Instrumentation & Measurement Magazine 25-7 - 12

◗ The AO block attends the request, falls into the " local
override " state, and waits some seconds-those specified
in the parameter " fault state time " -to see if the
fault persists. If the fault does not go away, the AO block
freezes the output or coerces it to a predefined value,
depending on whether the parameter " fault state type " is
set to " freeze " or " go to present value, " respectively.
If the actuator fails, or the connection PID ← AO is lost, the
BKCAL_IN input becomes " Bad " and the cascade breaks as
before, following the same steps. Thus, the BKCAL_IN signal
is to the actuator as the IN signal is to the sensor. In both cases,
the cascade is automatically resumed if the sensor or the actuator
or the communication links become healthy again.
To see all of this in action, we took the physical process represented
in Fig. 4 and tested it in terms of safety. We injected
a fault in the loop and saw how the process reacted. The test
went through the following steps:
1. The AO block was configured with " fault state type " =
" go to preset value, " " fault state value " = 0, and " fault
state time " = 5 seconds.
2. The process was started with setpoint = 50%, and
time was given for it to stabilize around that value.
The cascade was healthy, and all blocks were working
smoothly.
3. The output of the AI block was made " Bad " by enabling
transducer simulation and forcing a fault in the sensor
(OUT.Status.Sub-status = 0x10). The PID block issued
a " initiate fault state " request to the AO block, which
passed to the " local override " state immediately.
4. After 5 seconds, the AO block closed the control valve
(output coerced to 0).
5. The fault was removed by disabling transducer simulation
in the AI block, and the PID block removed the
" initiate fault state " request.
6. After 5 seconds (approximately), the AO block got out of
the fault state. The cascade was automatically resumed,
and the loop returned to working normally.
7. Sometime later the
process variable met
the setpoint again.
These seven events
are illustrated in Fig. 5 [7],
which plots the values of
the setpoint, process variable,
and manipulated
variable during the test. The
safety mechanisms worked
as expected, including all
timings and preset values.
Note that between instants
3 and 5, the water level falls
gradually from 50% to less
than 40%. That is not visible
in the graph because the PV
stops being updated while
the sensor is faulty.
12
Discussion
The two mechanisms seen above are effective in monitoring
the health of measurement equipment and, ultimately, in validating
the measurement data. The first mechanism-status
registers and service requests-follows a pull approach, in
the sense that the data consumer must query the status and
event registers to know the cause of the service request. In
other words, the external computer must search for the cause
of the interruption. This strategy is very efficient in allocating
resources because, when everything works fine, there is
no overhead in communications or processing. Only when
something goes wrong does the computer stop what it is doing
and goes to see what is happening, with some loss of performance.
However, this strategy is also more permissive because
it does not support systematic error handling, giving the data
consumer too many options to detect and treat abnormal situations
(in the limit he can simply ignore them). This opens the
door to ad-hoc, unconventional solutions that often do not follow
the best practices in terms of safety.
The second mechanism-status bits and safety locks-
follows a push strategy, in the sense that the measurement
instrument always provides status information, whether the
data consumer wants it or not. The external computer is never
interrupted because it always gets the measured data together
with the status information. This strategy is more wasteful in
terms of resources because, regardless of whether there are
errors or not (and usually there are not), the status information
Instrumentation & Measurement Magazine 25-7

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-7
