Not every tag that shows no data is broken in the same way. Some hold a wrong-but-plausible value, some stopped a few minutes ago, and some have simply never lived at all - configured, counted, and consuming resources, yet never receiving a single update. That last kind is the dead tag. This guide defines the dead tag precisely, distinguishes it from the frozen tag that holds a stuck value and the stale tag that recently stopped, and covers what causes tags to die, how to detect them by their last-update age, and how to clean them up.
Dead Tag in one line: A dead tag in SCADA is a tag that is configured and consuming system resources but has never received data, or has stopped updating indefinitely with no expectation of resuming. It differs from a frozen tag, which holds a plausible but unchanging last value, and from a stale tag, which stopped only recently; a dead tag is effectively defunct. Dead tags are found by their last-update age and removed through a decommissioning workflow.
These three conditions are easy to conflate but describe genuinely different failures, and telling them apart guides the response. A dead tag has no life to it: it either never received a value from the moment it was configured, or it stopped updating so long ago and so permanently that it is effectively defunct. It occupies a slot in the systems, is counted in the tag total, and may consume a license, but it delivers nothing. The defining trait is a total, indefinite absence of updates.
A frozen tag is different and in some ways more dangerous, because it looks alive. A frozen tag holds a plausible last value that no longer changes - a stuck transmitter that keeps reporting the same reading, so the number on the screen looks like a real measurement even though it is not updating. The hazard is that a frozen tag can pass a casual glance and even a range check, because its held value is within normal bounds; only by noticing that it never moves, or that its timestamp is not advancing, does the freeze reveal itself. A frozen tag is failing while appearing healthy.
A stale tag is a matter of degree and recency: it stopped updating recently, and the question is whether it will resume. A tag that has not updated for a short time might be experiencing a brief communication hiccup that will clear on its own, so it is stale rather than dead - its last value is aging but the situation may be temporary. The line between stale and dead is essentially how long the silence has lasted and whether recovery is expected: a recently silent tag is stale, and a tag that has been silent indefinitely with no prospect of returning is dead. The same underlying symptom - not updating - spans a spectrum from a momentary blip to a permanent death.
Dead tags arise from a handful of recurring causes, most of which trace back to a mismatch between the configuration and the physical reality it was supposed to represent. A common one is decommissioned equipment: an instrument or a whole skid is removed from the field, but its tags are left configured in the historian and HMI, so they sit there forever with nothing to report. Nobody meant to keep them; they simply were not cleaned up when the equipment left, and they quietly persist as dead points.
Another frequent cause is a bad or wrong address. A tag is configured to read from a device register or address that does not exist, was mistyped, or was never actually wired, so from the start it has nowhere to get data and never receives any. Such a tag was born dead - it looks complete in the configuration but was never connected to a real source. A deleted or replaced device does the same thing: if the underlying device is removed or its addressing changes and the tag is not updated to match, the tag loses its source and goes silent for good.
Project and template residue is a quieter contributor. Tags created for a temporary study, a test, or a commissioning exercise are often left behind after the work ends, and tags copied from a template for equipment that was planned but never installed exist without any device behind them. In all these cases the common thread is that the tag outlived, or never had, the physical thing it was meant to measure. Recognizing the cause matters because it determines the fix - a wrong address might be corrected, whereas a tag for removed equipment should simply be retired.
The reliable way to find dead tags is by their last-update age - how long it has been since each tag received a new value. A tag whose last update is very old, or which has no update ever recorded, stands out against tags that are refreshing normally. Ranking tags by the age of their most recent update surfaces the candidates: the ones silent for an extended, unexplained period are the likely dead tags, and those never updated at all are almost certainly configuration that was never connected. This age-based screening is what turns an invisible problem into a concrete list to investigate.
Detection has to be paired with judgment, because not every silent tag is dead. Some tags legitimately update only rarely - a manually entered value, a setpoint that changes seldom, an event that fires occasionally - so age alone can flag a healthy but slow tag as suspect. Investigating a candidate means asking whether it should be updating: is there equipment behind it, does that equipment still exist, and is it supposed to be reporting? Only tags that fail that test - silent because their source is gone or was never real - are genuinely dead and ready to be retired. The rest are explained and left alone.
Cleanup is a decommissioning workflow, and doing it deliberately is what keeps it safe. A tag confirmed dead should be retired through a controlled process - reviewed, approved, and removed or archived rather than deleted on a whim - because occasionally a tag that looks dead is one someone intends to revive. A cloud SCADA platform such as Merobix supports this end to end: it records each tag's last-update time so dead candidates surface automatically by age, it lets an operator see at a glance which points are silent and for how long, and it provides the governed change process to retire confirmed dead tags with a record of who approved the removal. Clearing out dead tags this way shrinks the point count that drives licensing, removes clutter that hides real problems, and keeps the tag database an honest reflection of what is actually being measured.
A dead tag has never received data or has stopped updating indefinitely, so it delivers nothing - its defining trait is a total, permanent absence of updates. A frozen tag holds a plausible last value that no longer changes, so it looks alive and can even pass a range check while actually being stuck. A dead tag is obviously defunct once you check its update age; a frozen tag is dangerous precisely because it appears healthy.
Common causes are decommissioned equipment whose tags were never cleaned up, a bad or mistyped address that was never connected to a real source, a deleted or replaced device that left the tag orphaned, and leftover tags from tests, projects, or templates for equipment that was never installed. The common thread is that the tag outlived or never had the physical thing it was meant to measure, so it has no source of data.
Dead tags are detected by their last-update age - tags silent for a long, unexplained period, or never updated at all, are the candidates. Because some tags update legitimately rarely, each candidate is investigated to confirm its source is truly gone before action. Confirmed dead tags are then retired through a controlled decommissioning workflow that reviews, approves, and records the removal, rather than being deleted casually in case one is meant to be revived.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.