Automatic Control of Flotation Columns - Advanced Control and Supervision of Mineral Processing Plants - page 270

tures of the UCT bubble sampler. This bubble size analyzer (HUT-BSA) consists of
a sampler tube, a viewing chamber and an image acquisition system. The sampler
tube is partially immersed in the liquid phase, below the froth-liquid interface, where
it is used to collect rising air bubbles, which are drawn into the viewing chamber.
The function of the chamber is to expose the bubbles to a monochrome CCD camera
which is connected to a personal computer, where the visual information is stored
and processed. The images captured during sampling are automatically processed
using commercial image analysis software and in-house visual basic user interface.
A major drawback of the present approach for analysing bubble distributions
obtained from both the McGill and the HTU sensors is the fact that the use of an
average bubble diameter ( D 10 and D 32 ) involves a compression of the whole BSD,
where information regarding the shape of the distribution, such as tails and the pres-
ence of multiple modes, is lost. Different probability density functions all having
the same D 10 and D 32 value can exist, which means that the use of a single aver-
age value would not allow the assessment of the multi-modal, narrowness and tails
characteristics of the actual distribution.
A method to on-line estimate the whole BSD represented by its bubble size prob-
ability density function (PDF) has been recently proposed at Universit´eLaval[54].
A Gaussian mixture model (GMM) is used to represent the bubble size PDF and
a modified version of the well-known expectation-maximization algorithm (EM) is
utilized to estimate the GMM parameters evolution in time. An appealing feature of
this method is that the mean diameters D 10 and D 32 can be directly obtained from
the GMM parameters, thus obtaining dynamic estimates.
Two other devices have been proposed by Brazilian researchers for the measure-
ment of very fine bubbles. The first sensor (LTM-BSizer) was developed for two-
phase systems only, but the authors mentioned future plans for three-phase system
testing [55]. The sample is driven from the flotation device (a laboratory column) to
a horizontal viewing chamber where bubbles decelerate and eventually stop when
the downstream pinch valve is closed. Images are then taken with the aid of a micro-
scope and a digital camera. When the pinch valve is opened the viewing chamber
is purged and refilled with a new sample and the process is repeated. Images are
processed using commercial software. The second device, the most recent one in
the literature, was designed to measure micro-BSDs on two-phase systems at labo-
ratory scale [56]. The sizing of the generated micro-bubbles was performed with a
Mastersizer 2000SM (Malvern Instruments) through light scattering, which allows
measurements down to 0.1 m. None of them seems, for the time being, to be good
candidates for industrial on-line application, at least for process control purposes.
6.3.5 Superficial Gas Velocity ( J g )
Another commonly measured variable whose importance for the assessment of the
hydrodynamic characteristics of flotation cells is undeniable, is the superficial gas
velocity ( J g ), defined as the ratio between the volumetric gas rate and the device