The liquid limit of peat and its application to the understanding of Irish blanket bog failures

Abstract

Catastrophic failures of blanket bogs, involving the escape and outflow of large volumes of semi-liquid basal peat, are well-known phenomena in Ireland but have only very rarely been reported from elsewhere in the world. Their precise causes and mechanisms are as yet unclear. The liquid limit (w _L) was identified as a potentially useful indicator of the susceptibility of peat to such failure because peat has extremely high natural water contents and, as an index property, w _L takes no account of the properties or structures of highly heterogeneous intact peat. However, the usual procedure for determining the w _L of peat is not fully standardised. Prepared samples will normally include potentially highly reactive particles of disrupted fibres and wood fragments that would not be present in such freshly disintegrated form in the field. This paper presents results from w _L determinations of peat obtained from the scar margins of three bog failures in northwest Ireland, using four different test procedures including a method involving wet-sieving of the peat to separate the humified <425-μm fraction for testing without incorporating artificially fragmented particles of fibres. The sampled peat was classified as H₈–H₁₀ according to the von Post humification scale. The fibre contents varied between the sites, but the ash contents were <3% in all but one test sample, and bulk densities (dry and field-wet) of the peat from all three sites were almost identical. w _L results from the wet-sieving method were 708–785%, compared with 633–980% from the standard method. The highest measured field water contents exceeded the wet-sieved w _L for all three of the field sites. Tests of cone penetration into intact peat cores demonstrated the influence of the reinforcing effect of in situ fibres. The results strongly suggest the need to adopt a fully standardised procedure for determining the w _L of peat. Additional shear vane measurements of intact and remoulded peat from a bog failure in Northern Ireland indicated a very high ‘strength sensitivity’. This leads to the suggestion that a slight disturbance of basal peat can lead to a loss of strength that rapidly propagates as local stresses change and cause further remoulding as water contents exceed w _L.

Appendix

Determination of the liquid limit using the drop-cone penetrometer method (after BSI 1998a).

Details of this method can be found in standard textbooks such as Head (1980) or Vickers (1983). Remoulded ‘soil’ (or other sample material), with particles >425 μm removed, is mixed to a paste with distilled (or de-ionised; BSI 1998b) water, then carefully packed (without trapping air) into a standard metal (usually brass) cup with a flat bottom of 55 mm diameter and vertical sides 40 mm high. The upper surface of the sample is leveled with the top of the cup. The tip of the cone of a drop-cone penetrometer is positioned so that it just touches the surface of the sample at its centre. The standard 35-mm long cone comprises 80 g of polished stainless steel forming a sharp point with an angle of 30°. The vertical position of the cone is recorded to the nearest 0.1 mm from the scale fixed to the penetrometer. The cone is then released to fall into the sample for 5 s, locking into its new position after this time. The new vertical position of the cone is recorded to the nearest 0.1 mm and the difference is recorded as the penetration in millimetres. The sample paste is then re-mixed, re-packed into the cup and the test is repeated. If both penetrations differ by ≤0.5 mm, the average is recorded and the water content (% dry mass) of ca.10 g subsample of the paste is determined by oven-drying to constant mass. If the difference between the two penetrations is >0.5 but ≤1.0 mm, then a third test is done. If all three penetrations fall within a range of ≤1.0 mm, the average is recorded and the water content of ca.10 g subsample is determined. If this criterion is not met, the entire test for this water content is void and must be restarted.

This procedure is repeated for different moisture contents of the original sample paste, with more distilled (or de-ionised) water being added to the paste each time. A range of penetration values should be obtained with at least two between 15 and 20 mm and two between 20 and 25 mm, evenly spaced, so that a graph of water content against penetration can be plotted. The liquid limit w _L is the water content that corresponds with a penetration of 20 mm on the best-fit straight line on the graph.