Environmental Engineering Reference
In-Depth Information
3.3.2
Particular construction issues in pyroclastics
Prebble (1983), Jones (1988) and Oborn (1988) describe problems which were encoun-
tered at the Ruahihi and Wheao hydroelectric projects, constructed in the Taupo Volcanic
Zone. Headrace canals for each project were constructed by cut and fill methods, on and
through highly variable ash and ignimbrite deposits. Both canals failed by piping and sub-
sequent collapse during early operation, apparently due to the high erodibility of some of
the soils, both
in situ
and when compacted, and to their brittle, non-healing nature which
enabled the development of erosion tunnels. Oborn (1988) suggests that some of the soils
at Ruahihi were probably dispersive and that accelerated rates of settlement after canal fill-
ing may have been due to the collapse of very low density soils on saturation after loading.
At the Wheao failure area, erodible ash soils were located above very high strength welded
ignimbrite with a columnar joint pattern. Near the upper surface of the ignimbrite, the
joints were “bridged” by infill soils from above, but below this they were open as much as
50 mm. This feature was apparently missed during the construction stage cleanup. During
operation, erodible ash soils were washed into these gaping joints, close to the penstock
intake structure as shown on
Figure 3.11
, taken from Jones (1988). Prebble (1983) and
Oborn (1988). Note also that the extreme sensitivity (up to about 60) of some of the alter-
ation products (allophane and halloysite clays) caused problems during construction of the
canals. Prebble predicts that these soils could collapse and liquefy when disturbed by earth-
quake loading or changed groundwater levels.
Jacquet (1990) describes the results of comprehensive laboratory tests on andesitic ash
soils from seven sites in New Zealand. The soils contained high proportions of allophane
or halloysite and all classified as MH in the Unified System. Sensitivities ranged from 5 to
55. Jacquet concluded that the sensitivity was associated with irreversible rupture of the
structural fabric of the soils and was not directly related to the clay mineralogy or classifi-
cation characteristics of the soils.
Not all weathered pyroclastic materials are sensitive. In weathered agglomerates at the
site of Sirinumu Dam in Papua New Guinea the matrix soils, which are clays of medium
to high plasticity with 40 to 50 percent moisture content, are very resistant to erosion.
The clay mineral types include halloysite, kaolin and allophane.
Rouse (1990) describes tropically weathered andesitic and dacitic ash soils from
Dominica, West Indies, which occur on generally stable slopes of 30° to more than 50°.
The soils are mainly allophane and halloysite clays, with very high residual friction angles
(most between 25° and 35°).
Some unweathered non-welded ash and weakly-welded ignimbrite materials can be
used as sand and gravel sized embankment filling, the weakly welded materials breaking
down readily during compaction. However based on the experience at the 60 m high
Matahina rockfill dam in New Zealand (Sherard, 1973) it is suggested that such materi-
als should not be used for filter zones unless it can be shown that they will remain cohe-
sionless in the long term. At Matahina weakly welded, partly weathered ignimbrite was
compacted to form “transition” zones between the impervious core and rockfill zones.
Subsequent excavation through the compacted ignimbrite showed it to have developed
appreciable cohesion and it appeared to behave like a very low strength rock. This
strength developed due to the interlocking of needle-shaped particles of glass. This cohe-
sive, brittle behaviour was an important contributing factor to the piping incident which
occurred during the first filling of the reservoir in January 1967. The main cause of this
incident, which resulted in the loss of more than 100 m
3
of core and transition materials
into the rockfill, was the failure to remove a 1.8 m projecting bench from the steeply slop-
ing foundation. This projection caused cracks to develop in the core and transition zones
as the embankment settled. Other factors, as described by Sherard (1972, 1973) and
Gillon and Newton (1988), included the possible reinforcement of the transition zones by
