Jean Doucet

Alcan International Limited

1188 Sherbrookde St. West

Montréal, Québec, Canada

H3A 3G2


This paper presents the results of our study on the effect of the addition of coarse sand particles on the rheological properties of red mud slurries at different red mud concentrations up to the paste consistency regime. The rheological properties of red mud slurries were measured over a very wide range of solids concentration. Sand consisting of particles having a median size of approximately 400 m were added to the fine mud particles and their impact on the rheological properties of the mud were measured. An empirical mathematical model is derived and the results are discussed in terms of their impact and potential benefits on the operation of mud washing and mud disposal circuits of Bayer plants.



Jean Doucet


The conveying of fine solid materials is most economically done by using a two-phase system such as a gas-solid or a liquid-solid system in a conduit, but the technique has its problems. The most important ones are erosion along the wall of the conduit and settling of the coarser material at the bottom of the pipe mainly when the solid material has either a very wide size distribution or has a bi-modal size distribution. This segregation eventually leads to severe blockages of the transfer lines.

In order to overcome this problem, at least in part, and also to increase the efficiency of the technique, high density phase systems are used. In such systems the amount of fluid (gas or liquid) is obviously less and the velocity in the line can normally be reduced. In the case of solids particles in a liquid, providing the particles are sufficiently fine (less than 50 microns), and the solid volume fraction is sufficiently high the system will start to behave like a paste. The addition of synthetic polymers that act as coagulants or flocculants will also contribute in reaching that state of consistency. It has been observed that when such a dense phase transport system is used and the solid material contains a relatively large proportion of coarse particles, the particles can be maintained in suspension up to a certain concentration. The question then is, how much and how coarse the material has to be before the system becomes disturbed. In addition, if the material exhibits non-Newtonian behaviour, what is the impact of adding coarse material to the yield stress of the system or in other words, what is the initial resistance of the material to its displacement?

Since Alcan has developed over the years a technology called Deep Thickening that allows processed residues to be thickened to very high solids concentration, knowledge about the rheological behaviour and pumping characteristics of the resulting slurry or paste is of the utmost importance. The Deep thickening technology has been found to be applicable on a very large numbers of processed solid residues in addition to the red mud; copper and zinc, gold, tar sands, etc. All these materials, including red mud can contain a relatively large amount of coarse particles called ² sand² and which can be defined as the particles having a diameter larger than 75 or 100 microns. Some Australian bauxite’s can contain as much as 45 to 50% of sand. Even bauxite’s such as the Boke have levels of sand that can reach 15 to 20%.

One of the most widely studied and documented system which is in use in the industry is the transportation of coal slurry in a water carrier. Panda et al. established a model for calculating the yield stress with respect to the addition of coarse particles to a slurry of fine coal particles in water by analysing the laminar slurry flow.

In this paper we will present the results of work on the determination of the effect of the addition of coarse particles to a slurry made up of fine Bayer red mud particles. We will determine the effect on the rheological properties of the resulting slurry. We will also present a mathematical formulation derived from these data which model the change in yield stress with the addition of coarse particles. Finally we will discuss the impact of these results on the operation of Bayer red mud circuits.




Most fluids, especially slurries, do not behave as Newtonian fluids, ie. shear stress is not directly proportional to the shear rate. In fact most slurries behave as non-Newtonian fluids and show a yield stress (t 0).

A model known as the Herschel-Bulkley equation describes well the behaviour of slurries having an initial yield stress and a non-linear flow curve.

t = t 0 + K (1)

where n represents the flow behaviour index and K is a consistency index.


3.1 Material Preparation

Ten (10) kg of red mud was sampled from last stage of the washing circuit at the underflow of the Deep Thickener. The mud sample digested from a mixture of 75% Trombetas bauxite and 25% Boke bauxite was then separated in two fractions according to its size distribution using the wet screen technique. One fraction contained material finer than 75 microns, the other fraction was made up of the material coarser than 75 microns. This fraction will be called sand fraction.

