Experimental Investigation


Site Contents





Experimental Investigation





Effects of swelling, dispersion and flocculation on the saturated hydraulic conductivities of different clay-sand mixtures.


The main mechanisms that can influence the hydraulic conductivity of a soil are related to the swelling, dispersion and flocculation characteristics of the clay species within the soil matrix. The hydraulic conductivity changes are due to clogging of the soil matrix and hence retardation of the ability of water to percolate freely through the pore spaces. This investigation was designed to re-visit the nature of clays and specifically make comparisons of the saturated hydraulic conductivity (K) between sand-clay mixtures of kaolinite, a 1:1, non-swelling clay, bentonite, a 2:1 swelling clay and red mud through comparing a number of different scenarios relevant to cumulative clay content, dispersion and, flocculation. (top)


Many investigations have already shown how the clay content can influence the hydraulic conductivity (K). These have shown that K depends on the chemical composition, electrolyte concentration and the structural state of the soil matrix. The main mechanisms that can influence the K of a soil are related to the swelling, dispersion and flocculation characteristics of the clay species within the soil (Bouwer, 1978, p.20).

Clay dispersion and swelling is well recognised to reduce flow. Parameters, such as the intrinsic permeability of the soil matrix and re-arrangement of clay particles during percolation have been shown to change the saturated hydraulic conductivity (Ksat) in steady state flow cases (Aringhiere & Capurro, 1994). In natural situations dispersion of soil clays and the associated breakdown of soil structure, have been implicated in the process of soil crusting, again ultimately reducing K.

The main factor determining the infiltration and run-off characteristics in a natural situation, is the stability of the structural units at the soil surface during the beating action of raindrops. These impacts are capable of dispersing soil particles, which then lodge in the soil matrix, clogging pores, and hence reducing infiltration. However, it has been shown that polyvalent cations can flocculate clays during raindrop impact, maintain a more porous surface layer and increase infiltration rates from the low infiltration associated with poor soil structure (Miller, 1998). Soil structure generally relies on the clay content being aggregated within the soil matrix, and to a degree, 'propping up’ the surrounding soil, so that water can percolate through the soil.

The aims of this investigation were: firstly, demonstrate the effects that clay has on the Ksat of a sand column and, secondly to measure the effects that dispersion, flocculation and swelling of clays has on flow rates and Ksat. This will be achieved through comparisons of the two main groups of clay minerals, the 1:1 clays represented by kaolinite and the 2:1 swelling clays represented by bentonite. Comparison with alkaloam (red mud) will also be included because red mud consists of a small percentage of clay minerals (~10% w/w) and an abundance of inert clay size particles( 40% w/w) (Roach, 1992). (top)

Theoretical Background

The theoretical background for these experiments can be obtained by reading the sections on Structure and the Properties of the clay minerals.

The Clay minerals used in this experiment

The properties specific to each of the 1:1 and 2:1 layer silicates are outlined in table 1. Bentonite can be characterised as a 2:1 Na-exchanged montmorillonite, with a large capacity for swelling and flocculation. Montmorillonite has a large surface area and swelling potential (Table 1), in the case of bentonite up to 800 metres squared and volume changes of between 1400-2000 % when immersed in water (Velde, 1992).

Saturated Hydraulic Conductivity (Ksat) and Darcy’s law

Darcy’s law was used to convert flow rates into hydraulic conductivity values by measuring the coefficient G, using Excel. The flow was then plotted against the head height, assuming the observed flow follows Darcy's law, the plot line is considered straight and the slope can be calculated from the gradient:

          G =Δ H(cm)                                                                                     ............... (1)
                Δ Q(cm³/min)

where H is the total head and Q is the flow rate.

And with Darcy's Law being;

         Q = K .(H)                                                                                        ............... (2)
                A  . L

         G = K .(A) = Δ H                                                                              ............... (3)
                      L      Δ Q


         K = G. L                                                                                            ............... (4)

where, the length of the soil column is L (cm) and A is the cross-sectional area (cm ²). Δ Q was measured for two different values of H. (top)

Materials and Methods


The kaolinite and bentonite were commercial products obtained from Rheochem Pty. Ltd. via the Murdoch University technical staff of the School of Environmental Science. The red mud was obtained directly from the Alcoa, Kwinana tailings site. The clay samples were mixed, under direction from Rheochem Pty Ltd, by blending the samples with distilled water, using a normal handheld ‘bar-mix’ blender. 2..5 + 0.02 grams of each sample was blended for a period of 2 minutes with 1 litre of distilled water. The blending was to ensure that the clay was reasonably dispersed (especially in the case of the red mud). As the nature of the experiment is only for rough comparison, no attempt was made to ensure complete dispersal.

