94 Intact rock
Finally, it should be noted that the required experimental data can be
monitored independently from the control system, as illustrated by the
lower left-hand box in Fig. 6.9(a).
The schematic diagram in Fig. 6.9@) illustrates this closed-loop control
with more direct reference to rock testing. Note the mode selector for the
feedback signal. If the output from the load cell were to be taken as
feedback and the programmed signal were to monotonically increase with
time, then we would be programming a stress-controlled test, which would
result in explosive failure at the peak of the complete stress-strain curve
as the machine attempted to increase the stress beyond the rock's
compressive strength. From the arguments already presented, it is the
displacement transducer output that would be used as the feedback signal
for an axial strain-controlled test.
The tests that can be conducted with the closed-loop control technique
are only limited by the imagination. The complete stressstrain curve can
be obtained in tension, by using displacement feedback. By utilizing the
load cell output and the displacement transducer output, we can program
a linear increase in energy to be supplied to the specimen. In fact, any
parameter or combination of parameters can be used as feedback.
Note that, in the complete stressstrain curves shown in Fig. 6.8, the Class
I1 curves do not monotonically increase in axial strain, and hence cannot be
obtained utilizing axial displacement (or axial strain) as the feedback signal.
To overcome this problem, and as a general principle, one takes as feedback
the parameter most sensitive to the failure that will occur in the test in
question: in this case the lateral displacement-which does monotonically
increase. The complete Class I1 stress-strain curve, which does not mono-
tonically increase, is then independently monitored as it is generated. The
lateral displacement is more sensitive to the axial cracking which occurs in
a uniaxial compression test. Conversely, the axial displacement is more
sensistive to the lateral cracking which occurs in a uniaxial tensile test.
Moreover, as the test configuration can be of any type, we will generally
choose the most sensitive indicator of failure as the feedback signal, For
example, to consider the mechanics of a hydraulic fracturing test in which
a hollow cylinder is internally pressurized to failure, the machine can be
programmed to linearly increase the circumference of the internal hole by
taking the output from a wire strain gauge bonded circumferentially
around the hole as feedback. The hydraulic pressure is then adjusted by
the closed-loop control such that the circumference linearly increases and
the fracturing is controlled. Figure 6.10 illustrates a suite of rock mechanics
tests and the corresponding optimal feedback signals.
With the ability to control failure and generate a failure locus for a variety
of testing configurations, the test can be stopped at any time to study stages
in failure development. For example, under stress control, if the machine
is programmed to 'hold the stress constant, a creep test is performed: the
analogue under strain control with a 'hold' is a relaxation test. Using stress
or strain control, the rock can be fatigued with any frequency and stress
or strain amplitude. It is even possible to record the three perpendicular
components of earthquake motion in the field and apply these through
three mutually perpendicular actuators under laboratory conditions. Even