Comparison
Operator access to historian trends on night shift
Short answer
Put routine historian trend access in front of operators: prebuilt HMI or SCADA trend screens for the recurring questions, point selection over already-collected points for everything else, and a read-only web portal as the alternate route. Keep collection configuration with engineering. No permission setting retrieves samples for a point the historian never collected, so that gap is closed by adding collection, not by access rights.
On this page
- Why the historian gatekeeper fails at night
- Five access routes compared
- Why the HMI trend client carries the routine load
- Worked example: night-shift coverage when prepared views answer 90 percent
- Dividing coverage between prepared views and selection
- Permissions: view and export, not configure
- Verify the path with an operator account
- The decision for each case
- Questions asked
When a night-shift operator needs a historical trend and the engineer who owns the historian is asleep, the plant loses either time or evidence: the operator phones someone at home, or makes the call on gut feeling. The fix is an access design, not overtime. Give operators read-only trend access through the HMI or SCADA trend client — prebuilt screens for the recurring questions plus point selection across every point already being collected — and keep a historian web portal as the second route. Keep point collection and historian configuration under engineering roles. One boundary decides the rest: operator rights can display and export collected history, but no right can retrieve samples for a point the historian was never collecting.
Why the historian gatekeeper fails at night #
In plants where one or two specialists own the historian, night shift waits or guesses: the interface is clunky, the tag names are cryptic, and the login rights stop at engineering. A tag like FIC_109A_PV tells a new operator nothing at 2 a.m., so the operator stops trying. The gut-feeling call is the dangerous outcome — process decisions made without data are how a small deviation grows into a 3 a.m. incident report.
Both halves of this are design problems. Access is fixed with role assignment, and usability is fixed with prepared views and operator-facing labels, not with an engineer on standby overnight.
Five access routes compared #
The routes below all appear in working plants. They differ in reach: how much of the collected tag set the operator can get to without engineering help.
| Route | Where the operator works | Best fit | Hard limit |
|---|---|---|---|
| Prebuilt HMI trend screens | HMI/SCADA screens | Recurring process views and quick checks | Only the views engineering prepared |
| Integrated trend client with point selection | HMI/SCADA | Prepared views plus any point already being collected | Cannot create collection for a point that is not collected |
| Historian web portal | Browser on the operator station | Broader queries when the portal interface is usable | Needs per-position accounts, permissions and a reachable station |
| Saved trend files in a network folder | Files that open in the historian client | Recurring investigations that can be anticipated | Presets go stale; no ad hoc point choice |
| Spreadsheet or script retrieval | A maintained local tool | Teams that already own such a method | Lives or dies with the tool, its connection and its owner |
The first two rows are one design, not alternatives: prepared screens buy speed, point selection buys reach. The portal is a legitimate second route when operators can open it mid-shift and the interface is usable on the station they have. Saved files and scripts are maintained extras — they cover anticipated questions and go stale when nobody owns them.
Why the HMI trend client carries the routine load #
HMI and SCADA platforms are built to sit on top of a historian: the historian stores the time-series data, the HMI presents it, and trend analysis happens at the machine instead of in a specialist's client. The pattern that removes the phone call is a live value on the point faceplate with a direct action that opens that point's historical graph or table — current state up front, history one click away.
Operators should be able to view and export. Adding or editing trend points stays with engineering, and it rarely needs to happen during an incident: adding a collection point is scheduled day work. For the display half of this pattern — putting a historian value into an operator screen — showing a historian maximum on an operator text field works through the same route.
Worked example: night-shift coverage when prepared views answer 90 percent #
Inputs, each with its source:
- 90% — the share of a regular operator's routine trend lookups that prepared trend views answer, reported by one plant running this design (a reported figure, not a measurement at your site).
- 99% — the same plant's reported share on overnight shifts, where the question set is narrower.
- 100 — prepared trend files that plant keeps in a network folder for the remainder.
- 0 — samples held for a point the historian was never configured to collect; a property of collection, not of access rights.
The rule in words: an access route answers a trend question only where samples exist for that point and period. Prepared views and saved files shorten the search; point selection extends the reachable set to every collected point; nothing extends it past the collected set.
