where froth stability is lower the coalescence activity is not constant but erratic and in special cases cyclic as discussed below.

Froth velocity is also different for the two cells of interest. Figure 20 shows a comparison of the froth vision and froth acoustic data. Several sets of data were acquired, each measurement lasting ten minutes; in Figure 20 every point represents the average signal processed every minute. During the survey period conditions were kept constant, hence these results represent steady state behaviour.

The first rougher cell has a constant and higher velocity than the slow cleaner cell. The acoustic emission and froth velocity signals seem to provide a similar diagnosis, that is, when the froth is stable it has lower coalescence (lower acoustic emissions) and flows more easily (higher velocity); on the other hand an instable froth (higher coalescence) increases drainage, therefore reducing the froth overflow.

Without wishing to be bound by theory, it is proposed that the stability of the froth is not just inversely proportional to the acoustic emission signals (represented by the average) but also to its erratic behaviour (represented by the standard deviation). Therefore, these two values, average and standard deviation, are chosen as the main outputs of the sensor system.

The sensor was also tested down the secondary rougher bank as shown in Figure 21. In this case, there is no complementary froth vision data that allows comparisons to be made. This bank receives material from the regrinding mill and depressant is added at the head of the bank. Generally, it is expected that the stability of the froth decreases down the bank as a response to the decreasing amount of solids entering the froth phase. However, there are cases in which there is an increasing amount unwanted floatable material reporting to the concentrate due to the depletion of depressant down the bank. These cases will not illustrate the traditional reduction of froth stability down the bank and appear also to be the explanation for the trend observed in Figure 21.

Results - Lead-Zinc concentrator

As in the platinum concentrator, the sensor was also tested down a flotation bank, in this case the zinc scavenger bank. In cells 7, 8 and 9 two measurements at different vertical positions were performed.

The nomenclature used in Figure 22 is the same as expressed in Table 6; the left side of Figure 22 corresponds to signals acquired closer to the froth phase interface (about 5 cm) and the right side to signals acquired closer to the froth surface (10 cm above the froth interface). For this case the expected trend of decreasing froth stability down the bank is observed for cells 6 to 9, with froth acoustic emission signals increasing as coalescence rate increases. In this bank, frother is added on the top of cell 5, close to the impeller inside the cone crowder. Figure 22 suggests that the frother is not being mixed perfectly in this cell and is carried to the next cell (Scv6) where it arrives better mixed, increasing froth stability. Seaman et al. International Journal of Mineral Processing, 74(1-4): 1-13, (2004), showed that bubble load of attached material decreased along the bank, decreasing the amount of attached material entering the froth phase and consequently solids concentration in the froth phase. The trend in the zinc scavenge bank agrees with Seaman's findings and supports the hypothesis that solids concentration in the froth enhances its stability (Johansson and Pugh, International Journal of Mineral Processing, 34(1-2): 1-21. 1992). It is interesting to note that the trend is maintained for measurements at different vertical locations in the froth; however, the acoustic emission activity decreases closer to the surface. It is important to note that the sensor only reacts to activity occurring immediately in front of the probe, and this could give an indication of where exactly most of the material is being released due to bubble coalescence along the froth.

As indicated earlier, the sensor was tested overnight in the first zinc scavenger cell and its data compared to other instrumentation on site. Figure 23, shows the sensor signal simultaneously plotted with other plant signals which have been scaled to identify interrelated trends. Table 7 shows the high and low values of the plant instrumentation signals for ease of interpretation. In Table 7, the zinc grade is of rougher tails which accounts for most of the feed flow of the cell.

Table 7. Low and high values of various signals of relevance in the Zn scavenger cell.

According to Figure 23 and Table 7, most of the instrumentation signals remained constant for the period of observation with the exception of collector addition (SIPX), which increased substantially after midnight.

The zinc grade of the rougher tails stream, which accounts for most of the feed entering the cell being monitored varied slightly around the 7% mark, but no trend can be clearly established and related to the sensor signal. In contrast, there is an evident correlation between the sensor signal and the collector addition. In this case the SIPX dosage was drastically increased just after midnight, corresponding to a decrease in the sensor signal. This could be interpreted as an increase in froth stability caused by an increase in solids entering the froth.

