Geotechnical News • June 2015
19
GEOTECHNICAL INSTRUMENTATION NEWS
ments expected in the dam. Each seg-
ment contains a triaxial gravity sensor
measuring tilt at every 0.5 m which
is automatically converted in settle-
ments. The readings are made using
a portable computer and a special-
ized program. The Romaine-2 SAA
is located next to INH-01 to compare
results from these two types of instru-
ment. Figure 3 (looking upstream)
shows the installation of INH-01 as
well as the SAA which needs to be
inserted into a protective PVC conduit.
The installation procedures for both
types of instrument are similar. They
have to be placed in an excavated
trench and protected from large fill
particles by using bedding sand, geo-
textiles and by controlling the grain
size distribution of surrounding soils.
Moreover, as for a INH, twisting of
the SAA must be avoided by care-
fully aligning cable markers since the
software used for data collecting and
processing is calibrated according to
this alignment.
Measurement and data
processing procedures
Measurements along the Romaine-2
SAA (and also INH-01) are taken rela-
tive to the near end (the downstream
end of the instruments i.e. near the
operator) and are corrected consid-
ering measured displacements of a
nearby survey point.
Readings of the 76 m-long INH-01
requires at least two persons for the
handling of bulky equipment and
cables. Vehicle accessibility to the
instrument is thus essential. Read-
ing time is in the order of hours and
can be greatly increased in adverse
weather conditions which can also
decrease measurement reliability. Only
basic checks of the reasonableness of
readings can be made in the field. Data
processing programs needed to be
customized for the Romaine-2 context.
Moreover, the INH probe was subject
to bias shift errors for which correc-
tions were not trivial since both ends
of these instruments are not consid-
ered fixed.
The Romaine-2 SAA has a length
of 76 m. The specified maximum
instrumented length of a typical SAA
cable is 100 m but multiple cables can
be joined to allow measurements for
greater lengths. Readings and data
processing are realized in minutes
using a portable computer. The actual
shape of the series of SAA segments
can be immediately viewed on screen.
Automated data acquisition and
transmission are also possible which
can alleviate instrument accessibil-
ity problems in Romaine-2 such as in
winter when the downstream face of
the dam is covered with snow.
Reported accuracy of
instruments
The reported random error for INH
measurements is approximately ± 1.4
mm per fifty readings. Considering
that this type of error tends to accumu-
late with the square root of the number
of readings, the expected random error
for INH-01 would be around ± 2.4
mm. However, systematic errors such
as those related to probe bias, depth
positioning and the effects of adverse
weather on the instrument (and the
operators) can be much higher and
cannot always be entirely corrected.
The reported accuracy deformation
value for a SAA is ± 1.5 mm per 32
m. This value tends to increase with
the square root of the length which
leads to an accuracy of ± 2.3 mm for
the SAA installed in the Romaine-2
dam. This value has been confirmed
Figure 3. Installation of INH-01 and SAA.
