Testwork and sizing for paste thickeners

As we approach the Paste06 conference in Ireland at the beginning of April, Ian Arbuthnot, a Director of Outokumpu Technology notes that thickening slurries to the maximum possible density (paste) is becoming increasingly common.  Applications such as Autoclave feed thickeners, paste backfill and some counter current washing (CCD) circuits are all examples of potential paste thickening applications.  The principle of paste thickeners is that the compression zone is much deeper than on either high rate (HRT) or high compression (HCT) units; generally 3 m or more. As the compression zone gets deeper, the underflow density for a given flux rate increases, as a result of two factors:

  • Longer bed settling time. Settling flocs move relative to one another and release interstitial fluid when they are in compression.  Increasing bed residence time achieves higher density underflow as more fluid is removed.
  • The net weight of the bed creates a compressive force on the settling flocs.  This force increases with depth, so that the flocs at the bottom of the bed experience the highest compressive force. This represents the driving force to expel interstitial fluid and hence increase bed density.

These two factors work together to increase pulp density, but it is generally not possible to separate the effect of each factor in a dynamic situation.  One of the challenges with paste thickening applications is the accurate testing and thus sizing of the units.

To size high rate and high compression thickeners, dynamic bench-scale or pilot-scale testwork is the preferred approach. However, static cylinder testwork and the application of the Talmage and Fitch construction is still sometimes used. The advantage of dynamic testwork is that a reliable relationship is established between the solids flux rate, underflow density and overflow clarity. The other significant parameter is flocculant dosage, and some attempt is usually made to optimize this as well.

It should also be understood that if the solids flux rate changes (i.e. an increased feedrate and underflow withdrawal rate for a given cross-sectional area), the liquid rise rate also changes, resulting in a different bed dynamic and a variation in settled density. When considering paste testwork, the additional bed depth used can be a problem as a much larger sample is required. To avoid this, attempts have been made to define a parameter termed ‘unit mud volume’ for scale-up purposes. Unit mud volume is defined as “the volume of thickener bed (mud) required to achieve a given underflow % solids at a throughput of 1 t/h of dry solids”.

For example, a laboratory or pilot-scale test unit might be operated to give a desired density at a flux rate of 0.3 t/m2.h and a bed depth of 2 m, giving a unit mud volume of 6.67 m3/t/h (bed depth/solids flux rate). This parameter is then used to calculate the required bed depth at a significantly higher flux rate, e.g. 6.67 m depth at a flux rate of 1.0 t/m2.h.  The problem with this approach is the need to extrapolate from one flux rate to another without having data on how the higher flux rate will affect the settling characteristics of the flocs.

To illustrate this further, Figure 1 shows the relationship between unit mud volume and solids flux rate at a constant underflow % solids for a copper tails sample-:                                        

We can see that the unit mud volume varies, in fact, increases, as the solids throughput of the thickener increases. Figure 2 shows the corresponding relationship between solids flux and bed depth.

Taking the simple example of a thickener with an area of 1m2 we see from Figure 2 that to treat 12 t/d (0.5t/m2.h flux rate) and produce a 59% solids underflow, we would require a mud bed depth of approximately 0.72 m. Increase throughput to 24 t/d  (1.0t/m2.h flux rate) and the required bed depth increases to 3.36 m to produce the same 59% solids. This represents an increase in bed depth of nearly 5 times. Finally, at a slightly higher throughput of 30 t/d (1.25t/m2.h flux rate) and the required mud bed depth increases to about 7 m.

 

Using the constant unit mud volume idea would suggest that the bed depth need only increase from 0.72 m at 12 t/d throughput to 1.8 m deep at 30 t/d throughput. Quite clearly this would create a situation where the required density of 59% solids would not be achieved.

Another way to review the data is to consider bed depth v % solids, Figure 3.  This figure indicates that an increase in bed depth allows either an increase in the % solids of the underflow or an increase in the throughput at a given underflow density.  Clearly these relationships are not linear and show a tendency to some maximum as bed depth increases. This illustrates the diminishing returns available from each increase in bed depth, again illustrating that the constant unit mud volume concept is flawed.

 

 

Figure 3. Bed depth vs % solids in the underflow for different flux rates (t/m2.h)

What this analysis clearly shows is that increasing the mud bed depth does provide a degree of control over the thickener underflow performance.  Deeper beds can be used to optimize the underflow density at a given throughput or optimize the throughput at a target underflow density.  What is also highly evident is that the mud bed volume for a given material is not constant.  It depends upon the solids flux rate and target underflow density.  As one would expect, the unit mud volume required to achieve a given underflow density increases as the thickener throughput increases.

 

This also leads to the conclusion that, inconvenient as it might be, the only reliable way to predict underflow density for a given flux rate and bed depth is to run small-scale tests at the flux rates and bed depths required. Since a difference in actual paste thickener performance of 1 or 2% in density can be crucial in some applications, a very limited amount of extrapolation is usually acceptable.

 

It is very often the case, e.g. for greenfield projects, that limited sample is available to carry out thickener tests. Hence, our approach would usually be to carry out initial testwork in a bench-scale high rate thickener rig to establish primary settling characteristics. These would normally include response of underflow density and overflow clarity to flocculant type, feed dilution, flocculant dosage and solids flux. Typically, with a sample size of around 10-12 kg of solids, a set of curves in the ‘high rate’ mode can be generated as below:

With this information, a set of operating conditions can be selected and one or two tests conducted in a deep bed test rig. This is a laboratory-scale unit with a diameter of 190 mm and height up to 4 m. It has been established that with the 190 mm diameter, ‘wall effects’ are negligible, and the results are equivalent to pilot scale. For data obtained, see below:

When underflow samples are taken from this rig, a vane rheometer is used to characterize the underflow material. These rheology measurements provide an important means of determining the practical configuration for a paste thickener on scale-up, including the mechanism and drive requirements. The flux rate/bed depth/underflow % of solids data are used to determine tank diameter and depth.

The concept of a constant unit mud volume appears to be very attractive when scaling up thickeners.  However, the reality is that the unit mud volume is neither constant nor predictable at different throughput rates. Testing at the desired throughput and bed depth conditions is the only way to reduce the risk associated with scaling up paste thickeners.