Today alarm

[Peter Kravtsov <pkravt@gmail.com>]

Dear colleagues,

By the Marat's request I would try to explain alarm which happened
today morning.
We have a special alarm in the control block for low pressure P2-P1
difference, which was set by Marat to 0.01 bar.
At 5:45 P2 pressure droped below this P1+0.01 limit, and that caused
alarm situation. This results to cuting off the TPC and closing the bypass valve
(all valves were closed). This immediately causes P2 pressure rise
and alarm flag was reset by normal P2-P1 difference.
This is normal alarm behaviour. The only problems in this sequence are:

1. Last year I calibrated our P1 sensor by Claude's sensor on TPC to
    have equal readings. All other sensors (including P2) are calculated
    by their full scale range and output signal 4-20 mA. The control
    block itself doesn't have any pressure calibration, it works in terms of
    ADC units.
    This leads to small error in P1 pressure shown on the screen,
    and P1 pressure signal (in ADC units) used for alarm handling by
    the control block. So in the database we see only 0.28 bar difference,
    which is obviously more than 0.01 threshold. But, when I recalculate
    these readings to ADC units, I got difference of 12 which is very close
    to setpoint (0.01 bar corresponds to 8 ADC units). This is explainable,
    if we take into account pressure rise and time delay between
    alarm trigger and computer readout.
    This is minor problem and I think we have to calibrate all our pressure
    sensors (and may be flow controllers as well) as we planed. Unfortunately
    we were busy with humidity sensor and nitrogen filling this autumn and
    had no time to do it.

2. Second problem is more critical, as it leads to "silent" alarm processing.
    The reason for this is time period of computer data acquisition.
    Computer reads all parameters only once per 5 seconds (or 1 second
    during system adjustment). Control block is much faster.
    So today morning control block handled "dP low" alarm, and
    after some time of pressure rise, it reset the alarm flag, because
    pressure difference returned to normal condition.
    All this operations took place between computer readings,
    so computer did not see dP Low  alarm flag.
    At the previous version of alarm handler I implemented a
    hysteresis for alarm recovery, but then we decided to get rid of
    the automatic alarms recovery and I also removed the hysteresis.
    Now there is only one threshold for P1 and (P2-P1) alarms.
    If readings exceeds the threshold => alarm, if not - alarm flag
    resets.
    This should be corrected in control block software.
    Either I put back some hysteresis or I can implement
    only manual alarm acknowledge from computer.

3. Third problem is misconfiguration of dP High alarm threshold.
     It was set to 7 bar, which is more than release valve setpoint.
     With this settings dP High alarm is useless, because
     release valve opens first.

So for current CHUPS work I would recommend to track pressure
difference more than at least 0.3 bar and set dP High setpoint
to less than release valve threshold. In this case after dP Low alarm
control block will open bypass valve after P2 pressure rise and
prevent protium flow to release volume.

--
Best regards,
Peter.