It is noteworthy to mention that the bentonite began swelling directly on contact with water and began flocculating after 2 -3 minutes. Secondly, no attempt was made to ensure that the solutions added to the sand was entirely made up of particles of size less than 2 µm (by definition clay sized particles). The red mud, which has a graded particle size from 150 -1 µm (Roach, 1992), much larger than the average particle size of the bentonite or kaolinite (Table 1), was allowed to settle for a period of ten minutes to allow the coarser particles to settle out before any experimentation was undertaken.(top)


For the flow experiments standard fixed head permeameters were used, they were made from Perspex with an internal diameter of 9.5 cm (A=70.88cm²). The receiving container was attached to the column with 1.5 cm plastic tubing. Graded gravel was added to the columns of to minimise the loss of sand from the columns.

Four columns were set up by layering 400 grams of 4 different pre-washed gravel. The first layer consisted of angular granite approximately 1- 1.5cm diameter. The other three layers consisted of progressively smaller gravel, with the final layer consisting of fine angular gravel approximately 0.3-0.5 cm diameter. On top of the stratified gravel layers was added 400 grams of pre-washed, coarse quartz sand of which 90 percent was between 3.2 mm and 2.6 mm diameter, obtained from Cook Industrial Minerals Pty Ltd.

The sand used in the final layer for mixing with the clay was also obtained from Cook Industrial Minerals and consisted of 99.5 percent graded quartz of which 90 percent was between 2 mm and 1.8 mm diameter. The sand was thoroughly pre-washed by agitation and flushing in a set of buckets, and was allowed to drain for a period of 24 hours. Each column had 1000 grams of this fine sand added to the top.The columns were then filled with water at a rate that did not allow much overflow. For a photograph of the permeameter and sand column set up click here.

Three of the columns had cumulative additions of red mud, kaolinite and, bentonite respectively, mixed with the sand. The mixing at each stage of the cumulative additions of clay was achieved by pouring the required amount of clay solution into the columns. For example, an initial addition of 0.5 grams of clay at a solution of 2.5 g/l required 200 millilitres to be added to the water column. The addition of the clay water suspension was carefully done to ensure clay was not lost in overflow. The flow rate was continued until the water colour had cleared, indicating that the clay had penetrated into the sand column. A ‘burst' of back-flushing, at a rate of 650cm³/min, was enough to mix the fine sand and clay together, without disturbing the layer of coarser sand. In this way the sand clay mix was achieved, and the mixture settled into a reasonable homogenous mixture with a flat surface.

The experiments were conducted by measuring the flow rates at two different head differences ΔH) of 50 + 0.02 cm and 64 + 0.02 cm for solutions of electrolytes namely, 9 litres of NaCl (Sodium chloride) at 16g/l, 400 millilitres of calgon (Sodium metahexaphosphate) at 5% w/w and, 1 litre of gypsum (Calcium sulphate) at 2g/l. The NaCl was initially run through the columns to flush them and, to ensure that reasonably uniform pH and ion species were present in each column, initial flow rates were then measured at this point (unconsolidated – NaCl). The columns were then left to consolidate for a period of 24 hours and the flow rates were then re-recorded (consolidated – NaCl). The solutions of calgon and gypsum were run ‘into’(were not flushed through) the columns and left to stand for a period of 24 hours, whereby the flow rates were then remeasured, (consolidated – calgon and, consolidated – gypsum respectively). Three replicates were measured of each flow and an average taken for the G and K calculations. Comparisons were made of the saturated hydraulic conductivity (Ksat) between consolidated- NaCl and unconsolidated –NaCl scenarios. The additions of NaCl (a weak dispersant at low to medium concentrations), calgon (a strong dispersant) and gypsum (a flocculent) were noted, to show the comparisons of the effects of these addition on the hydraulic conductivity of the clay sand mixtures.(top)

Rough Error calculation for K

% error for K = %error (H) + %error (Q) + %error (L) + %error (A)     ..............(5)
= 1 + 0.1 + 1 + 2
= 4.1%


The effects of increasing clay content for the red mud and bentonite decrease the overall flow rates, shown in Figure 1(a) and Figure 1 (b). Kaolinite shows no overall changes in flow at either difference in head of 50 cm or 64 cm. In all cases, bentonite produced a lower flow rate than either of the other two clays. Fluctuations in flow can be seen at lower additions of clay; specifically, the red mud shows an increase in flow at 0.5 grams; the kaolinite shows a slight trend upward at this amount as well.

The composite hydraulic conductivities are shown in Figure 2. The hydraulic conductivity show little overall change as more clay was added to the columns. All three clays show fluctuations in Ksat up to 1 gram of clay. Between 1 gram and 2.5 grams bentonite shows a progressive downward trend in K; Red mud shows a slight increase in Ksat and the kaolinite shows no dramatic change.

