Gravity thickeners are typically used to separate the liquid and particulate components of a slurry by the dynamic creation and removal of a portion of the mud bed at the base of the thickener.

A portion of the mud bed is conventionally extracted from the base of the thickener using an underflow pump, and the level of the mud bed is indirectly monitored by measuring the underflow density and comparing it with a density set point value, with the underflow pump speed being adjusted accordingly.

Ideally, the mud bed needs to be kept at a substantially constant level, which means that the underflow pump needs to extract a portion of the mud bed from the base thereof at a rate which essentially corresponds to the rate of formation of the mud bed.

Over the past few years, thickener technology has accelerated rapidly with the advent and general acceptance in the industry of so-called"deep cone"or "high compression"thickeners. These devices have the ability to generate high density underflows which approach paste-like consistencies, and have allowed the previously unattainable concept of thickened tailings disposal to become a reality. With the advent of these thickeners, the control of the level of the mud bed has become a critical problem due to the relatively high internal rise rate (typically above 11 metres per hour) associated with such thickeners.

Underflow density measurement as a means of monitoring mud bed level has some serious shortcomings. There are many factors which influence the underflow density, including slimes-to-grits ratio in the feed, the degree of flocculation, settling rate of the flocculated slimes and the like. As a result, underflow density is not a uniform or predictable function of mud bed level. In the past, the control of the mud bed level on the basis of underflow density has led to mishaps in which the internal structures of high compression thickeners have been damaged and have catastrophically failed under the pressure of an overflowing mud bed. Failure to accurately detect the level of the mud bed has also resulted in the entire mud bed and overlying water or diluted slurry being drawn down through the underflow pump, resulting in a new mud bed having to be formed, and considerably reducing the efficiency of the thickener.

SUMMARY OF THE INVENTION According to a first aspect of the invention there is provided an apparatus for detecting and controlling the level of a mud bed within a gravity thickener, the apparatus comprising a first vibrating probe which vibrates at a predetermined amplitude and frequency and first mounting means for mounting the probe substantially vertically in the thickener at a predetermined position in which it is arranged to detect a preset mud bed level on the basis of vibration dampening and to generate a first control signal in response thereto, the first control signal being arranged to adjust the throughflow of mud via an underflow pump so as to control the level of the mud bed within the column thickener.

Preferably, the first probe is arranged to detect a preset high mud bed level and the first control signal is a"high"signal, which is arranged to increase the throughflow of mud via the underflow pump so as to reduce the level of the mud bed.

Advantageously, the apparatus includes a second vibrating probe which vibrates at a predetermined amplitude and frequency, and second mounting means for the second probe mounting substantially vertically in the thickener at a predetermined position in which it is arranged to detect a preset low mud bed level, and to generate a second control signal in response thereto, the second control signal being arranged to decrease the throughflow of mud via the underflow pump to maintain, in conjunction with the first control signal, the level of the mud bed between the high and low mud bed levels.

The first and second control signals may be arranged respectively to turn the underflow pump on and off or to increase and decrease the speed of the pump.

One or more intermediate vibrating probes may be located between the first upper and second lower probes, the intermediate probes being arranged to generate control signals to vary the mud bed level by increasing or decreasing the speed of the underflow pump.

Alternatively, the first vibrating probe may be arranged to deliver an analog signal which is a function of the mud bed level along the length of the probe, whereby the signal is arranged to vary the speed of the underflow pump in response to measured changes in mud bed level.

Advantageously, the vibrating probes are provided with adjustable switching set points arranged to generate control signals.

Conveniently, the probes are arranged to sense differences in SG of as little as 0.04, and to switch between a setpoint in response thereto.

The thickener is preferably in the form of a high compression thickener. By the term"high compression thickener"is meant a thickener having an internal rise rate which is generally greater than 11 metres per hour, under a relatively high duty or processing rate of twenty to 50 tons per square metre per day. Such thickeners have relatively small diameter ranging from 1 to 15 metres, and a shallow feed well. Flocculant is required, and the mud bed which is created within the thickener is relatively deep, having a height of up to 10 metres.

Such thickeners may or may not be provided with a secondary dewatering device in the form of a"rake"system. Typically, in cases where the underflow density is less than or in the region of 1.6 kilograms per litre, no secondary dewatering device is required. Where the underflow density is greater than 1.7 kilograms per litre, a secondary dewatering device is generally used.