In order to provide an output proportional to froth stability, the original sensor signal has been inverted and scaled, this new output signal is plotted simultaneously with the SIPX dosage signal in Figure 24. In this case the higher the value of the signal the higher the froth stability, which seems to correlate well with the increasing amount of collector used during the monitored period.

This test suggests that the sensor is capable of uninterrupted functioning for periods of time up to a day.

Finally, tests in a 3m³ flotation cell were performed with different frother types, MIBC, Frother B and Frother C. Experiments were carried out at different concentrations and froth depths. For MIBC the froth depths were 20, 30 and

40 cm. However, due to the high froth carrying capacity at high concentrations of frothers B and C, the flotation cell was not able to maintain a shallow froth; consequently, experiments were performed at 25, 30 and 40 cm. For the sake of brevity and clarity, only results at 30cm for MIBC are shown here, since similar trends are observed for the other frothers.

Figure 25 shows sensor signals for MIBC at different concentrations, at a froth depth of.30 cm. For all the experiments performed in this cell, the sensor was kept at a constant position of 13 cm below the surface.

As expected, increasing the concentration for all frothers results in a lower FAE signal or a higher stability of the froth. There are however differences in the amplitude of the signals between frothers that seem to be related to water carrying capacity, as shown in Table 8.

It is interesting to note that at lower frother concentrations there is an oscillation in the sensor signal. The period of this oscillation is approximately one minute and cannot be attributed directly to the cell mechanics. The oscillation diminishes as the frother concentration increases, suggesting that it is linked to the coalescence behaviour in the froth.

Without wishing to be bound by theory, the oscillatory coalescence process in the froth phase with the presence of particles could be explained as follows. As bubbles rise and coalesce, solids are detached and enter the plateau borders in a downward flow; this is the period of high coalescence or high sensor values. Accumulation of previously detached solids in the bottom of the froth phase consequently stabilises this part of the froth, which results in a reduction of coalescence and increasing the upward froth flow, corresponding to the period of low coalescence or low sensor values. At this point more solids are leaving than entering the froth; therefore coalescence is promoted again until it reaches a peak with solids being liberated into the plateau borders, starting a new cycle. Increasing frother concentration reduces coalescence in the high period, thus reducing the amplitude of the cycle. When higher amounts of frother are introduced the coalescence process becomes more constant in time. The cycle previously described is also visible at the concentrate overflow, and is characterised by a wave like behaviour.

Table 8. Sensor and metallurgical response to changes in frother concentration and type.

Overall, increasing frother concentration increases frother stability by reducing its coalescence. This results in higher recovery but at the same time higher entrainment which ultimately reduces the concentrate grade.

There is a clear correlation between the sensor signal and water recovery which is dependant on froth coalescence. Figure 27 shows the inverted sensor signal being linearly related to water recovery, demonstrating the effectiveness of the sensor in producing an on-line measurement of froth stability.

It will of course be realised that the above has been given only by way of illustrative example of the invention and that all such modifications and variations thereto as would be apparent to persons skilled in the art are deemed to fall within the broad scope and ambit of the invention as herein set forth.

Claims

CLAIMS:

1. A method for monitoring a froth phase of a liquid system including: detecting acoustic emissions in the ultrasonic frequency range using a sensor located within the froth phase; and analysing the acoustic emissions.

2. A method according to claim 1 , wherein acoustic emissions are detected in the ultrasonic frequency range of at least 20 kHz.

3. A method according to claim 2, wherein acoustic emissions are detected in the ultrasonic frequency range of from 30 to 50 kHz.

4. A method according to claim 3, wherein acoustic emissions are detected at an ultrasonic frequency of 40 kHz.

5. A method according to claim 1 , wherein acoustic emissions are detected at a plurality of positions within the froth phase of the liquid system.

