How to reduce alarm chatter

If it can’t be eliminated at the source, here’s how to use features such as filtering, deadband and time delays without compromising operation or control.

By Kevin Brown

1 of 2 < 1 | 2 View on one page

There are times that an alarm can’t be removed or an alarm setpoint can’t be changed to resolve a noisy process signal causing a chattering alarm. When this occurs, there are techniques that can be applied to the process signal or alarm to reduce alarm chattering. It’s important to understand the effect each method will have on the alarm or process measurement prior to implementing it. This article discusses each technique and the results when applied.

Understand the issue

These techniques should never be applied blindly because they’ll likely not have the positive impact that was desired and could cause measurement issues or mask an alarm or delay a response by the operator. Before considering which method to apply, the reason(s) for false positive alarms must be determined for all states of the operation (startup, trip, normal, etc.). Solving the alarm issue for one operating condition may mask an alarm that’s real in another condition. Once the issues have been defined, then consider which technique will solve most of the false alarms.

A frequent mistake is increasing the masking effect because the applied technique did not resolve most of the false positive alarms. An example of this would be to increase the delay time from five seconds to five minutes. This will reduce more false positive alarms, but can lead to masking real alarms by delaying activation of the alarm until the five-minute delay has expired. The correct solution may be to apply a different technique or apply more than one technique to get the desired result of reduced false-positive alarms and minimized masking of real alarms.

Filtering smoothes measurement

Normally, filtering is applied to reduce control action on a noisy process signal to improve the control of the process. Filtering can also have a positive impact on reducing false alarms. Filtering minimizes a noisy process signal by smoothing the measurement, which reduces the false positive alarms. Applied directly to the measurement, filtering will affect control action on a closed-loop point, which must be considered before adding it.

Download Control's new ebook: Testing analog and digital I/O

Figure 1 shows that the filtered signal (output) doesn’t exceed the high alarm setpoint, while the unfiltered signal (input) does exceed the high alarm limit. On the first incident, the unfiltered and filtered signals cross the low alarm setpoint, but later, the filtered signal doesn’t cross the low alarm setpoint when the unfiltered signal does cross.

Advantages of filtering are:

  • Reduces false positive alarms
  • Alarm delay activation is minimized on fast changing process
  • Reduces alarms when noisy process is operating close to the alarm setpoint

Disadvantages of filtering are:

  • Delays alarm activation on slow moving process
  • Can hide an alarm when the process is operating around the alarm setpoint
  • Has minimal effect on very noisy processes
  • Can negatively impact control performance

Deadband adds alarm hysteresis

Alarm deadband is the change required in the process signal to either activate the alarm or return the alarm to normal (Figure 2). Some computer control systems allow deadband to be applied to either the activation or return of the alarm, while others allow it only on the return of the alarm. Deadband is typically applied by entering a percentage of the measurement, which can lead to latching of the alarm when the measurement range is large.

For example, if the flow range is 100,000 and a 2% deadband is applied, the alarm will not clear until the measurement is 2,000 below the alarm limit. Using these numbers for an example where the process normally runs at 68,000 and the alarm limit is set to 69,000, once the alarm is activated, it will not return to normal until the process goes below 67,000, which is below the operating setpoint and the process measurement. In this example, the alarm will never clear and the operator will never get another alarm when the process rises above the alarm limit. Here, if the computer control system allows, it would be best to set an absolute value for the deadband.

Deadband advantages are:

  • Reduces false positive alarms
  • Immediate activation of the alarm when deadband is only applied to the return to normal
  • Typically the easiest to apply or configure in a computer control system

Deadband disadvantages are:

  • Percentage applied deadband could apply a setting too large, latching the alarm on or off
  • False positive alarm activation holds the alarm high, which could cause an operator to miss a real alarm if the deadband level isn’t cleared before the next activation of the alarm
  • Slow operator response on a real alarm when the deadband is applied to activation of the alarm, and not on the return to normal

1 of 2 < 1 | 2 View on one page
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments