Troubleshooting
Fix PI calculations on slash-containing DeltaV historian tags
Short answer
Quote the entire point name — 'FIC101/PID1/ PV.CV' — in single quotes and put the arithmetic outside it: 'FIC101/PID1/ PV.CV' / 2. The slashes inside the quotes belong to the stored name; the parser divides only at the slash after the closing quote. Renaming points is a later decision, not the fix.
On this page
A PI calculation that references a point named FIC101/PID1/ PV.CV fails because the unquoted slashes are read as division operators. The fix is to put the entire point name in single quotes and keep the arithmetic outside them: 'FIC101/PID1/ PV.CV' / 2. Inside the quotes, the slashes belong to the name; the slash after the closing quote is the division. Historization and transfer to the PI Server already work with these names, so nothing upstream needs to change, and renaming healthy points is a separate, higher-risk decision rather than the fix.
Why the slash breaks the expression #
The auto-generation in the Advanced Continuous Historian builds point names from the DeltaV parameter path, so the stored identifier carries the path's slashes and its embedded space: FIC101/PID1/ PV.CV. Both the Advanced Continuous Historian and the Enterprise Historian use AVEVA PI System technology, so on the PI Server these are ordinary point names, and collection, historization and transfer accept them. The failure appears only when the name is typed into a calculation without quotes.
In PI expression syntax, a point reference is delimited by single quotes, and every character between them — the slashes and the space included — is part of the identifier the lookup receives. Outside the quotes, the slash is the division operator. Written unquoted, FIC101/PID1/ PV.CV / 2 hands the parser a chain of operands and operators instead of a name, the stored point is never assembled, and the expression fails validation.
Confirm where the calculation actually runs #
Quoting is the documented syntax for calculations that execute on the PI Server, so test it in the engine that actually runs the expression. A calculation that displays PI data can run server-side, in a client application, or in a third-party system consuming the tag, and where the data lives does not decide the expression language. Carrying the same unquoted text into a different editor never fixes that mismatch.
| Symptom | What it points to | Action |
|---|---|---|
| Expression rejected at validation in a PI editor | Reference syntax | Quote the full point name |
| Quoted reference validates but the point is not found | Name spelling or server selection | Copy the exact name from point search on the calculation's server |
| Point found, but values are stale or bad | Collection or transfer path | Trace the value back to the historian and the controller |
| Reference resolves, but the result is wrong | Arithmetic, units or evaluation time | Compare against the source value at the same timestamp |
Fix the reference in five steps #
- Open point search on the same PI Server the calculation is configured against; a wildcard pattern such as
FIC101*lists every candidate spelling. - Copy the stored name exactly —
FIC101/PID1/ PV.CV, with the space between the second slash andPV.CV, and no leading or trailing whitespace. Retyping it from a screenshot is how mismatched names get created. - Enter the reference alone,
'FIC101/PID1/ PV.CV', and evaluate it with no arithmetic; it should return a value, timestamp and quality that match the point's own display. - Add the arithmetic outside the quotes:
'FIC101/PID1/ PV.CV' / 2. - Evaluate the complete expression and check it against half the source value at the same timestamp — a current-value display and a historical calculation can legitimately read different samples.
If step 3 succeeds, the binding is correct and any remaining fault is arithmetic or data quality. If the editor rejects the quoted form outright, the calculation is not running in a PI expression engine: find the product that owns it and use its documented reference syntax or its reference picker.
Match the stored name and the server #
Quotes solve the parsing problem only. They do not correct a misspelled name, choose the right server, or refresh a stale client binding. When the quoted reference still fails, check three things:
- The name, character for character: the embedded space is real, and deleting it while typing produces a different identifier that point search will not find.
- The server: the calculation's data source and the client where you verified the point must be the same PI Server, or the lookup runs against a system that never received the tag.
- Whitespace at the ends of a pasted name, which survives visually and breaks the match invisibly.
When the point itself shows no history or has stopped updating, the fault is upstream of the expression — trace the transfer path the way moving DeltaV Continuous Historian data into a PI system describes before touching calculation syntax.
Divide the rate, or totalize it #
Requests like "totalize half of this flow" hide two different calculations. Dividing a flow rate by 2 halves the instantaneous rate and leaves the units as a rate; it accumulates nothing. A total integrates rate over time and returns a volume.
Illustrative example, arithmetic shown: the point reads 73.6 m³/h and holds that value for one hour.
- Half the rate: 73.6 / 2 = 36.8, so the division returns 36.8 m³/h at any timestamp in that hour.
- Volume accumulated at half flow over the hour: 36.8 m³/h × 1 h = 36.8 m³.
- Over a 24-hour day at the same constant rate: 36.8 m³/h × 24 h = 883.2 m³.
The decision: if the consumer wants a number in cubic meters at the end of a shift, the division alone is the wrong calculation — use the engine's totalization over the time range and halve the result, or halve the rate feeding the totalizer. If the point carries a raw integer rather than engineering units, scale it first, because dividing raw counts halves counts, not flow; the steps in historian tag scaling for an unscaled PLC integer value apply.
Keep the point names, fix the expressions #
For a working installation, leave the auto-generated names alone and quote them wherever a calculation references them. Renaming points to remove the slashes touches every trend, display, interface, report and downstream calculation on both the DeltaV and PI sides, while existing archives keep their original identity — a rename is a naming-standards project with a full reference inventory and a test window, not a workaround for expression syntax. Quote the reference today, and rename only if the site standard demands it and the change is planned as one.
Questions asked
Will double quotes work instead of single quotes in the calculation?
No. In PI expression syntax the single quotes mark a point reference; double quotes delimit a text string, so a double-quoted name is handled as text rather than looked up as a tag. Keep single quotes around the complete name, slashes and embedded space included, and put any arithmetic outside them.
Is it safe to rename the auto-generated tags to remove the slashes?
Not as a quick fix. Every trend, display, interface, report and downstream calculation that references the point must be updated on both the DeltaV and PI sides, and existing archives keep their original identity. Quote the point name in the calculation instead, and treat any rename as a planned naming-standards change with a full reference inventory and a test window.
Where does the space in FIC101/PID1/ PV.CV come from?
The auto-generation built the point name from the DeltaV parameter path, and the space between the second slash and PV.CV is part of the stored identifier. Do not retype the name from memory or a screenshot: copy it from point search on the server the calculation uses, preserving punctuation, spaces and case exactly.
The calculation still fails with the quotes in place. What next?
Verify three things in order: that the calculation really executes on the PI Server and not in a client or third-party tool with different syntax; that the quoted text matches the stored name character for character; and that the calculation's data source is the same server where the point was found. A quoted reference that validates but cannot find the point is a name or server fault, not a syntax fault.