The mud bed level detecting method and apparatus of the invention may also be used with so-called"high rate"and"conventional"gravity thickeners. A high rate thickener typically has an internal rise rate from 4 to 10 metres per hour, and has a diameter ranging from 10 to 60 metres. A thickener of this type requires a floculant and generally utilizes a deep feed well. A mud bed of up to 5 metres in height is created, and such thickeners are capable of processing from 12 to 19 tons per square metre per day. The underflow density of the mud from a high rate thickener can typically go up to 1.55 kilograms per litre.

A conventional thickener typically has an internal rise rate of less than 3 metres per hour and may be up to 100 metres or more in diameter. Floculant is not essential, and a relatively shallow mud bed of up to 2 metres is utilized.

Conventional thickeners have a relatively low processing rate of 5 to 7 tons per square metre per day, and are able to handle an underflow density of 1.3 to 1.4 kilograms per litre.

The invention extends to a method of controlling the level of a mud bed within a gravity thickener comprising the steps of sensing the level of the mud bed using a first vibrating probe immersed in the mud bed, generating a first control signal based on the extent of vibration dampening of the first vibrating probe, and using the first control signal to adjust the rate of extraction of mud so as to control the level of the mud bed within the thickener.

Preferably, the method includes the steps of sensing a high mud level using the first vibrating probe and using the first control signal to increase the throughflow of mud via an underflow pump to reduce the level of the mud bed.

The method conveniently includes the steps of sensing a low mud bed level using a second vibrating probe, generating a second control signal based on the extend of vibration dampening of the second vibrating probe, and using the second control signal to decrease the throughflow of mud via the underflow pump to maintain the level of the mud bed between the high and low mud bed levels in conjunction with the first vibrating probe.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 shows a schematic cross-sectional side view of a first prototype embodiment of a mud bed level controller of the invention fitted to a high compression gravity thickener; Figure 2 shows a graph of interface density versus time, indicating the operation of the mud bed level controller of the invention; Figure 3 shows a more detailed cross-sectional view of the manner in which the vibrating probes are fitted within the high compression thickener; Figure 4 shows a top plan view of the thickener of Figure 3; and Figure 5 shows a detail of the manner in which the vibrating probe is located within a guide pipe within the thickener.

DESCRIPTION OF EMBODIMENTS As was mentioned in the background of the invention, it has been accepted for some time that the underflow density measurement-based control philosophy has serious shortcomings, in particular with respect to high compression thickeners. Numerous different level detection instruments were investigated on a laboratory scale before a pilot plant was set up to conduct a more detailed evaluation. Measuring instruments for measuring material levels on the basis of capacitance, conductivity, torque, ultrasound speed, ultrasound dampening, turbidity and finally amplitude of vibration were eventually shortlisted for further evaluation. All of the off-the-shelf instruments selected were specifically designed to measure liquid levels by detecting an air and liquid interface. In mud bed level detection, it will be appreciated that the properties of two different slurry layers of marginally differing densities are far more akin and less easily measurable than the disparate properties of air and water, where numerous different measurement techniques can and have been used.

Initially, all of the instruments were evaluated by a dip-in method in slimes of different densities, temperatures and solids slimes-to-grits ratios. The susceptibility of the respective measurements with regards to these parameters were investigated. A three-by-three matrix of nine samples were generated, having SG densities of 1.1,1.2 and 1.3, and a slimes-to-grits ratios including no grits, a 1: 2 ratio and a 1: 1 ratio, with grits being defined as having an average particle size ranging from 300 micrometers to 1mm, and slimes having an average particle size of less than 300 micrometers. Temperature changes were simulated by placing the samples in a water bath controlled at one of 20°C, 30°C and 40°C.

Capacitance, conductivity, torque, amplitude of vibration, ultrasound speed, ultrasound dampening and turbidity measurements were all investigated in depth. The end result was that amplitude of vibration using an off-the-shelf VEGAVIB 52 vibrating probe was found to be the most satisfactory level detection means. This probe vibrates at a fixed frequency, with the amplitude and the frequency of vibration changing according to the amount of dampening that occurs. The probe does not have an analog output, but is rather provided with an adjustable switch point, so that the specific amplitude at which the switching will take place is adjustable. It was surprisingly found that an extremely sensitive switch point adjustment was possible with this probe, in that it recognized a difference in amplitude between respective sample densities of 1.1 SG, 1.2SG and 1.3SG, as well as between sample densities of 1.14SG and 1.18SG. It was also found that the amplitude of vibration was primarily SG-and viscosity-dependent, and was not affected by the above mentioned temperature and size distribution variables.

