EE - Forum Styles

deviation alarms not activating properly

I have the deviation limits set at +/- 0.20 pH, and have them enabled.  The PV is at 6.5 pH and the SP is at 7.5, but the deviation alarm is not active.  However, if I go into the pH detail panel -> alarms/limits -> limits, and click on the dev lo lim and change it from -0.20 to -0.20 (that's right, not actually changing the value, just inputing the same value), and click ok, then the deviation alarm is active.  If I reset it, by changing the setpoint to 6.5, the deviation alarm deactivates.  But if I change the setpoint back to 7.5, the deviation alarm does not activate, and requires the previously mentioned process to make the alarm active.  Any idea why this is, and how to fix it?

8 Replies

  • I think you will find this is by design, and is something that has misled me in the past. If you look at BOL for the PID block it says  "Note: Deviation alarms are suppressed on SP changes. When the PV comes within the deviation limits or if the status of OUT or BKCAL_IN becomes limited, the deviation alarm is enabled again."

    The other thing I find confusing is that the deviation low limit has to be a negative value.

  • Have you got a hysteresis band on the alarm set-point? Typically, the PV will need to leave the hysteresis band first and then return to the SP.
    This has happened to me on other systems.
  • In reply to Cedric Dawnhawk:

    Doesn't the deviation low limit need to be a negative value, because it is a deviation, not an absolute value? So, if you are deviating low, the feedback (PV) must be below the setpoint (SP)--that is, a negative deviation.

    It does not make sense to have two positive values. If A = SP, and Low deviation = B and High deviation = C, and B and C > 0, then the alarm would occur above PV = A+C, turn off between PV = A+C and PV = A+B, then turn on below PV = A+B, including when PV = A (deviation = 0) and any lower PV.

    The only reason you have two different high or low deviation setpoints, is because in certain circumstances, the threat of being above or below the setpoint is more hazardous or acceptable than the alternative. If your tank level is normally high, you would be more concerned if the deviation threatened to overflow the tank, rather than the tank level falling lower. If the tank runs relatively low and you had the ability to deal with overflow, but running out will destroy your pump, maybe you want to tighten up your negative deviation and leave the positive deviation loose.
  • In reply to Jeffrey Mach:

    Yes you are quite correct. It is logical and consistent with other alarms that the low deviation alarm should be activated when the value being monitored (in this case PV-SP) is lower than the alarm limit. As PV-SP is negative the limit most be negative. I perhaps should have said this was confusing but only to muddle-headed people like myself who assume without thought that the low deviation limit to be the absolute amount of deviation below the setpoint.

    I can see why there are two deviation limits. But just supposing there was only one, which applied to positive and negative deviations. Would it be a positive number or a negative number ? Don't bother to answer that!

  • In reply to Sumeet Chimalkar:

    Yes, and it's set to 0.5%. Would setting this to 0.0 fix this for me?
  • In reply to Cedric Dawnhawk:

    Hmm, this is interesting. On all my other systems, when I do the same thing, the deviation alarm activates...
  • In reply to Cedric Dawnhawk:

    So is there a workaround for this? For example, say the setpoint is changed, but then the PV never returns to within the deviation limits, then I would never get the deviation alarm notifications.
  • In reply to rummeln:

    It sounds like you need to configure a different alarm to handle that situation, that a Deviation Alarm is not intended to monitor for. Depending on your process response, your control system would need to be configured to alarm if the process variable is not within an acceptable deviation of the setpoint after a certain number of seconds. This would catch and alarm after a setpoint change fails to be completed on time, but if the initial change to the new setpoint is successful, once you are within the deviation band, the deviation alarm would be enabled and any deviation from setpoint beyond the deviation band would be caught immediately.

    This sort of alarming is useful for a process that tends to oscillate around the Setpoint.  You set a reasonable deviation band around the Setpoint that permits the normal oscillation, but would trigger on anything dangerously abnormal.  You then set a separate alarm with a tighter error band, but a delay on initiation, such that it will catch when the system fails to move to the new Setpoint in time, but will turn off as the Process Variable crosses the Setpoint.  As the process oscillates, it will keep crossing the Setpoint, or at least, within the tighter delayed-initiation error band, that there is not enough time for the delay to time out and trigger the alarm.