The flows were plotted for each head height for each scenario in EXCEL. G was calculated. The K values were calculated using equation (4) and displayed in Table 3. (L) for the unconsolidated columns was 8.5 + 0.02 cm, and after 24 hours consolidation was 7.75 + 0.02 cm and, A was 70.88 cm².

The flow rates for the NaCl scenario after consolidation were lower than those for the NaCl unconsolidated scenario (Table 2). However, the Ksat for the control and red mud showed little change while kaolinite showed an increase in Ksat, from 2 to 2.71 cm per minute and bentonite showed a decrease in Ksat, from 2.4 to 1.85 cm per minute (Figure 3).

The additions of different cations had a dramatic effect on Ksat for each clay type, shown in Table 2 and Figure 4. The control showed little change with the addition of calgon, however the addition of gypsum caused a dramatic reduction in flow and Ksat. Red mud showed a decrease in Ksat with the addition of calgon but little change between calgon and gypsum (Figure 4). However, flow rates drop overall (Table 2). Kaolinite in each case showed a decrease in Ksat with the greatest effect (as with the control) with the addition of gypsum. Bentonite showed progressively lower Ksat values - the addition of gypsum caused the greatest drop. In comparison with the other clays, bentonite showed reduced Ksat and flow in all scenarios. (top)


Increasing the clay fraction of the columns was observed to reduce the saturated hydraulic conductivity. This is due to clogging of the sand matrix, which can be caused by swelling of clays and/or the dispersion or flocculation of clays. Clogging occurs mostly because of the blocking of pore spaces in the soil matrix, preventing or reducing flow. The kaolinite and bentonite observations reflect the large differences between these two clay minerals; most importantly, the swelling and surface area characteristics associated with cation adsorption.

A possible reason for the fluctuations in the effect of cumulative clay additions (Figure 1(a) and Figure 1(b), with the kaolinite and red mud mixtures, was a through flow of some clay. This was observed in a slight discoloration of the effluent, after each cumulative addition of clay. Also, the addition of Calgon, a strong dispersant, 24 hours after stabilisation caused further discoloration of the effluent. These observations are consistent with the double layer theory that the addition of monovalent cations (Na+) expand the diffuse layer causing repelling of the clay micelles. The polyvalent anions (metahexaphosphate) help to keep the particles together. These observations suggest that the pore sizes in the sand matrix are too large to totally contain the dispersed clay material, which is flushed out of the column.

The addition of Gypsum caused further reductions in the saturated hydraulic conductivity. Though the K values for the control also dropped off, the effect is not satisfactorily explained so easily. The further reductions observed in hydraulic conductivity with the addition of gypsum also agrees with the double layer theory. In this case clogging can be attributed to compression of the double layer combined with percolation causing entrainment and re-arrangement of the clay fraction in the soil, forming agglomerations of a size that can lodge in the macro-pores of the soil matrix.

The swelling of clays may well take between 24 to 48 hours to reach their full volume (reference). Nevertheless, bentonite was observed to start swelling directly on contact with water. This process, although not observed directly, could still have continued over the next 48 hours, while the clay was contained within the sand column, giving lower overall flow rates (Table 2) and K values (Table 3).

In conclusion, in a constant head steady state system the ‘propping up’ of pore spaces and structural re-arrangement due to flocculation, swelling and dispersion could be possibly inhibited by the following conditions. These conditions are firstly, the applied pressure to the soil matrix from the head of water may compress the soil matrix. Secondly, the direction of flow is conducive to transporting of entrained particles and further lodgment within the sand matrix. Finally, any re-structuring of the clay particles under the first two conditions may serve to cause further clogging, and hence decreased rates of infiltration, rather than the increases that may be observed in conditions with 'heavy' soils. Furthermore, the laboratory trials involved mixing uniform and regular clay and sand sized particles. This cannot be a true representation of the ‘real world’ as few sizes and shapes of particles are included in this investigation. Further studies are necessary to see the effects that different sand size distribution, clay particle sizes and soil structure will have on the hydraulic conductivity. (top)


In no particular order I would like to thank, Alcoa Australia for supplying the red mud. Hayden Gardner from Rheochem Pty. Ltd. for providing his expertise on the methods of mixing and setting up the experiment. Professor Lawrie Davidson of Murdoch University and Theo Bazen for their direction and provocation. Dr. Bill Scott of Murdoch University for allowing me the opportunity to begin this study in the first place, for which I am grateful. (top)

This web site was constructed by Benjamin.K. Galton-Fenzi. Last Updated: 16 May, 2003 1:08 PM
© 2003

Page Contents



Theoretical Background

Materials and Methods