6. A method according to claim 5, wherein acoustic emissions are detected in an array of positions in at least one horizontal and/or vertical plane of the froth phase, the positions being at a distance of up to 5 cm from one another.

7. A method according to claim 1 , wherein the detected acoustic emissions are amplified prior to analysis.

8. A method according to claim 1, wherein analysis of the acoustic emissions includes filtering of acoustic emissions signals detected at the resonant frequency of the sensor and calculation of Root Mean Squared (RMS) values in cycles of 1 second.

9. A method according to claim 8, wherein the acoustic emission signals are acquired at a sampling frequency of at least 10 S/s.

10. A method according to claim 1 , wherein the analysis of the acoustic emissions includes digitally processing acoustic emission signals to filter the signals and calculate RMS values.

11. A method according to claim 10, wherein the acoustic emission signals are acquired at a high sampling frequency of at least 150 kS/s,

12. A method according to claim 1 , wherein the analysis includes filtering of acoustic emission signals at the resonant frequency of the sensor, calculation of RMS values in cycles of 1 second and processing the RMS values using a microcontroller.

13. A liquid system including: a vessel containing a froth phase; a sensor for detecting acoustic emissions in the ultrasonic frequency range, the sensor being located within the froth phase; and means for analysing the detected acoustic emissions.

14. A system according to claim 13, wherein the vessel is a froth flotation vessel including a froth phase and a liquid phase.

15. A system according to claim 13, wherein the sensor includes an ultrasonic transducer and an amplifier.

16. A system according to claim 13, wherein the means for analysing the detected acoustic emissions includes a data acquisition module and a computer.

17. A system according to claim 13, including a plurality of sensors located at a plurality of positions within the froth phase of the liquid system.

18. A system according to claim 17, wherein the sensors are located in an array of positions in at least one horizontal and/or vertical plane of the froth phase, the positions being at a distance of up to 5 cm from one another.

19. A system according to claim 13, wherein the sensor is mechanically isolated from any vibrations resulting from operation of the system.

20. A system according to claim 13, further including one or more vibration attenuation devices.

21. A method for controlling operation of a liquid system having a froth phase, the method including: detecting acoustic emissions in the ultrasonic frequency range using a sensor located within the froth phase; analysing the acoustic emissions; and adjusting at least one operating parameter of the liquid system based on the analysis of the acoustic emissions.

22. A method according to claim 21 , wherein acoustic emissions are detected in the ultrasonic frequency range of at least 20 kHz.

23. A method according to claim 21, wherein acoustic emissions are detected in the ultrasonic frequency range of from 30 to 50 kHz.

24. A method according to claim 23, wherein acoustic emissions are detected at an ultrasonic frequency of 40 kHz.

25. A method according to claim 24, wherein acoustic emissions are detected at a plurality of positions within the froth phase of the liquid system.

26. A method according to claim 25, wherein acoustic emissions are detected in an array of positions in at least one horizontal and/or vertical plane of the froth phase, the positions being at a distance of up to 5 cm from one another.

27. A method according to claim 21 , wherein the detected acoustic emissions are amplified prior to analysis.

28. A method according to claim 21 , wherein analysis of the acoustic emissions includes filtering of acoustic emission signals detected at the resonant frequency of the sensor and calculation of the Root Mean Squared (RMS) values in cycles of 1 second.

29. A method according to claim 28, wherein the acoustic emission signals are acquired at a sampling frequency of at least 10 S/s.

30. A method according to claim 21 , wherein the analysis of the acoustic emissions includes digitally processing acoustic emission signals to filter the signals and calculate RMS values.

31. A method according to claim 30, wherein the acoustic emission signals are acquired at a sampling frequency of at least 150 kS/s,

32. A method according to claim 21 , wherein the analysis includes filtering of acoustic emission signals at the resonant frequency of the sensor, calculation of RMS values in cycles of 1 second and processing the RMS values using a microcontroller.

33. A method according to claim 21 , wherein the liquid system is a froth flotation system and the adjustment of at least one operating parameter of the system includes adjusting at least one of feed rate, frother addition rate, collector addition rate, impeller speed and air flow rate.