The arithmetic: 100% − 90% = 10% of routine lookups fall outside the prepared views, and overnight 100% − 99% = 1%. Point selection over the collected tag set answers that residual, so coverage of collected points reaches 100%. The 100 saved files sit inside the same residual as maintained backups and add no coverage beyond it.
The result: prepared views are a search-time optimization; the collected tag set is the coverage boundary.
The decision, with its threshold: keep point selection enabled across collected points so the 10% residual is reachable without a phone call, and audit the residual for questions about uncollected points — for those, coverage through access alone is 0%, so add them to collection through the engineering change process before night shift needs them. No permission, portal or saved file recovers a single sample that was never collected.
Dividing coverage between prepared views and selection #
- Start from the decisions operators routinely make and build a prepared trend for each, with point description, units, time context and a route back to the process display.
- Allow point selection across everything the historian already collects, and say the boundary out loud in training: the client shows collected history, and a point that was never collected has nothing to show.
- Keep saved trend files and scripts as maintained extras with a named owner, and retire presets nobody uses.
- Train every shift on four moves: open a prepared trend, select a collected point, set the time range, and export where allowed.
- Do not present a folder of saved files as full ad hoc access; if the operator cannot pick another collected point, coverage stops at the folder.
Permissions: view and export, not configure #
Assign operator accounts the rights to view and, where reporting needs it, export historical data. Point creation, collection settings and tag administration stay with engineering roles: historians split this with built-in role-based access, so use those roles instead of sharing an engineering login, and never grant configuration rights merely to make trends visible. If the portal is part of the design, provision a read-only portal account for every operator position before it is needed; an account created during a night incident is an account that does not exist.
Verify the path with an operator account #
- Log into a standard operator station with an operator-role account — an engineering or administrator login proves nothing about operator access.
- Open a prepared trend and confirm samples appear for a period known to contain data.
- Select a collected point and confirm history plots over the intended range.
- Select a point known not to be collected and confirm the client shows no samples; proving that boundary beats explaining it after night shift hits it.
- Export a trend if operators are expected to export.
- Confirm the same account cannot reach historian configuration or point creation.
- If no point shows data at all, the fault is the client-to-historian path, not the design — troubleshooting historian trend client connection failures works that path step by step.
The decision for each case #
- The same questions every shift: prebuilt HMI trend screens.
- Questions beyond the prepared set: the HMI trend client with point selection over collected points.
- HMI screens frozen or vendor-locked: a read-only historian web portal, one account per operator position.
- A recurring question about a point with no collection: an engineering change to add collection first — no permission, portal or saved file retrieves samples that do not exist.
- Rare but recurring analyses: maintained saved trend files or a script, with a named owner.
Set up this way, the night shift answers its own trend questions, and the call to the engineer who owns the historian stops being part of the design.
Questions asked
Can an operator add a point to historian collection when it is missing?
No. Viewing rights display and export samples that already exist; they cannot create collection. Adding a point is an engineering change: define the tag, collection rate and scaling, add it to the historian, then verify the trend. Treat it as scheduled day work, because it almost never needs to happen while the night shift is handling an incident.
How long should the historian retain data for operator trends?
Retention is set per point or per class of points in the historian configuration. Size it to the longest period any operator workflow or prepared view actually reviews: if a shift compares against the previous week, one week of retention is a hard floor, and monthly comparisons need months. Record the chosen values so engineering and operations share the same expectation.
Should the web portal replace the HMI trend client?
No, the two routes coexist. The HMI client handles routine checks with the fewest clicks; the portal supports broader or less frequent analysis and serves stations where the HMI screens cannot be modified. If you run both, verify the browser client actually opens on the operator station and that every position holds a working read-only account before night shift depends on it.
The trend screen opens but shows no samples. What is checked first?
Check the selected time range first, then compare with a point and period you know contain data. If a known-good point plots and only one point is empty, that point is not being collected, and no permission setting will fill it. If every point is empty, the client cannot reach the historian, which is a connection fault rather than an access-design fault.
How do you make cryptic tag names usable at 2 a.m.?
Leave the historian tag as it is and add an operator-facing description and units wherever the client shows point identity: in prepared views, in the point-selection list and on the faceplate action that opens the trend. A new operator does not need to decode the tag naming convention if the description states what the point is.