In the subsequently established pilot plant, a commercially available Ultrasep23 high compression thickener supplied by Bateman Process Equipment (Pty) Ltd of Barlett Road, Boksburg North, Gauteng, South Africa, was used.

Unflocculated thickener feed slimes from Premier mine were used in the operation thereof. As the natural pH of the slimes was relatively high, at 10.25, the pH was dropped to below 8 by means of adding H2SO4 to the feed slimes to assist in flocculation. The floculant used was AD2, which was supplied by Ore Pro Consultants.

The average density of the feed material to the pilot plant was in the region of 1.04 to 1.05SG, with the flocculated feed material having an average density of approximately 1.1SG. The switch point of the VEGAVIB 52 vibrating probe was adjusted to switch in densities higher than the flocculated feed density in an attempt to detect the mud bed level, which is typically in the region of 1.2 to 1.3SG.

Referring now to Figure 1, an U ! trasep@ high compression thickener 10 is schematically shown fitted with a first VEGAVIB 52 vibrating probe 12. The U ! trasep@ thickener includes an uppermost slurry inlet pipe 13A and a lowermost mud outlet pipe 13B which extends from the base of a conical sump portion 14 of the thickener. An upper flanged tank insert 15 is inserted between two main sections of the tank. The insert 15 incorporates a recess 16 within which the first probe 12 was vertically mounted. A sampling point 18 was installed at a position close to the tip of the probe 12. The probe was initially adjusted not to switch on the flocculated feed material having a density of 1.06SG. On switching of the probe, a sample was manually extracted from the sampling point 18 and the density was measured to be 1.3SG. At this point, the underflow at a discharge end 20 of an underflow pump 22 was measured by an underflow density measuring meter 23 as having a density of 1.5SG. The mud bed level was then dropped by manually starting the underflow pump 22, as a result of which the probe 12 reverted to its original non-switched state.

As the VEGAVIB probe is arranged to deliver an adjustable switching signal, and is incapable of delivering a variable analog signal, a control philosophy was adopted in which the high level probe 12 was supplemented with a low level probe 24. The low level probe 24 was similarly fitted in a recess 26 within a second flanged pipe insert 28 located directly below the first pipe insert 15.

When the mud bed reached a level 30, this caused the vibrating probe 12 to switch, thereby delivering a signal via a control line 32 to switch on the underflow pump 22. The vibrating probe 24 was used to detect a lower mud bed level 34. At this level, the vibrating probe 24 was adjusted to transmit a signal via control line 36, thereby switching the pump 22 off and allowing the mud bed level to increase.

Referring now to Figure 2, a graph of density (SG) vs time is shown. A lowermost broken outline plot 38 indicates the on/off status of the underflow pump 22, and an intermediate plot 40 indicates the interface density or actual density measurements of the mud bed at both the high level 42 when the underflow pump 22 starts and the low level 44 when the underflow pump stops. It was found that the high and low level density measurements were not exactly the same, as would theoretically have been expected. This was probably due to the fact that the mud bed is not uniformiy level, with the sampling points 18 possibly producing a sample that was not fully representative of the material surrounding either of the vibrating probes 12 and 24. Nevertheless, it was visually observed that the mud bed level never exceeded the high level position, thereby successfully protecting the internal structures of the Ultrasep0 unit. By way of reference, an uppermost plot 46 showing underflow density was included. This plot shows a gradual reduction in density, but is not sufficiently sensitive or responsive to indicate variations in mud bed level shown by the plot 40.

In Figures 3 and 4, a large commercial Ultrasep high compression unit 50 is shown. This includes a dissipator cone 52, an inner feed well 54 leading to a re-circulator cone 56, and an overflow launder 58. High and low level guide pipes 60 and 62 respectively extend vertically from a walkway 63 at the top of the unit. As can more clearly be seen in Figure 5, the vibrating probe 12 is fitted within the guide pipe 62, with the probe head 64 sitting on gusset plates 66 located at the base of the pipe for centering the probe. The probe cable 68 extends to a probe control box 70. It is clear from Figure 5 how the pipe 60 extends through an opening in the dissipator cone 52.

Certain critical levels have been identified in the Uttrasep@ high compression thickener unit 50. These include a so-called"high high level"74, which is 100mm below the re-circulator cone 56, a"high level"76, which is 100mm below the top of a Trioid0 78, and a"low level", which is the lowermost operating level 80 just below the Trioid (D. A suitable additional guide pipe 82 may also be installed, to which a lowermost vibrating probe is fitted to detect the lowermost level 80.