Multi user Multi area Alarming PLC -> HMI/Scada

Hi, Not specific to any manufacturer of PLC or HMI/Scada but looking for best approach.

To ensure a HMI/Scada receives the alarm I assume most people latch alarms in the PLC, so I ask.

If you latch the alarm in the PLC when do you unlatch?

• If an alarm is latched, the HMI will not be able to determine if the alarm condition has gone without operator input, or using handshaking bits for each alarm.
• If operator input is required do you have multiple alarm acknowledge bits for user/areas to prevent users from acknowledging alarms from an area they are not authorised?
• How do you determine how long an alarm condition was active if alarm is latched
• Do you only latch the alarm for xx Seconds?

If alarm is not latched how do you ensure the HMI receives the alarm?

• If alarm is not latched in the PLC, the HMI/Scada could miss alarms due to comms
• HMI/Scada will know alarm condition has gone without operator input
• Alarm active times will be accurate for trending/reporting systems.

Thanks in advance.

Sometimes its all of the above, sometimes none. It depends on what the customer wants.

It's a bit of a 'How long is a bit of string' question.

Generally major alarms will latch, with an unlatch from the HMI, either a password protected area or not.

Some alarms wont latch, don't stop the process and act as 'info only' and everything in between.

There are guidelines for this and there isn't a one answer approach.

Certain applications will alarm and the system reset itself and restart one or two times (I know of only one such case, but it exists).

Then it also comes into implementation, some systems rely on the SCADA to calculate unacknowledged alarms and lately more and more that is done by the PLC itself.

You can have alarm bits recorded in the historian, not the latch but the trigger bit. You can then audit what happened in a trend with your analogue values.
Certain alarm systems latch on the SCADA, so the SCADA will see alarm condition, alarm condition gone and acknowledge. This way you know when something happened and went and when the operator did something about it.

In our remote location projects are alarms are setup to function this way.

We qualify each alarm,(meaning it really has to be a problem not just a bounce!!), once the alarm condition has been qualified we then LATCH the alarm and start sending out notice to the proper authority. Different level of alarms goto different levels of authority.

These alarms will continue to be sent until someone with proper authority "Acknowledges" it. (So someone from another dept. can't acknowledge!)

Now once the alarm has been acknowledged this will just stops the notice sending. If the error condition has not been resolved this pattern will start all over again.

We have put all kinds of condition into this routine from request of the customer.