The red mud sample was then left to settle to end up at a concentration of approximately 48% solids or 780 g/l and then centrifuged to reach a concentration of approximately 60% solids. The liquor was kept for further dilution. The other mud solids concentration were obtained by dilution of this mud with the overflow liquor. The dried sand was then added in quantities equivalent varying fromto 10 to 50% of the total solid fraction in the final mix. For a mix originally containing 48% of fine solids and a final sand content of 20% per total weight of solids, the final solids concentration is 59%.


3.2 Size Distribution

The size distribution of the two size fractions was measured using two different techniques. The size distribution of the sand or coarse fraction was measured using the vibrating screen technique (Rotap). The size distribution of the finer size fraction was measured using the Laser Coulter size analyser model LS 230. This last fraction was washed with water and methanol and then exposed to ultra-sound prior to the determination of its size distribution in order to deflocculate the fine agglomerated particles.

3.3 Rheological Properties

The visco-elastic properties, flow and constancy indices, of the mud samples were determined using a Haake VT500 rotational viscometer equipped with a cylindrical Sensor System SV1.

Separate yield stress measurements were carried out using the star shaped Sensor Rotor FL100. The method when applied to red mud slurries allows the determination of the constants of the Herschel-Bulkley rheological model as discussed in the previous section.

3.4 Sand Segregation Test

At various mud solids concentration varying between 30 and 45%, 15% of sand by weight of solids were added to the suspended slurry in a 500 mL cylinder. The system was left to settle for 30 minutes. The bottom half of the material contained in the cylinder was sampled and the sand fraction was determined.

In the case of the 38% solids mud sample, the test was conducted in two different ways. The first time it was done according to the method described above. The second time, the sand was well mixed with the mud. The cylinder containing the mixture was installed on the top of a vibrator used for wet screening for a period of 30 minutes. After the treatment, the top and bottom portions of the cylinder were analysed for their sand content.


4.1 Material Characterisation - Size Distribution

The median size of the fine mud particles is 3 m with a spread of about 1.6 m and no material coarser than 75m . The sand size distribution curve shows a median size of 420 m .

4.2 Yield Stress vs. Sand Addition

The yield stress of red mud samples from which the sand was removed, was measured over the range of 30 to 60% solids. The results are plotted and presented graphically in Figure1.


    Portions of sand particles were then added to this mud and the yield stress was again measured. The total % solids concentration was plotted against the corresponding yield stress for each given set of % sand addition. The resulting curves are shown in the graph of Figure 2.

    From the graph of Figure 2, we can see that the addition of sand to a slurry of fine particles obviously increases the solids concentration but does not significantly change the yield stress value of the resulting slurry. This can be seen by the curve in the graph of Figure 3 which represents the yield stress of a 36% solids sand-free slurry to which up to 50% of sand has been added.


  • 4.3 Rheological Behaviour of Mud

    To a mud sample containing 48% solids prior to the sand addition, amounts of sand varying from 10 to 60% of the total final weight solids content were added. For each of these mixes, the shear stress was measured at various shear rates. The resulting curves are plotted in the graph of Figure 4.


    The Herschel-Bulkley curves have been superimposed on the respective curves. Table 1 gives the parameters of the Herschel-Bulkley equation corresponding to these curves.

    At low shear rates, the experimental data points are usually not very reliable with the type of instrument used to measure the shear stress at different shear rates and are not shown on the graph of Figure 4. When the sand content in the solids fraction of the slurry reached 55% or more, the apparatus behaved very erratically and did not give reproducible results; for that reason they are not presented on the graph. No conclusion can be drawn on the behaviour of red mud slurries containing sand fractions of 55% or more.

  • Table 1

    Parameters of Herschel-Bulkley Equation For Red Mud With Sand

    Sand fraction


    Yield stress




























    4.4 Sand Segregation and Initial Concentration for Paste Consistency

    As indicated in section 3.4, two methods were used to determine the solids concentration required to maintain the sand in suspension in the slurry. The first method consisted in adding sand to the top of the suspended slurry and monitor the behaviour of the sand. Table 2 summarises the results obtained.

