The challenge
A brief door opening and a persistent equipment fault may both change a graph, but they may need different responses. Too much delay can also postpone awareness of a significant event. Product sensitivity, equipment behaviour and the team's practical response time belong in the decision.
Separate four settings
A threshold defines the condition that can trigger an alert. A delay defines how long it must persist before notification where that function is supported. A recovery condition defines return to the normal state. Escalation defines the next responsible person when the first response is insufficient, subject to the supplied system and procedure.
An alarm delay does not change a recorded temperature or make an excursion acceptable. A recovery message does not settle the consequences of the event. Records and follow-up still need attention.
Build the process from the site
- Have the responsible specialist define the product and process requirements.
- Review sensor position and normal measurement behaviour.
- Define conditions requiring action and faults affecting monitoring itself.
- Assign recipients, coverage periods and deputies.
- Run a controlled acceptance exercise that checks notification and documented response.
A test should not require exposing actual product to an excursion. Plan a separate, authorised method appropriate to the equipment. Make sure it checks the relevant chain rather than only a “send test message” button.
Ask the useful timing question
Instead of asking for a universally recommended delay, ask how much time passes between the start of an event, its detection, delivery of the message and the arrival of someone able to act. Each part of that chain matters. One setting cannot be assumed suitable for every room and product.
Review checklist
Can the recipient identify the sensor and location? Who receives messages after hours? Is a missing reading distinguished from a temperature excursion? Does the recipient know how to document the check? Are configuration changes recorded and verified?
Can every site share one threshold? Only where process requirements and risk support it. Consistent naming does not require identical numerical settings.
Does an alarm prove product damage? No. It identifies a condition requiring investigation; product assessment is a separate decision.
Define the monitoring response with ENDISENSE, from the reading to the person responsible for action.
