The Split‐Sample Method 91 

The main issue with the previous method was the inability to control the position of the seeding particles. Once the desired area density was created on the steel Z-plane, it could not easily be kept undisturbed throughout the preparation stage since the removal of the steel plane introduced a significant amount of particle displacement, creating densely seeded areas (too dense for PIV) at the top and bottom of the box, and relatively empty areas near the middle. After these initial attempts, an alternative method was proposed, and used in a scale model of final dimension 80x80x180mm (L:W:H). In this new method which constructs the soil model horizontally, two Perspex boxes were used. The first, (Box A, dimensions 180x120x80mm, L:W:H) was laid horizontally on a level surface for the soil cuboid to be made up layer by layer and consolidated within it, before being removed by way of a metal trowel and slid into a second Perspex box (Box B, dimensions 80x80x180mm) – to be placed vertically – whence the tube samples are to be extracted at a later stage (Figure 3.11).

Box A and trowel Construction sequence: light: slurry, dark: consolidated

soil, blue: particle plane Box B, filled Figure 3.11 - Split-Sample Method

92 First, some transparent soil slurry was mixed, de-aired and placed within Box A, which it filled to a height of approximately 80mm (to allow for a reduction of volume of approximately half during consolidation – this was to give a half-sample of dimensions 80x40x180mm). This was left to consolidate under self-weight for 48 hours, during which time a noticeable change in volume occurred. Oil was observed to have leaked out in majority through the connections between the sides of the Perspex box, not through the top of the box, as was designed. After the initial consolidation, a pressure plate and weights were put in place on top of the sample.

After consolidation of the first half-sample, the excess oil was removed from the surface of the soil, onto which dyed sand was then sprinkled. An additional slurry layer of approximately 5mm was created on top of the seeding particles. The difficulty in this step is the tendency of the particles to move in the direction of material flow. Since the transparent soil must be placed in a manner which discourages the entrapment of air bubbles, one method was to run the fluid soil continuously down the side of the container. This worked reasonably well but did create a horizontal flow of material which significantly disturbed the seeding particles. This was also the case when the soil was carefully introduced anywhere in the box by way of a baster. When the second layer was fully settled, the box was subjected to a small amount of horizontal shaking, which equalised the seeding area density, but has the unwanted effect of lodging seeding particles in the imperfections in the sample. (This is not a very important issue as long as a good visibility of the seeding plane is maintained at different positions around the box. This effect tends to disturb the edges of the sample where visibility during preparation is important, but this would not affect the later stages of sampler driving and PIV analysis). During shaking, particles may move slightly out of plane, but the majority tend to settle back to the surface of the consolidated layer. Any particles which remain out-of-plane can be removed with a small aperture pipette without disturbing the rest of the plane. After the second layer was consolidated under self-weight, the third layer could be created with the remainder of the slurry, and consolidated in the same manner as the first layer. When consolidation was complete, the trowel was lifted with care to remove the whole sample, which was inserted into the second box where it was allowed to finish consolidating in the correct orientation (In natural soils, excluding those severely

93 affected by tectonic effects such as folding strata, consolidation occurs in one direction, and this affects particle orientation. To keep in line with this, the majority of the consolidation would have to be in the vertical direction, therefore any horizontal consolidation should be small when compared to vertical consolidation - here horizontal consolidation stresses were kept at 10% of vertical consolidation stresses).

The creation of the plane of seeded particles was successful and well controlled throughout the process. The area density of the particles could be chosen and put in place very accurately, and no changes were observed at any point during preparation. It was however noted that during removal of the trowel, because of a loose fit and lack of support to the soil at the edge of the trowel, the consolidated soil rippled slightly and deformed in a non- recoverable manner (Figure 3.12a). It was, however, possible to retrieve the soil as a whole with some disturbance to the edges, but with no apparent effect on the seeded plane.

a b

Figure 3.12 - Split-Sample Method: a) Rippling and b) Tearing

The transfer of the soil to Box B was not as successful. The sample required trimming before being placed into the second container. It was decided that as the material was extremely soft, the extra soil would be left to shear itself off as the soil entered the new box. The friction between the transparent soil and the box tore two faces of the cuboid in places, resulting in large deformations on the outside faces of the soil. During removal the soil refused to slide off the trowel, even when oiled to decrease friction. Because of the shape of the trowel, it was not possible to oil the parts of the metal which refused to yield the transparent soil. This resulted in tearing of the soil sample (Figure 3.12b). It was concluded that while the layering of the transparent soil was a viable idea (since it produced a level and predictably seeded plane), the use of the trowel was not. The main problematic issue was that of transferring the

94 sample from one box to the other, and doubts were cast on the scaling up of this technique. It was therefore decided to develop the horizontal consolidation method without the trowel.