    Table 2

    Sand Segregation Determination

    Mud solids


    Sand fraction (bottom portion)

    33 %


    38 %

    0 %


    0 %

    A second set of experiments were done using a mud containing originally 38% solids. This time the mud was well mixed with the sand and the mixture vibrated for 30 minutes. Samples collected in the top half and bottom half indicated that the sand concentration was uniform throughout the tested material.

    The results from these two experiments indicate that the system begins to exhibit a change in behaviour and can maintain the sand fraction in suspension when the mud sample reaches a solids concentration of approximately 35%. This corresponds to a yield stress value of about 20 Pa


    5.0 Discussion

    5.1 Modelling Test Results

    It can be seen from the results presented in section 4 that despite the increase in the total % solids of the slurry with the addition of sand particles, the rheological properties of the mud are not largely affected at least when up to 50% of sand is present. In addition, it is mainly the yield stress component of the rheology equation, which is affected as noted by Panda et al.

    Our review of the literature did not allow us to find a mathematical model that fit our experimental data, hence we developed our own empirical model which gives an excellent correlation with the observed value.

    • Sand is the sand fraction in the solids expressed as a percentage of the total weight of the solids
    • %Tsolids represents the total solids concentration in the sample (fine mud+sand).
    • The constants a, b, c and d where determined by curve fitting as well as the exponent n which was found to be equal to 1.5.


    The graph of Figure 5 shows some experimental data to which have been superimposed the calculated curves obtained from equation 2. To obtain these curves the constant of the equation have been set to the following value:

    a = 0.3 b = 11.3

    c = 600 d = 20

  • Experimental Yield stress data have been plotted against the calculated value for various total % solids and sand fraction added to the slurry. The resulteds are plotted ion the graph of Figure 6 and show a correlation coefficient R2 equal to 0.93.


    5.2 Impact on Mud Circuit

    An important consequence of these results on the operation of a mud washing circuit is the possibility of retaining coarse sand particles with the rest of the slurry when a paste consistency regime is established. It then becomes possible to bypass the cyclones or "sand trap" in a mud circuit equipped with decanters operating with sufficiently high % solids so that a paste regime can be established.

    The benefits are:

    5.3 Impact on Pumping

    Boeger et al have shown that the pumping Energy is related to the yield stress of the mud. When the mud flows in a laminar regime, which is the case when the mud has reached a concentration such that it exhibits the characteristics of a paste, based on Boger observations, there is roughly a linear relationship between the yield stress of the mud and the pumping energy requirement as it can be seen on the graph of Figure 7 deduced from Boeger’s et al data.

    The slope of these curves beyond 50 Pa, which corresponds to the yield stress where the mud exhibits a laminar flow, was calculated and found to be of the order of 5 W/m Pa. From this number, we can estimate the additional power requirement due to the addition of sand to a mud sample, which is already at sufficiently high solids concentration to maintain sand in suspension in the slurry. The following table gives the calculated incremental power requirement due to the addition of sand.

    Since we have already determined that the yield stress increases exponentially with the solids concentration when the increase in the % solids is due to the addition of sand (see fig. 3), the additional pumping requirement will also increase exponentially with the addition of sand to a given mud slurry.


  • Table 3

    Calculated Pumping Requirement

    Original mud concentration


    Sand fraction


    Final solids concentration


    Yield stress


    Incremental Pumping Req.



























    However, this increase in pumping requirement is anticipated to be less than what would be required if the incremental solids concentration came from equivalent amount of very fine mud. A sand-free mud having 52% solids, pumped at a flow rate of 150 kg/s (dry basis) is estimated to require a pumping energy of the order of 300 to 400 W/m according to Boeger et al. data. On the other hand, a mud having a total solids concentration of 52%, 15% of the solids fraction being constituted by sand, is estimated to require only between 150 and 250 W/m at the same pumping flow rate.



    From the results obtained, we can conclude that it is primarily the fine particles of the mud, those below 10 m in size that essentially dictate the rheological behaviour of red mud slurries. The yield stress of red mud slurries are much less affected by the increase of solids concentration due to the addition of large sand particles than the addition of fine mud particles.

    From the results obtained, it has also been shown that above a given yield stress, the slurries start to behave like pastes and can support at least up to 50% of coarse sand particles without significant segregation.