IM 2016 June 2016 | Page 77

PASTE WESTECH_proof 25/05/2016 09:45 Page 2 Paste Supplement In the lab, movement of the sensor through the samples provided a measurable response over a range of solids concentrations that would be expected in a thickener bed rake arm at an elevation approximately 600 mm above the bottom of the tank wall. A circular disk on a tubular arm transmits the drag force to the load cell where the output signal is generated. Wiring from the sensors was routed up the support structure to the transceiver, located in a protective enclosure on top of the support and above the liquid level of the thickener. The thickener is a 22.86 m diameter, high density thickener with a 3.66 m tank sidewall, producing 40-60 Pa yield stress underflow. The process dewaters insoluble material from soda ash production for either underground or surface deposition. The thickener has a peripheral walkway that allowed convenient access to the instrument and provided the ability to make manual bed level readings at any angular position around the tank. Current thickener control is based on the operator conducting manual bed level reading every four hours. These bed readings were used to control the underflow discharge rate. Additional manual readings were taken at other times as needed to correlate with instrument data. The plant procedure for making manual measurements was to lower a long PVC pipe into the tank at a shallow angle until the operator could feel the bed. Then maintaining the feeling of the bed, the pipe would be transitioned to vertical along the tank wall. Markings on the pipe allowed operators to measure the depth of the bed below the liquid level. Testing of the prototype instrument started with the focus of detecting bed level. The instrument consisted of three sensors spaced 300 mm apart vertically. The output signals from the sensors as the mud bed level varied, during normal operation, demonstrated that the bed level interface could be clearly identified. Sensor signals were seen to increase from zero to maximum output before the bed level was even detected by the next higher sensor, 300 mm above it. This indicates a very distinct bed level interface and that the instrument was able to detect it. Correlation was demonstrated to exist between manual bed level measurements and instrument signals. Sensor signals are proportional to bed yield stress. Test data indicates that as bed level rises above the level of a sensor, the yield stress at that sensor also rises. This indicates that a relationship exists between yield stress at a position in the settled bed and the depth of the bed above that position. Potentially a single sensor may be able to detect not only when the bed level reaches the elevation of the sensor but also estimate the bed level as it rises above the elevation of the sensor. However, this will need to be confirmed through site specific operating experience. Accurate bed level sensing will enhance the ability to control the thickener. Data from testing the prototype instrument detected significant angular variability around the thickener. The cyclic nature of this data correlated to the period of rotation of the mechanism. Plotting the data of one sensor (one elevation) showed the settled bed had areas of high yield stress while other areas had low yield stress, indicating a significant variation in solids concentration and particle size distribution in the bed as a function of angular position. Further investigation revealed the cause was due to poor distribution from the feedwell. A current was carrying coarse solids to one side of the tank. This segregation within the tank was happening even though the manually measured bed level showed very little variation at different locations around the tank. The observed segregation within a thickener is known as an island and its presence was previously unknown. Having this amount of information about the state of a thickener bed has important implications for thickener design, operation and control. Instrument potential in thickening Developing a bed profile, as was demonstrated above, has the potential of detecting other operating problems within thickeners that are often difficult to diagnose. This may be done by using both vertical and horizontal sensor arrays. The ailments that may be possible to observe include rat-holes, shortcircuiting flows, doughnuts and rotating beds. In addition, this instrument can show how the bed becomes denser with depth and how that changes with varying operating conditions. A vertical bed profile that is out of the ordinary may indicate changes to the flocculating characteristics of the feed or changes in particle size distribution. The potential of this instrument as a diagnostic tool for thickeners is significant. Testing clearly shows the potential of this instrument goes far beyond measurement of the mud bed level. The ability to measure and observe the dynamics of the thickening process opens up possibilities in the fields of thickener control, process optimisation, thickener design and theoretical modelling of thickeners. The common thickener theory describing a bed as radially homogeneous is already in jeopardy. The data suggest non-homogeneous distributions and variable angular, horizontal and vertical density gradients are more likely. The magnitude of sensor signals was found to have a linear relationship to the yield stress of the material. The instrument described provided data from which the bed level interface inside a thickener tank can be determined. It also detected increasing yield stress as the bed level increased above the elevation of the sensor. This provides unprecedented information from within the thickener bed that could be used for early warning of reduced thickener performan